freebsd-nq/sys/dev/aic7xxx/aic79xx.seq

2291 lines
71 KiB
Plaintext
Raw Normal View History

/*-
* Adaptec U320 device driver firmware for Linux and FreeBSD.
*
Correct a very rare case where command ordering could be compromised by a transaction performing a driver handled message sequence (an scb with the MK_MESSAGE flag set). SCBs that perform host managed messaging must always be at the head of their per-target selection queue so that the firmware knows to manually assert ATN if the current negotiation agreement is packetized. In the past we guaranteed this by queuing these SCBs separarately in the execution queue. This exposes the system to potential command reordering in two cases: 1) Another SCB for the same ITL nexus is queued that does not have the MK_MESSAGE flag set. This SCB will be queued to the per-target list which can be serviced before the MK_MESSAGE scb that preceeded it. 2) If the target cannot accept all of the commands in the per-target selection queue in one selection, the remainder is queued to the tail of the selection queues so as to effect round-robin scheduling. This could allow the MK_MESSAGE scb to be sent to the target before the requeued commands. This commit changes the firmware policy to defer queuing MK_MESSAGE SCBs into the selection queues until this can be done without affecting order. This means that the target's selection queue is either empty, or the last SCB on the execution queue is also a MK_MESSAGE SCB. During any wait, the firmware halts the download of new SCBs so only a single "holding location" is required. Luckily, MK_MESSAGE SCBs are rare and typically occur only during CAM's bus probe where only one command is outstanding at a time. However, during some recovery scenarios, the reordering *could* occur. aic79xx.c: Update ahd_search_qinfifo() and helper routines to search for pending MK_MESSAGE scbs and properly restitch the execution queue if either the MK_MESSAGE SCB is being aborted, or the MK_MESSAGE SCB can be queued due to the execution queue draining due to aborts. Enable LQOBUSFREE status to assert an interrupt. This should be redundant since a BUSFREE interrupt should always occur along with an LQOBUSFREE event, but on the Rev A, this doesn't seem to be guaranteed. When a PPR request is rejected when a previously existing packetized agreement is in place, assume that the target has been reset without our knowledge and revert to async/narrow transfers. This corrects two issues: the stale ENATNO setting that was used to send the PPR is cleared so the firmware is not confused by a future packetized selection with ATN asserted but no MK_MESSAGE flag in the SCB and it speeds up recovery by aborting any pending packetized transactions that by definition are now dead. When re-queueing SCBs after a failed negotiation attempt, ensure command ordering by freezing the device queue first. Traverse the list of pending SCBs rather than the whole SCB array on the controller when pushing MK_MESSAGE flag changes out to the controller. The original code was optimized for the aic7xxx controllers where there are fewer controller slots then pending SCBs and the firmware picks SCB slots. For the U320 controller, the hope is that we have fewer pending SCBs then the 512 slots on the controller. Enhance some diagnostics. Factor out some common code. aic79xx.h: Add prototype for new ahd_done_with_status() that is used to factor out some commone code. aic79xx.reg: Add definisions for the pending MK_MESSAGE SCB. aic79xx.seq: Defer MK_MESSAGE SCB queing to the execution queue so as to preserve command ordering. Re-arrange some of the selection processing code so the above change had no performance impact on the common code path. Close a few critical section holes. When entering a non-packetized phase, manually enable busfree interrupts, since the controller hardware does not do this automatically. aic79xx_inline.h: Enhance logging for queued SCBs. aic79xx_osm.c: Add new a new DDB ahd command, ahd_dump, which invokes the ahd_dump_card_state() routine on the unit specified with the ahd_sunit DDB command. aic79xx_pci.c: Turn on the BUSFREEREV bug for the Rev B. controller. This is required to close the busfree during non-packetized phase hole.
2004-08-04 17:55:34 +00:00
* Copyright (c) 1994-2001, 2004 Justin T. Gibbs.
* Copyright (c) 2000-2002 Adaptec Inc.
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions, and the following disclaimer,
* without modification.
* 2. Redistributions in binary form must reproduce at minimum a disclaimer
* substantially similar to the "NO WARRANTY" disclaimer below
* ("Disclaimer") and any redistribution must be conditioned upon
* including a substantially similar Disclaimer requirement for further
* binary redistribution.
* 3. Neither the names of the above-listed copyright holders nor the names
* of any contributors may be used to endorse or promote products derived
* from this software without specific prior written permission.
*
* Alternatively, this software may be distributed under the terms of the
* GNU General Public License ("GPL") version 2 as published by the Free
* Software Foundation.
*
* NO WARRANTY
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
* "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
* LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTIBILITY AND FITNESS FOR
* A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
* HOLDERS OR CONTRIBUTORS BE LIABLE FOR SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
* STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING
* IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
* POSSIBILITY OF SUCH DAMAGES.
*
* $FreeBSD$
*/
aic79xx.c: Allow 500us between pauses in ahd_pause_and_flushwork(). The maximum we will wait is now 500ms. In the same routine, remove any attempt to clear ENSELO. Let the firmware do it once the current selection has completed. This avoids some race conditions having to do with non-packetized completions and the auto-clearing of ENSELO on packetized completions. Also avoid attempts to clear critical sections when interrups are pending. We are going to loop again anyway, so clearing critical sections is a waste of time. It also may not be possible to clear a critical section if the source of the interrupt was a SEQINT. aic79xx_pci.c: Use the Generic 9005 mask when looking for generic 7901B parts. This allows the driver to attach to 7901B parts on motherboards using a non-Adaptec subvendor ID. aic79xx_inline.h: Test for the SCBRAM_RD_BUG against the bugs field, not the flags field in the softc. aic79xx.c: Cancel pending transactions on devices that respond with a selection timeout. This decreases the duration of timeout recovery when a device disappears. aic79xx.c: Don't bother forcing renegotiation on a selection timeout now that we use the device reset handler to abort any pending commands on the target. The device reset handler already takes us down to async narrow and forces a renegotiation. In the device reset handlers, only send a BDR sent async event if the status is not CAM_SEL_TIMEOUT. This avoids sending this event in the selection timeout case aic79xx.c: Modify the Core timeout handler to verify that another command has the potential to timeout before passing off a command timeout as due to some other command. This safety measure is added in response to a timeout recovery failure on H2B where it appears that incoming reselection status was lost during a drive pull test. In that case, the recovery handler continued to wait for the command that was active on the bus indefinetly. While the root cause of the above issue is still being determined seems a prudent safeguard. aic79xx_pci.c: Add a specific probe entry for the Dell OEM 39320(B). aic79xx.c: aic79xx.h: aic79xx.reg: aic79xx.seq: Modify the aic79xx firmware to never cross a cacheline or ADB boundary when DMA'ing completion entries to the host. In PCI mode, at least in 32/33 configurations, the SCB DMA engine may lose its place in the data-stream should the target force a retry on something other than an 8byte aligned boundary. In PCI-X mode, we do this to avoid split transactions since many chipsets seem to be unable to format proper split completions to continue the data transfer. The above change allows us to drop our completion entries from 8 bytes to 4. We were using 8 byte entries to ensure that PCI retries could only occur on an 8byte aligned boundary. Now that the sequencer guarantees this by splitting up completions, we can safely drop the size to 4 bytes (2 byte tag, one byte SG_RESID, one byte pad). Both the split-completion and PCI retry problems only show up under high tag load when interrupt coalescing is being especially effective. The switch from a 2byte completion entry to an 8 byte entry to solve the PCI problem increased the chance of incurring a split in PCI-X mode when multiple transactions were completed at once. Dropping the completion size to 4 bytes also means that we can complete more commands in a single DMA (128byte FIFO -> 32 commands instead of 16). aic79xx.c: Modify the SCSIINT handler to defer clearing sequencer critical sections to the individual interrupt handlers. This allows us to immediately disable any outgoing selections in the case of an unexpected busfree so we don't inadvertantly clear ENSELO *after* a new selection has started. Doing so may cause the sequencer to miss a successful selection. In ahd_update_pending_scbs(), only clear ENSELO if the bus is currently busy and a selection is not already in progress or the sequencer has yet to handle a pending selection. While we want to ensure that the selection for the SCB at the head of the selection queue is restarted so that any change in negotiation request can take effect, we can't clobber pending selection state without confusing the sequencer into missing a selection.
2004-05-11 20:46:05 +00:00
VERSION = "$Id: //depot/aic7xxx/aic7xxx/aic79xx.seq#119 $"
PATCH_ARG_LIST = "struct ahd_softc *ahd"
PREFIX = "ahd_"
#include "aic79xx.reg"
#include "scsi_message.h"
restart:
if ((ahd->bugs & AHD_INTCOLLISION_BUG) != 0) {
test SEQINTCODE, 0xFF jz idle_loop;
SET_SEQINTCODE(NO_SEQINT)
}
idle_loop:
if ((ahd->bugs & AHD_INTCOLLISION_BUG) != 0) {
/*
* Convert ERROR status into a sequencer
* interrupt to handle the case of an
* interrupt collision on the hardware
* setting of HWERR.
*/
test ERROR, 0xFF jz no_error_set;
SET_SEQINTCODE(SAW_HWERR)
no_error_set:
}
Upgrade to version 1.1 of the aic79xx U320 driver. aic79xx.c: o Remove redundant ahd_update_modes() call. o Correct panic in diagnostic should state corruption cause the SCB Id to be invalid during a selection timeout. o Add workaround for missing BUSFREEREV feature in Rev A silicon. o Corect formatting nits. o Use register pretty printing in more places. o Save and restore our SCB pointer when updating the waiting queue list for an "expected" LQ-out busfree. o In ahd_clear_intstat, deal with the missing autoclear in the CLRLQO* registers. o BE fixup in a diagnostic printf. o Make sure that we are in the proper mode before disabling selections in ahd_update_pending_scbs. o Add more diagnostics. o task_attribute_nonpkt_tag -> task_attribute: we don't need a nonpkt_tag field anymore for allowing all 512 SCBs to be used in non-packetized connections. o Negotiate HOLD_MCS to U320 devices. o Add a few additional mode assertions. o Restore the chip mode after clearing out the qinfifo so that code using ahd_abort_scbs sees a consistent mode. o Simplify the DMA engine shutdown routine prior to performing a bus reset. o Perform the sequencer restart after a chip reset prior to setting up our timer to poll for the reset to be complete. On some OSes, the timer could actually pre-empt us and order is important here. o Have our "reset poller" set the expected mode since there is no guarantee of what mode will be in force when we are called from the OS timer. o Save and restore the SCB pointer in ahd_dump_card_state(). This routine must not modify card state. o Ditto for ahd_dump_scbs(). aic79xx.h: o Add a few more chip bug definitions. o Align our tag on a 32bit boundary. aic79xx.reg: aic79xx.seq: o Start work on removing workarounds for Rev B. o Use a special location in scratch from for stroring our SCBPTR during legacy FIFO allocations. This corrects problems in mixed packetized/non-packetized configurations where calling into a FIFO task corrupted our SCBPTR. o Don't rely on DMA priority to guarantee that all data in our FIFOs will flush prior to a command completion notification going out of the command channel. We've never seen this assumption fail, but better safe than sorry. o Deal with missing BUSFREEREV feature in H2A. o Simplify disconnect list code now that the list will always have only a single entry. o Implement the AHD_REG_SLOW_SETTLE_BUG workaround. o Swith to using "REG_ISR" for local mode scratch during our ISR. o Add a missing jmp to the data_group_dma_loop after our data pointers have been re-initialized by the kernel. o Correct test in the bitbucket code so that we actually wait for the bitbucket to complete before signaling the kernel of the overrun condition. o Reposition pkt_saveptrs to avoid a jmp instruction. o Update a comment to reflect that the code now waits for a FIFO to drain prior to issuing a CLRCHN. aic79xx_inline.h: o Remove unused untagged queue handling code. o Don't attempt to htole64 what could be a 32bit value. aic79xx_pci.c: o Set additional bug flags for rev A chips.
2002-09-26 22:54:00 +00:00
SET_MODE(M_SCSI, M_SCSI)
test SCSISEQ0, ENSELO|ENARBO jnz idle_loop_checkbus;
aic79xx.c: aic79xx.seq: Convert the COMPLETE_DMA_SCB list to an "stailq". This allows us to safely keep the SCB that is currently being DMA'ed back the host on the head of the list while processing completions off of the bus. The newly completed SCBs are appended to the tail of the queue. In the past, we just dequeued the SCB that was in flight from the list, but this could result in a lost completion should the host perform certain types of error recovery that must cancel all in-flight SCB DMA operations. Switch from using a 16bit completion entry, holding just the tag and the completion valid bit, to a 64bit completion entry that also contains a "status packet valid" indicator. This solves two problems: o The SCB DMA engine on at least Rev B. silicon does not properly deal with a PCI disconnect that occurs at a non-64bit aligned offset in the chips "source buffer". When the transfer is resumed, the DMA engine continues at the correct offset, but may wrap to the head of the buffer causing duplicate completions to be reported to the host. By using a completion buffer in host memory that is 64bit aligned and using 64bit completion entries, such disconnects should only occur at aligned addresses. This assumes that the host bridge will only disconnect on cache-line boundaries and that cache-lines are multpiles of 64bits. o By embedding the status information in the completion entry we can avoid an extra memory reference to the HSCB for commands that complete without error. Use the comparison of a "host freeze count" and a "sequencer freeze count" to allow the host to process most SCBs that complete with non-zero status without having to clear critical sections. Instead the host can just pause the sequencer, performs any necessary cleanup in the waiting for selection list, increments its freeze count on the controller, and unpauses. This is only possible because the sequencer defers completions of SCBs with bad status until after all pending selections have completed. The sequencer then avoids referencing any data structures the host may touch during completion of the SCB until the freeze counts match. aic79xx.c: Change the strategy for allocating our sentinal HSCB for the QINFIFO. In the past, this allocation was tacked onto the QOUTFIFO allocation. Now that the qoutfifo has grown to accomodate larger completion entries, the old approach will result in a 64byte allocation that costs an extra page of coherent memory. We now do this extra allocation via ahd_alloc_scbs() where the "unused space" can be used to allocate "normal" HSCBs. In our packetized busfree handler, use the ENSELO bit to differentiate between packetized and non-packetized unexpected busfree events that occur just after selection, but before the sequencer has had the oportunity to service the selection. When cleaning out the waiting for selection list, use the SCSI mode instead of the command channel mode. The SCB pointer in the command channel mode may be referenced by the SCB dma engine even while the sequencer is paused, whereas the SCSI mode SCB pointer is only accessed by the sequencer. Print the "complete on qfreeze" sequencer SCB completion list in ahd_dump_card_state(). This list holds all SCB completions that are deferred until a pending select-out qfreeze event has taken effect. aic79xx.h: Add definitions and structures to handle the new SCB completion scheme. Add a controller flag that indicates if the controller is in HostRAID mode. aic79xx.reg: Remove macros used for toggling from one data fifo mode to the other. They have not been in use for some time. Add scratch ram fields for our new qfreeze count scheme, converting the complete dma list into an "stailq", and providing for the "complete on qfreeze" SCB completion list. Some other fields were moved to retain proper field alignment (alignment >= field size in bytes). aic79xx.seq: Add code to our idle loop to: o Process deferred completions once a qfreeze event has taken full effect. o Thaw the queue once the sequencer and host qfreeze counts match. Generate 64bit completion entries passing the SCB_SGPTR field as the "good status" indicator. The first bit in this field is only set if we have a valid status packet to send to the host. Convert the COMPLETE_DMA_SCB list to an "stailq". When using "setjmp" to register an idle loop handler, do not combine the "ret" with the block move to pop the stack address in the same instruction. At least on the A, this results in a return to the setjmp caller, not to the new address at the top of the stack. Since we want the latter (we want the newly registered handler to only be invoked from the idle loop), we must use a separate ret instruction. Add a few missing critical sections. Close a race condition that can occur on Rev A. silicon. If both FIFOs happen to be allocated before the sequencer has a chance to service the FIFO that was allocated first, we must take special care to service the FIFO that is not active on the SCSI bus first. This guarantees that a FIFO will be freed to handle any snapshot requests for the FIFO that is still on the bus. Chosing the incorrect FIFO will result in deadlock. Update comments. aic79xx_inline.h Correct the offset calculation for the syncing of our qoutfifo. Update ahd_check_cmdcmpltqueues() for the larger completion entries. aic79xx_pci.c: Attach to HostRAID controllers by default. In the future I may add a sysctl to modify the behavior, but since FreeBSD does not have any HostRAID drivers, failing to attach just results in more email and bug reports for the author. MFC After: 1week
2004-02-04 16:38:38 +00:00
test SEQ_FLAGS2, SELECTOUT_QFROZEN jz check_waiting_list;
/*
* If the kernel has caught up with us, thaw the queue.
*/
mov A, KERNEL_QFREEZE_COUNT;
cmp QFREEZE_COUNT, A jne check_frozen_completions;
mov A, KERNEL_QFREEZE_COUNT[1];
cmp QFREEZE_COUNT[1], A jne check_frozen_completions;
and SEQ_FLAGS2, ~SELECTOUT_QFROZEN;
jmp check_waiting_list;
check_frozen_completions:
test SSTAT0, SELDO|SELINGO jnz idle_loop_checkbus;
BEGIN_CRITICAL;
/*
* If we have completions stalled waiting for the qfreeze
* to take effect, move them over to the complete_scb list
* now that no selections are pending.
*/
cmp COMPLETE_ON_QFREEZE_HEAD[1],SCB_LIST_NULL je idle_loop_checkbus;
/*
* Find the end of the qfreeze list. The first element has
* to be treated specially.
*/
bmov SCBPTR, COMPLETE_ON_QFREEZE_HEAD, 2;
cmp SCB_NEXT_COMPLETE[1], SCB_LIST_NULL je join_lists;
/*
* Now the normal loop.
*/
bmov SCBPTR, SCB_NEXT_COMPLETE, 2;
cmp SCB_NEXT_COMPLETE[1], SCB_LIST_NULL jne . - 1;
join_lists:
bmov SCB_NEXT_COMPLETE, COMPLETE_SCB_HEAD, 2;
bmov COMPLETE_SCB_HEAD, COMPLETE_ON_QFREEZE_HEAD, 2;
mvi COMPLETE_ON_QFREEZE_HEAD[1], SCB_LIST_NULL;
jmp idle_loop_checkbus;
check_waiting_list:
cmp WAITING_TID_HEAD[1], SCB_LIST_NULL je idle_loop_checkbus;
/*
* ENSELO is cleared by a SELDO, so we must test for SELDO
* one last time.
*/
test SSTAT0, SELDO jnz select_out;
call start_selection;
idle_loop_checkbus:
test SSTAT0, SELDO jnz select_out;
END_CRITICAL;
test SSTAT0, SELDI jnz select_in;
test SCSIPHASE, ~DATA_PHASE_MASK jz idle_loop_check_nonpackreq;
test SCSISIGO, ATNO jz idle_loop_check_nonpackreq;
call unexpected_nonpkt_phase_find_ctxt;
idle_loop_check_nonpackreq:
test SSTAT2, NONPACKREQ jz . + 2;
call unexpected_nonpkt_phase_find_ctxt;
Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-05-04 00:20:07 +00:00
if ((ahd->bugs & AHD_FAINT_LED_BUG) != 0) {
ahc_eisa.c: ahc_pci.c: ahd_pci.c: aic7xxx.c: aic79xx.c: aic_osm_lib.c: aic_osm_lib.h: Use common OSM routines from aic_osm_lib for bus dma operations, delay routines, accessing CCBs, byte swapping, etc. aic7xxx_pci.c: Provide a better description for the 2915/30LP on attach. aic7xxx.c: aic79xx.c: aic7770.c: aic79xx_pci.c: aic7xxx_pci.c: aic7xxx_93cx6.c: Move FBSDID behind an ifdef so that these core files will still compile under other OSes. aic79xx.h: aic79xx_pci.c: aic79xx.seq: To speed up non-packetized CDB delivery in Rev B, all CDB acks are "released" to the output sync as soon as the command phase starts. There is only one problem with this approach. If the target changes phase before all data are sent, we have left over acks that can go out on the bus in a data phase. Due to other chip contraints, this only happens if the target goes to data-in, but if the acks go out before we can test SDONE, we'll think that the transfer has completed successfully. Work around this by taking advantage of the 400ns or 800ns dead time between command phase and the REQ of the new phase. If the transfer has completed successfully, SCSIEN should fall *long* before we see a phase change. We thus treat any phasemiss that occurs before SCSIEN falls as an incomplete transfer. aic79xx.h: Add the AHD_FAST_CDB_DELIVERY feature. aic79xx_pci.c: Set AHD_FAST_CDB_DELIVERY for all Rev. B parts. aic79xx.seq: Test for PHASEMIS in the command phase for all AHD_FAST_CDB_DELIVERY controlelrs. ahd_pci.c: ahc_pci.c: aic7xxx.h: aic79xx.h: Move definition of controller BAR offsets to core header files. aic7xxx.c: aic79xx.c: In the softc free routine, leave removal of a softc from the global list of softcs to the OSM (the caller of this routine). This allows us to avoid holding the softc list_lock during device destruction where we may have to sleep waiting for our recovery thread to halt. ahc_pci.c: Use ahc_pci_test_register access to validate I/O mapped in addition to the tests already performed for memory mapped access. Remove unused ahc_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. ahd_pci.c: Remove reduntant definition of controller BAR offsets. These are also defined in aic79xx.h. Remove unused ahd_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. aic7xxx.c: aic79xx.c: aic79xx.h: aic7xxx.h: aic7xxx_osm.c: aic79xx_osm.c: Move timeout handling to the driver cores. In the case of the aic79xx driver, the algorithm has been enhanced to try target resets before performing a bus reset. For the aic7xxx driver, the algorithm is unchanged. Although the drivers do not currently sleep during recovery (recovery is timeout driven), the cores do expect all processing to be performed via a recovery thread. Our timeout handlers are now little stubs that wakeup the recovery thread. aic79xx.c: aic79xx.h: aic79xx_inline.h: Change shared_data allocation to use a map_node so that the sentinel hscb can use this map node in ahd_swap_with_next_hscb. This routine now swaps the hscb_map pointer in additon to the hscb contents so that any sync operations occur on the correct map. physaddr -> busaddr Pointed out by: Jason Thorpe <thorpej@wasabisystems.com> aic79xx.c: Make more use of the in/out/w/l/q macros for accessing byte registers in the chip. Correct some issues in the ahd_flush_qoutfifo() routine. o Run the qoutfifo only once the command channel DMA engine has been halted. This closes a window where we might have missed some entries. o Change ahd_run_data_fifo() to not loop to completion. If we happen to start on the wrong FIFO and the other FIFO has a snapshot savepointers, we might deadlock. This required our delay between FIFO tests to be moved to the ahd_flush_qoutfifo() routine. o Update/add comments. o Remove spurious test for COMPLETE_DMA list being empty when completing transactions from the GSFIFO with residuals. The SCB must be put on the COMPLETE_DMA scb list unconditionally. o When halting command channel DMA activity, we must disable the DMA channel in all cases but an update of the QOUTFIFO. The latter case is required so that the sequencer will update its position in the QOUTFIFO. Previously, we left the channel enabled for all "push" DMAs. This left us vulnerable to the sequencer handling an SCB push long after that SCB was already processed manually by this routine. o Correct the polarity of tests involving ahd_scb_active_in_fifo(). This routine returns non-zero for true. Return to processing bad status completions through the qoutfifo. This reduces the time that the sequencer is kept paused when handling transactions with bad status or underruns. When waiting for the controller to quiece selections, add a delay to our loop. Otherwise we may fail to wait long enough for the sequencer to comply. On H2A4 hardware, use the slow slewrate for non-paced transfers. This mirrors what the Adaptec Windows drivers do. On the Rev B. only slow down the CRC timing for older U160 devices that might need the slower timing. We define "older" as devices that do not support packetized protocol. Wait up to 5000 * 5us for the SEEPROM to become unbusy. Write ops seem to take much longer than read ops. aic79xx.seq: For controllers with the FAINT_LED bug, turn the diagnostic led feature on during selection and reselection. This covers the non-packetized case. The LED will be disabled for non-packetized transfers once we return to the top level idle loop. Add more comments about the busy LED workaround. Extend a critical section around the entire command channel idle loop process. Previously the portion of this handler that directly manipulated the linked list of completed SCBs was not protected. This is the likely cause of the recent reports of commands being completed twice by the driver. Extend critical sections across the test for, and the longjump to, longjump routines. This prevents the firmware from trying to jump to a longjmp handler that was just cleared by the host. Improve the locations of several critical section begin and end points. Typically these changes remove instructions that did not need to be inside a critical section. Close the "busfree after selection, but before busfree interrupts can be enabled" race to just a single sequencer instruction. We now test the BSY line explicitly before clearing the busfree status and enabling the busfree interrupt. Close a race condition in the processing of HS_MAILBOX updates. We now clear the "updated" status before the copy. This ensures that we don't accidentally clear the status incorrectly when the host sneaks in an update just after our last copy, but before we clear the status. This race has never been observed. Don't re-enable SCSIEN if we lose the race to disable SCSIEN in our interrupt handler's workaround for the RevA data-valid too early issue. aic79xx_inline.h: Add comments indicating that the order in which bytes are read or written in ahd_inw and ahd_outw is important. This allows us to use these inlines when accessing registers with side-effects. aic79xx_pci.c: The 29320 and the 29320B are 7902 not 7901 based products. Correct the driver banner. aic7xxx.h: Enable the use of the auto-access pause feature on the aic7870 and aic7880. It was disabled due to an oversight. aic7xxx.reg: Move TARG_IMMEDIATE_SCB to alias LAST_MSG to avoid leaving garbage in MWI_RESIDUAL. This prevents spurious overflows whn operating target mode on controllers that require the MWI_RESIDUAL work-around. aic7xxx.seq: AHC_TMODE_WIDEODD_BUG is a bug, not a softc flag. Reference the correct softc field when testing for its presence. Set the NOT_IDENTIFIED and NO_CDB_SENT bits in SEQ_FLAGS to indicate that the nexus is invalid in await busfree. aic7xxx_93cx6.c: Add support for the C56/C66 versions of the EWEN and EWDS commands. aic7xxx.c: aic7xxx_pci.c: Move test for the validity of left over BIOS data to ahc_test_register_access(). This guarantees that any left over CHIPRST value is not clobbered by our register access test and lost to the test that was in ahc_reset.
2003-12-17 00:02:10 +00:00
/*
* On Rev A. hardware, the busy LED is only
* turned on automaically during selections
* and re-selections. Make the LED status
* more useful by forcing it to be on so
* long as one of our data FIFOs is active.
*/
Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-05-04 00:20:07 +00:00
and A, FIFO0FREE|FIFO1FREE, DFFSTAT;
cmp A, FIFO0FREE|FIFO1FREE jne . + 3;
and SBLKCTL, ~DIAGLEDEN|DIAGLEDON;
jmp . + 2;
or SBLKCTL, DIAGLEDEN|DIAGLEDON;
}
call idle_loop_gsfifo_in_scsi_mode;
call idle_loop_service_fifos;
call idle_loop_cchan;
jmp idle_loop;
idle_loop_gsfifo:
SET_MODE(M_SCSI, M_SCSI)
ahc_eisa.c: ahc_pci.c: ahd_pci.c: aic7xxx.c: aic79xx.c: aic_osm_lib.c: aic_osm_lib.h: Use common OSM routines from aic_osm_lib for bus dma operations, delay routines, accessing CCBs, byte swapping, etc. aic7xxx_pci.c: Provide a better description for the 2915/30LP on attach. aic7xxx.c: aic79xx.c: aic7770.c: aic79xx_pci.c: aic7xxx_pci.c: aic7xxx_93cx6.c: Move FBSDID behind an ifdef so that these core files will still compile under other OSes. aic79xx.h: aic79xx_pci.c: aic79xx.seq: To speed up non-packetized CDB delivery in Rev B, all CDB acks are "released" to the output sync as soon as the command phase starts. There is only one problem with this approach. If the target changes phase before all data are sent, we have left over acks that can go out on the bus in a data phase. Due to other chip contraints, this only happens if the target goes to data-in, but if the acks go out before we can test SDONE, we'll think that the transfer has completed successfully. Work around this by taking advantage of the 400ns or 800ns dead time between command phase and the REQ of the new phase. If the transfer has completed successfully, SCSIEN should fall *long* before we see a phase change. We thus treat any phasemiss that occurs before SCSIEN falls as an incomplete transfer. aic79xx.h: Add the AHD_FAST_CDB_DELIVERY feature. aic79xx_pci.c: Set AHD_FAST_CDB_DELIVERY for all Rev. B parts. aic79xx.seq: Test for PHASEMIS in the command phase for all AHD_FAST_CDB_DELIVERY controlelrs. ahd_pci.c: ahc_pci.c: aic7xxx.h: aic79xx.h: Move definition of controller BAR offsets to core header files. aic7xxx.c: aic79xx.c: In the softc free routine, leave removal of a softc from the global list of softcs to the OSM (the caller of this routine). This allows us to avoid holding the softc list_lock during device destruction where we may have to sleep waiting for our recovery thread to halt. ahc_pci.c: Use ahc_pci_test_register access to validate I/O mapped in addition to the tests already performed for memory mapped access. Remove unused ahc_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. ahd_pci.c: Remove reduntant definition of controller BAR offsets. These are also defined in aic79xx.h. Remove unused ahd_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. aic7xxx.c: aic79xx.c: aic79xx.h: aic7xxx.h: aic7xxx_osm.c: aic79xx_osm.c: Move timeout handling to the driver cores. In the case of the aic79xx driver, the algorithm has been enhanced to try target resets before performing a bus reset. For the aic7xxx driver, the algorithm is unchanged. Although the drivers do not currently sleep during recovery (recovery is timeout driven), the cores do expect all processing to be performed via a recovery thread. Our timeout handlers are now little stubs that wakeup the recovery thread. aic79xx.c: aic79xx.h: aic79xx_inline.h: Change shared_data allocation to use a map_node so that the sentinel hscb can use this map node in ahd_swap_with_next_hscb. This routine now swaps the hscb_map pointer in additon to the hscb contents so that any sync operations occur on the correct map. physaddr -> busaddr Pointed out by: Jason Thorpe <thorpej@wasabisystems.com> aic79xx.c: Make more use of the in/out/w/l/q macros for accessing byte registers in the chip. Correct some issues in the ahd_flush_qoutfifo() routine. o Run the qoutfifo only once the command channel DMA engine has been halted. This closes a window where we might have missed some entries. o Change ahd_run_data_fifo() to not loop to completion. If we happen to start on the wrong FIFO and the other FIFO has a snapshot savepointers, we might deadlock. This required our delay between FIFO tests to be moved to the ahd_flush_qoutfifo() routine. o Update/add comments. o Remove spurious test for COMPLETE_DMA list being empty when completing transactions from the GSFIFO with residuals. The SCB must be put on the COMPLETE_DMA scb list unconditionally. o When halting command channel DMA activity, we must disable the DMA channel in all cases but an update of the QOUTFIFO. The latter case is required so that the sequencer will update its position in the QOUTFIFO. Previously, we left the channel enabled for all "push" DMAs. This left us vulnerable to the sequencer handling an SCB push long after that SCB was already processed manually by this routine. o Correct the polarity of tests involving ahd_scb_active_in_fifo(). This routine returns non-zero for true. Return to processing bad status completions through the qoutfifo. This reduces the time that the sequencer is kept paused when handling transactions with bad status or underruns. When waiting for the controller to quiece selections, add a delay to our loop. Otherwise we may fail to wait long enough for the sequencer to comply. On H2A4 hardware, use the slow slewrate for non-paced transfers. This mirrors what the Adaptec Windows drivers do. On the Rev B. only slow down the CRC timing for older U160 devices that might need the slower timing. We define "older" as devices that do not support packetized protocol. Wait up to 5000 * 5us for the SEEPROM to become unbusy. Write ops seem to take much longer than read ops. aic79xx.seq: For controllers with the FAINT_LED bug, turn the diagnostic led feature on during selection and reselection. This covers the non-packetized case. The LED will be disabled for non-packetized transfers once we return to the top level idle loop. Add more comments about the busy LED workaround. Extend a critical section around the entire command channel idle loop process. Previously the portion of this handler that directly manipulated the linked list of completed SCBs was not protected. This is the likely cause of the recent reports of commands being completed twice by the driver. Extend critical sections across the test for, and the longjump to, longjump routines. This prevents the firmware from trying to jump to a longjmp handler that was just cleared by the host. Improve the locations of several critical section begin and end points. Typically these changes remove instructions that did not need to be inside a critical section. Close the "busfree after selection, but before busfree interrupts can be enabled" race to just a single sequencer instruction. We now test the BSY line explicitly before clearing the busfree status and enabling the busfree interrupt. Close a race condition in the processing of HS_MAILBOX updates. We now clear the "updated" status before the copy. This ensures that we don't accidentally clear the status incorrectly when the host sneaks in an update just after our last copy, but before we clear the status. This race has never been observed. Don't re-enable SCSIEN if we lose the race to disable SCSIEN in our interrupt handler's workaround for the RevA data-valid too early issue. aic79xx_inline.h: Add comments indicating that the order in which bytes are read or written in ahd_inw and ahd_outw is important. This allows us to use these inlines when accessing registers with side-effects. aic79xx_pci.c: The 29320 and the 29320B are 7902 not 7901 based products. Correct the driver banner. aic7xxx.h: Enable the use of the auto-access pause feature on the aic7870 and aic7880. It was disabled due to an oversight. aic7xxx.reg: Move TARG_IMMEDIATE_SCB to alias LAST_MSG to avoid leaving garbage in MWI_RESIDUAL. This prevents spurious overflows whn operating target mode on controllers that require the MWI_RESIDUAL work-around. aic7xxx.seq: AHC_TMODE_WIDEODD_BUG is a bug, not a softc flag. Reference the correct softc field when testing for its presence. Set the NOT_IDENTIFIED and NO_CDB_SENT bits in SEQ_FLAGS to indicate that the nexus is invalid in await busfree. aic7xxx_93cx6.c: Add support for the C56/C66 versions of the EWEN and EWDS commands. aic7xxx.c: aic7xxx_pci.c: Move test for the validity of left over BIOS data to ahc_test_register_access(). This guarantees that any left over CHIPRST value is not clobbered by our register access test and lost to the test that was in ahc_reset.
2003-12-17 00:02:10 +00:00
BEGIN_CRITICAL;
idle_loop_gsfifo_in_scsi_mode:
test LQISTAT2, LQIGSAVAIL jz return;
/*
* We have received good status for this transaction. There may
Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-05-04 00:20:07 +00:00
* still be data in our FIFOs draining to the host. Complete
* the SCB only if all data has transferred to the host.
*/
good_status_IU_done:
bmov SCBPTR, GSFIFO, 2;
clr SCB_SCSI_STATUS;
aic79xx.c: Implement the SCB_SILENT flag. This is useful for hushing up the driver during DV or other operations that we expect to cause transmission errors. The messages will still print if the SHOW_MASKED_ERRORS debug option is enabled. Save and restore the NEGOADDR address when setting new transfer settings. The sequencer performs lookups in the negotiation table too and it expects NEGOADDR to remain consistent across pause/unpause sessions. Consistently use "offset" instead of "period" to determine if we are running sync or not. Add a SHOW_MESSAGES diagnostic for when we assert ATN during message processing. Print out IU, QAS, and RTI features when showing transfer options. Limit the syncrate after all option conformance changes have taken place in ahd_devlimited_syncrate. Changes in options may change the final syncrate we accept. Keep a copy of the hs_mailbox in our softc so that we can perform read/modify/write operations on the hs_mailbox without having to pause the sequencer to read the last written value. Use the ENINT_COALESS flag in the hs_mailbox to toggle interrupt coalessing. Add entrypoints for enabling interrupt coalessing and setting both a timeout (how long to wait for commands to be coalessed) and a maximum commands to coaless value. Add a statistics timer that decides when to enable or disable interrupt coalessing based on load. Add a routine, ahd_reset_cmds_pending() which is used to update the CMDS_PENDING sequencer variable whenever error recovery compeltes SCBs without notifying the sequencer. Since ahd_reset_cmds_pending is called during ahd_unpause() only if we've aborted SCBs, its call to ahd_flush_qoutfifo should not cause recursion through ahd_run_qoutfifo(). A panic has been added to ensure that this recursion does not occur. In ahd_search_qinfifo, update the CMDS_PENDING sequencer variable directly. ahd_search_qinififo can be called in situations where using ahd_reset_cmds_pending() might cause recursion. Since we can safely determine the exact number to reduce CMDS_PENDING by in this scenario without running the qoutfifo, the manual update is sufficient. Clean up diagnostics. Add ahd_flush_qoutfifo() which will run the qoutfifo as well as complete any commands sitting on the sequencer's COMPLETE_SCB lists or the good status FIFO. Use this routine in several places that did similar things in an add-hoc, but incomplete, fashion. A call to this routine was also added to ahd_abort_scbs() to close a race. In ahd_pause_and_flushwork() only return once selections are safely disabled. Flush all completed commands via ahd_flush_qoutfifo(). Remove "Now packetized" diagnostic now that this information is incorperated into the actual negotiation messages that are displayed. When forcing renegotiation, don't clober the current ppr_options. Much of the driver uses this information to determine if we are currently packetized or not. Remove some stray spaces at column 1 in ahd_set_tags. When complaining about getting a host message loop request with no pending messages, print out the SCB_CONTROL register down on the card. Modify the ahd_sent_msg() routine to handle a search for an outgoing identify message. Use this to detect a msg reject on an identify message which typically indicates that the target thought we were packetized. Force a renegotiation in this case. In ahd_search_qinfifo(), wait more effectively for SCB DMA activities to cease. We also disable SCB fetch operations since we are about to change the qinfifo and any fetch in progress will likely be invalidated. In ahd_qinfifo_count(), fix the qinfifo empty case. In ahd_dump_card_state(), print out CCSCBCTL in the correct mode. If we are a narrow controller, don't set the current width to unknown when forcing a future negotiation. This just confuses the code into attempting a wide negotiation on a narrow bus. Add support for task management function completions. Modify ahd_handle_devreset so that it can handle lun resets in addition to target resets. Use ahd_handle_devreset for lun and target reset task management functions. Handle the abort task TMF race case better. We now wait until any current selections are over and then set the TMF back to zero. This should cause the sequencer to ignore the abort TMF completion should it occur. Correct a bug in the illegal phase handler that caused us to drop down to narrow when handling the unexpected command phase case after 3rd party reset of a packetized device. Indicate the features, bugs, and flags set in the softc that are used to control firmware patch download when booting verbose. aic79xx.h: Add coalessing and HS_MAILBOX fields. Add per-softc variables for the stats "daemon". Add a debug option for interrupt coalessing activities. Add two new softc flags: o AHD_UPDATE_PEND_CMDS Run ahd_reset_cmds_pending() on the next unpause. o AHD_RUNNING_QOUTFIFO Used to catch recursion through ahd_run_qoutfifo(). aic79xx.reg: Correct register addresses related to the software timer and the DFDBCTL register. Add constants paramaterizing the software timer. Add scratch ram locations for storing interrupt coalessing tunables. Break INTMASK in SEQITNCTL out into INTMASK1 and INTMASK2. In at least the REV A, these are writable bits. We make use of that for a swtimer workaround in the sequencer. Since HS_MAILBOX autoclears, provide a sequencer variable to store its contents. Add SEQINT codes for handling task management completions. aic79xx.seq: Correct ignore wide residue processing check for a wide negotiation being in effect. We must be in the SCSI register window in order to access the negotiation table. Use the software timer and a commands completed count to implement interrupt coalessing. The command complete is deferred until either the maximum command threshold or a the expiration of a command deferral timer. If we have more SCBs to complete to the host (sitting in COMPLETE_SCB lists), always try to coaless them up to our coalessing limit. If coalessing is enabled, but we have fewer commands oustanting than the host's min coalessing limit, complete the command immediately. Add code to track the number of commands outstanding. Commands are outstanding from the time they are placed into the execution queue until the DMA to post completion is setup. Add a workaround for intvec_2 interrupts on the H2A4. In H2A4, the mode pointer is not saved for intvec2, but is restored on iret. This can lead to the restoration of a bogus mode ptr. Manually clear the intmask bits and do a normal return to compensate. We use intvec_2 to track interrupt coalessing timeouts. Since we cannot disable the swtimer's countdown, simply mask its interrupt once we no longer care about it firing. In idle_loop_cchan, update LOCAL_HS_MAILBOX everytime we are notified of an HS_MAILBOX update via the HS_MAILBOX_ACT bit in QOFF_CTLSTA. We have to use a local copy of persistant portions of the HS_MAILBOX as the mailbox auto-clears on any read. Move the test for the cfg4istat interrupt up an instruction to hopefully close a race between the next outgoing selection and our disabling of selections. Add a missing ret to the last instruction in load_overrun_buf. Add notifications to the host of task management completions as well as the completions for commands that completed successfully before their corresponding TMF could be sent. Hold a critical section during select-out processing until we have a fully identified connection. This removes a race condition with the legacy abort handler. Correct a few spelling errors in some comments. aic79xx_inline.h: Call ahd_reset_cmds_pending() in ahd_unpause if required. Update cmdcmplt interrupt statistics in our interrupt handler. Allow callers to ahd_send_scb() to set the task management function. aic79xx_pci.c: Disable SERR and pause the controller prior to performing our mmapped I/O test. The U320 controllers do not support "auto-access-pause". aic79xx_osm.c: Set the task management function now that ahd_send_scb() doesn't do it for us. We also perform a lun reset in response to BDR requests to packetized devices.
2003-01-20 20:17:35 +00:00
/*
* If a command completed before an attempted task management
* function completed, notify the host after disabling any
* pending select-outs.
*/
test SCB_TASK_MANAGEMENT, 0xFF jz gsfifo_complete_normally;
test SSTAT0, SELDO|SELINGO jnz . + 2;
and SCSISEQ0, ~ENSELO;
SET_SEQINTCODE(TASKMGMT_CMD_CMPLT_OKAY)
gsfifo_complete_normally:
or SCB_CONTROL, STATUS_RCVD;
/*
* Since this status did not consume a FIFO, we have to
* be a bit more dilligent in how we check for FIFOs pertaining
aic79xx.c: Clear the LQICRC_NLQ status should it pop up after we have already handled the SCSIPERR. During some streaming operations this status can be delayed until the stream ends. Without this change, the driver would complain about a "Missing case in ahd_handle_scsiint". In the LQOBUSFREE handler... Don't return the LQOMGR back to the idle state until after we have cleaned up ENSELO and any status related to this selection. The last thing we need is the LQO manager starting another select-out before we have updated the execution queue. It is not clear whether the LQOMGR would, or would not start a new selection early. Make sure ENSELO is off prior to clearing SELDO by flushing device writes. Move assignment of the next target SCB pointer inside of an if to make the code clearer. The effect is the same. Dump card state in both "Unexpected PKT busfree" paths. In ahd_reset(), set the chip to SCSI mode before reading SXFRCTL1. That register only exists in the SCSI mode. Also set the mode explicitly to the SCSI mode after chip reset due to paranoia. Re-arrange code so that SXFRCTL1 is restored as quickly after the chip reset as possible. S/G structurs must be 8byte aligned. Make this official by saying so in our DMA tag. Disable CIO bus stretch on MDFFSTAT if SHVALID is about to come true. This can cause a CIO bus lockup if a PCI or PCI-X error occurs while the stretch is occurring - the host cannot service the PCI-X error since the CIO bus is locked out and SHVALID will never resolve. The stretch was added in the Rev B to simplify the wait for SHVALID to resolve, but the code to do this in the open source sequencer is so simple it was never removed. Consistently use MAX_OFFSET for the user max syncrate set from non-volatile storage. This ensures that the offset does not conflict with AH?_OFFSET_UNKNOWN. Have ahd_pause_and_flushwork set the mode to ensure that it has access to the registers it checks. Also modify the checking of intstat so that the check against 0xFF can actually succeed if the INT_PEND mask is something other than 0xFF. Although there are no cardbus U320 controllers, this check may be needed to recover from a hot-plug PCI removal that occurs without informing the driver. Fix a typo. sg_prefetch_cnt -> sg_prefetch_align. This fixes an infinite loop at card initialization if the cacheline size is 0. aic79xx.h: Add AHD_EARLY_REQ_BUG bug flag. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. Change AH?_OFFSET_UNKNOWN to 0xFF. This is a value that the curr->offset can never be, unlike '0' which we previously used. This fixes code that only checks for a non-zero offset to determine if a sync negotiation is required since it will fire in the unknown case even if the goal is async. aic79xx.reg: Add comments for LQISTAT bits indicating their names in the 7902 data book. We use slightly different and more descriptive names in the firmware. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. aic79xx.seq: Update comments regarding rundown of the GSFIFO to reflect reality. Fix spelling errors. Since we use an 8byte address and 1 byte length, shorten the size of a block move for the legacy DMA'ed CDB case from 11 to 9 bytes. Remove code that, assuming the abort pending feature worked, would set MK_MESSAGE in the SCB's control byte on completion to catch invalid reselections. Since we don't see interrupts for completed selections, this status update could occur prior to us noticing the SELDO. The "select-out" queue logic will get confused by the MK_MESSAGE bit being set as this is used to catch packatized connections where we select-out with ATN. Since the abort pending feature doesn't work on any released controllers yet, this code was never executed. Add support for the AHD_EARLY_REQ_BUG. Don't ignore persistent REQ assertions just because they were asserted within the bus settle delay window. This allows us to tolerate devices like the GEM318 that violate the SCSI spec. Remove unintentional settnig of SG_CACHE_AVAIL. Writing this bit should have no effect, but who knows... On the Rev A, we must wait for HDMAENACK before loading additional segments to avoid clobbering the address of the first segment in the S/G FIFO. This resolves data-corruption issues with certain IBM (now Hitachi) and Fujitsu U320 drives. Rearrange calc_residual to avoid an extra jmp instruction. On RevA Silicon, if the target returns us to data-out after we have already trained for data-out, it is possible for us to transition the free running clock to data-valid before the required 100ns P1 setup time (8 P1 assertions in fast-160 mode). This will only happen if this L-Q is a continuation of a data transfer for which we have already prefetched data into our FIFO (LQ/Data followed by LQ/Data for the same write transaction). This can cause some target implementations to miss the first few data transfers on the bus. We detect this situation by noticing that this is the first data transfer after an LQ (LQIWORKONLQ true), that the data transfer is a continuation of a transfer already setup in our FIFO (SAVEPTRS interrupt), and that the transaction is a write (DIRECTION set in DFCNTRL). The delay is performed by disabling SCSIEN until we see the first REQ from the target. Only compile in snapshot savepointers handler for RevA silicon where it is enabled. Handle the cfg4icmd packetized interrupt. We just need to load the address and count, start the DMA, and CLRCHN once the transfer is complete. Fix an oversight in the overrun handler for packetized status operations. We need to wait for either CTXTDONE or an overrun when checking for an overrun. The previous code did not wait and thus could decide that no overrun had occurred even though an overrun will occur on the next data-valid req. Add some comment to this section for clarity. Use LAST_SEG_DONE instead of LASTSDONE for testing transfer completion in the packetized status case. LASTSDONE may come up more quickly since it only records completion on the SCSI side, but since LAST_SEG_DONE is used everywhere else (and needs to be), this is less confusing. Add a missing invalidation of the longjmp address in the non-pack handler. This code needs additional review. aic79xx_inline.h: Fix spelling error. aic79xx_osm.c: Set the cdb length for CDBs dma'ed from host memory. Add a comment indicating that, should CAM start supporting cdbs larger than 16bytes, the driver could store the CDB in the status buffer. aic79xx_pci.c: Add a table entry for the 39320A. Added a missing comma to an error string table. Fix spelling errors.
2003-02-27 23:23:16 +00:00
* to this transaction. There are two states that a FIFO still
* transferring data may be in.
*
aic79xx.c: Clear the LQICRC_NLQ status should it pop up after we have already handled the SCSIPERR. During some streaming operations this status can be delayed until the stream ends. Without this change, the driver would complain about a "Missing case in ahd_handle_scsiint". In the LQOBUSFREE handler... Don't return the LQOMGR back to the idle state until after we have cleaned up ENSELO and any status related to this selection. The last thing we need is the LQO manager starting another select-out before we have updated the execution queue. It is not clear whether the LQOMGR would, or would not start a new selection early. Make sure ENSELO is off prior to clearing SELDO by flushing device writes. Move assignment of the next target SCB pointer inside of an if to make the code clearer. The effect is the same. Dump card state in both "Unexpected PKT busfree" paths. In ahd_reset(), set the chip to SCSI mode before reading SXFRCTL1. That register only exists in the SCSI mode. Also set the mode explicitly to the SCSI mode after chip reset due to paranoia. Re-arrange code so that SXFRCTL1 is restored as quickly after the chip reset as possible. S/G structurs must be 8byte aligned. Make this official by saying so in our DMA tag. Disable CIO bus stretch on MDFFSTAT if SHVALID is about to come true. This can cause a CIO bus lockup if a PCI or PCI-X error occurs while the stretch is occurring - the host cannot service the PCI-X error since the CIO bus is locked out and SHVALID will never resolve. The stretch was added in the Rev B to simplify the wait for SHVALID to resolve, but the code to do this in the open source sequencer is so simple it was never removed. Consistently use MAX_OFFSET for the user max syncrate set from non-volatile storage. This ensures that the offset does not conflict with AH?_OFFSET_UNKNOWN. Have ahd_pause_and_flushwork set the mode to ensure that it has access to the registers it checks. Also modify the checking of intstat so that the check against 0xFF can actually succeed if the INT_PEND mask is something other than 0xFF. Although there are no cardbus U320 controllers, this check may be needed to recover from a hot-plug PCI removal that occurs without informing the driver. Fix a typo. sg_prefetch_cnt -> sg_prefetch_align. This fixes an infinite loop at card initialization if the cacheline size is 0. aic79xx.h: Add AHD_EARLY_REQ_BUG bug flag. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. Change AH?_OFFSET_UNKNOWN to 0xFF. This is a value that the curr->offset can never be, unlike '0' which we previously used. This fixes code that only checks for a non-zero offset to determine if a sync negotiation is required since it will fire in the unknown case even if the goal is async. aic79xx.reg: Add comments for LQISTAT bits indicating their names in the 7902 data book. We use slightly different and more descriptive names in the firmware. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. aic79xx.seq: Update comments regarding rundown of the GSFIFO to reflect reality. Fix spelling errors. Since we use an 8byte address and 1 byte length, shorten the size of a block move for the legacy DMA'ed CDB case from 11 to 9 bytes. Remove code that, assuming the abort pending feature worked, would set MK_MESSAGE in the SCB's control byte on completion to catch invalid reselections. Since we don't see interrupts for completed selections, this status update could occur prior to us noticing the SELDO. The "select-out" queue logic will get confused by the MK_MESSAGE bit being set as this is used to catch packatized connections where we select-out with ATN. Since the abort pending feature doesn't work on any released controllers yet, this code was never executed. Add support for the AHD_EARLY_REQ_BUG. Don't ignore persistent REQ assertions just because they were asserted within the bus settle delay window. This allows us to tolerate devices like the GEM318 that violate the SCSI spec. Remove unintentional settnig of SG_CACHE_AVAIL. Writing this bit should have no effect, but who knows... On the Rev A, we must wait for HDMAENACK before loading additional segments to avoid clobbering the address of the first segment in the S/G FIFO. This resolves data-corruption issues with certain IBM (now Hitachi) and Fujitsu U320 drives. Rearrange calc_residual to avoid an extra jmp instruction. On RevA Silicon, if the target returns us to data-out after we have already trained for data-out, it is possible for us to transition the free running clock to data-valid before the required 100ns P1 setup time (8 P1 assertions in fast-160 mode). This will only happen if this L-Q is a continuation of a data transfer for which we have already prefetched data into our FIFO (LQ/Data followed by LQ/Data for the same write transaction). This can cause some target implementations to miss the first few data transfers on the bus. We detect this situation by noticing that this is the first data transfer after an LQ (LQIWORKONLQ true), that the data transfer is a continuation of a transfer already setup in our FIFO (SAVEPTRS interrupt), and that the transaction is a write (DIRECTION set in DFCNTRL). The delay is performed by disabling SCSIEN until we see the first REQ from the target. Only compile in snapshot savepointers handler for RevA silicon where it is enabled. Handle the cfg4icmd packetized interrupt. We just need to load the address and count, start the DMA, and CLRCHN once the transfer is complete. Fix an oversight in the overrun handler for packetized status operations. We need to wait for either CTXTDONE or an overrun when checking for an overrun. The previous code did not wait and thus could decide that no overrun had occurred even though an overrun will occur on the next data-valid req. Add some comment to this section for clarity. Use LAST_SEG_DONE instead of LASTSDONE for testing transfer completion in the packetized status case. LASTSDONE may come up more quickly since it only records completion on the SCSI side, but since LAST_SEG_DONE is used everywhere else (and needs to be), this is less confusing. Add a missing invalidation of the longjmp address in the non-pack handler. This code needs additional review. aic79xx_inline.h: Fix spelling error. aic79xx_osm.c: Set the cdb length for CDBs dma'ed from host memory. Add a comment indicating that, should CAM start supporting cdbs larger than 16bytes, the driver could store the CDB in the status buffer. aic79xx_pci.c: Add a table entry for the 39320A. Added a missing comma to an error string table. Fix spelling errors.
2003-02-27 23:23:16 +00:00
* 1) Configured and draining to the host, with a FIFO handler.
* 2) Pending cfg4data, fifo not empty.
*
Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-05-04 00:20:07 +00:00
* Case 1 can be detected by noticing a non-zero FIFO active
* count in the SCB. In this case, we allow the routine servicing
* the FIFO to complete the SCB.
*
aic79xx.c: Clear the LQICRC_NLQ status should it pop up after we have already handled the SCSIPERR. During some streaming operations this status can be delayed until the stream ends. Without this change, the driver would complain about a "Missing case in ahd_handle_scsiint". In the LQOBUSFREE handler... Don't return the LQOMGR back to the idle state until after we have cleaned up ENSELO and any status related to this selection. The last thing we need is the LQO manager starting another select-out before we have updated the execution queue. It is not clear whether the LQOMGR would, or would not start a new selection early. Make sure ENSELO is off prior to clearing SELDO by flushing device writes. Move assignment of the next target SCB pointer inside of an if to make the code clearer. The effect is the same. Dump card state in both "Unexpected PKT busfree" paths. In ahd_reset(), set the chip to SCSI mode before reading SXFRCTL1. That register only exists in the SCSI mode. Also set the mode explicitly to the SCSI mode after chip reset due to paranoia. Re-arrange code so that SXFRCTL1 is restored as quickly after the chip reset as possible. S/G structurs must be 8byte aligned. Make this official by saying so in our DMA tag. Disable CIO bus stretch on MDFFSTAT if SHVALID is about to come true. This can cause a CIO bus lockup if a PCI or PCI-X error occurs while the stretch is occurring - the host cannot service the PCI-X error since the CIO bus is locked out and SHVALID will never resolve. The stretch was added in the Rev B to simplify the wait for SHVALID to resolve, but the code to do this in the open source sequencer is so simple it was never removed. Consistently use MAX_OFFSET for the user max syncrate set from non-volatile storage. This ensures that the offset does not conflict with AH?_OFFSET_UNKNOWN. Have ahd_pause_and_flushwork set the mode to ensure that it has access to the registers it checks. Also modify the checking of intstat so that the check against 0xFF can actually succeed if the INT_PEND mask is something other than 0xFF. Although there are no cardbus U320 controllers, this check may be needed to recover from a hot-plug PCI removal that occurs without informing the driver. Fix a typo. sg_prefetch_cnt -> sg_prefetch_align. This fixes an infinite loop at card initialization if the cacheline size is 0. aic79xx.h: Add AHD_EARLY_REQ_BUG bug flag. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. Change AH?_OFFSET_UNKNOWN to 0xFF. This is a value that the curr->offset can never be, unlike '0' which we previously used. This fixes code that only checks for a non-zero offset to determine if a sync negotiation is required since it will fire in the unknown case even if the goal is async. aic79xx.reg: Add comments for LQISTAT bits indicating their names in the 7902 data book. We use slightly different and more descriptive names in the firmware. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. aic79xx.seq: Update comments regarding rundown of the GSFIFO to reflect reality. Fix spelling errors. Since we use an 8byte address and 1 byte length, shorten the size of a block move for the legacy DMA'ed CDB case from 11 to 9 bytes. Remove code that, assuming the abort pending feature worked, would set MK_MESSAGE in the SCB's control byte on completion to catch invalid reselections. Since we don't see interrupts for completed selections, this status update could occur prior to us noticing the SELDO. The "select-out" queue logic will get confused by the MK_MESSAGE bit being set as this is used to catch packatized connections where we select-out with ATN. Since the abort pending feature doesn't work on any released controllers yet, this code was never executed. Add support for the AHD_EARLY_REQ_BUG. Don't ignore persistent REQ assertions just because they were asserted within the bus settle delay window. This allows us to tolerate devices like the GEM318 that violate the SCSI spec. Remove unintentional settnig of SG_CACHE_AVAIL. Writing this bit should have no effect, but who knows... On the Rev A, we must wait for HDMAENACK before loading additional segments to avoid clobbering the address of the first segment in the S/G FIFO. This resolves data-corruption issues with certain IBM (now Hitachi) and Fujitsu U320 drives. Rearrange calc_residual to avoid an extra jmp instruction. On RevA Silicon, if the target returns us to data-out after we have already trained for data-out, it is possible for us to transition the free running clock to data-valid before the required 100ns P1 setup time (8 P1 assertions in fast-160 mode). This will only happen if this L-Q is a continuation of a data transfer for which we have already prefetched data into our FIFO (LQ/Data followed by LQ/Data for the same write transaction). This can cause some target implementations to miss the first few data transfers on the bus. We detect this situation by noticing that this is the first data transfer after an LQ (LQIWORKONLQ true), that the data transfer is a continuation of a transfer already setup in our FIFO (SAVEPTRS interrupt), and that the transaction is a write (DIRECTION set in DFCNTRL). The delay is performed by disabling SCSIEN until we see the first REQ from the target. Only compile in snapshot savepointers handler for RevA silicon where it is enabled. Handle the cfg4icmd packetized interrupt. We just need to load the address and count, start the DMA, and CLRCHN once the transfer is complete. Fix an oversight in the overrun handler for packetized status operations. We need to wait for either CTXTDONE or an overrun when checking for an overrun. The previous code did not wait and thus could decide that no overrun had occurred even though an overrun will occur on the next data-valid req. Add some comment to this section for clarity. Use LAST_SEG_DONE instead of LASTSDONE for testing transfer completion in the packetized status case. LASTSDONE may come up more quickly since it only records completion on the SCSI side, but since LAST_SEG_DONE is used everywhere else (and needs to be), this is less confusing. Add a missing invalidation of the longjmp address in the non-pack handler. This code needs additional review. aic79xx_inline.h: Fix spelling error. aic79xx_osm.c: Set the cdb length for CDBs dma'ed from host memory. Add a comment indicating that, should CAM start supporting cdbs larger than 16bytes, the driver could store the CDB in the status buffer. aic79xx_pci.c: Add a table entry for the 39320A. Added a missing comma to an error string table. Fix spelling errors.
2003-02-27 23:23:16 +00:00
* Case 2 implies either a pending or yet to occur save data
* pointers for this same context in the other FIFO. So, if
aic79xx.c: Clear the LQICRC_NLQ status should it pop up after we have already handled the SCSIPERR. During some streaming operations this status can be delayed until the stream ends. Without this change, the driver would complain about a "Missing case in ahd_handle_scsiint". In the LQOBUSFREE handler... Don't return the LQOMGR back to the idle state until after we have cleaned up ENSELO and any status related to this selection. The last thing we need is the LQO manager starting another select-out before we have updated the execution queue. It is not clear whether the LQOMGR would, or would not start a new selection early. Make sure ENSELO is off prior to clearing SELDO by flushing device writes. Move assignment of the next target SCB pointer inside of an if to make the code clearer. The effect is the same. Dump card state in both "Unexpected PKT busfree" paths. In ahd_reset(), set the chip to SCSI mode before reading SXFRCTL1. That register only exists in the SCSI mode. Also set the mode explicitly to the SCSI mode after chip reset due to paranoia. Re-arrange code so that SXFRCTL1 is restored as quickly after the chip reset as possible. S/G structurs must be 8byte aligned. Make this official by saying so in our DMA tag. Disable CIO bus stretch on MDFFSTAT if SHVALID is about to come true. This can cause a CIO bus lockup if a PCI or PCI-X error occurs while the stretch is occurring - the host cannot service the PCI-X error since the CIO bus is locked out and SHVALID will never resolve. The stretch was added in the Rev B to simplify the wait for SHVALID to resolve, but the code to do this in the open source sequencer is so simple it was never removed. Consistently use MAX_OFFSET for the user max syncrate set from non-volatile storage. This ensures that the offset does not conflict with AH?_OFFSET_UNKNOWN. Have ahd_pause_and_flushwork set the mode to ensure that it has access to the registers it checks. Also modify the checking of intstat so that the check against 0xFF can actually succeed if the INT_PEND mask is something other than 0xFF. Although there are no cardbus U320 controllers, this check may be needed to recover from a hot-plug PCI removal that occurs without informing the driver. Fix a typo. sg_prefetch_cnt -> sg_prefetch_align. This fixes an infinite loop at card initialization if the cacheline size is 0. aic79xx.h: Add AHD_EARLY_REQ_BUG bug flag. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. Change AH?_OFFSET_UNKNOWN to 0xFF. This is a value that the curr->offset can never be, unlike '0' which we previously used. This fixes code that only checks for a non-zero offset to determine if a sync negotiation is required since it will fire in the unknown case even if the goal is async. aic79xx.reg: Add comments for LQISTAT bits indicating their names in the 7902 data book. We use slightly different and more descriptive names in the firmware. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. aic79xx.seq: Update comments regarding rundown of the GSFIFO to reflect reality. Fix spelling errors. Since we use an 8byte address and 1 byte length, shorten the size of a block move for the legacy DMA'ed CDB case from 11 to 9 bytes. Remove code that, assuming the abort pending feature worked, would set MK_MESSAGE in the SCB's control byte on completion to catch invalid reselections. Since we don't see interrupts for completed selections, this status update could occur prior to us noticing the SELDO. The "select-out" queue logic will get confused by the MK_MESSAGE bit being set as this is used to catch packatized connections where we select-out with ATN. Since the abort pending feature doesn't work on any released controllers yet, this code was never executed. Add support for the AHD_EARLY_REQ_BUG. Don't ignore persistent REQ assertions just because they were asserted within the bus settle delay window. This allows us to tolerate devices like the GEM318 that violate the SCSI spec. Remove unintentional settnig of SG_CACHE_AVAIL. Writing this bit should have no effect, but who knows... On the Rev A, we must wait for HDMAENACK before loading additional segments to avoid clobbering the address of the first segment in the S/G FIFO. This resolves data-corruption issues with certain IBM (now Hitachi) and Fujitsu U320 drives. Rearrange calc_residual to avoid an extra jmp instruction. On RevA Silicon, if the target returns us to data-out after we have already trained for data-out, it is possible for us to transition the free running clock to data-valid before the required 100ns P1 setup time (8 P1 assertions in fast-160 mode). This will only happen if this L-Q is a continuation of a data transfer for which we have already prefetched data into our FIFO (LQ/Data followed by LQ/Data for the same write transaction). This can cause some target implementations to miss the first few data transfers on the bus. We detect this situation by noticing that this is the first data transfer after an LQ (LQIWORKONLQ true), that the data transfer is a continuation of a transfer already setup in our FIFO (SAVEPTRS interrupt), and that the transaction is a write (DIRECTION set in DFCNTRL). The delay is performed by disabling SCSIEN until we see the first REQ from the target. Only compile in snapshot savepointers handler for RevA silicon where it is enabled. Handle the cfg4icmd packetized interrupt. We just need to load the address and count, start the DMA, and CLRCHN once the transfer is complete. Fix an oversight in the overrun handler for packetized status operations. We need to wait for either CTXTDONE or an overrun when checking for an overrun. The previous code did not wait and thus could decide that no overrun had occurred even though an overrun will occur on the next data-valid req. Add some comment to this section for clarity. Use LAST_SEG_DONE instead of LASTSDONE for testing transfer completion in the packetized status case. LASTSDONE may come up more quickly since it only records completion on the SCSI side, but since LAST_SEG_DONE is used everywhere else (and needs to be), this is less confusing. Add a missing invalidation of the longjmp address in the non-pack handler. This code needs additional review. aic79xx_inline.h: Fix spelling error. aic79xx_osm.c: Set the cdb length for CDBs dma'ed from host memory. Add a comment indicating that, should CAM start supporting cdbs larger than 16bytes, the driver could store the CDB in the status buffer. aic79xx_pci.c: Add a table entry for the 39320A. Added a missing comma to an error string table. Fix spelling errors.
2003-02-27 23:23:16 +00:00
* we detect case 1, we will properly defer the post of the SCB
* and achieve the desired result. The pending cfg4data will
* notice that status has been received and complete the SCB.
*/
Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-05-04 00:20:07 +00:00
test SCB_FIFO_USE_COUNT, 0xFF jnz idle_loop_gsfifo_in_scsi_mode;
call complete;
END_CRITICAL;
jmp idle_loop_gsfifo_in_scsi_mode;
idle_loop_service_fifos:
Upgrade to version 1.1 of the aic79xx U320 driver. aic79xx.c: o Remove redundant ahd_update_modes() call. o Correct panic in diagnostic should state corruption cause the SCB Id to be invalid during a selection timeout. o Add workaround for missing BUSFREEREV feature in Rev A silicon. o Corect formatting nits. o Use register pretty printing in more places. o Save and restore our SCB pointer when updating the waiting queue list for an "expected" LQ-out busfree. o In ahd_clear_intstat, deal with the missing autoclear in the CLRLQO* registers. o BE fixup in a diagnostic printf. o Make sure that we are in the proper mode before disabling selections in ahd_update_pending_scbs. o Add more diagnostics. o task_attribute_nonpkt_tag -> task_attribute: we don't need a nonpkt_tag field anymore for allowing all 512 SCBs to be used in non-packetized connections. o Negotiate HOLD_MCS to U320 devices. o Add a few additional mode assertions. o Restore the chip mode after clearing out the qinfifo so that code using ahd_abort_scbs sees a consistent mode. o Simplify the DMA engine shutdown routine prior to performing a bus reset. o Perform the sequencer restart after a chip reset prior to setting up our timer to poll for the reset to be complete. On some OSes, the timer could actually pre-empt us and order is important here. o Have our "reset poller" set the expected mode since there is no guarantee of what mode will be in force when we are called from the OS timer. o Save and restore the SCB pointer in ahd_dump_card_state(). This routine must not modify card state. o Ditto for ahd_dump_scbs(). aic79xx.h: o Add a few more chip bug definitions. o Align our tag on a 32bit boundary. aic79xx.reg: aic79xx.seq: o Start work on removing workarounds for Rev B. o Use a special location in scratch from for stroring our SCBPTR during legacy FIFO allocations. This corrects problems in mixed packetized/non-packetized configurations where calling into a FIFO task corrupted our SCBPTR. o Don't rely on DMA priority to guarantee that all data in our FIFOs will flush prior to a command completion notification going out of the command channel. We've never seen this assumption fail, but better safe than sorry. o Deal with missing BUSFREEREV feature in H2A. o Simplify disconnect list code now that the list will always have only a single entry. o Implement the AHD_REG_SLOW_SETTLE_BUG workaround. o Swith to using "REG_ISR" for local mode scratch during our ISR. o Add a missing jmp to the data_group_dma_loop after our data pointers have been re-initialized by the kernel. o Correct test in the bitbucket code so that we actually wait for the bitbucket to complete before signaling the kernel of the overrun condition. o Reposition pkt_saveptrs to avoid a jmp instruction. o Update a comment to reflect that the code now waits for a FIFO to drain prior to issuing a CLRCHN. aic79xx_inline.h: o Remove unused untagged queue handling code. o Don't attempt to htole64 what could be a 32bit value. aic79xx_pci.c: o Set additional bug flags for rev A chips.
2002-09-26 22:54:00 +00:00
SET_MODE(M_DFF0, M_DFF0)
ahc_eisa.c: ahc_pci.c: ahd_pci.c: aic7xxx.c: aic79xx.c: aic_osm_lib.c: aic_osm_lib.h: Use common OSM routines from aic_osm_lib for bus dma operations, delay routines, accessing CCBs, byte swapping, etc. aic7xxx_pci.c: Provide a better description for the 2915/30LP on attach. aic7xxx.c: aic79xx.c: aic7770.c: aic79xx_pci.c: aic7xxx_pci.c: aic7xxx_93cx6.c: Move FBSDID behind an ifdef so that these core files will still compile under other OSes. aic79xx.h: aic79xx_pci.c: aic79xx.seq: To speed up non-packetized CDB delivery in Rev B, all CDB acks are "released" to the output sync as soon as the command phase starts. There is only one problem with this approach. If the target changes phase before all data are sent, we have left over acks that can go out on the bus in a data phase. Due to other chip contraints, this only happens if the target goes to data-in, but if the acks go out before we can test SDONE, we'll think that the transfer has completed successfully. Work around this by taking advantage of the 400ns or 800ns dead time between command phase and the REQ of the new phase. If the transfer has completed successfully, SCSIEN should fall *long* before we see a phase change. We thus treat any phasemiss that occurs before SCSIEN falls as an incomplete transfer. aic79xx.h: Add the AHD_FAST_CDB_DELIVERY feature. aic79xx_pci.c: Set AHD_FAST_CDB_DELIVERY for all Rev. B parts. aic79xx.seq: Test for PHASEMIS in the command phase for all AHD_FAST_CDB_DELIVERY controlelrs. ahd_pci.c: ahc_pci.c: aic7xxx.h: aic79xx.h: Move definition of controller BAR offsets to core header files. aic7xxx.c: aic79xx.c: In the softc free routine, leave removal of a softc from the global list of softcs to the OSM (the caller of this routine). This allows us to avoid holding the softc list_lock during device destruction where we may have to sleep waiting for our recovery thread to halt. ahc_pci.c: Use ahc_pci_test_register access to validate I/O mapped in addition to the tests already performed for memory mapped access. Remove unused ahc_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. ahd_pci.c: Remove reduntant definition of controller BAR offsets. These are also defined in aic79xx.h. Remove unused ahd_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. aic7xxx.c: aic79xx.c: aic79xx.h: aic7xxx.h: aic7xxx_osm.c: aic79xx_osm.c: Move timeout handling to the driver cores. In the case of the aic79xx driver, the algorithm has been enhanced to try target resets before performing a bus reset. For the aic7xxx driver, the algorithm is unchanged. Although the drivers do not currently sleep during recovery (recovery is timeout driven), the cores do expect all processing to be performed via a recovery thread. Our timeout handlers are now little stubs that wakeup the recovery thread. aic79xx.c: aic79xx.h: aic79xx_inline.h: Change shared_data allocation to use a map_node so that the sentinel hscb can use this map node in ahd_swap_with_next_hscb. This routine now swaps the hscb_map pointer in additon to the hscb contents so that any sync operations occur on the correct map. physaddr -> busaddr Pointed out by: Jason Thorpe <thorpej@wasabisystems.com> aic79xx.c: Make more use of the in/out/w/l/q macros for accessing byte registers in the chip. Correct some issues in the ahd_flush_qoutfifo() routine. o Run the qoutfifo only once the command channel DMA engine has been halted. This closes a window where we might have missed some entries. o Change ahd_run_data_fifo() to not loop to completion. If we happen to start on the wrong FIFO and the other FIFO has a snapshot savepointers, we might deadlock. This required our delay between FIFO tests to be moved to the ahd_flush_qoutfifo() routine. o Update/add comments. o Remove spurious test for COMPLETE_DMA list being empty when completing transactions from the GSFIFO with residuals. The SCB must be put on the COMPLETE_DMA scb list unconditionally. o When halting command channel DMA activity, we must disable the DMA channel in all cases but an update of the QOUTFIFO. The latter case is required so that the sequencer will update its position in the QOUTFIFO. Previously, we left the channel enabled for all "push" DMAs. This left us vulnerable to the sequencer handling an SCB push long after that SCB was already processed manually by this routine. o Correct the polarity of tests involving ahd_scb_active_in_fifo(). This routine returns non-zero for true. Return to processing bad status completions through the qoutfifo. This reduces the time that the sequencer is kept paused when handling transactions with bad status or underruns. When waiting for the controller to quiece selections, add a delay to our loop. Otherwise we may fail to wait long enough for the sequencer to comply. On H2A4 hardware, use the slow slewrate for non-paced transfers. This mirrors what the Adaptec Windows drivers do. On the Rev B. only slow down the CRC timing for older U160 devices that might need the slower timing. We define "older" as devices that do not support packetized protocol. Wait up to 5000 * 5us for the SEEPROM to become unbusy. Write ops seem to take much longer than read ops. aic79xx.seq: For controllers with the FAINT_LED bug, turn the diagnostic led feature on during selection and reselection. This covers the non-packetized case. The LED will be disabled for non-packetized transfers once we return to the top level idle loop. Add more comments about the busy LED workaround. Extend a critical section around the entire command channel idle loop process. Previously the portion of this handler that directly manipulated the linked list of completed SCBs was not protected. This is the likely cause of the recent reports of commands being completed twice by the driver. Extend critical sections across the test for, and the longjump to, longjump routines. This prevents the firmware from trying to jump to a longjmp handler that was just cleared by the host. Improve the locations of several critical section begin and end points. Typically these changes remove instructions that did not need to be inside a critical section. Close the "busfree after selection, but before busfree interrupts can be enabled" race to just a single sequencer instruction. We now test the BSY line explicitly before clearing the busfree status and enabling the busfree interrupt. Close a race condition in the processing of HS_MAILBOX updates. We now clear the "updated" status before the copy. This ensures that we don't accidentally clear the status incorrectly when the host sneaks in an update just after our last copy, but before we clear the status. This race has never been observed. Don't re-enable SCSIEN if we lose the race to disable SCSIEN in our interrupt handler's workaround for the RevA data-valid too early issue. aic79xx_inline.h: Add comments indicating that the order in which bytes are read or written in ahd_inw and ahd_outw is important. This allows us to use these inlines when accessing registers with side-effects. aic79xx_pci.c: The 29320 and the 29320B are 7902 not 7901 based products. Correct the driver banner. aic7xxx.h: Enable the use of the auto-access pause feature on the aic7870 and aic7880. It was disabled due to an oversight. aic7xxx.reg: Move TARG_IMMEDIATE_SCB to alias LAST_MSG to avoid leaving garbage in MWI_RESIDUAL. This prevents spurious overflows whn operating target mode on controllers that require the MWI_RESIDUAL work-around. aic7xxx.seq: AHC_TMODE_WIDEODD_BUG is a bug, not a softc flag. Reference the correct softc field when testing for its presence. Set the NOT_IDENTIFIED and NO_CDB_SENT bits in SEQ_FLAGS to indicate that the nexus is invalid in await busfree. aic7xxx_93cx6.c: Add support for the C56/C66 versions of the EWEN and EWDS commands. aic7xxx.c: aic7xxx_pci.c: Move test for the validity of left over BIOS data to ahc_test_register_access(). This guarantees that any left over CHIPRST value is not clobbered by our register access test and lost to the test that was in ahc_reset.
2003-12-17 00:02:10 +00:00
BEGIN_CRITICAL;
test LONGJMP_ADDR[1], INVALID_ADDR jnz idle_loop_next_fifo;
call longjmp;
ahc_eisa.c: ahc_pci.c: ahd_pci.c: aic7xxx.c: aic79xx.c: aic_osm_lib.c: aic_osm_lib.h: Use common OSM routines from aic_osm_lib for bus dma operations, delay routines, accessing CCBs, byte swapping, etc. aic7xxx_pci.c: Provide a better description for the 2915/30LP on attach. aic7xxx.c: aic79xx.c: aic7770.c: aic79xx_pci.c: aic7xxx_pci.c: aic7xxx_93cx6.c: Move FBSDID behind an ifdef so that these core files will still compile under other OSes. aic79xx.h: aic79xx_pci.c: aic79xx.seq: To speed up non-packetized CDB delivery in Rev B, all CDB acks are "released" to the output sync as soon as the command phase starts. There is only one problem with this approach. If the target changes phase before all data are sent, we have left over acks that can go out on the bus in a data phase. Due to other chip contraints, this only happens if the target goes to data-in, but if the acks go out before we can test SDONE, we'll think that the transfer has completed successfully. Work around this by taking advantage of the 400ns or 800ns dead time between command phase and the REQ of the new phase. If the transfer has completed successfully, SCSIEN should fall *long* before we see a phase change. We thus treat any phasemiss that occurs before SCSIEN falls as an incomplete transfer. aic79xx.h: Add the AHD_FAST_CDB_DELIVERY feature. aic79xx_pci.c: Set AHD_FAST_CDB_DELIVERY for all Rev. B parts. aic79xx.seq: Test for PHASEMIS in the command phase for all AHD_FAST_CDB_DELIVERY controlelrs. ahd_pci.c: ahc_pci.c: aic7xxx.h: aic79xx.h: Move definition of controller BAR offsets to core header files. aic7xxx.c: aic79xx.c: In the softc free routine, leave removal of a softc from the global list of softcs to the OSM (the caller of this routine). This allows us to avoid holding the softc list_lock during device destruction where we may have to sleep waiting for our recovery thread to halt. ahc_pci.c: Use ahc_pci_test_register access to validate I/O mapped in addition to the tests already performed for memory mapped access. Remove unused ahc_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. ahd_pci.c: Remove reduntant definition of controller BAR offsets. These are also defined in aic79xx.h. Remove unused ahd_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. aic7xxx.c: aic79xx.c: aic79xx.h: aic7xxx.h: aic7xxx_osm.c: aic79xx_osm.c: Move timeout handling to the driver cores. In the case of the aic79xx driver, the algorithm has been enhanced to try target resets before performing a bus reset. For the aic7xxx driver, the algorithm is unchanged. Although the drivers do not currently sleep during recovery (recovery is timeout driven), the cores do expect all processing to be performed via a recovery thread. Our timeout handlers are now little stubs that wakeup the recovery thread. aic79xx.c: aic79xx.h: aic79xx_inline.h: Change shared_data allocation to use a map_node so that the sentinel hscb can use this map node in ahd_swap_with_next_hscb. This routine now swaps the hscb_map pointer in additon to the hscb contents so that any sync operations occur on the correct map. physaddr -> busaddr Pointed out by: Jason Thorpe <thorpej@wasabisystems.com> aic79xx.c: Make more use of the in/out/w/l/q macros for accessing byte registers in the chip. Correct some issues in the ahd_flush_qoutfifo() routine. o Run the qoutfifo only once the command channel DMA engine has been halted. This closes a window where we might have missed some entries. o Change ahd_run_data_fifo() to not loop to completion. If we happen to start on the wrong FIFO and the other FIFO has a snapshot savepointers, we might deadlock. This required our delay between FIFO tests to be moved to the ahd_flush_qoutfifo() routine. o Update/add comments. o Remove spurious test for COMPLETE_DMA list being empty when completing transactions from the GSFIFO with residuals. The SCB must be put on the COMPLETE_DMA scb list unconditionally. o When halting command channel DMA activity, we must disable the DMA channel in all cases but an update of the QOUTFIFO. The latter case is required so that the sequencer will update its position in the QOUTFIFO. Previously, we left the channel enabled for all "push" DMAs. This left us vulnerable to the sequencer handling an SCB push long after that SCB was already processed manually by this routine. o Correct the polarity of tests involving ahd_scb_active_in_fifo(). This routine returns non-zero for true. Return to processing bad status completions through the qoutfifo. This reduces the time that the sequencer is kept paused when handling transactions with bad status or underruns. When waiting for the controller to quiece selections, add a delay to our loop. Otherwise we may fail to wait long enough for the sequencer to comply. On H2A4 hardware, use the slow slewrate for non-paced transfers. This mirrors what the Adaptec Windows drivers do. On the Rev B. only slow down the CRC timing for older U160 devices that might need the slower timing. We define "older" as devices that do not support packetized protocol. Wait up to 5000 * 5us for the SEEPROM to become unbusy. Write ops seem to take much longer than read ops. aic79xx.seq: For controllers with the FAINT_LED bug, turn the diagnostic led feature on during selection and reselection. This covers the non-packetized case. The LED will be disabled for non-packetized transfers once we return to the top level idle loop. Add more comments about the busy LED workaround. Extend a critical section around the entire command channel idle loop process. Previously the portion of this handler that directly manipulated the linked list of completed SCBs was not protected. This is the likely cause of the recent reports of commands being completed twice by the driver. Extend critical sections across the test for, and the longjump to, longjump routines. This prevents the firmware from trying to jump to a longjmp handler that was just cleared by the host. Improve the locations of several critical section begin and end points. Typically these changes remove instructions that did not need to be inside a critical section. Close the "busfree after selection, but before busfree interrupts can be enabled" race to just a single sequencer instruction. We now test the BSY line explicitly before clearing the busfree status and enabling the busfree interrupt. Close a race condition in the processing of HS_MAILBOX updates. We now clear the "updated" status before the copy. This ensures that we don't accidentally clear the status incorrectly when the host sneaks in an update just after our last copy, but before we clear the status. This race has never been observed. Don't re-enable SCSIEN if we lose the race to disable SCSIEN in our interrupt handler's workaround for the RevA data-valid too early issue. aic79xx_inline.h: Add comments indicating that the order in which bytes are read or written in ahd_inw and ahd_outw is important. This allows us to use these inlines when accessing registers with side-effects. aic79xx_pci.c: The 29320 and the 29320B are 7902 not 7901 based products. Correct the driver banner. aic7xxx.h: Enable the use of the auto-access pause feature on the aic7870 and aic7880. It was disabled due to an oversight. aic7xxx.reg: Move TARG_IMMEDIATE_SCB to alias LAST_MSG to avoid leaving garbage in MWI_RESIDUAL. This prevents spurious overflows whn operating target mode on controllers that require the MWI_RESIDUAL work-around. aic7xxx.seq: AHC_TMODE_WIDEODD_BUG is a bug, not a softc flag. Reference the correct softc field when testing for its presence. Set the NOT_IDENTIFIED and NO_CDB_SENT bits in SEQ_FLAGS to indicate that the nexus is invalid in await busfree. aic7xxx_93cx6.c: Add support for the C56/C66 versions of the EWEN and EWDS commands. aic7xxx.c: aic7xxx_pci.c: Move test for the validity of left over BIOS data to ahc_test_register_access(). This guarantees that any left over CHIPRST value is not clobbered by our register access test and lost to the test that was in ahc_reset.
2003-12-17 00:02:10 +00:00
END_CRITICAL;
idle_loop_next_fifo:
Upgrade to version 1.1 of the aic79xx U320 driver. aic79xx.c: o Remove redundant ahd_update_modes() call. o Correct panic in diagnostic should state corruption cause the SCB Id to be invalid during a selection timeout. o Add workaround for missing BUSFREEREV feature in Rev A silicon. o Corect formatting nits. o Use register pretty printing in more places. o Save and restore our SCB pointer when updating the waiting queue list for an "expected" LQ-out busfree. o In ahd_clear_intstat, deal with the missing autoclear in the CLRLQO* registers. o BE fixup in a diagnostic printf. o Make sure that we are in the proper mode before disabling selections in ahd_update_pending_scbs. o Add more diagnostics. o task_attribute_nonpkt_tag -> task_attribute: we don't need a nonpkt_tag field anymore for allowing all 512 SCBs to be used in non-packetized connections. o Negotiate HOLD_MCS to U320 devices. o Add a few additional mode assertions. o Restore the chip mode after clearing out the qinfifo so that code using ahd_abort_scbs sees a consistent mode. o Simplify the DMA engine shutdown routine prior to performing a bus reset. o Perform the sequencer restart after a chip reset prior to setting up our timer to poll for the reset to be complete. On some OSes, the timer could actually pre-empt us and order is important here. o Have our "reset poller" set the expected mode since there is no guarantee of what mode will be in force when we are called from the OS timer. o Save and restore the SCB pointer in ahd_dump_card_state(). This routine must not modify card state. o Ditto for ahd_dump_scbs(). aic79xx.h: o Add a few more chip bug definitions. o Align our tag on a 32bit boundary. aic79xx.reg: aic79xx.seq: o Start work on removing workarounds for Rev B. o Use a special location in scratch from for stroring our SCBPTR during legacy FIFO allocations. This corrects problems in mixed packetized/non-packetized configurations where calling into a FIFO task corrupted our SCBPTR. o Don't rely on DMA priority to guarantee that all data in our FIFOs will flush prior to a command completion notification going out of the command channel. We've never seen this assumption fail, but better safe than sorry. o Deal with missing BUSFREEREV feature in H2A. o Simplify disconnect list code now that the list will always have only a single entry. o Implement the AHD_REG_SLOW_SETTLE_BUG workaround. o Swith to using "REG_ISR" for local mode scratch during our ISR. o Add a missing jmp to the data_group_dma_loop after our data pointers have been re-initialized by the kernel. o Correct test in the bitbucket code so that we actually wait for the bitbucket to complete before signaling the kernel of the overrun condition. o Reposition pkt_saveptrs to avoid a jmp instruction. o Update a comment to reflect that the code now waits for a FIFO to drain prior to issuing a CLRCHN. aic79xx_inline.h: o Remove unused untagged queue handling code. o Don't attempt to htole64 what could be a 32bit value. aic79xx_pci.c: o Set additional bug flags for rev A chips.
2002-09-26 22:54:00 +00:00
SET_MODE(M_DFF1, M_DFF1)
ahc_eisa.c: ahc_pci.c: ahd_pci.c: aic7xxx.c: aic79xx.c: aic_osm_lib.c: aic_osm_lib.h: Use common OSM routines from aic_osm_lib for bus dma operations, delay routines, accessing CCBs, byte swapping, etc. aic7xxx_pci.c: Provide a better description for the 2915/30LP on attach. aic7xxx.c: aic79xx.c: aic7770.c: aic79xx_pci.c: aic7xxx_pci.c: aic7xxx_93cx6.c: Move FBSDID behind an ifdef so that these core files will still compile under other OSes. aic79xx.h: aic79xx_pci.c: aic79xx.seq: To speed up non-packetized CDB delivery in Rev B, all CDB acks are "released" to the output sync as soon as the command phase starts. There is only one problem with this approach. If the target changes phase before all data are sent, we have left over acks that can go out on the bus in a data phase. Due to other chip contraints, this only happens if the target goes to data-in, but if the acks go out before we can test SDONE, we'll think that the transfer has completed successfully. Work around this by taking advantage of the 400ns or 800ns dead time between command phase and the REQ of the new phase. If the transfer has completed successfully, SCSIEN should fall *long* before we see a phase change. We thus treat any phasemiss that occurs before SCSIEN falls as an incomplete transfer. aic79xx.h: Add the AHD_FAST_CDB_DELIVERY feature. aic79xx_pci.c: Set AHD_FAST_CDB_DELIVERY for all Rev. B parts. aic79xx.seq: Test for PHASEMIS in the command phase for all AHD_FAST_CDB_DELIVERY controlelrs. ahd_pci.c: ahc_pci.c: aic7xxx.h: aic79xx.h: Move definition of controller BAR offsets to core header files. aic7xxx.c: aic79xx.c: In the softc free routine, leave removal of a softc from the global list of softcs to the OSM (the caller of this routine). This allows us to avoid holding the softc list_lock during device destruction where we may have to sleep waiting for our recovery thread to halt. ahc_pci.c: Use ahc_pci_test_register access to validate I/O mapped in addition to the tests already performed for memory mapped access. Remove unused ahc_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. ahd_pci.c: Remove reduntant definition of controller BAR offsets. These are also defined in aic79xx.h. Remove unused ahd_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. aic7xxx.c: aic79xx.c: aic79xx.h: aic7xxx.h: aic7xxx_osm.c: aic79xx_osm.c: Move timeout handling to the driver cores. In the case of the aic79xx driver, the algorithm has been enhanced to try target resets before performing a bus reset. For the aic7xxx driver, the algorithm is unchanged. Although the drivers do not currently sleep during recovery (recovery is timeout driven), the cores do expect all processing to be performed via a recovery thread. Our timeout handlers are now little stubs that wakeup the recovery thread. aic79xx.c: aic79xx.h: aic79xx_inline.h: Change shared_data allocation to use a map_node so that the sentinel hscb can use this map node in ahd_swap_with_next_hscb. This routine now swaps the hscb_map pointer in additon to the hscb contents so that any sync operations occur on the correct map. physaddr -> busaddr Pointed out by: Jason Thorpe <thorpej@wasabisystems.com> aic79xx.c: Make more use of the in/out/w/l/q macros for accessing byte registers in the chip. Correct some issues in the ahd_flush_qoutfifo() routine. o Run the qoutfifo only once the command channel DMA engine has been halted. This closes a window where we might have missed some entries. o Change ahd_run_data_fifo() to not loop to completion. If we happen to start on the wrong FIFO and the other FIFO has a snapshot savepointers, we might deadlock. This required our delay between FIFO tests to be moved to the ahd_flush_qoutfifo() routine. o Update/add comments. o Remove spurious test for COMPLETE_DMA list being empty when completing transactions from the GSFIFO with residuals. The SCB must be put on the COMPLETE_DMA scb list unconditionally. o When halting command channel DMA activity, we must disable the DMA channel in all cases but an update of the QOUTFIFO. The latter case is required so that the sequencer will update its position in the QOUTFIFO. Previously, we left the channel enabled for all "push" DMAs. This left us vulnerable to the sequencer handling an SCB push long after that SCB was already processed manually by this routine. o Correct the polarity of tests involving ahd_scb_active_in_fifo(). This routine returns non-zero for true. Return to processing bad status completions through the qoutfifo. This reduces the time that the sequencer is kept paused when handling transactions with bad status or underruns. When waiting for the controller to quiece selections, add a delay to our loop. Otherwise we may fail to wait long enough for the sequencer to comply. On H2A4 hardware, use the slow slewrate for non-paced transfers. This mirrors what the Adaptec Windows drivers do. On the Rev B. only slow down the CRC timing for older U160 devices that might need the slower timing. We define "older" as devices that do not support packetized protocol. Wait up to 5000 * 5us for the SEEPROM to become unbusy. Write ops seem to take much longer than read ops. aic79xx.seq: For controllers with the FAINT_LED bug, turn the diagnostic led feature on during selection and reselection. This covers the non-packetized case. The LED will be disabled for non-packetized transfers once we return to the top level idle loop. Add more comments about the busy LED workaround. Extend a critical section around the entire command channel idle loop process. Previously the portion of this handler that directly manipulated the linked list of completed SCBs was not protected. This is the likely cause of the recent reports of commands being completed twice by the driver. Extend critical sections across the test for, and the longjump to, longjump routines. This prevents the firmware from trying to jump to a longjmp handler that was just cleared by the host. Improve the locations of several critical section begin and end points. Typically these changes remove instructions that did not need to be inside a critical section. Close the "busfree after selection, but before busfree interrupts can be enabled" race to just a single sequencer instruction. We now test the BSY line explicitly before clearing the busfree status and enabling the busfree interrupt. Close a race condition in the processing of HS_MAILBOX updates. We now clear the "updated" status before the copy. This ensures that we don't accidentally clear the status incorrectly when the host sneaks in an update just after our last copy, but before we clear the status. This race has never been observed. Don't re-enable SCSIEN if we lose the race to disable SCSIEN in our interrupt handler's workaround for the RevA data-valid too early issue. aic79xx_inline.h: Add comments indicating that the order in which bytes are read or written in ahd_inw and ahd_outw is important. This allows us to use these inlines when accessing registers with side-effects. aic79xx_pci.c: The 29320 and the 29320B are 7902 not 7901 based products. Correct the driver banner. aic7xxx.h: Enable the use of the auto-access pause feature on the aic7870 and aic7880. It was disabled due to an oversight. aic7xxx.reg: Move TARG_IMMEDIATE_SCB to alias LAST_MSG to avoid leaving garbage in MWI_RESIDUAL. This prevents spurious overflows whn operating target mode on controllers that require the MWI_RESIDUAL work-around. aic7xxx.seq: AHC_TMODE_WIDEODD_BUG is a bug, not a softc flag. Reference the correct softc field when testing for its presence. Set the NOT_IDENTIFIED and NO_CDB_SENT bits in SEQ_FLAGS to indicate that the nexus is invalid in await busfree. aic7xxx_93cx6.c: Add support for the C56/C66 versions of the EWEN and EWDS commands. aic7xxx.c: aic7xxx_pci.c: Move test for the validity of left over BIOS data to ahc_test_register_access(). This guarantees that any left over CHIPRST value is not clobbered by our register access test and lost to the test that was in ahc_reset.
2003-12-17 00:02:10 +00:00
BEGIN_CRITICAL;
test LONGJMP_ADDR[1], INVALID_ADDR jz longjmp;
ahc_eisa.c: ahc_pci.c: ahd_pci.c: aic7xxx.c: aic79xx.c: aic_osm_lib.c: aic_osm_lib.h: Use common OSM routines from aic_osm_lib for bus dma operations, delay routines, accessing CCBs, byte swapping, etc. aic7xxx_pci.c: Provide a better description for the 2915/30LP on attach. aic7xxx.c: aic79xx.c: aic7770.c: aic79xx_pci.c: aic7xxx_pci.c: aic7xxx_93cx6.c: Move FBSDID behind an ifdef so that these core files will still compile under other OSes. aic79xx.h: aic79xx_pci.c: aic79xx.seq: To speed up non-packetized CDB delivery in Rev B, all CDB acks are "released" to the output sync as soon as the command phase starts. There is only one problem with this approach. If the target changes phase before all data are sent, we have left over acks that can go out on the bus in a data phase. Due to other chip contraints, this only happens if the target goes to data-in, but if the acks go out before we can test SDONE, we'll think that the transfer has completed successfully. Work around this by taking advantage of the 400ns or 800ns dead time between command phase and the REQ of the new phase. If the transfer has completed successfully, SCSIEN should fall *long* before we see a phase change. We thus treat any phasemiss that occurs before SCSIEN falls as an incomplete transfer. aic79xx.h: Add the AHD_FAST_CDB_DELIVERY feature. aic79xx_pci.c: Set AHD_FAST_CDB_DELIVERY for all Rev. B parts. aic79xx.seq: Test for PHASEMIS in the command phase for all AHD_FAST_CDB_DELIVERY controlelrs. ahd_pci.c: ahc_pci.c: aic7xxx.h: aic79xx.h: Move definition of controller BAR offsets to core header files. aic7xxx.c: aic79xx.c: In the softc free routine, leave removal of a softc from the global list of softcs to the OSM (the caller of this routine). This allows us to avoid holding the softc list_lock during device destruction where we may have to sleep waiting for our recovery thread to halt. ahc_pci.c: Use ahc_pci_test_register access to validate I/O mapped in addition to the tests already performed for memory mapped access. Remove unused ahc_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. ahd_pci.c: Remove reduntant definition of controller BAR offsets. These are also defined in aic79xx.h. Remove unused ahd_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. aic7xxx.c: aic79xx.c: aic79xx.h: aic7xxx.h: aic7xxx_osm.c: aic79xx_osm.c: Move timeout handling to the driver cores. In the case of the aic79xx driver, the algorithm has been enhanced to try target resets before performing a bus reset. For the aic7xxx driver, the algorithm is unchanged. Although the drivers do not currently sleep during recovery (recovery is timeout driven), the cores do expect all processing to be performed via a recovery thread. Our timeout handlers are now little stubs that wakeup the recovery thread. aic79xx.c: aic79xx.h: aic79xx_inline.h: Change shared_data allocation to use a map_node so that the sentinel hscb can use this map node in ahd_swap_with_next_hscb. This routine now swaps the hscb_map pointer in additon to the hscb contents so that any sync operations occur on the correct map. physaddr -> busaddr Pointed out by: Jason Thorpe <thorpej@wasabisystems.com> aic79xx.c: Make more use of the in/out/w/l/q macros for accessing byte registers in the chip. Correct some issues in the ahd_flush_qoutfifo() routine. o Run the qoutfifo only once the command channel DMA engine has been halted. This closes a window where we might have missed some entries. o Change ahd_run_data_fifo() to not loop to completion. If we happen to start on the wrong FIFO and the other FIFO has a snapshot savepointers, we might deadlock. This required our delay between FIFO tests to be moved to the ahd_flush_qoutfifo() routine. o Update/add comments. o Remove spurious test for COMPLETE_DMA list being empty when completing transactions from the GSFIFO with residuals. The SCB must be put on the COMPLETE_DMA scb list unconditionally. o When halting command channel DMA activity, we must disable the DMA channel in all cases but an update of the QOUTFIFO. The latter case is required so that the sequencer will update its position in the QOUTFIFO. Previously, we left the channel enabled for all "push" DMAs. This left us vulnerable to the sequencer handling an SCB push long after that SCB was already processed manually by this routine. o Correct the polarity of tests involving ahd_scb_active_in_fifo(). This routine returns non-zero for true. Return to processing bad status completions through the qoutfifo. This reduces the time that the sequencer is kept paused when handling transactions with bad status or underruns. When waiting for the controller to quiece selections, add a delay to our loop. Otherwise we may fail to wait long enough for the sequencer to comply. On H2A4 hardware, use the slow slewrate for non-paced transfers. This mirrors what the Adaptec Windows drivers do. On the Rev B. only slow down the CRC timing for older U160 devices that might need the slower timing. We define "older" as devices that do not support packetized protocol. Wait up to 5000 * 5us for the SEEPROM to become unbusy. Write ops seem to take much longer than read ops. aic79xx.seq: For controllers with the FAINT_LED bug, turn the diagnostic led feature on during selection and reselection. This covers the non-packetized case. The LED will be disabled for non-packetized transfers once we return to the top level idle loop. Add more comments about the busy LED workaround. Extend a critical section around the entire command channel idle loop process. Previously the portion of this handler that directly manipulated the linked list of completed SCBs was not protected. This is the likely cause of the recent reports of commands being completed twice by the driver. Extend critical sections across the test for, and the longjump to, longjump routines. This prevents the firmware from trying to jump to a longjmp handler that was just cleared by the host. Improve the locations of several critical section begin and end points. Typically these changes remove instructions that did not need to be inside a critical section. Close the "busfree after selection, but before busfree interrupts can be enabled" race to just a single sequencer instruction. We now test the BSY line explicitly before clearing the busfree status and enabling the busfree interrupt. Close a race condition in the processing of HS_MAILBOX updates. We now clear the "updated" status before the copy. This ensures that we don't accidentally clear the status incorrectly when the host sneaks in an update just after our last copy, but before we clear the status. This race has never been observed. Don't re-enable SCSIEN if we lose the race to disable SCSIEN in our interrupt handler's workaround for the RevA data-valid too early issue. aic79xx_inline.h: Add comments indicating that the order in which bytes are read or written in ahd_inw and ahd_outw is important. This allows us to use these inlines when accessing registers with side-effects. aic79xx_pci.c: The 29320 and the 29320B are 7902 not 7901 based products. Correct the driver banner. aic7xxx.h: Enable the use of the auto-access pause feature on the aic7870 and aic7880. It was disabled due to an oversight. aic7xxx.reg: Move TARG_IMMEDIATE_SCB to alias LAST_MSG to avoid leaving garbage in MWI_RESIDUAL. This prevents spurious overflows whn operating target mode on controllers that require the MWI_RESIDUAL work-around. aic7xxx.seq: AHC_TMODE_WIDEODD_BUG is a bug, not a softc flag. Reference the correct softc field when testing for its presence. Set the NOT_IDENTIFIED and NO_CDB_SENT bits in SEQ_FLAGS to indicate that the nexus is invalid in await busfree. aic7xxx_93cx6.c: Add support for the C56/C66 versions of the EWEN and EWDS commands. aic7xxx.c: aic7xxx_pci.c: Move test for the validity of left over BIOS data to ahc_test_register_access(). This guarantees that any left over CHIPRST value is not clobbered by our register access test and lost to the test that was in ahc_reset.
2003-12-17 00:02:10 +00:00
END_CRITICAL;
Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-05-04 00:20:07 +00:00
return:
ret;
idle_loop_cchan:
Upgrade to version 1.1 of the aic79xx U320 driver. aic79xx.c: o Remove redundant ahd_update_modes() call. o Correct panic in diagnostic should state corruption cause the SCB Id to be invalid during a selection timeout. o Add workaround for missing BUSFREEREV feature in Rev A silicon. o Corect formatting nits. o Use register pretty printing in more places. o Save and restore our SCB pointer when updating the waiting queue list for an "expected" LQ-out busfree. o In ahd_clear_intstat, deal with the missing autoclear in the CLRLQO* registers. o BE fixup in a diagnostic printf. o Make sure that we are in the proper mode before disabling selections in ahd_update_pending_scbs. o Add more diagnostics. o task_attribute_nonpkt_tag -> task_attribute: we don't need a nonpkt_tag field anymore for allowing all 512 SCBs to be used in non-packetized connections. o Negotiate HOLD_MCS to U320 devices. o Add a few additional mode assertions. o Restore the chip mode after clearing out the qinfifo so that code using ahd_abort_scbs sees a consistent mode. o Simplify the DMA engine shutdown routine prior to performing a bus reset. o Perform the sequencer restart after a chip reset prior to setting up our timer to poll for the reset to be complete. On some OSes, the timer could actually pre-empt us and order is important here. o Have our "reset poller" set the expected mode since there is no guarantee of what mode will be in force when we are called from the OS timer. o Save and restore the SCB pointer in ahd_dump_card_state(). This routine must not modify card state. o Ditto for ahd_dump_scbs(). aic79xx.h: o Add a few more chip bug definitions. o Align our tag on a 32bit boundary. aic79xx.reg: aic79xx.seq: o Start work on removing workarounds for Rev B. o Use a special location in scratch from for stroring our SCBPTR during legacy FIFO allocations. This corrects problems in mixed packetized/non-packetized configurations where calling into a FIFO task corrupted our SCBPTR. o Don't rely on DMA priority to guarantee that all data in our FIFOs will flush prior to a command completion notification going out of the command channel. We've never seen this assumption fail, but better safe than sorry. o Deal with missing BUSFREEREV feature in H2A. o Simplify disconnect list code now that the list will always have only a single entry. o Implement the AHD_REG_SLOW_SETTLE_BUG workaround. o Swith to using "REG_ISR" for local mode scratch during our ISR. o Add a missing jmp to the data_group_dma_loop after our data pointers have been re-initialized by the kernel. o Correct test in the bitbucket code so that we actually wait for the bitbucket to complete before signaling the kernel of the overrun condition. o Reposition pkt_saveptrs to avoid a jmp instruction. o Update a comment to reflect that the code now waits for a FIFO to drain prior to issuing a CLRCHN. aic79xx_inline.h: o Remove unused untagged queue handling code. o Don't attempt to htole64 what could be a 32bit value. aic79xx_pci.c: o Set additional bug flags for rev A chips.
2002-09-26 22:54:00 +00:00
SET_MODE(M_CCHAN, M_CCHAN)
aic79xx.c: Implement the SCB_SILENT flag. This is useful for hushing up the driver during DV or other operations that we expect to cause transmission errors. The messages will still print if the SHOW_MASKED_ERRORS debug option is enabled. Save and restore the NEGOADDR address when setting new transfer settings. The sequencer performs lookups in the negotiation table too and it expects NEGOADDR to remain consistent across pause/unpause sessions. Consistently use "offset" instead of "period" to determine if we are running sync or not. Add a SHOW_MESSAGES diagnostic for when we assert ATN during message processing. Print out IU, QAS, and RTI features when showing transfer options. Limit the syncrate after all option conformance changes have taken place in ahd_devlimited_syncrate. Changes in options may change the final syncrate we accept. Keep a copy of the hs_mailbox in our softc so that we can perform read/modify/write operations on the hs_mailbox without having to pause the sequencer to read the last written value. Use the ENINT_COALESS flag in the hs_mailbox to toggle interrupt coalessing. Add entrypoints for enabling interrupt coalessing and setting both a timeout (how long to wait for commands to be coalessed) and a maximum commands to coaless value. Add a statistics timer that decides when to enable or disable interrupt coalessing based on load. Add a routine, ahd_reset_cmds_pending() which is used to update the CMDS_PENDING sequencer variable whenever error recovery compeltes SCBs without notifying the sequencer. Since ahd_reset_cmds_pending is called during ahd_unpause() only if we've aborted SCBs, its call to ahd_flush_qoutfifo should not cause recursion through ahd_run_qoutfifo(). A panic has been added to ensure that this recursion does not occur. In ahd_search_qinfifo, update the CMDS_PENDING sequencer variable directly. ahd_search_qinififo can be called in situations where using ahd_reset_cmds_pending() might cause recursion. Since we can safely determine the exact number to reduce CMDS_PENDING by in this scenario without running the qoutfifo, the manual update is sufficient. Clean up diagnostics. Add ahd_flush_qoutfifo() which will run the qoutfifo as well as complete any commands sitting on the sequencer's COMPLETE_SCB lists or the good status FIFO. Use this routine in several places that did similar things in an add-hoc, but incomplete, fashion. A call to this routine was also added to ahd_abort_scbs() to close a race. In ahd_pause_and_flushwork() only return once selections are safely disabled. Flush all completed commands via ahd_flush_qoutfifo(). Remove "Now packetized" diagnostic now that this information is incorperated into the actual negotiation messages that are displayed. When forcing renegotiation, don't clober the current ppr_options. Much of the driver uses this information to determine if we are currently packetized or not. Remove some stray spaces at column 1 in ahd_set_tags. When complaining about getting a host message loop request with no pending messages, print out the SCB_CONTROL register down on the card. Modify the ahd_sent_msg() routine to handle a search for an outgoing identify message. Use this to detect a msg reject on an identify message which typically indicates that the target thought we were packetized. Force a renegotiation in this case. In ahd_search_qinfifo(), wait more effectively for SCB DMA activities to cease. We also disable SCB fetch operations since we are about to change the qinfifo and any fetch in progress will likely be invalidated. In ahd_qinfifo_count(), fix the qinfifo empty case. In ahd_dump_card_state(), print out CCSCBCTL in the correct mode. If we are a narrow controller, don't set the current width to unknown when forcing a future negotiation. This just confuses the code into attempting a wide negotiation on a narrow bus. Add support for task management function completions. Modify ahd_handle_devreset so that it can handle lun resets in addition to target resets. Use ahd_handle_devreset for lun and target reset task management functions. Handle the abort task TMF race case better. We now wait until any current selections are over and then set the TMF back to zero. This should cause the sequencer to ignore the abort TMF completion should it occur. Correct a bug in the illegal phase handler that caused us to drop down to narrow when handling the unexpected command phase case after 3rd party reset of a packetized device. Indicate the features, bugs, and flags set in the softc that are used to control firmware patch download when booting verbose. aic79xx.h: Add coalessing and HS_MAILBOX fields. Add per-softc variables for the stats "daemon". Add a debug option for interrupt coalessing activities. Add two new softc flags: o AHD_UPDATE_PEND_CMDS Run ahd_reset_cmds_pending() on the next unpause. o AHD_RUNNING_QOUTFIFO Used to catch recursion through ahd_run_qoutfifo(). aic79xx.reg: Correct register addresses related to the software timer and the DFDBCTL register. Add constants paramaterizing the software timer. Add scratch ram locations for storing interrupt coalessing tunables. Break INTMASK in SEQITNCTL out into INTMASK1 and INTMASK2. In at least the REV A, these are writable bits. We make use of that for a swtimer workaround in the sequencer. Since HS_MAILBOX autoclears, provide a sequencer variable to store its contents. Add SEQINT codes for handling task management completions. aic79xx.seq: Correct ignore wide residue processing check for a wide negotiation being in effect. We must be in the SCSI register window in order to access the negotiation table. Use the software timer and a commands completed count to implement interrupt coalessing. The command complete is deferred until either the maximum command threshold or a the expiration of a command deferral timer. If we have more SCBs to complete to the host (sitting in COMPLETE_SCB lists), always try to coaless them up to our coalessing limit. If coalessing is enabled, but we have fewer commands oustanting than the host's min coalessing limit, complete the command immediately. Add code to track the number of commands outstanding. Commands are outstanding from the time they are placed into the execution queue until the DMA to post completion is setup. Add a workaround for intvec_2 interrupts on the H2A4. In H2A4, the mode pointer is not saved for intvec2, but is restored on iret. This can lead to the restoration of a bogus mode ptr. Manually clear the intmask bits and do a normal return to compensate. We use intvec_2 to track interrupt coalessing timeouts. Since we cannot disable the swtimer's countdown, simply mask its interrupt once we no longer care about it firing. In idle_loop_cchan, update LOCAL_HS_MAILBOX everytime we are notified of an HS_MAILBOX update via the HS_MAILBOX_ACT bit in QOFF_CTLSTA. We have to use a local copy of persistant portions of the HS_MAILBOX as the mailbox auto-clears on any read. Move the test for the cfg4istat interrupt up an instruction to hopefully close a race between the next outgoing selection and our disabling of selections. Add a missing ret to the last instruction in load_overrun_buf. Add notifications to the host of task management completions as well as the completions for commands that completed successfully before their corresponding TMF could be sent. Hold a critical section during select-out processing until we have a fully identified connection. This removes a race condition with the legacy abort handler. Correct a few spelling errors in some comments. aic79xx_inline.h: Call ahd_reset_cmds_pending() in ahd_unpause if required. Update cmdcmplt interrupt statistics in our interrupt handler. Allow callers to ahd_send_scb() to set the task management function. aic79xx_pci.c: Disable SERR and pause the controller prior to performing our mmapped I/O test. The U320 controllers do not support "auto-access-pause". aic79xx_osm.c: Set the task management function now that ahd_send_scb() doesn't do it for us. We also perform a lun reset in response to BDR requests to packetized devices.
2003-01-20 20:17:35 +00:00
test QOFF_CTLSTA, HS_MAILBOX_ACT jz hs_mailbox_empty;
or QOFF_CTLSTA, HS_MAILBOX_ACT;
ahc_eisa.c: ahc_pci.c: ahd_pci.c: aic7xxx.c: aic79xx.c: aic_osm_lib.c: aic_osm_lib.h: Use common OSM routines from aic_osm_lib for bus dma operations, delay routines, accessing CCBs, byte swapping, etc. aic7xxx_pci.c: Provide a better description for the 2915/30LP on attach. aic7xxx.c: aic79xx.c: aic7770.c: aic79xx_pci.c: aic7xxx_pci.c: aic7xxx_93cx6.c: Move FBSDID behind an ifdef so that these core files will still compile under other OSes. aic79xx.h: aic79xx_pci.c: aic79xx.seq: To speed up non-packetized CDB delivery in Rev B, all CDB acks are "released" to the output sync as soon as the command phase starts. There is only one problem with this approach. If the target changes phase before all data are sent, we have left over acks that can go out on the bus in a data phase. Due to other chip contraints, this only happens if the target goes to data-in, but if the acks go out before we can test SDONE, we'll think that the transfer has completed successfully. Work around this by taking advantage of the 400ns or 800ns dead time between command phase and the REQ of the new phase. If the transfer has completed successfully, SCSIEN should fall *long* before we see a phase change. We thus treat any phasemiss that occurs before SCSIEN falls as an incomplete transfer. aic79xx.h: Add the AHD_FAST_CDB_DELIVERY feature. aic79xx_pci.c: Set AHD_FAST_CDB_DELIVERY for all Rev. B parts. aic79xx.seq: Test for PHASEMIS in the command phase for all AHD_FAST_CDB_DELIVERY controlelrs. ahd_pci.c: ahc_pci.c: aic7xxx.h: aic79xx.h: Move definition of controller BAR offsets to core header files. aic7xxx.c: aic79xx.c: In the softc free routine, leave removal of a softc from the global list of softcs to the OSM (the caller of this routine). This allows us to avoid holding the softc list_lock during device destruction where we may have to sleep waiting for our recovery thread to halt. ahc_pci.c: Use ahc_pci_test_register access to validate I/O mapped in addition to the tests already performed for memory mapped access. Remove unused ahc_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. ahd_pci.c: Remove reduntant definition of controller BAR offsets. These are also defined in aic79xx.h. Remove unused ahd_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. aic7xxx.c: aic79xx.c: aic79xx.h: aic7xxx.h: aic7xxx_osm.c: aic79xx_osm.c: Move timeout handling to the driver cores. In the case of the aic79xx driver, the algorithm has been enhanced to try target resets before performing a bus reset. For the aic7xxx driver, the algorithm is unchanged. Although the drivers do not currently sleep during recovery (recovery is timeout driven), the cores do expect all processing to be performed via a recovery thread. Our timeout handlers are now little stubs that wakeup the recovery thread. aic79xx.c: aic79xx.h: aic79xx_inline.h: Change shared_data allocation to use a map_node so that the sentinel hscb can use this map node in ahd_swap_with_next_hscb. This routine now swaps the hscb_map pointer in additon to the hscb contents so that any sync operations occur on the correct map. physaddr -> busaddr Pointed out by: Jason Thorpe <thorpej@wasabisystems.com> aic79xx.c: Make more use of the in/out/w/l/q macros for accessing byte registers in the chip. Correct some issues in the ahd_flush_qoutfifo() routine. o Run the qoutfifo only once the command channel DMA engine has been halted. This closes a window where we might have missed some entries. o Change ahd_run_data_fifo() to not loop to completion. If we happen to start on the wrong FIFO and the other FIFO has a snapshot savepointers, we might deadlock. This required our delay between FIFO tests to be moved to the ahd_flush_qoutfifo() routine. o Update/add comments. o Remove spurious test for COMPLETE_DMA list being empty when completing transactions from the GSFIFO with residuals. The SCB must be put on the COMPLETE_DMA scb list unconditionally. o When halting command channel DMA activity, we must disable the DMA channel in all cases but an update of the QOUTFIFO. The latter case is required so that the sequencer will update its position in the QOUTFIFO. Previously, we left the channel enabled for all "push" DMAs. This left us vulnerable to the sequencer handling an SCB push long after that SCB was already processed manually by this routine. o Correct the polarity of tests involving ahd_scb_active_in_fifo(). This routine returns non-zero for true. Return to processing bad status completions through the qoutfifo. This reduces the time that the sequencer is kept paused when handling transactions with bad status or underruns. When waiting for the controller to quiece selections, add a delay to our loop. Otherwise we may fail to wait long enough for the sequencer to comply. On H2A4 hardware, use the slow slewrate for non-paced transfers. This mirrors what the Adaptec Windows drivers do. On the Rev B. only slow down the CRC timing for older U160 devices that might need the slower timing. We define "older" as devices that do not support packetized protocol. Wait up to 5000 * 5us for the SEEPROM to become unbusy. Write ops seem to take much longer than read ops. aic79xx.seq: For controllers with the FAINT_LED bug, turn the diagnostic led feature on during selection and reselection. This covers the non-packetized case. The LED will be disabled for non-packetized transfers once we return to the top level idle loop. Add more comments about the busy LED workaround. Extend a critical section around the entire command channel idle loop process. Previously the portion of this handler that directly manipulated the linked list of completed SCBs was not protected. This is the likely cause of the recent reports of commands being completed twice by the driver. Extend critical sections across the test for, and the longjump to, longjump routines. This prevents the firmware from trying to jump to a longjmp handler that was just cleared by the host. Improve the locations of several critical section begin and end points. Typically these changes remove instructions that did not need to be inside a critical section. Close the "busfree after selection, but before busfree interrupts can be enabled" race to just a single sequencer instruction. We now test the BSY line explicitly before clearing the busfree status and enabling the busfree interrupt. Close a race condition in the processing of HS_MAILBOX updates. We now clear the "updated" status before the copy. This ensures that we don't accidentally clear the status incorrectly when the host sneaks in an update just after our last copy, but before we clear the status. This race has never been observed. Don't re-enable SCSIEN if we lose the race to disable SCSIEN in our interrupt handler's workaround for the RevA data-valid too early issue. aic79xx_inline.h: Add comments indicating that the order in which bytes are read or written in ahd_inw and ahd_outw is important. This allows us to use these inlines when accessing registers with side-effects. aic79xx_pci.c: The 29320 and the 29320B are 7902 not 7901 based products. Correct the driver banner. aic7xxx.h: Enable the use of the auto-access pause feature on the aic7870 and aic7880. It was disabled due to an oversight. aic7xxx.reg: Move TARG_IMMEDIATE_SCB to alias LAST_MSG to avoid leaving garbage in MWI_RESIDUAL. This prevents spurious overflows whn operating target mode on controllers that require the MWI_RESIDUAL work-around. aic7xxx.seq: AHC_TMODE_WIDEODD_BUG is a bug, not a softc flag. Reference the correct softc field when testing for its presence. Set the NOT_IDENTIFIED and NO_CDB_SENT bits in SEQ_FLAGS to indicate that the nexus is invalid in await busfree. aic7xxx_93cx6.c: Add support for the C56/C66 versions of the EWEN and EWDS commands. aic7xxx.c: aic7xxx_pci.c: Move test for the validity of left over BIOS data to ahc_test_register_access(). This guarantees that any left over CHIPRST value is not clobbered by our register access test and lost to the test that was in ahc_reset.
2003-12-17 00:02:10 +00:00
mov LOCAL_HS_MAILBOX, HS_MAILBOX;
aic79xx.c: Implement the SCB_SILENT flag. This is useful for hushing up the driver during DV or other operations that we expect to cause transmission errors. The messages will still print if the SHOW_MASKED_ERRORS debug option is enabled. Save and restore the NEGOADDR address when setting new transfer settings. The sequencer performs lookups in the negotiation table too and it expects NEGOADDR to remain consistent across pause/unpause sessions. Consistently use "offset" instead of "period" to determine if we are running sync or not. Add a SHOW_MESSAGES diagnostic for when we assert ATN during message processing. Print out IU, QAS, and RTI features when showing transfer options. Limit the syncrate after all option conformance changes have taken place in ahd_devlimited_syncrate. Changes in options may change the final syncrate we accept. Keep a copy of the hs_mailbox in our softc so that we can perform read/modify/write operations on the hs_mailbox without having to pause the sequencer to read the last written value. Use the ENINT_COALESS flag in the hs_mailbox to toggle interrupt coalessing. Add entrypoints for enabling interrupt coalessing and setting both a timeout (how long to wait for commands to be coalessed) and a maximum commands to coaless value. Add a statistics timer that decides when to enable or disable interrupt coalessing based on load. Add a routine, ahd_reset_cmds_pending() which is used to update the CMDS_PENDING sequencer variable whenever error recovery compeltes SCBs without notifying the sequencer. Since ahd_reset_cmds_pending is called during ahd_unpause() only if we've aborted SCBs, its call to ahd_flush_qoutfifo should not cause recursion through ahd_run_qoutfifo(). A panic has been added to ensure that this recursion does not occur. In ahd_search_qinfifo, update the CMDS_PENDING sequencer variable directly. ahd_search_qinififo can be called in situations where using ahd_reset_cmds_pending() might cause recursion. Since we can safely determine the exact number to reduce CMDS_PENDING by in this scenario without running the qoutfifo, the manual update is sufficient. Clean up diagnostics. Add ahd_flush_qoutfifo() which will run the qoutfifo as well as complete any commands sitting on the sequencer's COMPLETE_SCB lists or the good status FIFO. Use this routine in several places that did similar things in an add-hoc, but incomplete, fashion. A call to this routine was also added to ahd_abort_scbs() to close a race. In ahd_pause_and_flushwork() only return once selections are safely disabled. Flush all completed commands via ahd_flush_qoutfifo(). Remove "Now packetized" diagnostic now that this information is incorperated into the actual negotiation messages that are displayed. When forcing renegotiation, don't clober the current ppr_options. Much of the driver uses this information to determine if we are currently packetized or not. Remove some stray spaces at column 1 in ahd_set_tags. When complaining about getting a host message loop request with no pending messages, print out the SCB_CONTROL register down on the card. Modify the ahd_sent_msg() routine to handle a search for an outgoing identify message. Use this to detect a msg reject on an identify message which typically indicates that the target thought we were packetized. Force a renegotiation in this case. In ahd_search_qinfifo(), wait more effectively for SCB DMA activities to cease. We also disable SCB fetch operations since we are about to change the qinfifo and any fetch in progress will likely be invalidated. In ahd_qinfifo_count(), fix the qinfifo empty case. In ahd_dump_card_state(), print out CCSCBCTL in the correct mode. If we are a narrow controller, don't set the current width to unknown when forcing a future negotiation. This just confuses the code into attempting a wide negotiation on a narrow bus. Add support for task management function completions. Modify ahd_handle_devreset so that it can handle lun resets in addition to target resets. Use ahd_handle_devreset for lun and target reset task management functions. Handle the abort task TMF race case better. We now wait until any current selections are over and then set the TMF back to zero. This should cause the sequencer to ignore the abort TMF completion should it occur. Correct a bug in the illegal phase handler that caused us to drop down to narrow when handling the unexpected command phase case after 3rd party reset of a packetized device. Indicate the features, bugs, and flags set in the softc that are used to control firmware patch download when booting verbose. aic79xx.h: Add coalessing and HS_MAILBOX fields. Add per-softc variables for the stats "daemon". Add a debug option for interrupt coalessing activities. Add two new softc flags: o AHD_UPDATE_PEND_CMDS Run ahd_reset_cmds_pending() on the next unpause. o AHD_RUNNING_QOUTFIFO Used to catch recursion through ahd_run_qoutfifo(). aic79xx.reg: Correct register addresses related to the software timer and the DFDBCTL register. Add constants paramaterizing the software timer. Add scratch ram locations for storing interrupt coalessing tunables. Break INTMASK in SEQITNCTL out into INTMASK1 and INTMASK2. In at least the REV A, these are writable bits. We make use of that for a swtimer workaround in the sequencer. Since HS_MAILBOX autoclears, provide a sequencer variable to store its contents. Add SEQINT codes for handling task management completions. aic79xx.seq: Correct ignore wide residue processing check for a wide negotiation being in effect. We must be in the SCSI register window in order to access the negotiation table. Use the software timer and a commands completed count to implement interrupt coalessing. The command complete is deferred until either the maximum command threshold or a the expiration of a command deferral timer. If we have more SCBs to complete to the host (sitting in COMPLETE_SCB lists), always try to coaless them up to our coalessing limit. If coalessing is enabled, but we have fewer commands oustanting than the host's min coalessing limit, complete the command immediately. Add code to track the number of commands outstanding. Commands are outstanding from the time they are placed into the execution queue until the DMA to post completion is setup. Add a workaround for intvec_2 interrupts on the H2A4. In H2A4, the mode pointer is not saved for intvec2, but is restored on iret. This can lead to the restoration of a bogus mode ptr. Manually clear the intmask bits and do a normal return to compensate. We use intvec_2 to track interrupt coalessing timeouts. Since we cannot disable the swtimer's countdown, simply mask its interrupt once we no longer care about it firing. In idle_loop_cchan, update LOCAL_HS_MAILBOX everytime we are notified of an HS_MAILBOX update via the HS_MAILBOX_ACT bit in QOFF_CTLSTA. We have to use a local copy of persistant portions of the HS_MAILBOX as the mailbox auto-clears on any read. Move the test for the cfg4istat interrupt up an instruction to hopefully close a race between the next outgoing selection and our disabling of selections. Add a missing ret to the last instruction in load_overrun_buf. Add notifications to the host of task management completions as well as the completions for commands that completed successfully before their corresponding TMF could be sent. Hold a critical section during select-out processing until we have a fully identified connection. This removes a race condition with the legacy abort handler. Correct a few spelling errors in some comments. aic79xx_inline.h: Call ahd_reset_cmds_pending() in ahd_unpause if required. Update cmdcmplt interrupt statistics in our interrupt handler. Allow callers to ahd_send_scb() to set the task management function. aic79xx_pci.c: Disable SERR and pause the controller prior to performing our mmapped I/O test. The U320 controllers do not support "auto-access-pause". aic79xx_osm.c: Set the task management function now that ahd_send_scb() doesn't do it for us. We also perform a lun reset in response to BDR requests to packetized devices.
2003-01-20 20:17:35 +00:00
hs_mailbox_empty:
Upgrade to version 1.1 of the aic79xx U320 driver. aic79xx.c: o Remove redundant ahd_update_modes() call. o Correct panic in diagnostic should state corruption cause the SCB Id to be invalid during a selection timeout. o Add workaround for missing BUSFREEREV feature in Rev A silicon. o Corect formatting nits. o Use register pretty printing in more places. o Save and restore our SCB pointer when updating the waiting queue list for an "expected" LQ-out busfree. o In ahd_clear_intstat, deal with the missing autoclear in the CLRLQO* registers. o BE fixup in a diagnostic printf. o Make sure that we are in the proper mode before disabling selections in ahd_update_pending_scbs. o Add more diagnostics. o task_attribute_nonpkt_tag -> task_attribute: we don't need a nonpkt_tag field anymore for allowing all 512 SCBs to be used in non-packetized connections. o Negotiate HOLD_MCS to U320 devices. o Add a few additional mode assertions. o Restore the chip mode after clearing out the qinfifo so that code using ahd_abort_scbs sees a consistent mode. o Simplify the DMA engine shutdown routine prior to performing a bus reset. o Perform the sequencer restart after a chip reset prior to setting up our timer to poll for the reset to be complete. On some OSes, the timer could actually pre-empt us and order is important here. o Have our "reset poller" set the expected mode since there is no guarantee of what mode will be in force when we are called from the OS timer. o Save and restore the SCB pointer in ahd_dump_card_state(). This routine must not modify card state. o Ditto for ahd_dump_scbs(). aic79xx.h: o Add a few more chip bug definitions. o Align our tag on a 32bit boundary. aic79xx.reg: aic79xx.seq: o Start work on removing workarounds for Rev B. o Use a special location in scratch from for stroring our SCBPTR during legacy FIFO allocations. This corrects problems in mixed packetized/non-packetized configurations where calling into a FIFO task corrupted our SCBPTR. o Don't rely on DMA priority to guarantee that all data in our FIFOs will flush prior to a command completion notification going out of the command channel. We've never seen this assumption fail, but better safe than sorry. o Deal with missing BUSFREEREV feature in H2A. o Simplify disconnect list code now that the list will always have only a single entry. o Implement the AHD_REG_SLOW_SETTLE_BUG workaround. o Swith to using "REG_ISR" for local mode scratch during our ISR. o Add a missing jmp to the data_group_dma_loop after our data pointers have been re-initialized by the kernel. o Correct test in the bitbucket code so that we actually wait for the bitbucket to complete before signaling the kernel of the overrun condition. o Reposition pkt_saveptrs to avoid a jmp instruction. o Update a comment to reflect that the code now waits for a FIFO to drain prior to issuing a CLRCHN. aic79xx_inline.h: o Remove unused untagged queue handling code. o Don't attempt to htole64 what could be a 32bit value. aic79xx_pci.c: o Set additional bug flags for rev A chips.
2002-09-26 22:54:00 +00:00
BEGIN_CRITICAL;
test CCSCBCTL, CCARREN|CCSCBEN jz scbdma_idle;
test CCSCBCTL, CCSCBDIR jnz fetch_new_scb_inprog;
test CCSCBCTL, CCSCBDONE jz return;
/* FALLTHROUGH */
scbdma_tohost_done:
test CCSCBCTL, CCARREN jz fill_qoutfifo_dmadone;
/*
Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-05-04 00:20:07 +00:00
* An SCB has been succesfully uploaded to the host.
* If the SCB was uploaded for some reason other than
* bad SCSI status (currently only for underruns), we
* queue the SCB for normal completion. Otherwise, we
* wait until any select-out activity has halted, and
ahc_eisa.c: ahc_pci.c: ahd_pci.c: aic7xxx.c: aic79xx.c: aic_osm_lib.c: aic_osm_lib.h: Use common OSM routines from aic_osm_lib for bus dma operations, delay routines, accessing CCBs, byte swapping, etc. aic7xxx_pci.c: Provide a better description for the 2915/30LP on attach. aic7xxx.c: aic79xx.c: aic7770.c: aic79xx_pci.c: aic7xxx_pci.c: aic7xxx_93cx6.c: Move FBSDID behind an ifdef so that these core files will still compile under other OSes. aic79xx.h: aic79xx_pci.c: aic79xx.seq: To speed up non-packetized CDB delivery in Rev B, all CDB acks are "released" to the output sync as soon as the command phase starts. There is only one problem with this approach. If the target changes phase before all data are sent, we have left over acks that can go out on the bus in a data phase. Due to other chip contraints, this only happens if the target goes to data-in, but if the acks go out before we can test SDONE, we'll think that the transfer has completed successfully. Work around this by taking advantage of the 400ns or 800ns dead time between command phase and the REQ of the new phase. If the transfer has completed successfully, SCSIEN should fall *long* before we see a phase change. We thus treat any phasemiss that occurs before SCSIEN falls as an incomplete transfer. aic79xx.h: Add the AHD_FAST_CDB_DELIVERY feature. aic79xx_pci.c: Set AHD_FAST_CDB_DELIVERY for all Rev. B parts. aic79xx.seq: Test for PHASEMIS in the command phase for all AHD_FAST_CDB_DELIVERY controlelrs. ahd_pci.c: ahc_pci.c: aic7xxx.h: aic79xx.h: Move definition of controller BAR offsets to core header files. aic7xxx.c: aic79xx.c: In the softc free routine, leave removal of a softc from the global list of softcs to the OSM (the caller of this routine). This allows us to avoid holding the softc list_lock during device destruction where we may have to sleep waiting for our recovery thread to halt. ahc_pci.c: Use ahc_pci_test_register access to validate I/O mapped in addition to the tests already performed for memory mapped access. Remove unused ahc_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. ahd_pci.c: Remove reduntant definition of controller BAR offsets. These are also defined in aic79xx.h. Remove unused ahd_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. aic7xxx.c: aic79xx.c: aic79xx.h: aic7xxx.h: aic7xxx_osm.c: aic79xx_osm.c: Move timeout handling to the driver cores. In the case of the aic79xx driver, the algorithm has been enhanced to try target resets before performing a bus reset. For the aic7xxx driver, the algorithm is unchanged. Although the drivers do not currently sleep during recovery (recovery is timeout driven), the cores do expect all processing to be performed via a recovery thread. Our timeout handlers are now little stubs that wakeup the recovery thread. aic79xx.c: aic79xx.h: aic79xx_inline.h: Change shared_data allocation to use a map_node so that the sentinel hscb can use this map node in ahd_swap_with_next_hscb. This routine now swaps the hscb_map pointer in additon to the hscb contents so that any sync operations occur on the correct map. physaddr -> busaddr Pointed out by: Jason Thorpe <thorpej@wasabisystems.com> aic79xx.c: Make more use of the in/out/w/l/q macros for accessing byte registers in the chip. Correct some issues in the ahd_flush_qoutfifo() routine. o Run the qoutfifo only once the command channel DMA engine has been halted. This closes a window where we might have missed some entries. o Change ahd_run_data_fifo() to not loop to completion. If we happen to start on the wrong FIFO and the other FIFO has a snapshot savepointers, we might deadlock. This required our delay between FIFO tests to be moved to the ahd_flush_qoutfifo() routine. o Update/add comments. o Remove spurious test for COMPLETE_DMA list being empty when completing transactions from the GSFIFO with residuals. The SCB must be put on the COMPLETE_DMA scb list unconditionally. o When halting command channel DMA activity, we must disable the DMA channel in all cases but an update of the QOUTFIFO. The latter case is required so that the sequencer will update its position in the QOUTFIFO. Previously, we left the channel enabled for all "push" DMAs. This left us vulnerable to the sequencer handling an SCB push long after that SCB was already processed manually by this routine. o Correct the polarity of tests involving ahd_scb_active_in_fifo(). This routine returns non-zero for true. Return to processing bad status completions through the qoutfifo. This reduces the time that the sequencer is kept paused when handling transactions with bad status or underruns. When waiting for the controller to quiece selections, add a delay to our loop. Otherwise we may fail to wait long enough for the sequencer to comply. On H2A4 hardware, use the slow slewrate for non-paced transfers. This mirrors what the Adaptec Windows drivers do. On the Rev B. only slow down the CRC timing for older U160 devices that might need the slower timing. We define "older" as devices that do not support packetized protocol. Wait up to 5000 * 5us for the SEEPROM to become unbusy. Write ops seem to take much longer than read ops. aic79xx.seq: For controllers with the FAINT_LED bug, turn the diagnostic led feature on during selection and reselection. This covers the non-packetized case. The LED will be disabled for non-packetized transfers once we return to the top level idle loop. Add more comments about the busy LED workaround. Extend a critical section around the entire command channel idle loop process. Previously the portion of this handler that directly manipulated the linked list of completed SCBs was not protected. This is the likely cause of the recent reports of commands being completed twice by the driver. Extend critical sections across the test for, and the longjump to, longjump routines. This prevents the firmware from trying to jump to a longjmp handler that was just cleared by the host. Improve the locations of several critical section begin and end points. Typically these changes remove instructions that did not need to be inside a critical section. Close the "busfree after selection, but before busfree interrupts can be enabled" race to just a single sequencer instruction. We now test the BSY line explicitly before clearing the busfree status and enabling the busfree interrupt. Close a race condition in the processing of HS_MAILBOX updates. We now clear the "updated" status before the copy. This ensures that we don't accidentally clear the status incorrectly when the host sneaks in an update just after our last copy, but before we clear the status. This race has never been observed. Don't re-enable SCSIEN if we lose the race to disable SCSIEN in our interrupt handler's workaround for the RevA data-valid too early issue. aic79xx_inline.h: Add comments indicating that the order in which bytes are read or written in ahd_inw and ahd_outw is important. This allows us to use these inlines when accessing registers with side-effects. aic79xx_pci.c: The 29320 and the 29320B are 7902 not 7901 based products. Correct the driver banner. aic7xxx.h: Enable the use of the auto-access pause feature on the aic7870 and aic7880. It was disabled due to an oversight. aic7xxx.reg: Move TARG_IMMEDIATE_SCB to alias LAST_MSG to avoid leaving garbage in MWI_RESIDUAL. This prevents spurious overflows whn operating target mode on controllers that require the MWI_RESIDUAL work-around. aic7xxx.seq: AHC_TMODE_WIDEODD_BUG is a bug, not a softc flag. Reference the correct softc field when testing for its presence. Set the NOT_IDENTIFIED and NO_CDB_SENT bits in SEQ_FLAGS to indicate that the nexus is invalid in await busfree. aic7xxx_93cx6.c: Add support for the C56/C66 versions of the EWEN and EWDS commands. aic7xxx.c: aic7xxx_pci.c: Move test for the validity of left over BIOS data to ahc_test_register_access(). This guarantees that any left over CHIPRST value is not clobbered by our register access test and lost to the test that was in ahc_reset.
2003-12-17 00:02:10 +00:00
* then queue the completion.
*/
Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-05-04 00:20:07 +00:00
and CCSCBCTL, ~(CCARREN|CCSCBEN);
bmov COMPLETE_DMA_SCB_HEAD, SCB_NEXT_COMPLETE, 2;
aic79xx.c: aic79xx.seq: Convert the COMPLETE_DMA_SCB list to an "stailq". This allows us to safely keep the SCB that is currently being DMA'ed back the host on the head of the list while processing completions off of the bus. The newly completed SCBs are appended to the tail of the queue. In the past, we just dequeued the SCB that was in flight from the list, but this could result in a lost completion should the host perform certain types of error recovery that must cancel all in-flight SCB DMA operations. Switch from using a 16bit completion entry, holding just the tag and the completion valid bit, to a 64bit completion entry that also contains a "status packet valid" indicator. This solves two problems: o The SCB DMA engine on at least Rev B. silicon does not properly deal with a PCI disconnect that occurs at a non-64bit aligned offset in the chips "source buffer". When the transfer is resumed, the DMA engine continues at the correct offset, but may wrap to the head of the buffer causing duplicate completions to be reported to the host. By using a completion buffer in host memory that is 64bit aligned and using 64bit completion entries, such disconnects should only occur at aligned addresses. This assumes that the host bridge will only disconnect on cache-line boundaries and that cache-lines are multpiles of 64bits. o By embedding the status information in the completion entry we can avoid an extra memory reference to the HSCB for commands that complete without error. Use the comparison of a "host freeze count" and a "sequencer freeze count" to allow the host to process most SCBs that complete with non-zero status without having to clear critical sections. Instead the host can just pause the sequencer, performs any necessary cleanup in the waiting for selection list, increments its freeze count on the controller, and unpauses. This is only possible because the sequencer defers completions of SCBs with bad status until after all pending selections have completed. The sequencer then avoids referencing any data structures the host may touch during completion of the SCB until the freeze counts match. aic79xx.c: Change the strategy for allocating our sentinal HSCB for the QINFIFO. In the past, this allocation was tacked onto the QOUTFIFO allocation. Now that the qoutfifo has grown to accomodate larger completion entries, the old approach will result in a 64byte allocation that costs an extra page of coherent memory. We now do this extra allocation via ahd_alloc_scbs() where the "unused space" can be used to allocate "normal" HSCBs. In our packetized busfree handler, use the ENSELO bit to differentiate between packetized and non-packetized unexpected busfree events that occur just after selection, but before the sequencer has had the oportunity to service the selection. When cleaning out the waiting for selection list, use the SCSI mode instead of the command channel mode. The SCB pointer in the command channel mode may be referenced by the SCB dma engine even while the sequencer is paused, whereas the SCSI mode SCB pointer is only accessed by the sequencer. Print the "complete on qfreeze" sequencer SCB completion list in ahd_dump_card_state(). This list holds all SCB completions that are deferred until a pending select-out qfreeze event has taken effect. aic79xx.h: Add definitions and structures to handle the new SCB completion scheme. Add a controller flag that indicates if the controller is in HostRAID mode. aic79xx.reg: Remove macros used for toggling from one data fifo mode to the other. They have not been in use for some time. Add scratch ram fields for our new qfreeze count scheme, converting the complete dma list into an "stailq", and providing for the "complete on qfreeze" SCB completion list. Some other fields were moved to retain proper field alignment (alignment >= field size in bytes). aic79xx.seq: Add code to our idle loop to: o Process deferred completions once a qfreeze event has taken full effect. o Thaw the queue once the sequencer and host qfreeze counts match. Generate 64bit completion entries passing the SCB_SGPTR field as the "good status" indicator. The first bit in this field is only set if we have a valid status packet to send to the host. Convert the COMPLETE_DMA_SCB list to an "stailq". When using "setjmp" to register an idle loop handler, do not combine the "ret" with the block move to pop the stack address in the same instruction. At least on the A, this results in a return to the setjmp caller, not to the new address at the top of the stack. Since we want the latter (we want the newly registered handler to only be invoked from the idle loop), we must use a separate ret instruction. Add a few missing critical sections. Close a race condition that can occur on Rev A. silicon. If both FIFOs happen to be allocated before the sequencer has a chance to service the FIFO that was allocated first, we must take special care to service the FIFO that is not active on the SCSI bus first. This guarantees that a FIFO will be freed to handle any snapshot requests for the FIFO that is still on the bus. Chosing the incorrect FIFO will result in deadlock. Update comments. aic79xx_inline.h Correct the offset calculation for the syncing of our qoutfifo. Update ahd_check_cmdcmpltqueues() for the larger completion entries. aic79xx_pci.c: Attach to HostRAID controllers by default. In the future I may add a sysctl to modify the behavior, but since FreeBSD does not have any HostRAID drivers, failing to attach just results in more email and bug reports for the author. MFC After: 1week
2004-02-04 16:38:38 +00:00
cmp SCB_NEXT_COMPLETE[1], SCB_LIST_NULL jne . + 2;
mvi COMPLETE_DMA_SCB_TAIL[1], SCB_LIST_NULL;
test SCB_SCSI_STATUS, 0xff jz scbdma_queue_completion;
bmov SCB_NEXT_COMPLETE, COMPLETE_ON_QFREEZE_HEAD, 2;
bmov COMPLETE_ON_QFREEZE_HEAD, SCBPTR, 2 ret;
scbdma_queue_completion:
ahc_eisa.c: ahc_pci.c: ahd_pci.c: aic7xxx.c: aic79xx.c: aic_osm_lib.c: aic_osm_lib.h: Use common OSM routines from aic_osm_lib for bus dma operations, delay routines, accessing CCBs, byte swapping, etc. aic7xxx_pci.c: Provide a better description for the 2915/30LP on attach. aic7xxx.c: aic79xx.c: aic7770.c: aic79xx_pci.c: aic7xxx_pci.c: aic7xxx_93cx6.c: Move FBSDID behind an ifdef so that these core files will still compile under other OSes. aic79xx.h: aic79xx_pci.c: aic79xx.seq: To speed up non-packetized CDB delivery in Rev B, all CDB acks are "released" to the output sync as soon as the command phase starts. There is only one problem with this approach. If the target changes phase before all data are sent, we have left over acks that can go out on the bus in a data phase. Due to other chip contraints, this only happens if the target goes to data-in, but if the acks go out before we can test SDONE, we'll think that the transfer has completed successfully. Work around this by taking advantage of the 400ns or 800ns dead time between command phase and the REQ of the new phase. If the transfer has completed successfully, SCSIEN should fall *long* before we see a phase change. We thus treat any phasemiss that occurs before SCSIEN falls as an incomplete transfer. aic79xx.h: Add the AHD_FAST_CDB_DELIVERY feature. aic79xx_pci.c: Set AHD_FAST_CDB_DELIVERY for all Rev. B parts. aic79xx.seq: Test for PHASEMIS in the command phase for all AHD_FAST_CDB_DELIVERY controlelrs. ahd_pci.c: ahc_pci.c: aic7xxx.h: aic79xx.h: Move definition of controller BAR offsets to core header files. aic7xxx.c: aic79xx.c: In the softc free routine, leave removal of a softc from the global list of softcs to the OSM (the caller of this routine). This allows us to avoid holding the softc list_lock during device destruction where we may have to sleep waiting for our recovery thread to halt. ahc_pci.c: Use ahc_pci_test_register access to validate I/O mapped in addition to the tests already performed for memory mapped access. Remove unused ahc_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. ahd_pci.c: Remove reduntant definition of controller BAR offsets. These are also defined in aic79xx.h. Remove unused ahd_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. aic7xxx.c: aic79xx.c: aic79xx.h: aic7xxx.h: aic7xxx_osm.c: aic79xx_osm.c: Move timeout handling to the driver cores. In the case of the aic79xx driver, the algorithm has been enhanced to try target resets before performing a bus reset. For the aic7xxx driver, the algorithm is unchanged. Although the drivers do not currently sleep during recovery (recovery is timeout driven), the cores do expect all processing to be performed via a recovery thread. Our timeout handlers are now little stubs that wakeup the recovery thread. aic79xx.c: aic79xx.h: aic79xx_inline.h: Change shared_data allocation to use a map_node so that the sentinel hscb can use this map node in ahd_swap_with_next_hscb. This routine now swaps the hscb_map pointer in additon to the hscb contents so that any sync operations occur on the correct map. physaddr -> busaddr Pointed out by: Jason Thorpe <thorpej@wasabisystems.com> aic79xx.c: Make more use of the in/out/w/l/q macros for accessing byte registers in the chip. Correct some issues in the ahd_flush_qoutfifo() routine. o Run the qoutfifo only once the command channel DMA engine has been halted. This closes a window where we might have missed some entries. o Change ahd_run_data_fifo() to not loop to completion. If we happen to start on the wrong FIFO and the other FIFO has a snapshot savepointers, we might deadlock. This required our delay between FIFO tests to be moved to the ahd_flush_qoutfifo() routine. o Update/add comments. o Remove spurious test for COMPLETE_DMA list being empty when completing transactions from the GSFIFO with residuals. The SCB must be put on the COMPLETE_DMA scb list unconditionally. o When halting command channel DMA activity, we must disable the DMA channel in all cases but an update of the QOUTFIFO. The latter case is required so that the sequencer will update its position in the QOUTFIFO. Previously, we left the channel enabled for all "push" DMAs. This left us vulnerable to the sequencer handling an SCB push long after that SCB was already processed manually by this routine. o Correct the polarity of tests involving ahd_scb_active_in_fifo(). This routine returns non-zero for true. Return to processing bad status completions through the qoutfifo. This reduces the time that the sequencer is kept paused when handling transactions with bad status or underruns. When waiting for the controller to quiece selections, add a delay to our loop. Otherwise we may fail to wait long enough for the sequencer to comply. On H2A4 hardware, use the slow slewrate for non-paced transfers. This mirrors what the Adaptec Windows drivers do. On the Rev B. only slow down the CRC timing for older U160 devices that might need the slower timing. We define "older" as devices that do not support packetized protocol. Wait up to 5000 * 5us for the SEEPROM to become unbusy. Write ops seem to take much longer than read ops. aic79xx.seq: For controllers with the FAINT_LED bug, turn the diagnostic led feature on during selection and reselection. This covers the non-packetized case. The LED will be disabled for non-packetized transfers once we return to the top level idle loop. Add more comments about the busy LED workaround. Extend a critical section around the entire command channel idle loop process. Previously the portion of this handler that directly manipulated the linked list of completed SCBs was not protected. This is the likely cause of the recent reports of commands being completed twice by the driver. Extend critical sections across the test for, and the longjump to, longjump routines. This prevents the firmware from trying to jump to a longjmp handler that was just cleared by the host. Improve the locations of several critical section begin and end points. Typically these changes remove instructions that did not need to be inside a critical section. Close the "busfree after selection, but before busfree interrupts can be enabled" race to just a single sequencer instruction. We now test the BSY line explicitly before clearing the busfree status and enabling the busfree interrupt. Close a race condition in the processing of HS_MAILBOX updates. We now clear the "updated" status before the copy. This ensures that we don't accidentally clear the status incorrectly when the host sneaks in an update just after our last copy, but before we clear the status. This race has never been observed. Don't re-enable SCSIEN if we lose the race to disable SCSIEN in our interrupt handler's workaround for the RevA data-valid too early issue. aic79xx_inline.h: Add comments indicating that the order in which bytes are read or written in ahd_inw and ahd_outw is important. This allows us to use these inlines when accessing registers with side-effects. aic79xx_pci.c: The 29320 and the 29320B are 7902 not 7901 based products. Correct the driver banner. aic7xxx.h: Enable the use of the auto-access pause feature on the aic7870 and aic7880. It was disabled due to an oversight. aic7xxx.reg: Move TARG_IMMEDIATE_SCB to alias LAST_MSG to avoid leaving garbage in MWI_RESIDUAL. This prevents spurious overflows whn operating target mode on controllers that require the MWI_RESIDUAL work-around. aic7xxx.seq: AHC_TMODE_WIDEODD_BUG is a bug, not a softc flag. Reference the correct softc field when testing for its presence. Set the NOT_IDENTIFIED and NO_CDB_SENT bits in SEQ_FLAGS to indicate that the nexus is invalid in await busfree. aic7xxx_93cx6.c: Add support for the C56/C66 versions of the EWEN and EWDS commands. aic7xxx.c: aic7xxx_pci.c: Move test for the validity of left over BIOS data to ahc_test_register_access(). This guarantees that any left over CHIPRST value is not clobbered by our register access test and lost to the test that was in ahc_reset.
2003-12-17 00:02:10 +00:00
bmov SCB_NEXT_COMPLETE, COMPLETE_SCB_HEAD, 2;
bmov COMPLETE_SCB_HEAD, SCBPTR, 2 ret;
fill_qoutfifo_dmadone:
and CCSCBCTL, ~(CCARREN|CCSCBEN);
aic79xx.c: Implement the SCB_SILENT flag. This is useful for hushing up the driver during DV or other operations that we expect to cause transmission errors. The messages will still print if the SHOW_MASKED_ERRORS debug option is enabled. Save and restore the NEGOADDR address when setting new transfer settings. The sequencer performs lookups in the negotiation table too and it expects NEGOADDR to remain consistent across pause/unpause sessions. Consistently use "offset" instead of "period" to determine if we are running sync or not. Add a SHOW_MESSAGES diagnostic for when we assert ATN during message processing. Print out IU, QAS, and RTI features when showing transfer options. Limit the syncrate after all option conformance changes have taken place in ahd_devlimited_syncrate. Changes in options may change the final syncrate we accept. Keep a copy of the hs_mailbox in our softc so that we can perform read/modify/write operations on the hs_mailbox without having to pause the sequencer to read the last written value. Use the ENINT_COALESS flag in the hs_mailbox to toggle interrupt coalessing. Add entrypoints for enabling interrupt coalessing and setting both a timeout (how long to wait for commands to be coalessed) and a maximum commands to coaless value. Add a statistics timer that decides when to enable or disable interrupt coalessing based on load. Add a routine, ahd_reset_cmds_pending() which is used to update the CMDS_PENDING sequencer variable whenever error recovery compeltes SCBs without notifying the sequencer. Since ahd_reset_cmds_pending is called during ahd_unpause() only if we've aborted SCBs, its call to ahd_flush_qoutfifo should not cause recursion through ahd_run_qoutfifo(). A panic has been added to ensure that this recursion does not occur. In ahd_search_qinfifo, update the CMDS_PENDING sequencer variable directly. ahd_search_qinififo can be called in situations where using ahd_reset_cmds_pending() might cause recursion. Since we can safely determine the exact number to reduce CMDS_PENDING by in this scenario without running the qoutfifo, the manual update is sufficient. Clean up diagnostics. Add ahd_flush_qoutfifo() which will run the qoutfifo as well as complete any commands sitting on the sequencer's COMPLETE_SCB lists or the good status FIFO. Use this routine in several places that did similar things in an add-hoc, but incomplete, fashion. A call to this routine was also added to ahd_abort_scbs() to close a race. In ahd_pause_and_flushwork() only return once selections are safely disabled. Flush all completed commands via ahd_flush_qoutfifo(). Remove "Now packetized" diagnostic now that this information is incorperated into the actual negotiation messages that are displayed. When forcing renegotiation, don't clober the current ppr_options. Much of the driver uses this information to determine if we are currently packetized or not. Remove some stray spaces at column 1 in ahd_set_tags. When complaining about getting a host message loop request with no pending messages, print out the SCB_CONTROL register down on the card. Modify the ahd_sent_msg() routine to handle a search for an outgoing identify message. Use this to detect a msg reject on an identify message which typically indicates that the target thought we were packetized. Force a renegotiation in this case. In ahd_search_qinfifo(), wait more effectively for SCB DMA activities to cease. We also disable SCB fetch operations since we are about to change the qinfifo and any fetch in progress will likely be invalidated. In ahd_qinfifo_count(), fix the qinfifo empty case. In ahd_dump_card_state(), print out CCSCBCTL in the correct mode. If we are a narrow controller, don't set the current width to unknown when forcing a future negotiation. This just confuses the code into attempting a wide negotiation on a narrow bus. Add support for task management function completions. Modify ahd_handle_devreset so that it can handle lun resets in addition to target resets. Use ahd_handle_devreset for lun and target reset task management functions. Handle the abort task TMF race case better. We now wait until any current selections are over and then set the TMF back to zero. This should cause the sequencer to ignore the abort TMF completion should it occur. Correct a bug in the illegal phase handler that caused us to drop down to narrow when handling the unexpected command phase case after 3rd party reset of a packetized device. Indicate the features, bugs, and flags set in the softc that are used to control firmware patch download when booting verbose. aic79xx.h: Add coalessing and HS_MAILBOX fields. Add per-softc variables for the stats "daemon". Add a debug option for interrupt coalessing activities. Add two new softc flags: o AHD_UPDATE_PEND_CMDS Run ahd_reset_cmds_pending() on the next unpause. o AHD_RUNNING_QOUTFIFO Used to catch recursion through ahd_run_qoutfifo(). aic79xx.reg: Correct register addresses related to the software timer and the DFDBCTL register. Add constants paramaterizing the software timer. Add scratch ram locations for storing interrupt coalessing tunables. Break INTMASK in SEQITNCTL out into INTMASK1 and INTMASK2. In at least the REV A, these are writable bits. We make use of that for a swtimer workaround in the sequencer. Since HS_MAILBOX autoclears, provide a sequencer variable to store its contents. Add SEQINT codes for handling task management completions. aic79xx.seq: Correct ignore wide residue processing check for a wide negotiation being in effect. We must be in the SCSI register window in order to access the negotiation table. Use the software timer and a commands completed count to implement interrupt coalessing. The command complete is deferred until either the maximum command threshold or a the expiration of a command deferral timer. If we have more SCBs to complete to the host (sitting in COMPLETE_SCB lists), always try to coaless them up to our coalessing limit. If coalessing is enabled, but we have fewer commands oustanting than the host's min coalessing limit, complete the command immediately. Add code to track the number of commands outstanding. Commands are outstanding from the time they are placed into the execution queue until the DMA to post completion is setup. Add a workaround for intvec_2 interrupts on the H2A4. In H2A4, the mode pointer is not saved for intvec2, but is restored on iret. This can lead to the restoration of a bogus mode ptr. Manually clear the intmask bits and do a normal return to compensate. We use intvec_2 to track interrupt coalessing timeouts. Since we cannot disable the swtimer's countdown, simply mask its interrupt once we no longer care about it firing. In idle_loop_cchan, update LOCAL_HS_MAILBOX everytime we are notified of an HS_MAILBOX update via the HS_MAILBOX_ACT bit in QOFF_CTLSTA. We have to use a local copy of persistant portions of the HS_MAILBOX as the mailbox auto-clears on any read. Move the test for the cfg4istat interrupt up an instruction to hopefully close a race between the next outgoing selection and our disabling of selections. Add a missing ret to the last instruction in load_overrun_buf. Add notifications to the host of task management completions as well as the completions for commands that completed successfully before their corresponding TMF could be sent. Hold a critical section during select-out processing until we have a fully identified connection. This removes a race condition with the legacy abort handler. Correct a few spelling errors in some comments. aic79xx_inline.h: Call ahd_reset_cmds_pending() in ahd_unpause if required. Update cmdcmplt interrupt statistics in our interrupt handler. Allow callers to ahd_send_scb() to set the task management function. aic79xx_pci.c: Disable SERR and pause the controller prior to performing our mmapped I/O test. The U320 controllers do not support "auto-access-pause". aic79xx_osm.c: Set the task management function now that ahd_send_scb() doesn't do it for us. We also perform a lun reset in response to BDR requests to packetized devices.
2003-01-20 20:17:35 +00:00
call qoutfifo_updated;
mvi COMPLETE_SCB_DMAINPROG_HEAD[1], SCB_LIST_NULL;
bmov QOUTFIFO_NEXT_ADDR, SCBHADDR, 4;
test QOFF_CTLSTA, SDSCB_ROLLOVR jz return;
bmov QOUTFIFO_NEXT_ADDR, SHARED_DATA_ADDR, 4;
xor QOUTFIFO_ENTRY_VALID_TAG, QOUTFIFO_ENTRY_VALID_TOGGLE ret;
ahc_eisa.c: ahc_pci.c: ahd_pci.c: aic7xxx.c: aic79xx.c: aic_osm_lib.c: aic_osm_lib.h: Use common OSM routines from aic_osm_lib for bus dma operations, delay routines, accessing CCBs, byte swapping, etc. aic7xxx_pci.c: Provide a better description for the 2915/30LP on attach. aic7xxx.c: aic79xx.c: aic7770.c: aic79xx_pci.c: aic7xxx_pci.c: aic7xxx_93cx6.c: Move FBSDID behind an ifdef so that these core files will still compile under other OSes. aic79xx.h: aic79xx_pci.c: aic79xx.seq: To speed up non-packetized CDB delivery in Rev B, all CDB acks are "released" to the output sync as soon as the command phase starts. There is only one problem with this approach. If the target changes phase before all data are sent, we have left over acks that can go out on the bus in a data phase. Due to other chip contraints, this only happens if the target goes to data-in, but if the acks go out before we can test SDONE, we'll think that the transfer has completed successfully. Work around this by taking advantage of the 400ns or 800ns dead time between command phase and the REQ of the new phase. If the transfer has completed successfully, SCSIEN should fall *long* before we see a phase change. We thus treat any phasemiss that occurs before SCSIEN falls as an incomplete transfer. aic79xx.h: Add the AHD_FAST_CDB_DELIVERY feature. aic79xx_pci.c: Set AHD_FAST_CDB_DELIVERY for all Rev. B parts. aic79xx.seq: Test for PHASEMIS in the command phase for all AHD_FAST_CDB_DELIVERY controlelrs. ahd_pci.c: ahc_pci.c: aic7xxx.h: aic79xx.h: Move definition of controller BAR offsets to core header files. aic7xxx.c: aic79xx.c: In the softc free routine, leave removal of a softc from the global list of softcs to the OSM (the caller of this routine). This allows us to avoid holding the softc list_lock during device destruction where we may have to sleep waiting for our recovery thread to halt. ahc_pci.c: Use ahc_pci_test_register access to validate I/O mapped in addition to the tests already performed for memory mapped access. Remove unused ahc_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. ahd_pci.c: Remove reduntant definition of controller BAR offsets. These are also defined in aic79xx.h. Remove unused ahd_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. aic7xxx.c: aic79xx.c: aic79xx.h: aic7xxx.h: aic7xxx_osm.c: aic79xx_osm.c: Move timeout handling to the driver cores. In the case of the aic79xx driver, the algorithm has been enhanced to try target resets before performing a bus reset. For the aic7xxx driver, the algorithm is unchanged. Although the drivers do not currently sleep during recovery (recovery is timeout driven), the cores do expect all processing to be performed via a recovery thread. Our timeout handlers are now little stubs that wakeup the recovery thread. aic79xx.c: aic79xx.h: aic79xx_inline.h: Change shared_data allocation to use a map_node so that the sentinel hscb can use this map node in ahd_swap_with_next_hscb. This routine now swaps the hscb_map pointer in additon to the hscb contents so that any sync operations occur on the correct map. physaddr -> busaddr Pointed out by: Jason Thorpe <thorpej@wasabisystems.com> aic79xx.c: Make more use of the in/out/w/l/q macros for accessing byte registers in the chip. Correct some issues in the ahd_flush_qoutfifo() routine. o Run the qoutfifo only once the command channel DMA engine has been halted. This closes a window where we might have missed some entries. o Change ahd_run_data_fifo() to not loop to completion. If we happen to start on the wrong FIFO and the other FIFO has a snapshot savepointers, we might deadlock. This required our delay between FIFO tests to be moved to the ahd_flush_qoutfifo() routine. o Update/add comments. o Remove spurious test for COMPLETE_DMA list being empty when completing transactions from the GSFIFO with residuals. The SCB must be put on the COMPLETE_DMA scb list unconditionally. o When halting command channel DMA activity, we must disable the DMA channel in all cases but an update of the QOUTFIFO. The latter case is required so that the sequencer will update its position in the QOUTFIFO. Previously, we left the channel enabled for all "push" DMAs. This left us vulnerable to the sequencer handling an SCB push long after that SCB was already processed manually by this routine. o Correct the polarity of tests involving ahd_scb_active_in_fifo(). This routine returns non-zero for true. Return to processing bad status completions through the qoutfifo. This reduces the time that the sequencer is kept paused when handling transactions with bad status or underruns. When waiting for the controller to quiece selections, add a delay to our loop. Otherwise we may fail to wait long enough for the sequencer to comply. On H2A4 hardware, use the slow slewrate for non-paced transfers. This mirrors what the Adaptec Windows drivers do. On the Rev B. only slow down the CRC timing for older U160 devices that might need the slower timing. We define "older" as devices that do not support packetized protocol. Wait up to 5000 * 5us for the SEEPROM to become unbusy. Write ops seem to take much longer than read ops. aic79xx.seq: For controllers with the FAINT_LED bug, turn the diagnostic led feature on during selection and reselection. This covers the non-packetized case. The LED will be disabled for non-packetized transfers once we return to the top level idle loop. Add more comments about the busy LED workaround. Extend a critical section around the entire command channel idle loop process. Previously the portion of this handler that directly manipulated the linked list of completed SCBs was not protected. This is the likely cause of the recent reports of commands being completed twice by the driver. Extend critical sections across the test for, and the longjump to, longjump routines. This prevents the firmware from trying to jump to a longjmp handler that was just cleared by the host. Improve the locations of several critical section begin and end points. Typically these changes remove instructions that did not need to be inside a critical section. Close the "busfree after selection, but before busfree interrupts can be enabled" race to just a single sequencer instruction. We now test the BSY line explicitly before clearing the busfree status and enabling the busfree interrupt. Close a race condition in the processing of HS_MAILBOX updates. We now clear the "updated" status before the copy. This ensures that we don't accidentally clear the status incorrectly when the host sneaks in an update just after our last copy, but before we clear the status. This race has never been observed. Don't re-enable SCSIEN if we lose the race to disable SCSIEN in our interrupt handler's workaround for the RevA data-valid too early issue. aic79xx_inline.h: Add comments indicating that the order in which bytes are read or written in ahd_inw and ahd_outw is important. This allows us to use these inlines when accessing registers with side-effects. aic79xx_pci.c: The 29320 and the 29320B are 7902 not 7901 based products. Correct the driver banner. aic7xxx.h: Enable the use of the auto-access pause feature on the aic7870 and aic7880. It was disabled due to an oversight. aic7xxx.reg: Move TARG_IMMEDIATE_SCB to alias LAST_MSG to avoid leaving garbage in MWI_RESIDUAL. This prevents spurious overflows whn operating target mode on controllers that require the MWI_RESIDUAL work-around. aic7xxx.seq: AHC_TMODE_WIDEODD_BUG is a bug, not a softc flag. Reference the correct softc field when testing for its presence. Set the NOT_IDENTIFIED and NO_CDB_SENT bits in SEQ_FLAGS to indicate that the nexus is invalid in await busfree. aic7xxx_93cx6.c: Add support for the C56/C66 versions of the EWEN and EWDS commands. aic7xxx.c: aic7xxx_pci.c: Move test for the validity of left over BIOS data to ahc_test_register_access(). This guarantees that any left over CHIPRST value is not clobbered by our register access test and lost to the test that was in ahc_reset.
2003-12-17 00:02:10 +00:00
END_CRITICAL;
aic79xx.c: Implement the SCB_SILENT flag. This is useful for hushing up the driver during DV or other operations that we expect to cause transmission errors. The messages will still print if the SHOW_MASKED_ERRORS debug option is enabled. Save and restore the NEGOADDR address when setting new transfer settings. The sequencer performs lookups in the negotiation table too and it expects NEGOADDR to remain consistent across pause/unpause sessions. Consistently use "offset" instead of "period" to determine if we are running sync or not. Add a SHOW_MESSAGES diagnostic for when we assert ATN during message processing. Print out IU, QAS, and RTI features when showing transfer options. Limit the syncrate after all option conformance changes have taken place in ahd_devlimited_syncrate. Changes in options may change the final syncrate we accept. Keep a copy of the hs_mailbox in our softc so that we can perform read/modify/write operations on the hs_mailbox without having to pause the sequencer to read the last written value. Use the ENINT_COALESS flag in the hs_mailbox to toggle interrupt coalessing. Add entrypoints for enabling interrupt coalessing and setting both a timeout (how long to wait for commands to be coalessed) and a maximum commands to coaless value. Add a statistics timer that decides when to enable or disable interrupt coalessing based on load. Add a routine, ahd_reset_cmds_pending() which is used to update the CMDS_PENDING sequencer variable whenever error recovery compeltes SCBs without notifying the sequencer. Since ahd_reset_cmds_pending is called during ahd_unpause() only if we've aborted SCBs, its call to ahd_flush_qoutfifo should not cause recursion through ahd_run_qoutfifo(). A panic has been added to ensure that this recursion does not occur. In ahd_search_qinfifo, update the CMDS_PENDING sequencer variable directly. ahd_search_qinififo can be called in situations where using ahd_reset_cmds_pending() might cause recursion. Since we can safely determine the exact number to reduce CMDS_PENDING by in this scenario without running the qoutfifo, the manual update is sufficient. Clean up diagnostics. Add ahd_flush_qoutfifo() which will run the qoutfifo as well as complete any commands sitting on the sequencer's COMPLETE_SCB lists or the good status FIFO. Use this routine in several places that did similar things in an add-hoc, but incomplete, fashion. A call to this routine was also added to ahd_abort_scbs() to close a race. In ahd_pause_and_flushwork() only return once selections are safely disabled. Flush all completed commands via ahd_flush_qoutfifo(). Remove "Now packetized" diagnostic now that this information is incorperated into the actual negotiation messages that are displayed. When forcing renegotiation, don't clober the current ppr_options. Much of the driver uses this information to determine if we are currently packetized or not. Remove some stray spaces at column 1 in ahd_set_tags. When complaining about getting a host message loop request with no pending messages, print out the SCB_CONTROL register down on the card. Modify the ahd_sent_msg() routine to handle a search for an outgoing identify message. Use this to detect a msg reject on an identify message which typically indicates that the target thought we were packetized. Force a renegotiation in this case. In ahd_search_qinfifo(), wait more effectively for SCB DMA activities to cease. We also disable SCB fetch operations since we are about to change the qinfifo and any fetch in progress will likely be invalidated. In ahd_qinfifo_count(), fix the qinfifo empty case. In ahd_dump_card_state(), print out CCSCBCTL in the correct mode. If we are a narrow controller, don't set the current width to unknown when forcing a future negotiation. This just confuses the code into attempting a wide negotiation on a narrow bus. Add support for task management function completions. Modify ahd_handle_devreset so that it can handle lun resets in addition to target resets. Use ahd_handle_devreset for lun and target reset task management functions. Handle the abort task TMF race case better. We now wait until any current selections are over and then set the TMF back to zero. This should cause the sequencer to ignore the abort TMF completion should it occur. Correct a bug in the illegal phase handler that caused us to drop down to narrow when handling the unexpected command phase case after 3rd party reset of a packetized device. Indicate the features, bugs, and flags set in the softc that are used to control firmware patch download when booting verbose. aic79xx.h: Add coalessing and HS_MAILBOX fields. Add per-softc variables for the stats "daemon". Add a debug option for interrupt coalessing activities. Add two new softc flags: o AHD_UPDATE_PEND_CMDS Run ahd_reset_cmds_pending() on the next unpause. o AHD_RUNNING_QOUTFIFO Used to catch recursion through ahd_run_qoutfifo(). aic79xx.reg: Correct register addresses related to the software timer and the DFDBCTL register. Add constants paramaterizing the software timer. Add scratch ram locations for storing interrupt coalessing tunables. Break INTMASK in SEQITNCTL out into INTMASK1 and INTMASK2. In at least the REV A, these are writable bits. We make use of that for a swtimer workaround in the sequencer. Since HS_MAILBOX autoclears, provide a sequencer variable to store its contents. Add SEQINT codes for handling task management completions. aic79xx.seq: Correct ignore wide residue processing check for a wide negotiation being in effect. We must be in the SCSI register window in order to access the negotiation table. Use the software timer and a commands completed count to implement interrupt coalessing. The command complete is deferred until either the maximum command threshold or a the expiration of a command deferral timer. If we have more SCBs to complete to the host (sitting in COMPLETE_SCB lists), always try to coaless them up to our coalessing limit. If coalessing is enabled, but we have fewer commands oustanting than the host's min coalessing limit, complete the command immediately. Add code to track the number of commands outstanding. Commands are outstanding from the time they are placed into the execution queue until the DMA to post completion is setup. Add a workaround for intvec_2 interrupts on the H2A4. In H2A4, the mode pointer is not saved for intvec2, but is restored on iret. This can lead to the restoration of a bogus mode ptr. Manually clear the intmask bits and do a normal return to compensate. We use intvec_2 to track interrupt coalessing timeouts. Since we cannot disable the swtimer's countdown, simply mask its interrupt once we no longer care about it firing. In idle_loop_cchan, update LOCAL_HS_MAILBOX everytime we are notified of an HS_MAILBOX update via the HS_MAILBOX_ACT bit in QOFF_CTLSTA. We have to use a local copy of persistant portions of the HS_MAILBOX as the mailbox auto-clears on any read. Move the test for the cfg4istat interrupt up an instruction to hopefully close a race between the next outgoing selection and our disabling of selections. Add a missing ret to the last instruction in load_overrun_buf. Add notifications to the host of task management completions as well as the completions for commands that completed successfully before their corresponding TMF could be sent. Hold a critical section during select-out processing until we have a fully identified connection. This removes a race condition with the legacy abort handler. Correct a few spelling errors in some comments. aic79xx_inline.h: Call ahd_reset_cmds_pending() in ahd_unpause if required. Update cmdcmplt interrupt statistics in our interrupt handler. Allow callers to ahd_send_scb() to set the task management function. aic79xx_pci.c: Disable SERR and pause the controller prior to performing our mmapped I/O test. The U320 controllers do not support "auto-access-pause". aic79xx_osm.c: Set the task management function now that ahd_send_scb() doesn't do it for us. We also perform a lun reset in response to BDR requests to packetized devices.
2003-01-20 20:17:35 +00:00
qoutfifo_updated:
/*
* If there are more commands waiting to be dma'ed
* to the host, always coalesce. Otherwise honor the
aic79xx.c: Implement the SCB_SILENT flag. This is useful for hushing up the driver during DV or other operations that we expect to cause transmission errors. The messages will still print if the SHOW_MASKED_ERRORS debug option is enabled. Save and restore the NEGOADDR address when setting new transfer settings. The sequencer performs lookups in the negotiation table too and it expects NEGOADDR to remain consistent across pause/unpause sessions. Consistently use "offset" instead of "period" to determine if we are running sync or not. Add a SHOW_MESSAGES diagnostic for when we assert ATN during message processing. Print out IU, QAS, and RTI features when showing transfer options. Limit the syncrate after all option conformance changes have taken place in ahd_devlimited_syncrate. Changes in options may change the final syncrate we accept. Keep a copy of the hs_mailbox in our softc so that we can perform read/modify/write operations on the hs_mailbox without having to pause the sequencer to read the last written value. Use the ENINT_COALESS flag in the hs_mailbox to toggle interrupt coalessing. Add entrypoints for enabling interrupt coalessing and setting both a timeout (how long to wait for commands to be coalessed) and a maximum commands to coaless value. Add a statistics timer that decides when to enable or disable interrupt coalessing based on load. Add a routine, ahd_reset_cmds_pending() which is used to update the CMDS_PENDING sequencer variable whenever error recovery compeltes SCBs without notifying the sequencer. Since ahd_reset_cmds_pending is called during ahd_unpause() only if we've aborted SCBs, its call to ahd_flush_qoutfifo should not cause recursion through ahd_run_qoutfifo(). A panic has been added to ensure that this recursion does not occur. In ahd_search_qinfifo, update the CMDS_PENDING sequencer variable directly. ahd_search_qinififo can be called in situations where using ahd_reset_cmds_pending() might cause recursion. Since we can safely determine the exact number to reduce CMDS_PENDING by in this scenario without running the qoutfifo, the manual update is sufficient. Clean up diagnostics. Add ahd_flush_qoutfifo() which will run the qoutfifo as well as complete any commands sitting on the sequencer's COMPLETE_SCB lists or the good status FIFO. Use this routine in several places that did similar things in an add-hoc, but incomplete, fashion. A call to this routine was also added to ahd_abort_scbs() to close a race. In ahd_pause_and_flushwork() only return once selections are safely disabled. Flush all completed commands via ahd_flush_qoutfifo(). Remove "Now packetized" diagnostic now that this information is incorperated into the actual negotiation messages that are displayed. When forcing renegotiation, don't clober the current ppr_options. Much of the driver uses this information to determine if we are currently packetized or not. Remove some stray spaces at column 1 in ahd_set_tags. When complaining about getting a host message loop request with no pending messages, print out the SCB_CONTROL register down on the card. Modify the ahd_sent_msg() routine to handle a search for an outgoing identify message. Use this to detect a msg reject on an identify message which typically indicates that the target thought we were packetized. Force a renegotiation in this case. In ahd_search_qinfifo(), wait more effectively for SCB DMA activities to cease. We also disable SCB fetch operations since we are about to change the qinfifo and any fetch in progress will likely be invalidated. In ahd_qinfifo_count(), fix the qinfifo empty case. In ahd_dump_card_state(), print out CCSCBCTL in the correct mode. If we are a narrow controller, don't set the current width to unknown when forcing a future negotiation. This just confuses the code into attempting a wide negotiation on a narrow bus. Add support for task management function completions. Modify ahd_handle_devreset so that it can handle lun resets in addition to target resets. Use ahd_handle_devreset for lun and target reset task management functions. Handle the abort task TMF race case better. We now wait until any current selections are over and then set the TMF back to zero. This should cause the sequencer to ignore the abort TMF completion should it occur. Correct a bug in the illegal phase handler that caused us to drop down to narrow when handling the unexpected command phase case after 3rd party reset of a packetized device. Indicate the features, bugs, and flags set in the softc that are used to control firmware patch download when booting verbose. aic79xx.h: Add coalessing and HS_MAILBOX fields. Add per-softc variables for the stats "daemon". Add a debug option for interrupt coalessing activities. Add two new softc flags: o AHD_UPDATE_PEND_CMDS Run ahd_reset_cmds_pending() on the next unpause. o AHD_RUNNING_QOUTFIFO Used to catch recursion through ahd_run_qoutfifo(). aic79xx.reg: Correct register addresses related to the software timer and the DFDBCTL register. Add constants paramaterizing the software timer. Add scratch ram locations for storing interrupt coalessing tunables. Break INTMASK in SEQITNCTL out into INTMASK1 and INTMASK2. In at least the REV A, these are writable bits. We make use of that for a swtimer workaround in the sequencer. Since HS_MAILBOX autoclears, provide a sequencer variable to store its contents. Add SEQINT codes for handling task management completions. aic79xx.seq: Correct ignore wide residue processing check for a wide negotiation being in effect. We must be in the SCSI register window in order to access the negotiation table. Use the software timer and a commands completed count to implement interrupt coalessing. The command complete is deferred until either the maximum command threshold or a the expiration of a command deferral timer. If we have more SCBs to complete to the host (sitting in COMPLETE_SCB lists), always try to coaless them up to our coalessing limit. If coalessing is enabled, but we have fewer commands oustanting than the host's min coalessing limit, complete the command immediately. Add code to track the number of commands outstanding. Commands are outstanding from the time they are placed into the execution queue until the DMA to post completion is setup. Add a workaround for intvec_2 interrupts on the H2A4. In H2A4, the mode pointer is not saved for intvec2, but is restored on iret. This can lead to the restoration of a bogus mode ptr. Manually clear the intmask bits and do a normal return to compensate. We use intvec_2 to track interrupt coalessing timeouts. Since we cannot disable the swtimer's countdown, simply mask its interrupt once we no longer care about it firing. In idle_loop_cchan, update LOCAL_HS_MAILBOX everytime we are notified of an HS_MAILBOX update via the HS_MAILBOX_ACT bit in QOFF_CTLSTA. We have to use a local copy of persistant portions of the HS_MAILBOX as the mailbox auto-clears on any read. Move the test for the cfg4istat interrupt up an instruction to hopefully close a race between the next outgoing selection and our disabling of selections. Add a missing ret to the last instruction in load_overrun_buf. Add notifications to the host of task management completions as well as the completions for commands that completed successfully before their corresponding TMF could be sent. Hold a critical section during select-out processing until we have a fully identified connection. This removes a race condition with the legacy abort handler. Correct a few spelling errors in some comments. aic79xx_inline.h: Call ahd_reset_cmds_pending() in ahd_unpause if required. Update cmdcmplt interrupt statistics in our interrupt handler. Allow callers to ahd_send_scb() to set the task management function. aic79xx_pci.c: Disable SERR and pause the controller prior to performing our mmapped I/O test. The U320 controllers do not support "auto-access-pause". aic79xx_osm.c: Set the task management function now that ahd_send_scb() doesn't do it for us. We also perform a lun reset in response to BDR requests to packetized devices.
2003-01-20 20:17:35 +00:00
* host's wishes.
*/
cmp COMPLETE_DMA_SCB_HEAD[1], SCB_LIST_NULL jne coalesce_by_count;
cmp COMPLETE_SCB_HEAD[1], SCB_LIST_NULL jne coalesce_by_count;
test LOCAL_HS_MAILBOX, ENINT_COALESCE jz issue_cmdcmplt;
aic79xx.c: Implement the SCB_SILENT flag. This is useful for hushing up the driver during DV or other operations that we expect to cause transmission errors. The messages will still print if the SHOW_MASKED_ERRORS debug option is enabled. Save and restore the NEGOADDR address when setting new transfer settings. The sequencer performs lookups in the negotiation table too and it expects NEGOADDR to remain consistent across pause/unpause sessions. Consistently use "offset" instead of "period" to determine if we are running sync or not. Add a SHOW_MESSAGES diagnostic for when we assert ATN during message processing. Print out IU, QAS, and RTI features when showing transfer options. Limit the syncrate after all option conformance changes have taken place in ahd_devlimited_syncrate. Changes in options may change the final syncrate we accept. Keep a copy of the hs_mailbox in our softc so that we can perform read/modify/write operations on the hs_mailbox without having to pause the sequencer to read the last written value. Use the ENINT_COALESS flag in the hs_mailbox to toggle interrupt coalessing. Add entrypoints for enabling interrupt coalessing and setting both a timeout (how long to wait for commands to be coalessed) and a maximum commands to coaless value. Add a statistics timer that decides when to enable or disable interrupt coalessing based on load. Add a routine, ahd_reset_cmds_pending() which is used to update the CMDS_PENDING sequencer variable whenever error recovery compeltes SCBs without notifying the sequencer. Since ahd_reset_cmds_pending is called during ahd_unpause() only if we've aborted SCBs, its call to ahd_flush_qoutfifo should not cause recursion through ahd_run_qoutfifo(). A panic has been added to ensure that this recursion does not occur. In ahd_search_qinfifo, update the CMDS_PENDING sequencer variable directly. ahd_search_qinififo can be called in situations where using ahd_reset_cmds_pending() might cause recursion. Since we can safely determine the exact number to reduce CMDS_PENDING by in this scenario without running the qoutfifo, the manual update is sufficient. Clean up diagnostics. Add ahd_flush_qoutfifo() which will run the qoutfifo as well as complete any commands sitting on the sequencer's COMPLETE_SCB lists or the good status FIFO. Use this routine in several places that did similar things in an add-hoc, but incomplete, fashion. A call to this routine was also added to ahd_abort_scbs() to close a race. In ahd_pause_and_flushwork() only return once selections are safely disabled. Flush all completed commands via ahd_flush_qoutfifo(). Remove "Now packetized" diagnostic now that this information is incorperated into the actual negotiation messages that are displayed. When forcing renegotiation, don't clober the current ppr_options. Much of the driver uses this information to determine if we are currently packetized or not. Remove some stray spaces at column 1 in ahd_set_tags. When complaining about getting a host message loop request with no pending messages, print out the SCB_CONTROL register down on the card. Modify the ahd_sent_msg() routine to handle a search for an outgoing identify message. Use this to detect a msg reject on an identify message which typically indicates that the target thought we were packetized. Force a renegotiation in this case. In ahd_search_qinfifo(), wait more effectively for SCB DMA activities to cease. We also disable SCB fetch operations since we are about to change the qinfifo and any fetch in progress will likely be invalidated. In ahd_qinfifo_count(), fix the qinfifo empty case. In ahd_dump_card_state(), print out CCSCBCTL in the correct mode. If we are a narrow controller, don't set the current width to unknown when forcing a future negotiation. This just confuses the code into attempting a wide negotiation on a narrow bus. Add support for task management function completions. Modify ahd_handle_devreset so that it can handle lun resets in addition to target resets. Use ahd_handle_devreset for lun and target reset task management functions. Handle the abort task TMF race case better. We now wait until any current selections are over and then set the TMF back to zero. This should cause the sequencer to ignore the abort TMF completion should it occur. Correct a bug in the illegal phase handler that caused us to drop down to narrow when handling the unexpected command phase case after 3rd party reset of a packetized device. Indicate the features, bugs, and flags set in the softc that are used to control firmware patch download when booting verbose. aic79xx.h: Add coalessing and HS_MAILBOX fields. Add per-softc variables for the stats "daemon". Add a debug option for interrupt coalessing activities. Add two new softc flags: o AHD_UPDATE_PEND_CMDS Run ahd_reset_cmds_pending() on the next unpause. o AHD_RUNNING_QOUTFIFO Used to catch recursion through ahd_run_qoutfifo(). aic79xx.reg: Correct register addresses related to the software timer and the DFDBCTL register. Add constants paramaterizing the software timer. Add scratch ram locations for storing interrupt coalessing tunables. Break INTMASK in SEQITNCTL out into INTMASK1 and INTMASK2. In at least the REV A, these are writable bits. We make use of that for a swtimer workaround in the sequencer. Since HS_MAILBOX autoclears, provide a sequencer variable to store its contents. Add SEQINT codes for handling task management completions. aic79xx.seq: Correct ignore wide residue processing check for a wide negotiation being in effect. We must be in the SCSI register window in order to access the negotiation table. Use the software timer and a commands completed count to implement interrupt coalessing. The command complete is deferred until either the maximum command threshold or a the expiration of a command deferral timer. If we have more SCBs to complete to the host (sitting in COMPLETE_SCB lists), always try to coaless them up to our coalessing limit. If coalessing is enabled, but we have fewer commands oustanting than the host's min coalessing limit, complete the command immediately. Add code to track the number of commands outstanding. Commands are outstanding from the time they are placed into the execution queue until the DMA to post completion is setup. Add a workaround for intvec_2 interrupts on the H2A4. In H2A4, the mode pointer is not saved for intvec2, but is restored on iret. This can lead to the restoration of a bogus mode ptr. Manually clear the intmask bits and do a normal return to compensate. We use intvec_2 to track interrupt coalessing timeouts. Since we cannot disable the swtimer's countdown, simply mask its interrupt once we no longer care about it firing. In idle_loop_cchan, update LOCAL_HS_MAILBOX everytime we are notified of an HS_MAILBOX update via the HS_MAILBOX_ACT bit in QOFF_CTLSTA. We have to use a local copy of persistant portions of the HS_MAILBOX as the mailbox auto-clears on any read. Move the test for the cfg4istat interrupt up an instruction to hopefully close a race between the next outgoing selection and our disabling of selections. Add a missing ret to the last instruction in load_overrun_buf. Add notifications to the host of task management completions as well as the completions for commands that completed successfully before their corresponding TMF could be sent. Hold a critical section during select-out processing until we have a fully identified connection. This removes a race condition with the legacy abort handler. Correct a few spelling errors in some comments. aic79xx_inline.h: Call ahd_reset_cmds_pending() in ahd_unpause if required. Update cmdcmplt interrupt statistics in our interrupt handler. Allow callers to ahd_send_scb() to set the task management function. aic79xx_pci.c: Disable SERR and pause the controller prior to performing our mmapped I/O test. The U320 controllers do not support "auto-access-pause". aic79xx_osm.c: Set the task management function now that ahd_send_scb() doesn't do it for us. We also perform a lun reset in response to BDR requests to packetized devices.
2003-01-20 20:17:35 +00:00
/*
* If we have relatively few commands outstanding, don't
* bother waiting for another command to complete.
*/
test CMDS_PENDING[1], 0xFF jnz coalesce_by_count;
aic79xx.c: Implement the SCB_SILENT flag. This is useful for hushing up the driver during DV or other operations that we expect to cause transmission errors. The messages will still print if the SHOW_MASKED_ERRORS debug option is enabled. Save and restore the NEGOADDR address when setting new transfer settings. The sequencer performs lookups in the negotiation table too and it expects NEGOADDR to remain consistent across pause/unpause sessions. Consistently use "offset" instead of "period" to determine if we are running sync or not. Add a SHOW_MESSAGES diagnostic for when we assert ATN during message processing. Print out IU, QAS, and RTI features when showing transfer options. Limit the syncrate after all option conformance changes have taken place in ahd_devlimited_syncrate. Changes in options may change the final syncrate we accept. Keep a copy of the hs_mailbox in our softc so that we can perform read/modify/write operations on the hs_mailbox without having to pause the sequencer to read the last written value. Use the ENINT_COALESS flag in the hs_mailbox to toggle interrupt coalessing. Add entrypoints for enabling interrupt coalessing and setting both a timeout (how long to wait for commands to be coalessed) and a maximum commands to coaless value. Add a statistics timer that decides when to enable or disable interrupt coalessing based on load. Add a routine, ahd_reset_cmds_pending() which is used to update the CMDS_PENDING sequencer variable whenever error recovery compeltes SCBs without notifying the sequencer. Since ahd_reset_cmds_pending is called during ahd_unpause() only if we've aborted SCBs, its call to ahd_flush_qoutfifo should not cause recursion through ahd_run_qoutfifo(). A panic has been added to ensure that this recursion does not occur. In ahd_search_qinfifo, update the CMDS_PENDING sequencer variable directly. ahd_search_qinififo can be called in situations where using ahd_reset_cmds_pending() might cause recursion. Since we can safely determine the exact number to reduce CMDS_PENDING by in this scenario without running the qoutfifo, the manual update is sufficient. Clean up diagnostics. Add ahd_flush_qoutfifo() which will run the qoutfifo as well as complete any commands sitting on the sequencer's COMPLETE_SCB lists or the good status FIFO. Use this routine in several places that did similar things in an add-hoc, but incomplete, fashion. A call to this routine was also added to ahd_abort_scbs() to close a race. In ahd_pause_and_flushwork() only return once selections are safely disabled. Flush all completed commands via ahd_flush_qoutfifo(). Remove "Now packetized" diagnostic now that this information is incorperated into the actual negotiation messages that are displayed. When forcing renegotiation, don't clober the current ppr_options. Much of the driver uses this information to determine if we are currently packetized or not. Remove some stray spaces at column 1 in ahd_set_tags. When complaining about getting a host message loop request with no pending messages, print out the SCB_CONTROL register down on the card. Modify the ahd_sent_msg() routine to handle a search for an outgoing identify message. Use this to detect a msg reject on an identify message which typically indicates that the target thought we were packetized. Force a renegotiation in this case. In ahd_search_qinfifo(), wait more effectively for SCB DMA activities to cease. We also disable SCB fetch operations since we are about to change the qinfifo and any fetch in progress will likely be invalidated. In ahd_qinfifo_count(), fix the qinfifo empty case. In ahd_dump_card_state(), print out CCSCBCTL in the correct mode. If we are a narrow controller, don't set the current width to unknown when forcing a future negotiation. This just confuses the code into attempting a wide negotiation on a narrow bus. Add support for task management function completions. Modify ahd_handle_devreset so that it can handle lun resets in addition to target resets. Use ahd_handle_devreset for lun and target reset task management functions. Handle the abort task TMF race case better. We now wait until any current selections are over and then set the TMF back to zero. This should cause the sequencer to ignore the abort TMF completion should it occur. Correct a bug in the illegal phase handler that caused us to drop down to narrow when handling the unexpected command phase case after 3rd party reset of a packetized device. Indicate the features, bugs, and flags set in the softc that are used to control firmware patch download when booting verbose. aic79xx.h: Add coalessing and HS_MAILBOX fields. Add per-softc variables for the stats "daemon". Add a debug option for interrupt coalessing activities. Add two new softc flags: o AHD_UPDATE_PEND_CMDS Run ahd_reset_cmds_pending() on the next unpause. o AHD_RUNNING_QOUTFIFO Used to catch recursion through ahd_run_qoutfifo(). aic79xx.reg: Correct register addresses related to the software timer and the DFDBCTL register. Add constants paramaterizing the software timer. Add scratch ram locations for storing interrupt coalessing tunables. Break INTMASK in SEQITNCTL out into INTMASK1 and INTMASK2. In at least the REV A, these are writable bits. We make use of that for a swtimer workaround in the sequencer. Since HS_MAILBOX autoclears, provide a sequencer variable to store its contents. Add SEQINT codes for handling task management completions. aic79xx.seq: Correct ignore wide residue processing check for a wide negotiation being in effect. We must be in the SCSI register window in order to access the negotiation table. Use the software timer and a commands completed count to implement interrupt coalessing. The command complete is deferred until either the maximum command threshold or a the expiration of a command deferral timer. If we have more SCBs to complete to the host (sitting in COMPLETE_SCB lists), always try to coaless them up to our coalessing limit. If coalessing is enabled, but we have fewer commands oustanting than the host's min coalessing limit, complete the command immediately. Add code to track the number of commands outstanding. Commands are outstanding from the time they are placed into the execution queue until the DMA to post completion is setup. Add a workaround for intvec_2 interrupts on the H2A4. In H2A4, the mode pointer is not saved for intvec2, but is restored on iret. This can lead to the restoration of a bogus mode ptr. Manually clear the intmask bits and do a normal return to compensate. We use intvec_2 to track interrupt coalessing timeouts. Since we cannot disable the swtimer's countdown, simply mask its interrupt once we no longer care about it firing. In idle_loop_cchan, update LOCAL_HS_MAILBOX everytime we are notified of an HS_MAILBOX update via the HS_MAILBOX_ACT bit in QOFF_CTLSTA. We have to use a local copy of persistant portions of the HS_MAILBOX as the mailbox auto-clears on any read. Move the test for the cfg4istat interrupt up an instruction to hopefully close a race between the next outgoing selection and our disabling of selections. Add a missing ret to the last instruction in load_overrun_buf. Add notifications to the host of task management completions as well as the completions for commands that completed successfully before their corresponding TMF could be sent. Hold a critical section during select-out processing until we have a fully identified connection. This removes a race condition with the legacy abort handler. Correct a few spelling errors in some comments. aic79xx_inline.h: Call ahd_reset_cmds_pending() in ahd_unpause if required. Update cmdcmplt interrupt statistics in our interrupt handler. Allow callers to ahd_send_scb() to set the task management function. aic79xx_pci.c: Disable SERR and pause the controller prior to performing our mmapped I/O test. The U320 controllers do not support "auto-access-pause". aic79xx_osm.c: Set the task management function now that ahd_send_scb() doesn't do it for us. We also perform a lun reset in response to BDR requests to packetized devices.
2003-01-20 20:17:35 +00:00
/* Add -1 so that jnc means <= not just < */
add A, -1, INT_COALESCING_MINCMDS;
aic79xx.c: Implement the SCB_SILENT flag. This is useful for hushing up the driver during DV or other operations that we expect to cause transmission errors. The messages will still print if the SHOW_MASKED_ERRORS debug option is enabled. Save and restore the NEGOADDR address when setting new transfer settings. The sequencer performs lookups in the negotiation table too and it expects NEGOADDR to remain consistent across pause/unpause sessions. Consistently use "offset" instead of "period" to determine if we are running sync or not. Add a SHOW_MESSAGES diagnostic for when we assert ATN during message processing. Print out IU, QAS, and RTI features when showing transfer options. Limit the syncrate after all option conformance changes have taken place in ahd_devlimited_syncrate. Changes in options may change the final syncrate we accept. Keep a copy of the hs_mailbox in our softc so that we can perform read/modify/write operations on the hs_mailbox without having to pause the sequencer to read the last written value. Use the ENINT_COALESS flag in the hs_mailbox to toggle interrupt coalessing. Add entrypoints for enabling interrupt coalessing and setting both a timeout (how long to wait for commands to be coalessed) and a maximum commands to coaless value. Add a statistics timer that decides when to enable or disable interrupt coalessing based on load. Add a routine, ahd_reset_cmds_pending() which is used to update the CMDS_PENDING sequencer variable whenever error recovery compeltes SCBs without notifying the sequencer. Since ahd_reset_cmds_pending is called during ahd_unpause() only if we've aborted SCBs, its call to ahd_flush_qoutfifo should not cause recursion through ahd_run_qoutfifo(). A panic has been added to ensure that this recursion does not occur. In ahd_search_qinfifo, update the CMDS_PENDING sequencer variable directly. ahd_search_qinififo can be called in situations where using ahd_reset_cmds_pending() might cause recursion. Since we can safely determine the exact number to reduce CMDS_PENDING by in this scenario without running the qoutfifo, the manual update is sufficient. Clean up diagnostics. Add ahd_flush_qoutfifo() which will run the qoutfifo as well as complete any commands sitting on the sequencer's COMPLETE_SCB lists or the good status FIFO. Use this routine in several places that did similar things in an add-hoc, but incomplete, fashion. A call to this routine was also added to ahd_abort_scbs() to close a race. In ahd_pause_and_flushwork() only return once selections are safely disabled. Flush all completed commands via ahd_flush_qoutfifo(). Remove "Now packetized" diagnostic now that this information is incorperated into the actual negotiation messages that are displayed. When forcing renegotiation, don't clober the current ppr_options. Much of the driver uses this information to determine if we are currently packetized or not. Remove some stray spaces at column 1 in ahd_set_tags. When complaining about getting a host message loop request with no pending messages, print out the SCB_CONTROL register down on the card. Modify the ahd_sent_msg() routine to handle a search for an outgoing identify message. Use this to detect a msg reject on an identify message which typically indicates that the target thought we were packetized. Force a renegotiation in this case. In ahd_search_qinfifo(), wait more effectively for SCB DMA activities to cease. We also disable SCB fetch operations since we are about to change the qinfifo and any fetch in progress will likely be invalidated. In ahd_qinfifo_count(), fix the qinfifo empty case. In ahd_dump_card_state(), print out CCSCBCTL in the correct mode. If we are a narrow controller, don't set the current width to unknown when forcing a future negotiation. This just confuses the code into attempting a wide negotiation on a narrow bus. Add support for task management function completions. Modify ahd_handle_devreset so that it can handle lun resets in addition to target resets. Use ahd_handle_devreset for lun and target reset task management functions. Handle the abort task TMF race case better. We now wait until any current selections are over and then set the TMF back to zero. This should cause the sequencer to ignore the abort TMF completion should it occur. Correct a bug in the illegal phase handler that caused us to drop down to narrow when handling the unexpected command phase case after 3rd party reset of a packetized device. Indicate the features, bugs, and flags set in the softc that are used to control firmware patch download when booting verbose. aic79xx.h: Add coalessing and HS_MAILBOX fields. Add per-softc variables for the stats "daemon". Add a debug option for interrupt coalessing activities. Add two new softc flags: o AHD_UPDATE_PEND_CMDS Run ahd_reset_cmds_pending() on the next unpause. o AHD_RUNNING_QOUTFIFO Used to catch recursion through ahd_run_qoutfifo(). aic79xx.reg: Correct register addresses related to the software timer and the DFDBCTL register. Add constants paramaterizing the software timer. Add scratch ram locations for storing interrupt coalessing tunables. Break INTMASK in SEQITNCTL out into INTMASK1 and INTMASK2. In at least the REV A, these are writable bits. We make use of that for a swtimer workaround in the sequencer. Since HS_MAILBOX autoclears, provide a sequencer variable to store its contents. Add SEQINT codes for handling task management completions. aic79xx.seq: Correct ignore wide residue processing check for a wide negotiation being in effect. We must be in the SCSI register window in order to access the negotiation table. Use the software timer and a commands completed count to implement interrupt coalessing. The command complete is deferred until either the maximum command threshold or a the expiration of a command deferral timer. If we have more SCBs to complete to the host (sitting in COMPLETE_SCB lists), always try to coaless them up to our coalessing limit. If coalessing is enabled, but we have fewer commands oustanting than the host's min coalessing limit, complete the command immediately. Add code to track the number of commands outstanding. Commands are outstanding from the time they are placed into the execution queue until the DMA to post completion is setup. Add a workaround for intvec_2 interrupts on the H2A4. In H2A4, the mode pointer is not saved for intvec2, but is restored on iret. This can lead to the restoration of a bogus mode ptr. Manually clear the intmask bits and do a normal return to compensate. We use intvec_2 to track interrupt coalessing timeouts. Since we cannot disable the swtimer's countdown, simply mask its interrupt once we no longer care about it firing. In idle_loop_cchan, update LOCAL_HS_MAILBOX everytime we are notified of an HS_MAILBOX update via the HS_MAILBOX_ACT bit in QOFF_CTLSTA. We have to use a local copy of persistant portions of the HS_MAILBOX as the mailbox auto-clears on any read. Move the test for the cfg4istat interrupt up an instruction to hopefully close a race between the next outgoing selection and our disabling of selections. Add a missing ret to the last instruction in load_overrun_buf. Add notifications to the host of task management completions as well as the completions for commands that completed successfully before their corresponding TMF could be sent. Hold a critical section during select-out processing until we have a fully identified connection. This removes a race condition with the legacy abort handler. Correct a few spelling errors in some comments. aic79xx_inline.h: Call ahd_reset_cmds_pending() in ahd_unpause if required. Update cmdcmplt interrupt statistics in our interrupt handler. Allow callers to ahd_send_scb() to set the task management function. aic79xx_pci.c: Disable SERR and pause the controller prior to performing our mmapped I/O test. The U320 controllers do not support "auto-access-pause". aic79xx_osm.c: Set the task management function now that ahd_send_scb() doesn't do it for us. We also perform a lun reset in response to BDR requests to packetized devices.
2003-01-20 20:17:35 +00:00
add NONE, A, CMDS_PENDING;
jnc issue_cmdcmplt;
/*
* If coalescing, only coalesce up to the limit
aic79xx.c: Implement the SCB_SILENT flag. This is useful for hushing up the driver during DV or other operations that we expect to cause transmission errors. The messages will still print if the SHOW_MASKED_ERRORS debug option is enabled. Save and restore the NEGOADDR address when setting new transfer settings. The sequencer performs lookups in the negotiation table too and it expects NEGOADDR to remain consistent across pause/unpause sessions. Consistently use "offset" instead of "period" to determine if we are running sync or not. Add a SHOW_MESSAGES diagnostic for when we assert ATN during message processing. Print out IU, QAS, and RTI features when showing transfer options. Limit the syncrate after all option conformance changes have taken place in ahd_devlimited_syncrate. Changes in options may change the final syncrate we accept. Keep a copy of the hs_mailbox in our softc so that we can perform read/modify/write operations on the hs_mailbox without having to pause the sequencer to read the last written value. Use the ENINT_COALESS flag in the hs_mailbox to toggle interrupt coalessing. Add entrypoints for enabling interrupt coalessing and setting both a timeout (how long to wait for commands to be coalessed) and a maximum commands to coaless value. Add a statistics timer that decides when to enable or disable interrupt coalessing based on load. Add a routine, ahd_reset_cmds_pending() which is used to update the CMDS_PENDING sequencer variable whenever error recovery compeltes SCBs without notifying the sequencer. Since ahd_reset_cmds_pending is called during ahd_unpause() only if we've aborted SCBs, its call to ahd_flush_qoutfifo should not cause recursion through ahd_run_qoutfifo(). A panic has been added to ensure that this recursion does not occur. In ahd_search_qinfifo, update the CMDS_PENDING sequencer variable directly. ahd_search_qinififo can be called in situations where using ahd_reset_cmds_pending() might cause recursion. Since we can safely determine the exact number to reduce CMDS_PENDING by in this scenario without running the qoutfifo, the manual update is sufficient. Clean up diagnostics. Add ahd_flush_qoutfifo() which will run the qoutfifo as well as complete any commands sitting on the sequencer's COMPLETE_SCB lists or the good status FIFO. Use this routine in several places that did similar things in an add-hoc, but incomplete, fashion. A call to this routine was also added to ahd_abort_scbs() to close a race. In ahd_pause_and_flushwork() only return once selections are safely disabled. Flush all completed commands via ahd_flush_qoutfifo(). Remove "Now packetized" diagnostic now that this information is incorperated into the actual negotiation messages that are displayed. When forcing renegotiation, don't clober the current ppr_options. Much of the driver uses this information to determine if we are currently packetized or not. Remove some stray spaces at column 1 in ahd_set_tags. When complaining about getting a host message loop request with no pending messages, print out the SCB_CONTROL register down on the card. Modify the ahd_sent_msg() routine to handle a search for an outgoing identify message. Use this to detect a msg reject on an identify message which typically indicates that the target thought we were packetized. Force a renegotiation in this case. In ahd_search_qinfifo(), wait more effectively for SCB DMA activities to cease. We also disable SCB fetch operations since we are about to change the qinfifo and any fetch in progress will likely be invalidated. In ahd_qinfifo_count(), fix the qinfifo empty case. In ahd_dump_card_state(), print out CCSCBCTL in the correct mode. If we are a narrow controller, don't set the current width to unknown when forcing a future negotiation. This just confuses the code into attempting a wide negotiation on a narrow bus. Add support for task management function completions. Modify ahd_handle_devreset so that it can handle lun resets in addition to target resets. Use ahd_handle_devreset for lun and target reset task management functions. Handle the abort task TMF race case better. We now wait until any current selections are over and then set the TMF back to zero. This should cause the sequencer to ignore the abort TMF completion should it occur. Correct a bug in the illegal phase handler that caused us to drop down to narrow when handling the unexpected command phase case after 3rd party reset of a packetized device. Indicate the features, bugs, and flags set in the softc that are used to control firmware patch download when booting verbose. aic79xx.h: Add coalessing and HS_MAILBOX fields. Add per-softc variables for the stats "daemon". Add a debug option for interrupt coalessing activities. Add two new softc flags: o AHD_UPDATE_PEND_CMDS Run ahd_reset_cmds_pending() on the next unpause. o AHD_RUNNING_QOUTFIFO Used to catch recursion through ahd_run_qoutfifo(). aic79xx.reg: Correct register addresses related to the software timer and the DFDBCTL register. Add constants paramaterizing the software timer. Add scratch ram locations for storing interrupt coalessing tunables. Break INTMASK in SEQITNCTL out into INTMASK1 and INTMASK2. In at least the REV A, these are writable bits. We make use of that for a swtimer workaround in the sequencer. Since HS_MAILBOX autoclears, provide a sequencer variable to store its contents. Add SEQINT codes for handling task management completions. aic79xx.seq: Correct ignore wide residue processing check for a wide negotiation being in effect. We must be in the SCSI register window in order to access the negotiation table. Use the software timer and a commands completed count to implement interrupt coalessing. The command complete is deferred until either the maximum command threshold or a the expiration of a command deferral timer. If we have more SCBs to complete to the host (sitting in COMPLETE_SCB lists), always try to coaless them up to our coalessing limit. If coalessing is enabled, but we have fewer commands oustanting than the host's min coalessing limit, complete the command immediately. Add code to track the number of commands outstanding. Commands are outstanding from the time they are placed into the execution queue until the DMA to post completion is setup. Add a workaround for intvec_2 interrupts on the H2A4. In H2A4, the mode pointer is not saved for intvec2, but is restored on iret. This can lead to the restoration of a bogus mode ptr. Manually clear the intmask bits and do a normal return to compensate. We use intvec_2 to track interrupt coalessing timeouts. Since we cannot disable the swtimer's countdown, simply mask its interrupt once we no longer care about it firing. In idle_loop_cchan, update LOCAL_HS_MAILBOX everytime we are notified of an HS_MAILBOX update via the HS_MAILBOX_ACT bit in QOFF_CTLSTA. We have to use a local copy of persistant portions of the HS_MAILBOX as the mailbox auto-clears on any read. Move the test for the cfg4istat interrupt up an instruction to hopefully close a race between the next outgoing selection and our disabling of selections. Add a missing ret to the last instruction in load_overrun_buf. Add notifications to the host of task management completions as well as the completions for commands that completed successfully before their corresponding TMF could be sent. Hold a critical section during select-out processing until we have a fully identified connection. This removes a race condition with the legacy abort handler. Correct a few spelling errors in some comments. aic79xx_inline.h: Call ahd_reset_cmds_pending() in ahd_unpause if required. Update cmdcmplt interrupt statistics in our interrupt handler. Allow callers to ahd_send_scb() to set the task management function. aic79xx_pci.c: Disable SERR and pause the controller prior to performing our mmapped I/O test. The U320 controllers do not support "auto-access-pause". aic79xx_osm.c: Set the task management function now that ahd_send_scb() doesn't do it for us. We also perform a lun reset in response to BDR requests to packetized devices.
2003-01-20 20:17:35 +00:00
* provided by the host driver.
*/
coalesce_by_count:
mov A, INT_COALESCING_MAXCMDS;
add NONE, A, INT_COALESCING_CMDCOUNT;
aic79xx.c: Implement the SCB_SILENT flag. This is useful for hushing up the driver during DV or other operations that we expect to cause transmission errors. The messages will still print if the SHOW_MASKED_ERRORS debug option is enabled. Save and restore the NEGOADDR address when setting new transfer settings. The sequencer performs lookups in the negotiation table too and it expects NEGOADDR to remain consistent across pause/unpause sessions. Consistently use "offset" instead of "period" to determine if we are running sync or not. Add a SHOW_MESSAGES diagnostic for when we assert ATN during message processing. Print out IU, QAS, and RTI features when showing transfer options. Limit the syncrate after all option conformance changes have taken place in ahd_devlimited_syncrate. Changes in options may change the final syncrate we accept. Keep a copy of the hs_mailbox in our softc so that we can perform read/modify/write operations on the hs_mailbox without having to pause the sequencer to read the last written value. Use the ENINT_COALESS flag in the hs_mailbox to toggle interrupt coalessing. Add entrypoints for enabling interrupt coalessing and setting both a timeout (how long to wait for commands to be coalessed) and a maximum commands to coaless value. Add a statistics timer that decides when to enable or disable interrupt coalessing based on load. Add a routine, ahd_reset_cmds_pending() which is used to update the CMDS_PENDING sequencer variable whenever error recovery compeltes SCBs without notifying the sequencer. Since ahd_reset_cmds_pending is called during ahd_unpause() only if we've aborted SCBs, its call to ahd_flush_qoutfifo should not cause recursion through ahd_run_qoutfifo(). A panic has been added to ensure that this recursion does not occur. In ahd_search_qinfifo, update the CMDS_PENDING sequencer variable directly. ahd_search_qinififo can be called in situations where using ahd_reset_cmds_pending() might cause recursion. Since we can safely determine the exact number to reduce CMDS_PENDING by in this scenario without running the qoutfifo, the manual update is sufficient. Clean up diagnostics. Add ahd_flush_qoutfifo() which will run the qoutfifo as well as complete any commands sitting on the sequencer's COMPLETE_SCB lists or the good status FIFO. Use this routine in several places that did similar things in an add-hoc, but incomplete, fashion. A call to this routine was also added to ahd_abort_scbs() to close a race. In ahd_pause_and_flushwork() only return once selections are safely disabled. Flush all completed commands via ahd_flush_qoutfifo(). Remove "Now packetized" diagnostic now that this information is incorperated into the actual negotiation messages that are displayed. When forcing renegotiation, don't clober the current ppr_options. Much of the driver uses this information to determine if we are currently packetized or not. Remove some stray spaces at column 1 in ahd_set_tags. When complaining about getting a host message loop request with no pending messages, print out the SCB_CONTROL register down on the card. Modify the ahd_sent_msg() routine to handle a search for an outgoing identify message. Use this to detect a msg reject on an identify message which typically indicates that the target thought we were packetized. Force a renegotiation in this case. In ahd_search_qinfifo(), wait more effectively for SCB DMA activities to cease. We also disable SCB fetch operations since we are about to change the qinfifo and any fetch in progress will likely be invalidated. In ahd_qinfifo_count(), fix the qinfifo empty case. In ahd_dump_card_state(), print out CCSCBCTL in the correct mode. If we are a narrow controller, don't set the current width to unknown when forcing a future negotiation. This just confuses the code into attempting a wide negotiation on a narrow bus. Add support for task management function completions. Modify ahd_handle_devreset so that it can handle lun resets in addition to target resets. Use ahd_handle_devreset for lun and target reset task management functions. Handle the abort task TMF race case better. We now wait until any current selections are over and then set the TMF back to zero. This should cause the sequencer to ignore the abort TMF completion should it occur. Correct a bug in the illegal phase handler that caused us to drop down to narrow when handling the unexpected command phase case after 3rd party reset of a packetized device. Indicate the features, bugs, and flags set in the softc that are used to control firmware patch download when booting verbose. aic79xx.h: Add coalessing and HS_MAILBOX fields. Add per-softc variables for the stats "daemon". Add a debug option for interrupt coalessing activities. Add two new softc flags: o AHD_UPDATE_PEND_CMDS Run ahd_reset_cmds_pending() on the next unpause. o AHD_RUNNING_QOUTFIFO Used to catch recursion through ahd_run_qoutfifo(). aic79xx.reg: Correct register addresses related to the software timer and the DFDBCTL register. Add constants paramaterizing the software timer. Add scratch ram locations for storing interrupt coalessing tunables. Break INTMASK in SEQITNCTL out into INTMASK1 and INTMASK2. In at least the REV A, these are writable bits. We make use of that for a swtimer workaround in the sequencer. Since HS_MAILBOX autoclears, provide a sequencer variable to store its contents. Add SEQINT codes for handling task management completions. aic79xx.seq: Correct ignore wide residue processing check for a wide negotiation being in effect. We must be in the SCSI register window in order to access the negotiation table. Use the software timer and a commands completed count to implement interrupt coalessing. The command complete is deferred until either the maximum command threshold or a the expiration of a command deferral timer. If we have more SCBs to complete to the host (sitting in COMPLETE_SCB lists), always try to coaless them up to our coalessing limit. If coalessing is enabled, but we have fewer commands oustanting than the host's min coalessing limit, complete the command immediately. Add code to track the number of commands outstanding. Commands are outstanding from the time they are placed into the execution queue until the DMA to post completion is setup. Add a workaround for intvec_2 interrupts on the H2A4. In H2A4, the mode pointer is not saved for intvec2, but is restored on iret. This can lead to the restoration of a bogus mode ptr. Manually clear the intmask bits and do a normal return to compensate. We use intvec_2 to track interrupt coalessing timeouts. Since we cannot disable the swtimer's countdown, simply mask its interrupt once we no longer care about it firing. In idle_loop_cchan, update LOCAL_HS_MAILBOX everytime we are notified of an HS_MAILBOX update via the HS_MAILBOX_ACT bit in QOFF_CTLSTA. We have to use a local copy of persistant portions of the HS_MAILBOX as the mailbox auto-clears on any read. Move the test for the cfg4istat interrupt up an instruction to hopefully close a race between the next outgoing selection and our disabling of selections. Add a missing ret to the last instruction in load_overrun_buf. Add notifications to the host of task management completions as well as the completions for commands that completed successfully before their corresponding TMF could be sent. Hold a critical section during select-out processing until we have a fully identified connection. This removes a race condition with the legacy abort handler. Correct a few spelling errors in some comments. aic79xx_inline.h: Call ahd_reset_cmds_pending() in ahd_unpause if required. Update cmdcmplt interrupt statistics in our interrupt handler. Allow callers to ahd_send_scb() to set the task management function. aic79xx_pci.c: Disable SERR and pause the controller prior to performing our mmapped I/O test. The U320 controllers do not support "auto-access-pause". aic79xx_osm.c: Set the task management function now that ahd_send_scb() doesn't do it for us. We also perform a lun reset in response to BDR requests to packetized devices.
2003-01-20 20:17:35 +00:00
jc issue_cmdcmplt;
/*
* If the timer is not currently active,
* fire it up.
*/
test INTCTL, SWTMINTMASK jz return;
bmov SWTIMER, INT_COALESCING_TIMER, 2;
aic79xx.c: Implement the SCB_SILENT flag. This is useful for hushing up the driver during DV or other operations that we expect to cause transmission errors. The messages will still print if the SHOW_MASKED_ERRORS debug option is enabled. Save and restore the NEGOADDR address when setting new transfer settings. The sequencer performs lookups in the negotiation table too and it expects NEGOADDR to remain consistent across pause/unpause sessions. Consistently use "offset" instead of "period" to determine if we are running sync or not. Add a SHOW_MESSAGES diagnostic for when we assert ATN during message processing. Print out IU, QAS, and RTI features when showing transfer options. Limit the syncrate after all option conformance changes have taken place in ahd_devlimited_syncrate. Changes in options may change the final syncrate we accept. Keep a copy of the hs_mailbox in our softc so that we can perform read/modify/write operations on the hs_mailbox without having to pause the sequencer to read the last written value. Use the ENINT_COALESS flag in the hs_mailbox to toggle interrupt coalessing. Add entrypoints for enabling interrupt coalessing and setting both a timeout (how long to wait for commands to be coalessed) and a maximum commands to coaless value. Add a statistics timer that decides when to enable or disable interrupt coalessing based on load. Add a routine, ahd_reset_cmds_pending() which is used to update the CMDS_PENDING sequencer variable whenever error recovery compeltes SCBs without notifying the sequencer. Since ahd_reset_cmds_pending is called during ahd_unpause() only if we've aborted SCBs, its call to ahd_flush_qoutfifo should not cause recursion through ahd_run_qoutfifo(). A panic has been added to ensure that this recursion does not occur. In ahd_search_qinfifo, update the CMDS_PENDING sequencer variable directly. ahd_search_qinififo can be called in situations where using ahd_reset_cmds_pending() might cause recursion. Since we can safely determine the exact number to reduce CMDS_PENDING by in this scenario without running the qoutfifo, the manual update is sufficient. Clean up diagnostics. Add ahd_flush_qoutfifo() which will run the qoutfifo as well as complete any commands sitting on the sequencer's COMPLETE_SCB lists or the good status FIFO. Use this routine in several places that did similar things in an add-hoc, but incomplete, fashion. A call to this routine was also added to ahd_abort_scbs() to close a race. In ahd_pause_and_flushwork() only return once selections are safely disabled. Flush all completed commands via ahd_flush_qoutfifo(). Remove "Now packetized" diagnostic now that this information is incorperated into the actual negotiation messages that are displayed. When forcing renegotiation, don't clober the current ppr_options. Much of the driver uses this information to determine if we are currently packetized or not. Remove some stray spaces at column 1 in ahd_set_tags. When complaining about getting a host message loop request with no pending messages, print out the SCB_CONTROL register down on the card. Modify the ahd_sent_msg() routine to handle a search for an outgoing identify message. Use this to detect a msg reject on an identify message which typically indicates that the target thought we were packetized. Force a renegotiation in this case. In ahd_search_qinfifo(), wait more effectively for SCB DMA activities to cease. We also disable SCB fetch operations since we are about to change the qinfifo and any fetch in progress will likely be invalidated. In ahd_qinfifo_count(), fix the qinfifo empty case. In ahd_dump_card_state(), print out CCSCBCTL in the correct mode. If we are a narrow controller, don't set the current width to unknown when forcing a future negotiation. This just confuses the code into attempting a wide negotiation on a narrow bus. Add support for task management function completions. Modify ahd_handle_devreset so that it can handle lun resets in addition to target resets. Use ahd_handle_devreset for lun and target reset task management functions. Handle the abort task TMF race case better. We now wait until any current selections are over and then set the TMF back to zero. This should cause the sequencer to ignore the abort TMF completion should it occur. Correct a bug in the illegal phase handler that caused us to drop down to narrow when handling the unexpected command phase case after 3rd party reset of a packetized device. Indicate the features, bugs, and flags set in the softc that are used to control firmware patch download when booting verbose. aic79xx.h: Add coalessing and HS_MAILBOX fields. Add per-softc variables for the stats "daemon". Add a debug option for interrupt coalessing activities. Add two new softc flags: o AHD_UPDATE_PEND_CMDS Run ahd_reset_cmds_pending() on the next unpause. o AHD_RUNNING_QOUTFIFO Used to catch recursion through ahd_run_qoutfifo(). aic79xx.reg: Correct register addresses related to the software timer and the DFDBCTL register. Add constants paramaterizing the software timer. Add scratch ram locations for storing interrupt coalessing tunables. Break INTMASK in SEQITNCTL out into INTMASK1 and INTMASK2. In at least the REV A, these are writable bits. We make use of that for a swtimer workaround in the sequencer. Since HS_MAILBOX autoclears, provide a sequencer variable to store its contents. Add SEQINT codes for handling task management completions. aic79xx.seq: Correct ignore wide residue processing check for a wide negotiation being in effect. We must be in the SCSI register window in order to access the negotiation table. Use the software timer and a commands completed count to implement interrupt coalessing. The command complete is deferred until either the maximum command threshold or a the expiration of a command deferral timer. If we have more SCBs to complete to the host (sitting in COMPLETE_SCB lists), always try to coaless them up to our coalessing limit. If coalessing is enabled, but we have fewer commands oustanting than the host's min coalessing limit, complete the command immediately. Add code to track the number of commands outstanding. Commands are outstanding from the time they are placed into the execution queue until the DMA to post completion is setup. Add a workaround for intvec_2 interrupts on the H2A4. In H2A4, the mode pointer is not saved for intvec2, but is restored on iret. This can lead to the restoration of a bogus mode ptr. Manually clear the intmask bits and do a normal return to compensate. We use intvec_2 to track interrupt coalessing timeouts. Since we cannot disable the swtimer's countdown, simply mask its interrupt once we no longer care about it firing. In idle_loop_cchan, update LOCAL_HS_MAILBOX everytime we are notified of an HS_MAILBOX update via the HS_MAILBOX_ACT bit in QOFF_CTLSTA. We have to use a local copy of persistant portions of the HS_MAILBOX as the mailbox auto-clears on any read. Move the test for the cfg4istat interrupt up an instruction to hopefully close a race between the next outgoing selection and our disabling of selections. Add a missing ret to the last instruction in load_overrun_buf. Add notifications to the host of task management completions as well as the completions for commands that completed successfully before their corresponding TMF could be sent. Hold a critical section during select-out processing until we have a fully identified connection. This removes a race condition with the legacy abort handler. Correct a few spelling errors in some comments. aic79xx_inline.h: Call ahd_reset_cmds_pending() in ahd_unpause if required. Update cmdcmplt interrupt statistics in our interrupt handler. Allow callers to ahd_send_scb() to set the task management function. aic79xx_pci.c: Disable SERR and pause the controller prior to performing our mmapped I/O test. The U320 controllers do not support "auto-access-pause". aic79xx_osm.c: Set the task management function now that ahd_send_scb() doesn't do it for us. We also perform a lun reset in response to BDR requests to packetized devices.
2003-01-20 20:17:35 +00:00
mvi CLRSEQINTSTAT, CLRSEQ_SWTMRTO;
or INTCTL, SWTMINTEN|SWTIMER_START;
and INTCTL, ~SWTMINTMASK ret;
issue_cmdcmplt:
mvi INTSTAT, CMDCMPLT;
clr INT_COALESCING_CMDCOUNT;
aic79xx.c: Implement the SCB_SILENT flag. This is useful for hushing up the driver during DV or other operations that we expect to cause transmission errors. The messages will still print if the SHOW_MASKED_ERRORS debug option is enabled. Save and restore the NEGOADDR address when setting new transfer settings. The sequencer performs lookups in the negotiation table too and it expects NEGOADDR to remain consistent across pause/unpause sessions. Consistently use "offset" instead of "period" to determine if we are running sync or not. Add a SHOW_MESSAGES diagnostic for when we assert ATN during message processing. Print out IU, QAS, and RTI features when showing transfer options. Limit the syncrate after all option conformance changes have taken place in ahd_devlimited_syncrate. Changes in options may change the final syncrate we accept. Keep a copy of the hs_mailbox in our softc so that we can perform read/modify/write operations on the hs_mailbox without having to pause the sequencer to read the last written value. Use the ENINT_COALESS flag in the hs_mailbox to toggle interrupt coalessing. Add entrypoints for enabling interrupt coalessing and setting both a timeout (how long to wait for commands to be coalessed) and a maximum commands to coaless value. Add a statistics timer that decides when to enable or disable interrupt coalessing based on load. Add a routine, ahd_reset_cmds_pending() which is used to update the CMDS_PENDING sequencer variable whenever error recovery compeltes SCBs without notifying the sequencer. Since ahd_reset_cmds_pending is called during ahd_unpause() only if we've aborted SCBs, its call to ahd_flush_qoutfifo should not cause recursion through ahd_run_qoutfifo(). A panic has been added to ensure that this recursion does not occur. In ahd_search_qinfifo, update the CMDS_PENDING sequencer variable directly. ahd_search_qinififo can be called in situations where using ahd_reset_cmds_pending() might cause recursion. Since we can safely determine the exact number to reduce CMDS_PENDING by in this scenario without running the qoutfifo, the manual update is sufficient. Clean up diagnostics. Add ahd_flush_qoutfifo() which will run the qoutfifo as well as complete any commands sitting on the sequencer's COMPLETE_SCB lists or the good status FIFO. Use this routine in several places that did similar things in an add-hoc, but incomplete, fashion. A call to this routine was also added to ahd_abort_scbs() to close a race. In ahd_pause_and_flushwork() only return once selections are safely disabled. Flush all completed commands via ahd_flush_qoutfifo(). Remove "Now packetized" diagnostic now that this information is incorperated into the actual negotiation messages that are displayed. When forcing renegotiation, don't clober the current ppr_options. Much of the driver uses this information to determine if we are currently packetized or not. Remove some stray spaces at column 1 in ahd_set_tags. When complaining about getting a host message loop request with no pending messages, print out the SCB_CONTROL register down on the card. Modify the ahd_sent_msg() routine to handle a search for an outgoing identify message. Use this to detect a msg reject on an identify message which typically indicates that the target thought we were packetized. Force a renegotiation in this case. In ahd_search_qinfifo(), wait more effectively for SCB DMA activities to cease. We also disable SCB fetch operations since we are about to change the qinfifo and any fetch in progress will likely be invalidated. In ahd_qinfifo_count(), fix the qinfifo empty case. In ahd_dump_card_state(), print out CCSCBCTL in the correct mode. If we are a narrow controller, don't set the current width to unknown when forcing a future negotiation. This just confuses the code into attempting a wide negotiation on a narrow bus. Add support for task management function completions. Modify ahd_handle_devreset so that it can handle lun resets in addition to target resets. Use ahd_handle_devreset for lun and target reset task management functions. Handle the abort task TMF race case better. We now wait until any current selections are over and then set the TMF back to zero. This should cause the sequencer to ignore the abort TMF completion should it occur. Correct a bug in the illegal phase handler that caused us to drop down to narrow when handling the unexpected command phase case after 3rd party reset of a packetized device. Indicate the features, bugs, and flags set in the softc that are used to control firmware patch download when booting verbose. aic79xx.h: Add coalessing and HS_MAILBOX fields. Add per-softc variables for the stats "daemon". Add a debug option for interrupt coalessing activities. Add two new softc flags: o AHD_UPDATE_PEND_CMDS Run ahd_reset_cmds_pending() on the next unpause. o AHD_RUNNING_QOUTFIFO Used to catch recursion through ahd_run_qoutfifo(). aic79xx.reg: Correct register addresses related to the software timer and the DFDBCTL register. Add constants paramaterizing the software timer. Add scratch ram locations for storing interrupt coalessing tunables. Break INTMASK in SEQITNCTL out into INTMASK1 and INTMASK2. In at least the REV A, these are writable bits. We make use of that for a swtimer workaround in the sequencer. Since HS_MAILBOX autoclears, provide a sequencer variable to store its contents. Add SEQINT codes for handling task management completions. aic79xx.seq: Correct ignore wide residue processing check for a wide negotiation being in effect. We must be in the SCSI register window in order to access the negotiation table. Use the software timer and a commands completed count to implement interrupt coalessing. The command complete is deferred until either the maximum command threshold or a the expiration of a command deferral timer. If we have more SCBs to complete to the host (sitting in COMPLETE_SCB lists), always try to coaless them up to our coalessing limit. If coalessing is enabled, but we have fewer commands oustanting than the host's min coalessing limit, complete the command immediately. Add code to track the number of commands outstanding. Commands are outstanding from the time they are placed into the execution queue until the DMA to post completion is setup. Add a workaround for intvec_2 interrupts on the H2A4. In H2A4, the mode pointer is not saved for intvec2, but is restored on iret. This can lead to the restoration of a bogus mode ptr. Manually clear the intmask bits and do a normal return to compensate. We use intvec_2 to track interrupt coalessing timeouts. Since we cannot disable the swtimer's countdown, simply mask its interrupt once we no longer care about it firing. In idle_loop_cchan, update LOCAL_HS_MAILBOX everytime we are notified of an HS_MAILBOX update via the HS_MAILBOX_ACT bit in QOFF_CTLSTA. We have to use a local copy of persistant portions of the HS_MAILBOX as the mailbox auto-clears on any read. Move the test for the cfg4istat interrupt up an instruction to hopefully close a race between the next outgoing selection and our disabling of selections. Add a missing ret to the last instruction in load_overrun_buf. Add notifications to the host of task management completions as well as the completions for commands that completed successfully before their corresponding TMF could be sent. Hold a critical section during select-out processing until we have a fully identified connection. This removes a race condition with the legacy abort handler. Correct a few spelling errors in some comments. aic79xx_inline.h: Call ahd_reset_cmds_pending() in ahd_unpause if required. Update cmdcmplt interrupt statistics in our interrupt handler. Allow callers to ahd_send_scb() to set the task management function. aic79xx_pci.c: Disable SERR and pause the controller prior to performing our mmapped I/O test. The U320 controllers do not support "auto-access-pause". aic79xx_osm.c: Set the task management function now that ahd_send_scb() doesn't do it for us. We also perform a lun reset in response to BDR requests to packetized devices.
2003-01-20 20:17:35 +00:00
or INTCTL, SWTMINTMASK ret;
Upgrade to version 1.1 of the aic79xx U320 driver. aic79xx.c: o Remove redundant ahd_update_modes() call. o Correct panic in diagnostic should state corruption cause the SCB Id to be invalid during a selection timeout. o Add workaround for missing BUSFREEREV feature in Rev A silicon. o Corect formatting nits. o Use register pretty printing in more places. o Save and restore our SCB pointer when updating the waiting queue list for an "expected" LQ-out busfree. o In ahd_clear_intstat, deal with the missing autoclear in the CLRLQO* registers. o BE fixup in a diagnostic printf. o Make sure that we are in the proper mode before disabling selections in ahd_update_pending_scbs. o Add more diagnostics. o task_attribute_nonpkt_tag -> task_attribute: we don't need a nonpkt_tag field anymore for allowing all 512 SCBs to be used in non-packetized connections. o Negotiate HOLD_MCS to U320 devices. o Add a few additional mode assertions. o Restore the chip mode after clearing out the qinfifo so that code using ahd_abort_scbs sees a consistent mode. o Simplify the DMA engine shutdown routine prior to performing a bus reset. o Perform the sequencer restart after a chip reset prior to setting up our timer to poll for the reset to be complete. On some OSes, the timer could actually pre-empt us and order is important here. o Have our "reset poller" set the expected mode since there is no guarantee of what mode will be in force when we are called from the OS timer. o Save and restore the SCB pointer in ahd_dump_card_state(). This routine must not modify card state. o Ditto for ahd_dump_scbs(). aic79xx.h: o Add a few more chip bug definitions. o Align our tag on a 32bit boundary. aic79xx.reg: aic79xx.seq: o Start work on removing workarounds for Rev B. o Use a special location in scratch from for stroring our SCBPTR during legacy FIFO allocations. This corrects problems in mixed packetized/non-packetized configurations where calling into a FIFO task corrupted our SCBPTR. o Don't rely on DMA priority to guarantee that all data in our FIFOs will flush prior to a command completion notification going out of the command channel. We've never seen this assumption fail, but better safe than sorry. o Deal with missing BUSFREEREV feature in H2A. o Simplify disconnect list code now that the list will always have only a single entry. o Implement the AHD_REG_SLOW_SETTLE_BUG workaround. o Swith to using "REG_ISR" for local mode scratch during our ISR. o Add a missing jmp to the data_group_dma_loop after our data pointers have been re-initialized by the kernel. o Correct test in the bitbucket code so that we actually wait for the bitbucket to complete before signaling the kernel of the overrun condition. o Reposition pkt_saveptrs to avoid a jmp instruction. o Update a comment to reflect that the code now waits for a FIFO to drain prior to issuing a CLRCHN. aic79xx_inline.h: o Remove unused untagged queue handling code. o Don't attempt to htole64 what could be a 32bit value. aic79xx_pci.c: o Set additional bug flags for rev A chips.
2002-09-26 22:54:00 +00:00
BEGIN_CRITICAL;
fetch_new_scb_inprog:
test CCSCBCTL, ARRDONE jz return;
fetch_new_scb_done:
and CCSCBCTL, ~(CCARREN|CCSCBEN);
aic79xx.c: Implement the SCB_SILENT flag. This is useful for hushing up the driver during DV or other operations that we expect to cause transmission errors. The messages will still print if the SHOW_MASKED_ERRORS debug option is enabled. Save and restore the NEGOADDR address when setting new transfer settings. The sequencer performs lookups in the negotiation table too and it expects NEGOADDR to remain consistent across pause/unpause sessions. Consistently use "offset" instead of "period" to determine if we are running sync or not. Add a SHOW_MESSAGES diagnostic for when we assert ATN during message processing. Print out IU, QAS, and RTI features when showing transfer options. Limit the syncrate after all option conformance changes have taken place in ahd_devlimited_syncrate. Changes in options may change the final syncrate we accept. Keep a copy of the hs_mailbox in our softc so that we can perform read/modify/write operations on the hs_mailbox without having to pause the sequencer to read the last written value. Use the ENINT_COALESS flag in the hs_mailbox to toggle interrupt coalessing. Add entrypoints for enabling interrupt coalessing and setting both a timeout (how long to wait for commands to be coalessed) and a maximum commands to coaless value. Add a statistics timer that decides when to enable or disable interrupt coalessing based on load. Add a routine, ahd_reset_cmds_pending() which is used to update the CMDS_PENDING sequencer variable whenever error recovery compeltes SCBs without notifying the sequencer. Since ahd_reset_cmds_pending is called during ahd_unpause() only if we've aborted SCBs, its call to ahd_flush_qoutfifo should not cause recursion through ahd_run_qoutfifo(). A panic has been added to ensure that this recursion does not occur. In ahd_search_qinfifo, update the CMDS_PENDING sequencer variable directly. ahd_search_qinififo can be called in situations where using ahd_reset_cmds_pending() might cause recursion. Since we can safely determine the exact number to reduce CMDS_PENDING by in this scenario without running the qoutfifo, the manual update is sufficient. Clean up diagnostics. Add ahd_flush_qoutfifo() which will run the qoutfifo as well as complete any commands sitting on the sequencer's COMPLETE_SCB lists or the good status FIFO. Use this routine in several places that did similar things in an add-hoc, but incomplete, fashion. A call to this routine was also added to ahd_abort_scbs() to close a race. In ahd_pause_and_flushwork() only return once selections are safely disabled. Flush all completed commands via ahd_flush_qoutfifo(). Remove "Now packetized" diagnostic now that this information is incorperated into the actual negotiation messages that are displayed. When forcing renegotiation, don't clober the current ppr_options. Much of the driver uses this information to determine if we are currently packetized or not. Remove some stray spaces at column 1 in ahd_set_tags. When complaining about getting a host message loop request with no pending messages, print out the SCB_CONTROL register down on the card. Modify the ahd_sent_msg() routine to handle a search for an outgoing identify message. Use this to detect a msg reject on an identify message which typically indicates that the target thought we were packetized. Force a renegotiation in this case. In ahd_search_qinfifo(), wait more effectively for SCB DMA activities to cease. We also disable SCB fetch operations since we are about to change the qinfifo and any fetch in progress will likely be invalidated. In ahd_qinfifo_count(), fix the qinfifo empty case. In ahd_dump_card_state(), print out CCSCBCTL in the correct mode. If we are a narrow controller, don't set the current width to unknown when forcing a future negotiation. This just confuses the code into attempting a wide negotiation on a narrow bus. Add support for task management function completions. Modify ahd_handle_devreset so that it can handle lun resets in addition to target resets. Use ahd_handle_devreset for lun and target reset task management functions. Handle the abort task TMF race case better. We now wait until any current selections are over and then set the TMF back to zero. This should cause the sequencer to ignore the abort TMF completion should it occur. Correct a bug in the illegal phase handler that caused us to drop down to narrow when handling the unexpected command phase case after 3rd party reset of a packetized device. Indicate the features, bugs, and flags set in the softc that are used to control firmware patch download when booting verbose. aic79xx.h: Add coalessing and HS_MAILBOX fields. Add per-softc variables for the stats "daemon". Add a debug option for interrupt coalessing activities. Add two new softc flags: o AHD_UPDATE_PEND_CMDS Run ahd_reset_cmds_pending() on the next unpause. o AHD_RUNNING_QOUTFIFO Used to catch recursion through ahd_run_qoutfifo(). aic79xx.reg: Correct register addresses related to the software timer and the DFDBCTL register. Add constants paramaterizing the software timer. Add scratch ram locations for storing interrupt coalessing tunables. Break INTMASK in SEQITNCTL out into INTMASK1 and INTMASK2. In at least the REV A, these are writable bits. We make use of that for a swtimer workaround in the sequencer. Since HS_MAILBOX autoclears, provide a sequencer variable to store its contents. Add SEQINT codes for handling task management completions. aic79xx.seq: Correct ignore wide residue processing check for a wide negotiation being in effect. We must be in the SCSI register window in order to access the negotiation table. Use the software timer and a commands completed count to implement interrupt coalessing. The command complete is deferred until either the maximum command threshold or a the expiration of a command deferral timer. If we have more SCBs to complete to the host (sitting in COMPLETE_SCB lists), always try to coaless them up to our coalessing limit. If coalessing is enabled, but we have fewer commands oustanting than the host's min coalessing limit, complete the command immediately. Add code to track the number of commands outstanding. Commands are outstanding from the time they are placed into the execution queue until the DMA to post completion is setup. Add a workaround for intvec_2 interrupts on the H2A4. In H2A4, the mode pointer is not saved for intvec2, but is restored on iret. This can lead to the restoration of a bogus mode ptr. Manually clear the intmask bits and do a normal return to compensate. We use intvec_2 to track interrupt coalessing timeouts. Since we cannot disable the swtimer's countdown, simply mask its interrupt once we no longer care about it firing. In idle_loop_cchan, update LOCAL_HS_MAILBOX everytime we are notified of an HS_MAILBOX update via the HS_MAILBOX_ACT bit in QOFF_CTLSTA. We have to use a local copy of persistant portions of the HS_MAILBOX as the mailbox auto-clears on any read. Move the test for the cfg4istat interrupt up an instruction to hopefully close a race between the next outgoing selection and our disabling of selections. Add a missing ret to the last instruction in load_overrun_buf. Add notifications to the host of task management completions as well as the completions for commands that completed successfully before their corresponding TMF could be sent. Hold a critical section during select-out processing until we have a fully identified connection. This removes a race condition with the legacy abort handler. Correct a few spelling errors in some comments. aic79xx_inline.h: Call ahd_reset_cmds_pending() in ahd_unpause if required. Update cmdcmplt interrupt statistics in our interrupt handler. Allow callers to ahd_send_scb() to set the task management function. aic79xx_pci.c: Disable SERR and pause the controller prior to performing our mmapped I/O test. The U320 controllers do not support "auto-access-pause". aic79xx_osm.c: Set the task management function now that ahd_send_scb() doesn't do it for us. We also perform a lun reset in response to BDR requests to packetized devices.
2003-01-20 20:17:35 +00:00
clr A;
add CMDS_PENDING, 1;
adc CMDS_PENDING[1], A;
if ((ahd->bugs & AHD_PKT_LUN_BUG) != 0) {
/*
* "Short Luns" are not placed into outgoing LQ
* packets in the correct byte order. Use a full
* sized lun field instead and fill it with the
* one byte of lun information we support.
*/
mov SCB_PKT_LUN[6], SCB_LUN;
}
Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-05-04 00:20:07 +00:00
/*
* The FIFO use count field is shared with the
* tag set by the host so that our SCB dma engine
* knows the correct location to store the SCB.
* Set it to zero before processing the SCB.
*/
clr SCB_FIFO_USE_COUNT;
/* Update the next SCB address to download. */
bmov NEXT_QUEUED_SCB_ADDR, SCB_NEXT_SCB_BUSADDR, 4;
Correct a very rare case where command ordering could be compromised by a transaction performing a driver handled message sequence (an scb with the MK_MESSAGE flag set). SCBs that perform host managed messaging must always be at the head of their per-target selection queue so that the firmware knows to manually assert ATN if the current negotiation agreement is packetized. In the past we guaranteed this by queuing these SCBs separarately in the execution queue. This exposes the system to potential command reordering in two cases: 1) Another SCB for the same ITL nexus is queued that does not have the MK_MESSAGE flag set. This SCB will be queued to the per-target list which can be serviced before the MK_MESSAGE scb that preceeded it. 2) If the target cannot accept all of the commands in the per-target selection queue in one selection, the remainder is queued to the tail of the selection queues so as to effect round-robin scheduling. This could allow the MK_MESSAGE scb to be sent to the target before the requeued commands. This commit changes the firmware policy to defer queuing MK_MESSAGE SCBs into the selection queues until this can be done without affecting order. This means that the target's selection queue is either empty, or the last SCB on the execution queue is also a MK_MESSAGE SCB. During any wait, the firmware halts the download of new SCBs so only a single "holding location" is required. Luckily, MK_MESSAGE SCBs are rare and typically occur only during CAM's bus probe where only one command is outstanding at a time. However, during some recovery scenarios, the reordering *could* occur. aic79xx.c: Update ahd_search_qinfifo() and helper routines to search for pending MK_MESSAGE scbs and properly restitch the execution queue if either the MK_MESSAGE SCB is being aborted, or the MK_MESSAGE SCB can be queued due to the execution queue draining due to aborts. Enable LQOBUSFREE status to assert an interrupt. This should be redundant since a BUSFREE interrupt should always occur along with an LQOBUSFREE event, but on the Rev A, this doesn't seem to be guaranteed. When a PPR request is rejected when a previously existing packetized agreement is in place, assume that the target has been reset without our knowledge and revert to async/narrow transfers. This corrects two issues: the stale ENATNO setting that was used to send the PPR is cleared so the firmware is not confused by a future packetized selection with ATN asserted but no MK_MESSAGE flag in the SCB and it speeds up recovery by aborting any pending packetized transactions that by definition are now dead. When re-queueing SCBs after a failed negotiation attempt, ensure command ordering by freezing the device queue first. Traverse the list of pending SCBs rather than the whole SCB array on the controller when pushing MK_MESSAGE flag changes out to the controller. The original code was optimized for the aic7xxx controllers where there are fewer controller slots then pending SCBs and the firmware picks SCB slots. For the U320 controller, the hope is that we have fewer pending SCBs then the 512 slots on the controller. Enhance some diagnostics. Factor out some common code. aic79xx.h: Add prototype for new ahd_done_with_status() that is used to factor out some commone code. aic79xx.reg: Add definisions for the pending MK_MESSAGE SCB. aic79xx.seq: Defer MK_MESSAGE SCB queing to the execution queue so as to preserve command ordering. Re-arrange some of the selection processing code so the above change had no performance impact on the common code path. Close a few critical section holes. When entering a non-packetized phase, manually enable busfree interrupts, since the controller hardware does not do this automatically. aic79xx_inline.h: Enhance logging for queued SCBs. aic79xx_osm.c: Add new a new DDB ahd command, ahd_dump, which invokes the ahd_dump_card_state() routine on the unit specified with the ahd_sunit DDB command. aic79xx_pci.c: Turn on the BUSFREEREV bug for the Rev B. controller. This is required to close the busfree during non-packetized phase hole.
2004-08-04 17:55:34 +00:00
/*
* NULL out the SCB links since these fields
* occupy the same location as SCB_NEXT_SCB_BUSADDR.
*/
mvi SCB_NEXT[1], SCB_LIST_NULL;
mvi SCB_NEXT2[1], SCB_LIST_NULL;
/* Increment our position in the QINFIFO. */
mov NONE, SNSCB_QOFF;
Correct a very rare case where command ordering could be compromised by a transaction performing a driver handled message sequence (an scb with the MK_MESSAGE flag set). SCBs that perform host managed messaging must always be at the head of their per-target selection queue so that the firmware knows to manually assert ATN if the current negotiation agreement is packetized. In the past we guaranteed this by queuing these SCBs separarately in the execution queue. This exposes the system to potential command reordering in two cases: 1) Another SCB for the same ITL nexus is queued that does not have the MK_MESSAGE flag set. This SCB will be queued to the per-target list which can be serviced before the MK_MESSAGE scb that preceeded it. 2) If the target cannot accept all of the commands in the per-target selection queue in one selection, the remainder is queued to the tail of the selection queues so as to effect round-robin scheduling. This could allow the MK_MESSAGE scb to be sent to the target before the requeued commands. This commit changes the firmware policy to defer queuing MK_MESSAGE SCBs into the selection queues until this can be done without affecting order. This means that the target's selection queue is either empty, or the last SCB on the execution queue is also a MK_MESSAGE SCB. During any wait, the firmware halts the download of new SCBs so only a single "holding location" is required. Luckily, MK_MESSAGE SCBs are rare and typically occur only during CAM's bus probe where only one command is outstanding at a time. However, during some recovery scenarios, the reordering *could* occur. aic79xx.c: Update ahd_search_qinfifo() and helper routines to search for pending MK_MESSAGE scbs and properly restitch the execution queue if either the MK_MESSAGE SCB is being aborted, or the MK_MESSAGE SCB can be queued due to the execution queue draining due to aborts. Enable LQOBUSFREE status to assert an interrupt. This should be redundant since a BUSFREE interrupt should always occur along with an LQOBUSFREE event, but on the Rev A, this doesn't seem to be guaranteed. When a PPR request is rejected when a previously existing packetized agreement is in place, assume that the target has been reset without our knowledge and revert to async/narrow transfers. This corrects two issues: the stale ENATNO setting that was used to send the PPR is cleared so the firmware is not confused by a future packetized selection with ATN asserted but no MK_MESSAGE flag in the SCB and it speeds up recovery by aborting any pending packetized transactions that by definition are now dead. When re-queueing SCBs after a failed negotiation attempt, ensure command ordering by freezing the device queue first. Traverse the list of pending SCBs rather than the whole SCB array on the controller when pushing MK_MESSAGE flag changes out to the controller. The original code was optimized for the aic7xxx controllers where there are fewer controller slots then pending SCBs and the firmware picks SCB slots. For the U320 controller, the hope is that we have fewer pending SCBs then the 512 slots on the controller. Enhance some diagnostics. Factor out some common code. aic79xx.h: Add prototype for new ahd_done_with_status() that is used to factor out some commone code. aic79xx.reg: Add definisions for the pending MK_MESSAGE SCB. aic79xx.seq: Defer MK_MESSAGE SCB queing to the execution queue so as to preserve command ordering. Re-arrange some of the selection processing code so the above change had no performance impact on the common code path. Close a few critical section holes. When entering a non-packetized phase, manually enable busfree interrupts, since the controller hardware does not do this automatically. aic79xx_inline.h: Enhance logging for queued SCBs. aic79xx_osm.c: Add new a new DDB ahd command, ahd_dump, which invokes the ahd_dump_card_state() routine on the unit specified with the ahd_sunit DDB command. aic79xx_pci.c: Turn on the BUSFREEREV bug for the Rev B. controller. This is required to close the busfree during non-packetized phase hole.
2004-08-04 17:55:34 +00:00
/*
Correct a very rare case where command ordering could be compromised by a transaction performing a driver handled message sequence (an scb with the MK_MESSAGE flag set). SCBs that perform host managed messaging must always be at the head of their per-target selection queue so that the firmware knows to manually assert ATN if the current negotiation agreement is packetized. In the past we guaranteed this by queuing these SCBs separarately in the execution queue. This exposes the system to potential command reordering in two cases: 1) Another SCB for the same ITL nexus is queued that does not have the MK_MESSAGE flag set. This SCB will be queued to the per-target list which can be serviced before the MK_MESSAGE scb that preceeded it. 2) If the target cannot accept all of the commands in the per-target selection queue in one selection, the remainder is queued to the tail of the selection queues so as to effect round-robin scheduling. This could allow the MK_MESSAGE scb to be sent to the target before the requeued commands. This commit changes the firmware policy to defer queuing MK_MESSAGE SCBs into the selection queues until this can be done without affecting order. This means that the target's selection queue is either empty, or the last SCB on the execution queue is also a MK_MESSAGE SCB. During any wait, the firmware halts the download of new SCBs so only a single "holding location" is required. Luckily, MK_MESSAGE SCBs are rare and typically occur only during CAM's bus probe where only one command is outstanding at a time. However, during some recovery scenarios, the reordering *could* occur. aic79xx.c: Update ahd_search_qinfifo() and helper routines to search for pending MK_MESSAGE scbs and properly restitch the execution queue if either the MK_MESSAGE SCB is being aborted, or the MK_MESSAGE SCB can be queued due to the execution queue draining due to aborts. Enable LQOBUSFREE status to assert an interrupt. This should be redundant since a BUSFREE interrupt should always occur along with an LQOBUSFREE event, but on the Rev A, this doesn't seem to be guaranteed. When a PPR request is rejected when a previously existing packetized agreement is in place, assume that the target has been reset without our knowledge and revert to async/narrow transfers. This corrects two issues: the stale ENATNO setting that was used to send the PPR is cleared so the firmware is not confused by a future packetized selection with ATN asserted but no MK_MESSAGE flag in the SCB and it speeds up recovery by aborting any pending packetized transactions that by definition are now dead. When re-queueing SCBs after a failed negotiation attempt, ensure command ordering by freezing the device queue first. Traverse the list of pending SCBs rather than the whole SCB array on the controller when pushing MK_MESSAGE flag changes out to the controller. The original code was optimized for the aic7xxx controllers where there are fewer controller slots then pending SCBs and the firmware picks SCB slots. For the U320 controller, the hope is that we have fewer pending SCBs then the 512 slots on the controller. Enhance some diagnostics. Factor out some common code. aic79xx.h: Add prototype for new ahd_done_with_status() that is used to factor out some commone code. aic79xx.reg: Add definisions for the pending MK_MESSAGE SCB. aic79xx.seq: Defer MK_MESSAGE SCB queing to the execution queue so as to preserve command ordering. Re-arrange some of the selection processing code so the above change had no performance impact on the common code path. Close a few critical section holes. When entering a non-packetized phase, manually enable busfree interrupts, since the controller hardware does not do this automatically. aic79xx_inline.h: Enhance logging for queued SCBs. aic79xx_osm.c: Add new a new DDB ahd command, ahd_dump, which invokes the ahd_dump_card_state() routine on the unit specified with the ahd_sunit DDB command. aic79xx_pci.c: Turn on the BUSFREEREV bug for the Rev B. controller. This is required to close the busfree during non-packetized phase hole.
2004-08-04 17:55:34 +00:00
* Save SCBID of this SCB in REG0 since
* SCBPTR will be clobbered during target
* list updates. We also record the SCB's
* flags so that we can refer to them even
* after SCBPTR has been changed.
*/
bmov REG0, SCBPTR, 2;
mov A, SCB_CONTROL;
/*
* Find the tail SCB of the execution queue
* for this target.
*/
shr SINDEX, 3, SCB_SCSIID;
and SINDEX, ~0x1;
mvi SINDEX[1], (WAITING_SCB_TAILS >> 8);
bmov DINDEX, SINDEX, 2;
bmov SCBPTR, SINDIR, 2;
Correct a very rare case where command ordering could be compromised by a transaction performing a driver handled message sequence (an scb with the MK_MESSAGE flag set). SCBs that perform host managed messaging must always be at the head of their per-target selection queue so that the firmware knows to manually assert ATN if the current negotiation agreement is packetized. In the past we guaranteed this by queuing these SCBs separarately in the execution queue. This exposes the system to potential command reordering in two cases: 1) Another SCB for the same ITL nexus is queued that does not have the MK_MESSAGE flag set. This SCB will be queued to the per-target list which can be serviced before the MK_MESSAGE scb that preceeded it. 2) If the target cannot accept all of the commands in the per-target selection queue in one selection, the remainder is queued to the tail of the selection queues so as to effect round-robin scheduling. This could allow the MK_MESSAGE scb to be sent to the target before the requeued commands. This commit changes the firmware policy to defer queuing MK_MESSAGE SCBs into the selection queues until this can be done without affecting order. This means that the target's selection queue is either empty, or the last SCB on the execution queue is also a MK_MESSAGE SCB. During any wait, the firmware halts the download of new SCBs so only a single "holding location" is required. Luckily, MK_MESSAGE SCBs are rare and typically occur only during CAM's bus probe where only one command is outstanding at a time. However, during some recovery scenarios, the reordering *could* occur. aic79xx.c: Update ahd_search_qinfifo() and helper routines to search for pending MK_MESSAGE scbs and properly restitch the execution queue if either the MK_MESSAGE SCB is being aborted, or the MK_MESSAGE SCB can be queued due to the execution queue draining due to aborts. Enable LQOBUSFREE status to assert an interrupt. This should be redundant since a BUSFREE interrupt should always occur along with an LQOBUSFREE event, but on the Rev A, this doesn't seem to be guaranteed. When a PPR request is rejected when a previously existing packetized agreement is in place, assume that the target has been reset without our knowledge and revert to async/narrow transfers. This corrects two issues: the stale ENATNO setting that was used to send the PPR is cleared so the firmware is not confused by a future packetized selection with ATN asserted but no MK_MESSAGE flag in the SCB and it speeds up recovery by aborting any pending packetized transactions that by definition are now dead. When re-queueing SCBs after a failed negotiation attempt, ensure command ordering by freezing the device queue first. Traverse the list of pending SCBs rather than the whole SCB array on the controller when pushing MK_MESSAGE flag changes out to the controller. The original code was optimized for the aic7xxx controllers where there are fewer controller slots then pending SCBs and the firmware picks SCB slots. For the U320 controller, the hope is that we have fewer pending SCBs then the 512 slots on the controller. Enhance some diagnostics. Factor out some common code. aic79xx.h: Add prototype for new ahd_done_with_status() that is used to factor out some commone code. aic79xx.reg: Add definisions for the pending MK_MESSAGE SCB. aic79xx.seq: Defer MK_MESSAGE SCB queing to the execution queue so as to preserve command ordering. Re-arrange some of the selection processing code so the above change had no performance impact on the common code path. Close a few critical section holes. When entering a non-packetized phase, manually enable busfree interrupts, since the controller hardware does not do this automatically. aic79xx_inline.h: Enhance logging for queued SCBs. aic79xx_osm.c: Add new a new DDB ahd command, ahd_dump, which invokes the ahd_dump_card_state() routine on the unit specified with the ahd_sunit DDB command. aic79xx_pci.c: Turn on the BUSFREEREV bug for the Rev B. controller. This is required to close the busfree during non-packetized phase hole.
2004-08-04 17:55:34 +00:00
/*
* Update the tail to point to the new SCB.
*/
bmov DINDIR, REG0, 2;
Correct a very rare case where command ordering could be compromised by a transaction performing a driver handled message sequence (an scb with the MK_MESSAGE flag set). SCBs that perform host managed messaging must always be at the head of their per-target selection queue so that the firmware knows to manually assert ATN if the current negotiation agreement is packetized. In the past we guaranteed this by queuing these SCBs separarately in the execution queue. This exposes the system to potential command reordering in two cases: 1) Another SCB for the same ITL nexus is queued that does not have the MK_MESSAGE flag set. This SCB will be queued to the per-target list which can be serviced before the MK_MESSAGE scb that preceeded it. 2) If the target cannot accept all of the commands in the per-target selection queue in one selection, the remainder is queued to the tail of the selection queues so as to effect round-robin scheduling. This could allow the MK_MESSAGE scb to be sent to the target before the requeued commands. This commit changes the firmware policy to defer queuing MK_MESSAGE SCBs into the selection queues until this can be done without affecting order. This means that the target's selection queue is either empty, or the last SCB on the execution queue is also a MK_MESSAGE SCB. During any wait, the firmware halts the download of new SCBs so only a single "holding location" is required. Luckily, MK_MESSAGE SCBs are rare and typically occur only during CAM's bus probe where only one command is outstanding at a time. However, during some recovery scenarios, the reordering *could* occur. aic79xx.c: Update ahd_search_qinfifo() and helper routines to search for pending MK_MESSAGE scbs and properly restitch the execution queue if either the MK_MESSAGE SCB is being aborted, or the MK_MESSAGE SCB can be queued due to the execution queue draining due to aborts. Enable LQOBUSFREE status to assert an interrupt. This should be redundant since a BUSFREE interrupt should always occur along with an LQOBUSFREE event, but on the Rev A, this doesn't seem to be guaranteed. When a PPR request is rejected when a previously existing packetized agreement is in place, assume that the target has been reset without our knowledge and revert to async/narrow transfers. This corrects two issues: the stale ENATNO setting that was used to send the PPR is cleared so the firmware is not confused by a future packetized selection with ATN asserted but no MK_MESSAGE flag in the SCB and it speeds up recovery by aborting any pending packetized transactions that by definition are now dead. When re-queueing SCBs after a failed negotiation attempt, ensure command ordering by freezing the device queue first. Traverse the list of pending SCBs rather than the whole SCB array on the controller when pushing MK_MESSAGE flag changes out to the controller. The original code was optimized for the aic7xxx controllers where there are fewer controller slots then pending SCBs and the firmware picks SCB slots. For the U320 controller, the hope is that we have fewer pending SCBs then the 512 slots on the controller. Enhance some diagnostics. Factor out some common code. aic79xx.h: Add prototype for new ahd_done_with_status() that is used to factor out some commone code. aic79xx.reg: Add definisions for the pending MK_MESSAGE SCB. aic79xx.seq: Defer MK_MESSAGE SCB queing to the execution queue so as to preserve command ordering. Re-arrange some of the selection processing code so the above change had no performance impact on the common code path. Close a few critical section holes. When entering a non-packetized phase, manually enable busfree interrupts, since the controller hardware does not do this automatically. aic79xx_inline.h: Enhance logging for queued SCBs. aic79xx_osm.c: Add new a new DDB ahd command, ahd_dump, which invokes the ahd_dump_card_state() routine on the unit specified with the ahd_sunit DDB command. aic79xx_pci.c: Turn on the BUSFREEREV bug for the Rev B. controller. This is required to close the busfree during non-packetized phase hole.
2004-08-04 17:55:34 +00:00
/*
* If the queue was empty, queue this SCB as
* the first for this target.
*/
cmp SCBPTR[1], SCB_LIST_NULL je first_new_target_scb;
Correct a very rare case where command ordering could be compromised by a transaction performing a driver handled message sequence (an scb with the MK_MESSAGE flag set). SCBs that perform host managed messaging must always be at the head of their per-target selection queue so that the firmware knows to manually assert ATN if the current negotiation agreement is packetized. In the past we guaranteed this by queuing these SCBs separarately in the execution queue. This exposes the system to potential command reordering in two cases: 1) Another SCB for the same ITL nexus is queued that does not have the MK_MESSAGE flag set. This SCB will be queued to the per-target list which can be serviced before the MK_MESSAGE scb that preceeded it. 2) If the target cannot accept all of the commands in the per-target selection queue in one selection, the remainder is queued to the tail of the selection queues so as to effect round-robin scheduling. This could allow the MK_MESSAGE scb to be sent to the target before the requeued commands. This commit changes the firmware policy to defer queuing MK_MESSAGE SCBs into the selection queues until this can be done without affecting order. This means that the target's selection queue is either empty, or the last SCB on the execution queue is also a MK_MESSAGE SCB. During any wait, the firmware halts the download of new SCBs so only a single "holding location" is required. Luckily, MK_MESSAGE SCBs are rare and typically occur only during CAM's bus probe where only one command is outstanding at a time. However, during some recovery scenarios, the reordering *could* occur. aic79xx.c: Update ahd_search_qinfifo() and helper routines to search for pending MK_MESSAGE scbs and properly restitch the execution queue if either the MK_MESSAGE SCB is being aborted, or the MK_MESSAGE SCB can be queued due to the execution queue draining due to aborts. Enable LQOBUSFREE status to assert an interrupt. This should be redundant since a BUSFREE interrupt should always occur along with an LQOBUSFREE event, but on the Rev A, this doesn't seem to be guaranteed. When a PPR request is rejected when a previously existing packetized agreement is in place, assume that the target has been reset without our knowledge and revert to async/narrow transfers. This corrects two issues: the stale ENATNO setting that was used to send the PPR is cleared so the firmware is not confused by a future packetized selection with ATN asserted but no MK_MESSAGE flag in the SCB and it speeds up recovery by aborting any pending packetized transactions that by definition are now dead. When re-queueing SCBs after a failed negotiation attempt, ensure command ordering by freezing the device queue first. Traverse the list of pending SCBs rather than the whole SCB array on the controller when pushing MK_MESSAGE flag changes out to the controller. The original code was optimized for the aic7xxx controllers where there are fewer controller slots then pending SCBs and the firmware picks SCB slots. For the U320 controller, the hope is that we have fewer pending SCBs then the 512 slots on the controller. Enhance some diagnostics. Factor out some common code. aic79xx.h: Add prototype for new ahd_done_with_status() that is used to factor out some commone code. aic79xx.reg: Add definisions for the pending MK_MESSAGE SCB. aic79xx.seq: Defer MK_MESSAGE SCB queing to the execution queue so as to preserve command ordering. Re-arrange some of the selection processing code so the above change had no performance impact on the common code path. Close a few critical section holes. When entering a non-packetized phase, manually enable busfree interrupts, since the controller hardware does not do this automatically. aic79xx_inline.h: Enhance logging for queued SCBs. aic79xx_osm.c: Add new a new DDB ahd command, ahd_dump, which invokes the ahd_dump_card_state() routine on the unit specified with the ahd_sunit DDB command. aic79xx_pci.c: Turn on the BUSFREEREV bug for the Rev B. controller. This is required to close the busfree during non-packetized phase hole.
2004-08-04 17:55:34 +00:00
/*
* SCBs that want to send messages must always be
* at the head of their per-target queue so that
* ATN can be asserted even if the current
* negotiation agreement is packetized. If the
* target queue is empty, the SCB can be queued
* immediately. If the queue is not empty, we must
* wait for it to empty before entering this SCB
* into the waiting for selection queue. Otherwise
* our batching and round-robin selection scheme
* could allow commands to be queued out of order.
* To simplify the implementation, we stop pulling
* new commands from the host until the MK_MESSAGE
* SCB can be queued to the waiting for selection
* list.
*/
test A, MK_MESSAGE jz batch_scb;
/*
* If the last SCB is also a MK_MESSAGE SCB, then
* order is preserved even if we batch.
*/
test SCB_CONTROL, MK_MESSAGE jz batch_scb;
/*
* Defer this SCB and stop fetching new SCBs until
* it can be queued. Since the SCB_SCSIID of the
* tail SCB must be the same as that of the newly
* queued SCB, there is no need to restore the SCBID
* here.
*/
or SEQ_FLAGS2, PENDING_MK_MESSAGE;
bmov MK_MESSAGE_SCB, REG0, 2;
mov MK_MESSAGE_SCSIID, SCB_SCSIID ret;
batch_scb:
/*
* Otherwise just update the previous tail SCB to
* point to the new tail.
*/
bmov SCB_NEXT, REG0, 2 ret;
Correct a very rare case where command ordering could be compromised by a transaction performing a driver handled message sequence (an scb with the MK_MESSAGE flag set). SCBs that perform host managed messaging must always be at the head of their per-target selection queue so that the firmware knows to manually assert ATN if the current negotiation agreement is packetized. In the past we guaranteed this by queuing these SCBs separarately in the execution queue. This exposes the system to potential command reordering in two cases: 1) Another SCB for the same ITL nexus is queued that does not have the MK_MESSAGE flag set. This SCB will be queued to the per-target list which can be serviced before the MK_MESSAGE scb that preceeded it. 2) If the target cannot accept all of the commands in the per-target selection queue in one selection, the remainder is queued to the tail of the selection queues so as to effect round-robin scheduling. This could allow the MK_MESSAGE scb to be sent to the target before the requeued commands. This commit changes the firmware policy to defer queuing MK_MESSAGE SCBs into the selection queues until this can be done without affecting order. This means that the target's selection queue is either empty, or the last SCB on the execution queue is also a MK_MESSAGE SCB. During any wait, the firmware halts the download of new SCBs so only a single "holding location" is required. Luckily, MK_MESSAGE SCBs are rare and typically occur only during CAM's bus probe where only one command is outstanding at a time. However, during some recovery scenarios, the reordering *could* occur. aic79xx.c: Update ahd_search_qinfifo() and helper routines to search for pending MK_MESSAGE scbs and properly restitch the execution queue if either the MK_MESSAGE SCB is being aborted, or the MK_MESSAGE SCB can be queued due to the execution queue draining due to aborts. Enable LQOBUSFREE status to assert an interrupt. This should be redundant since a BUSFREE interrupt should always occur along with an LQOBUSFREE event, but on the Rev A, this doesn't seem to be guaranteed. When a PPR request is rejected when a previously existing packetized agreement is in place, assume that the target has been reset without our knowledge and revert to async/narrow transfers. This corrects two issues: the stale ENATNO setting that was used to send the PPR is cleared so the firmware is not confused by a future packetized selection with ATN asserted but no MK_MESSAGE flag in the SCB and it speeds up recovery by aborting any pending packetized transactions that by definition are now dead. When re-queueing SCBs after a failed negotiation attempt, ensure command ordering by freezing the device queue first. Traverse the list of pending SCBs rather than the whole SCB array on the controller when pushing MK_MESSAGE flag changes out to the controller. The original code was optimized for the aic7xxx controllers where there are fewer controller slots then pending SCBs and the firmware picks SCB slots. For the U320 controller, the hope is that we have fewer pending SCBs then the 512 slots on the controller. Enhance some diagnostics. Factor out some common code. aic79xx.h: Add prototype for new ahd_done_with_status() that is used to factor out some commone code. aic79xx.reg: Add definisions for the pending MK_MESSAGE SCB. aic79xx.seq: Defer MK_MESSAGE SCB queing to the execution queue so as to preserve command ordering. Re-arrange some of the selection processing code so the above change had no performance impact on the common code path. Close a few critical section holes. When entering a non-packetized phase, manually enable busfree interrupts, since the controller hardware does not do this automatically. aic79xx_inline.h: Enhance logging for queued SCBs. aic79xx_osm.c: Add new a new DDB ahd command, ahd_dump, which invokes the ahd_dump_card_state() routine on the unit specified with the ahd_sunit DDB command. aic79xx_pci.c: Turn on the BUSFREEREV bug for the Rev B. controller. This is required to close the busfree during non-packetized phase hole.
2004-08-04 17:55:34 +00:00
first_new_target_scb:
Correct a very rare case where command ordering could be compromised by a transaction performing a driver handled message sequence (an scb with the MK_MESSAGE flag set). SCBs that perform host managed messaging must always be at the head of their per-target selection queue so that the firmware knows to manually assert ATN if the current negotiation agreement is packetized. In the past we guaranteed this by queuing these SCBs separarately in the execution queue. This exposes the system to potential command reordering in two cases: 1) Another SCB for the same ITL nexus is queued that does not have the MK_MESSAGE flag set. This SCB will be queued to the per-target list which can be serviced before the MK_MESSAGE scb that preceeded it. 2) If the target cannot accept all of the commands in the per-target selection queue in one selection, the remainder is queued to the tail of the selection queues so as to effect round-robin scheduling. This could allow the MK_MESSAGE scb to be sent to the target before the requeued commands. This commit changes the firmware policy to defer queuing MK_MESSAGE SCBs into the selection queues until this can be done without affecting order. This means that the target's selection queue is either empty, or the last SCB on the execution queue is also a MK_MESSAGE SCB. During any wait, the firmware halts the download of new SCBs so only a single "holding location" is required. Luckily, MK_MESSAGE SCBs are rare and typically occur only during CAM's bus probe where only one command is outstanding at a time. However, during some recovery scenarios, the reordering *could* occur. aic79xx.c: Update ahd_search_qinfifo() and helper routines to search for pending MK_MESSAGE scbs and properly restitch the execution queue if either the MK_MESSAGE SCB is being aborted, or the MK_MESSAGE SCB can be queued due to the execution queue draining due to aborts. Enable LQOBUSFREE status to assert an interrupt. This should be redundant since a BUSFREE interrupt should always occur along with an LQOBUSFREE event, but on the Rev A, this doesn't seem to be guaranteed. When a PPR request is rejected when a previously existing packetized agreement is in place, assume that the target has been reset without our knowledge and revert to async/narrow transfers. This corrects two issues: the stale ENATNO setting that was used to send the PPR is cleared so the firmware is not confused by a future packetized selection with ATN asserted but no MK_MESSAGE flag in the SCB and it speeds up recovery by aborting any pending packetized transactions that by definition are now dead. When re-queueing SCBs after a failed negotiation attempt, ensure command ordering by freezing the device queue first. Traverse the list of pending SCBs rather than the whole SCB array on the controller when pushing MK_MESSAGE flag changes out to the controller. The original code was optimized for the aic7xxx controllers where there are fewer controller slots then pending SCBs and the firmware picks SCB slots. For the U320 controller, the hope is that we have fewer pending SCBs then the 512 slots on the controller. Enhance some diagnostics. Factor out some common code. aic79xx.h: Add prototype for new ahd_done_with_status() that is used to factor out some commone code. aic79xx.reg: Add definisions for the pending MK_MESSAGE SCB. aic79xx.seq: Defer MK_MESSAGE SCB queing to the execution queue so as to preserve command ordering. Re-arrange some of the selection processing code so the above change had no performance impact on the common code path. Close a few critical section holes. When entering a non-packetized phase, manually enable busfree interrupts, since the controller hardware does not do this automatically. aic79xx_inline.h: Enhance logging for queued SCBs. aic79xx_osm.c: Add new a new DDB ahd command, ahd_dump, which invokes the ahd_dump_card_state() routine on the unit specified with the ahd_sunit DDB command. aic79xx_pci.c: Turn on the BUSFREEREV bug for the Rev B. controller. This is required to close the busfree during non-packetized phase hole.
2004-08-04 17:55:34 +00:00
/*
* Append SCB to the tail of the waiting for
* selection list.
*/
cmp WAITING_TID_HEAD[1], SCB_LIST_NULL je first_new_scb;
bmov SCBPTR, WAITING_TID_TAIL, 2;
bmov SCB_NEXT2, REG0, 2;
bmov WAITING_TID_TAIL, REG0, 2 ret;
first_new_scb:
Correct a very rare case where command ordering could be compromised by a transaction performing a driver handled message sequence (an scb with the MK_MESSAGE flag set). SCBs that perform host managed messaging must always be at the head of their per-target selection queue so that the firmware knows to manually assert ATN if the current negotiation agreement is packetized. In the past we guaranteed this by queuing these SCBs separarately in the execution queue. This exposes the system to potential command reordering in two cases: 1) Another SCB for the same ITL nexus is queued that does not have the MK_MESSAGE flag set. This SCB will be queued to the per-target list which can be serviced before the MK_MESSAGE scb that preceeded it. 2) If the target cannot accept all of the commands in the per-target selection queue in one selection, the remainder is queued to the tail of the selection queues so as to effect round-robin scheduling. This could allow the MK_MESSAGE scb to be sent to the target before the requeued commands. This commit changes the firmware policy to defer queuing MK_MESSAGE SCBs into the selection queues until this can be done without affecting order. This means that the target's selection queue is either empty, or the last SCB on the execution queue is also a MK_MESSAGE SCB. During any wait, the firmware halts the download of new SCBs so only a single "holding location" is required. Luckily, MK_MESSAGE SCBs are rare and typically occur only during CAM's bus probe where only one command is outstanding at a time. However, during some recovery scenarios, the reordering *could* occur. aic79xx.c: Update ahd_search_qinfifo() and helper routines to search for pending MK_MESSAGE scbs and properly restitch the execution queue if either the MK_MESSAGE SCB is being aborted, or the MK_MESSAGE SCB can be queued due to the execution queue draining due to aborts. Enable LQOBUSFREE status to assert an interrupt. This should be redundant since a BUSFREE interrupt should always occur along with an LQOBUSFREE event, but on the Rev A, this doesn't seem to be guaranteed. When a PPR request is rejected when a previously existing packetized agreement is in place, assume that the target has been reset without our knowledge and revert to async/narrow transfers. This corrects two issues: the stale ENATNO setting that was used to send the PPR is cleared so the firmware is not confused by a future packetized selection with ATN asserted but no MK_MESSAGE flag in the SCB and it speeds up recovery by aborting any pending packetized transactions that by definition are now dead. When re-queueing SCBs after a failed negotiation attempt, ensure command ordering by freezing the device queue first. Traverse the list of pending SCBs rather than the whole SCB array on the controller when pushing MK_MESSAGE flag changes out to the controller. The original code was optimized for the aic7xxx controllers where there are fewer controller slots then pending SCBs and the firmware picks SCB slots. For the U320 controller, the hope is that we have fewer pending SCBs then the 512 slots on the controller. Enhance some diagnostics. Factor out some common code. aic79xx.h: Add prototype for new ahd_done_with_status() that is used to factor out some commone code. aic79xx.reg: Add definisions for the pending MK_MESSAGE SCB. aic79xx.seq: Defer MK_MESSAGE SCB queing to the execution queue so as to preserve command ordering. Re-arrange some of the selection processing code so the above change had no performance impact on the common code path. Close a few critical section holes. When entering a non-packetized phase, manually enable busfree interrupts, since the controller hardware does not do this automatically. aic79xx_inline.h: Enhance logging for queued SCBs. aic79xx_osm.c: Add new a new DDB ahd command, ahd_dump, which invokes the ahd_dump_card_state() routine on the unit specified with the ahd_sunit DDB command. aic79xx_pci.c: Turn on the BUSFREEREV bug for the Rev B. controller. This is required to close the busfree during non-packetized phase hole.
2004-08-04 17:55:34 +00:00
/*
* Whole list is empty, so the head of
* the list must be initialized too.
*/
bmov WAITING_TID_HEAD, REG0, 2;
bmov WAITING_TID_TAIL, REG0, 2 ret;
Upgrade to version 1.1 of the aic79xx U320 driver. aic79xx.c: o Remove redundant ahd_update_modes() call. o Correct panic in diagnostic should state corruption cause the SCB Id to be invalid during a selection timeout. o Add workaround for missing BUSFREEREV feature in Rev A silicon. o Corect formatting nits. o Use register pretty printing in more places. o Save and restore our SCB pointer when updating the waiting queue list for an "expected" LQ-out busfree. o In ahd_clear_intstat, deal with the missing autoclear in the CLRLQO* registers. o BE fixup in a diagnostic printf. o Make sure that we are in the proper mode before disabling selections in ahd_update_pending_scbs. o Add more diagnostics. o task_attribute_nonpkt_tag -> task_attribute: we don't need a nonpkt_tag field anymore for allowing all 512 SCBs to be used in non-packetized connections. o Negotiate HOLD_MCS to U320 devices. o Add a few additional mode assertions. o Restore the chip mode after clearing out the qinfifo so that code using ahd_abort_scbs sees a consistent mode. o Simplify the DMA engine shutdown routine prior to performing a bus reset. o Perform the sequencer restart after a chip reset prior to setting up our timer to poll for the reset to be complete. On some OSes, the timer could actually pre-empt us and order is important here. o Have our "reset poller" set the expected mode since there is no guarantee of what mode will be in force when we are called from the OS timer. o Save and restore the SCB pointer in ahd_dump_card_state(). This routine must not modify card state. o Ditto for ahd_dump_scbs(). aic79xx.h: o Add a few more chip bug definitions. o Align our tag on a 32bit boundary. aic79xx.reg: aic79xx.seq: o Start work on removing workarounds for Rev B. o Use a special location in scratch from for stroring our SCBPTR during legacy FIFO allocations. This corrects problems in mixed packetized/non-packetized configurations where calling into a FIFO task corrupted our SCBPTR. o Don't rely on DMA priority to guarantee that all data in our FIFOs will flush prior to a command completion notification going out of the command channel. We've never seen this assumption fail, but better safe than sorry. o Deal with missing BUSFREEREV feature in H2A. o Simplify disconnect list code now that the list will always have only a single entry. o Implement the AHD_REG_SLOW_SETTLE_BUG workaround. o Swith to using "REG_ISR" for local mode scratch during our ISR. o Add a missing jmp to the data_group_dma_loop after our data pointers have been re-initialized by the kernel. o Correct test in the bitbucket code so that we actually wait for the bitbucket to complete before signaling the kernel of the overrun condition. o Reposition pkt_saveptrs to avoid a jmp instruction. o Update a comment to reflect that the code now waits for a FIFO to drain prior to issuing a CLRCHN. aic79xx_inline.h: o Remove unused untagged queue handling code. o Don't attempt to htole64 what could be a 32bit value. aic79xx_pci.c: o Set additional bug flags for rev A chips.
2002-09-26 22:54:00 +00:00
END_CRITICAL;
scbdma_idle:
/*
Correct a very rare case where command ordering could be compromised by a transaction performing a driver handled message sequence (an scb with the MK_MESSAGE flag set). SCBs that perform host managed messaging must always be at the head of their per-target selection queue so that the firmware knows to manually assert ATN if the current negotiation agreement is packetized. In the past we guaranteed this by queuing these SCBs separarately in the execution queue. This exposes the system to potential command reordering in two cases: 1) Another SCB for the same ITL nexus is queued that does not have the MK_MESSAGE flag set. This SCB will be queued to the per-target list which can be serviced before the MK_MESSAGE scb that preceeded it. 2) If the target cannot accept all of the commands in the per-target selection queue in one selection, the remainder is queued to the tail of the selection queues so as to effect round-robin scheduling. This could allow the MK_MESSAGE scb to be sent to the target before the requeued commands. This commit changes the firmware policy to defer queuing MK_MESSAGE SCBs into the selection queues until this can be done without affecting order. This means that the target's selection queue is either empty, or the last SCB on the execution queue is also a MK_MESSAGE SCB. During any wait, the firmware halts the download of new SCBs so only a single "holding location" is required. Luckily, MK_MESSAGE SCBs are rare and typically occur only during CAM's bus probe where only one command is outstanding at a time. However, during some recovery scenarios, the reordering *could* occur. aic79xx.c: Update ahd_search_qinfifo() and helper routines to search for pending MK_MESSAGE scbs and properly restitch the execution queue if either the MK_MESSAGE SCB is being aborted, or the MK_MESSAGE SCB can be queued due to the execution queue draining due to aborts. Enable LQOBUSFREE status to assert an interrupt. This should be redundant since a BUSFREE interrupt should always occur along with an LQOBUSFREE event, but on the Rev A, this doesn't seem to be guaranteed. When a PPR request is rejected when a previously existing packetized agreement is in place, assume that the target has been reset without our knowledge and revert to async/narrow transfers. This corrects two issues: the stale ENATNO setting that was used to send the PPR is cleared so the firmware is not confused by a future packetized selection with ATN asserted but no MK_MESSAGE flag in the SCB and it speeds up recovery by aborting any pending packetized transactions that by definition are now dead. When re-queueing SCBs after a failed negotiation attempt, ensure command ordering by freezing the device queue first. Traverse the list of pending SCBs rather than the whole SCB array on the controller when pushing MK_MESSAGE flag changes out to the controller. The original code was optimized for the aic7xxx controllers where there are fewer controller slots then pending SCBs and the firmware picks SCB slots. For the U320 controller, the hope is that we have fewer pending SCBs then the 512 slots on the controller. Enhance some diagnostics. Factor out some common code. aic79xx.h: Add prototype for new ahd_done_with_status() that is used to factor out some commone code. aic79xx.reg: Add definisions for the pending MK_MESSAGE SCB. aic79xx.seq: Defer MK_MESSAGE SCB queing to the execution queue so as to preserve command ordering. Re-arrange some of the selection processing code so the above change had no performance impact on the common code path. Close a few critical section holes. When entering a non-packetized phase, manually enable busfree interrupts, since the controller hardware does not do this automatically. aic79xx_inline.h: Enhance logging for queued SCBs. aic79xx_osm.c: Add new a new DDB ahd command, ahd_dump, which invokes the ahd_dump_card_state() routine on the unit specified with the ahd_sunit DDB command. aic79xx_pci.c: Turn on the BUSFREEREV bug for the Rev B. controller. This is required to close the busfree during non-packetized phase hole.
2004-08-04 17:55:34 +00:00
* Don't bother downloading new SCBs to execute
* if select-outs are currently frozen or we have
* a MK_MESSAGE SCB waiting to enter the queue.
*/
Correct a very rare case where command ordering could be compromised by a transaction performing a driver handled message sequence (an scb with the MK_MESSAGE flag set). SCBs that perform host managed messaging must always be at the head of their per-target selection queue so that the firmware knows to manually assert ATN if the current negotiation agreement is packetized. In the past we guaranteed this by queuing these SCBs separarately in the execution queue. This exposes the system to potential command reordering in two cases: 1) Another SCB for the same ITL nexus is queued that does not have the MK_MESSAGE flag set. This SCB will be queued to the per-target list which can be serviced before the MK_MESSAGE scb that preceeded it. 2) If the target cannot accept all of the commands in the per-target selection queue in one selection, the remainder is queued to the tail of the selection queues so as to effect round-robin scheduling. This could allow the MK_MESSAGE scb to be sent to the target before the requeued commands. This commit changes the firmware policy to defer queuing MK_MESSAGE SCBs into the selection queues until this can be done without affecting order. This means that the target's selection queue is either empty, or the last SCB on the execution queue is also a MK_MESSAGE SCB. During any wait, the firmware halts the download of new SCBs so only a single "holding location" is required. Luckily, MK_MESSAGE SCBs are rare and typically occur only during CAM's bus probe where only one command is outstanding at a time. However, during some recovery scenarios, the reordering *could* occur. aic79xx.c: Update ahd_search_qinfifo() and helper routines to search for pending MK_MESSAGE scbs and properly restitch the execution queue if either the MK_MESSAGE SCB is being aborted, or the MK_MESSAGE SCB can be queued due to the execution queue draining due to aborts. Enable LQOBUSFREE status to assert an interrupt. This should be redundant since a BUSFREE interrupt should always occur along with an LQOBUSFREE event, but on the Rev A, this doesn't seem to be guaranteed. When a PPR request is rejected when a previously existing packetized agreement is in place, assume that the target has been reset without our knowledge and revert to async/narrow transfers. This corrects two issues: the stale ENATNO setting that was used to send the PPR is cleared so the firmware is not confused by a future packetized selection with ATN asserted but no MK_MESSAGE flag in the SCB and it speeds up recovery by aborting any pending packetized transactions that by definition are now dead. When re-queueing SCBs after a failed negotiation attempt, ensure command ordering by freezing the device queue first. Traverse the list of pending SCBs rather than the whole SCB array on the controller when pushing MK_MESSAGE flag changes out to the controller. The original code was optimized for the aic7xxx controllers where there are fewer controller slots then pending SCBs and the firmware picks SCB slots. For the U320 controller, the hope is that we have fewer pending SCBs then the 512 slots on the controller. Enhance some diagnostics. Factor out some common code. aic79xx.h: Add prototype for new ahd_done_with_status() that is used to factor out some commone code. aic79xx.reg: Add definisions for the pending MK_MESSAGE SCB. aic79xx.seq: Defer MK_MESSAGE SCB queing to the execution queue so as to preserve command ordering. Re-arrange some of the selection processing code so the above change had no performance impact on the common code path. Close a few critical section holes. When entering a non-packetized phase, manually enable busfree interrupts, since the controller hardware does not do this automatically. aic79xx_inline.h: Enhance logging for queued SCBs. aic79xx_osm.c: Add new a new DDB ahd command, ahd_dump, which invokes the ahd_dump_card_state() routine on the unit specified with the ahd_sunit DDB command. aic79xx_pci.c: Turn on the BUSFREEREV bug for the Rev B. controller. This is required to close the busfree during non-packetized phase hole.
2004-08-04 17:55:34 +00:00
test SEQ_FLAGS2, SELECTOUT_QFROZEN|PENDING_MK_MESSAGE
jnz scbdma_no_new_scbs;
BEGIN_CRITICAL;
test QOFF_CTLSTA, NEW_SCB_AVAIL jnz fetch_new_scb;
Correct a very rare case where command ordering could be compromised by a transaction performing a driver handled message sequence (an scb with the MK_MESSAGE flag set). SCBs that perform host managed messaging must always be at the head of their per-target selection queue so that the firmware knows to manually assert ATN if the current negotiation agreement is packetized. In the past we guaranteed this by queuing these SCBs separarately in the execution queue. This exposes the system to potential command reordering in two cases: 1) Another SCB for the same ITL nexus is queued that does not have the MK_MESSAGE flag set. This SCB will be queued to the per-target list which can be serviced before the MK_MESSAGE scb that preceeded it. 2) If the target cannot accept all of the commands in the per-target selection queue in one selection, the remainder is queued to the tail of the selection queues so as to effect round-robin scheduling. This could allow the MK_MESSAGE scb to be sent to the target before the requeued commands. This commit changes the firmware policy to defer queuing MK_MESSAGE SCBs into the selection queues until this can be done without affecting order. This means that the target's selection queue is either empty, or the last SCB on the execution queue is also a MK_MESSAGE SCB. During any wait, the firmware halts the download of new SCBs so only a single "holding location" is required. Luckily, MK_MESSAGE SCBs are rare and typically occur only during CAM's bus probe where only one command is outstanding at a time. However, during some recovery scenarios, the reordering *could* occur. aic79xx.c: Update ahd_search_qinfifo() and helper routines to search for pending MK_MESSAGE scbs and properly restitch the execution queue if either the MK_MESSAGE SCB is being aborted, or the MK_MESSAGE SCB can be queued due to the execution queue draining due to aborts. Enable LQOBUSFREE status to assert an interrupt. This should be redundant since a BUSFREE interrupt should always occur along with an LQOBUSFREE event, but on the Rev A, this doesn't seem to be guaranteed. When a PPR request is rejected when a previously existing packetized agreement is in place, assume that the target has been reset without our knowledge and revert to async/narrow transfers. This corrects two issues: the stale ENATNO setting that was used to send the PPR is cleared so the firmware is not confused by a future packetized selection with ATN asserted but no MK_MESSAGE flag in the SCB and it speeds up recovery by aborting any pending packetized transactions that by definition are now dead. When re-queueing SCBs after a failed negotiation attempt, ensure command ordering by freezing the device queue first. Traverse the list of pending SCBs rather than the whole SCB array on the controller when pushing MK_MESSAGE flag changes out to the controller. The original code was optimized for the aic7xxx controllers where there are fewer controller slots then pending SCBs and the firmware picks SCB slots. For the U320 controller, the hope is that we have fewer pending SCBs then the 512 slots on the controller. Enhance some diagnostics. Factor out some common code. aic79xx.h: Add prototype for new ahd_done_with_status() that is used to factor out some commone code. aic79xx.reg: Add definisions for the pending MK_MESSAGE SCB. aic79xx.seq: Defer MK_MESSAGE SCB queing to the execution queue so as to preserve command ordering. Re-arrange some of the selection processing code so the above change had no performance impact on the common code path. Close a few critical section holes. When entering a non-packetized phase, manually enable busfree interrupts, since the controller hardware does not do this automatically. aic79xx_inline.h: Enhance logging for queued SCBs. aic79xx_osm.c: Add new a new DDB ahd command, ahd_dump, which invokes the ahd_dump_card_state() routine on the unit specified with the ahd_sunit DDB command. aic79xx_pci.c: Turn on the BUSFREEREV bug for the Rev B. controller. This is required to close the busfree during non-packetized phase hole.
2004-08-04 17:55:34 +00:00
scbdma_no_new_scbs:
cmp COMPLETE_DMA_SCB_HEAD[1], SCB_LIST_NULL jne dma_complete_scb;
cmp COMPLETE_SCB_HEAD[1], SCB_LIST_NULL je return;
/* FALLTHROUGH */
fill_qoutfifo:
/*
* Keep track of the SCBs we are dmaing just
* in case the DMA fails or is aborted.
*/
bmov COMPLETE_SCB_DMAINPROG_HEAD, COMPLETE_SCB_HEAD, 2;
mvi CCSCBCTL, CCSCBRESET;
bmov SCBHADDR, QOUTFIFO_NEXT_ADDR, 4;
aic79xx.c: Allow 500us between pauses in ahd_pause_and_flushwork(). The maximum we will wait is now 500ms. In the same routine, remove any attempt to clear ENSELO. Let the firmware do it once the current selection has completed. This avoids some race conditions having to do with non-packetized completions and the auto-clearing of ENSELO on packetized completions. Also avoid attempts to clear critical sections when interrups are pending. We are going to loop again anyway, so clearing critical sections is a waste of time. It also may not be possible to clear a critical section if the source of the interrupt was a SEQINT. aic79xx_pci.c: Use the Generic 9005 mask when looking for generic 7901B parts. This allows the driver to attach to 7901B parts on motherboards using a non-Adaptec subvendor ID. aic79xx_inline.h: Test for the SCBRAM_RD_BUG against the bugs field, not the flags field in the softc. aic79xx.c: Cancel pending transactions on devices that respond with a selection timeout. This decreases the duration of timeout recovery when a device disappears. aic79xx.c: Don't bother forcing renegotiation on a selection timeout now that we use the device reset handler to abort any pending commands on the target. The device reset handler already takes us down to async narrow and forces a renegotiation. In the device reset handlers, only send a BDR sent async event if the status is not CAM_SEL_TIMEOUT. This avoids sending this event in the selection timeout case aic79xx.c: Modify the Core timeout handler to verify that another command has the potential to timeout before passing off a command timeout as due to some other command. This safety measure is added in response to a timeout recovery failure on H2B where it appears that incoming reselection status was lost during a drive pull test. In that case, the recovery handler continued to wait for the command that was active on the bus indefinetly. While the root cause of the above issue is still being determined seems a prudent safeguard. aic79xx_pci.c: Add a specific probe entry for the Dell OEM 39320(B). aic79xx.c: aic79xx.h: aic79xx.reg: aic79xx.seq: Modify the aic79xx firmware to never cross a cacheline or ADB boundary when DMA'ing completion entries to the host. In PCI mode, at least in 32/33 configurations, the SCB DMA engine may lose its place in the data-stream should the target force a retry on something other than an 8byte aligned boundary. In PCI-X mode, we do this to avoid split transactions since many chipsets seem to be unable to format proper split completions to continue the data transfer. The above change allows us to drop our completion entries from 8 bytes to 4. We were using 8 byte entries to ensure that PCI retries could only occur on an 8byte aligned boundary. Now that the sequencer guarantees this by splitting up completions, we can safely drop the size to 4 bytes (2 byte tag, one byte SG_RESID, one byte pad). Both the split-completion and PCI retry problems only show up under high tag load when interrupt coalescing is being especially effective. The switch from a 2byte completion entry to an 8 byte entry to solve the PCI problem increased the chance of incurring a split in PCI-X mode when multiple transactions were completed at once. Dropping the completion size to 4 bytes also means that we can complete more commands in a single DMA (128byte FIFO -> 32 commands instead of 16). aic79xx.c: Modify the SCSIINT handler to defer clearing sequencer critical sections to the individual interrupt handlers. This allows us to immediately disable any outgoing selections in the case of an unexpected busfree so we don't inadvertantly clear ENSELO *after* a new selection has started. Doing so may cause the sequencer to miss a successful selection. In ahd_update_pending_scbs(), only clear ENSELO if the bus is currently busy and a selection is not already in progress or the sequencer has yet to handle a pending selection. While we want to ensure that the selection for the SCB at the head of the selection queue is restarted so that any change in negotiation request can take effect, we can't clobber pending selection state without confusing the sequencer into missing a selection.
2004-05-11 20:46:05 +00:00
mov A, QOUTFIFO_NEXT_ADDR;
bmov SCBPTR, COMPLETE_SCB_HEAD, 2;
fill_qoutfifo_loop:
aic79xx.c: aic79xx.seq: Convert the COMPLETE_DMA_SCB list to an "stailq". This allows us to safely keep the SCB that is currently being DMA'ed back the host on the head of the list while processing completions off of the bus. The newly completed SCBs are appended to the tail of the queue. In the past, we just dequeued the SCB that was in flight from the list, but this could result in a lost completion should the host perform certain types of error recovery that must cancel all in-flight SCB DMA operations. Switch from using a 16bit completion entry, holding just the tag and the completion valid bit, to a 64bit completion entry that also contains a "status packet valid" indicator. This solves two problems: o The SCB DMA engine on at least Rev B. silicon does not properly deal with a PCI disconnect that occurs at a non-64bit aligned offset in the chips "source buffer". When the transfer is resumed, the DMA engine continues at the correct offset, but may wrap to the head of the buffer causing duplicate completions to be reported to the host. By using a completion buffer in host memory that is 64bit aligned and using 64bit completion entries, such disconnects should only occur at aligned addresses. This assumes that the host bridge will only disconnect on cache-line boundaries and that cache-lines are multpiles of 64bits. o By embedding the status information in the completion entry we can avoid an extra memory reference to the HSCB for commands that complete without error. Use the comparison of a "host freeze count" and a "sequencer freeze count" to allow the host to process most SCBs that complete with non-zero status without having to clear critical sections. Instead the host can just pause the sequencer, performs any necessary cleanup in the waiting for selection list, increments its freeze count on the controller, and unpauses. This is only possible because the sequencer defers completions of SCBs with bad status until after all pending selections have completed. The sequencer then avoids referencing any data structures the host may touch during completion of the SCB until the freeze counts match. aic79xx.c: Change the strategy for allocating our sentinal HSCB for the QINFIFO. In the past, this allocation was tacked onto the QOUTFIFO allocation. Now that the qoutfifo has grown to accomodate larger completion entries, the old approach will result in a 64byte allocation that costs an extra page of coherent memory. We now do this extra allocation via ahd_alloc_scbs() where the "unused space" can be used to allocate "normal" HSCBs. In our packetized busfree handler, use the ENSELO bit to differentiate between packetized and non-packetized unexpected busfree events that occur just after selection, but before the sequencer has had the oportunity to service the selection. When cleaning out the waiting for selection list, use the SCSI mode instead of the command channel mode. The SCB pointer in the command channel mode may be referenced by the SCB dma engine even while the sequencer is paused, whereas the SCSI mode SCB pointer is only accessed by the sequencer. Print the "complete on qfreeze" sequencer SCB completion list in ahd_dump_card_state(). This list holds all SCB completions that are deferred until a pending select-out qfreeze event has taken effect. aic79xx.h: Add definitions and structures to handle the new SCB completion scheme. Add a controller flag that indicates if the controller is in HostRAID mode. aic79xx.reg: Remove macros used for toggling from one data fifo mode to the other. They have not been in use for some time. Add scratch ram fields for our new qfreeze count scheme, converting the complete dma list into an "stailq", and providing for the "complete on qfreeze" SCB completion list. Some other fields were moved to retain proper field alignment (alignment >= field size in bytes). aic79xx.seq: Add code to our idle loop to: o Process deferred completions once a qfreeze event has taken full effect. o Thaw the queue once the sequencer and host qfreeze counts match. Generate 64bit completion entries passing the SCB_SGPTR field as the "good status" indicator. The first bit in this field is only set if we have a valid status packet to send to the host. Convert the COMPLETE_DMA_SCB list to an "stailq". When using "setjmp" to register an idle loop handler, do not combine the "ret" with the block move to pop the stack address in the same instruction. At least on the A, this results in a return to the setjmp caller, not to the new address at the top of the stack. Since we want the latter (we want the newly registered handler to only be invoked from the idle loop), we must use a separate ret instruction. Add a few missing critical sections. Close a race condition that can occur on Rev A. silicon. If both FIFOs happen to be allocated before the sequencer has a chance to service the FIFO that was allocated first, we must take special care to service the FIFO that is not active on the SCSI bus first. This guarantees that a FIFO will be freed to handle any snapshot requests for the FIFO that is still on the bus. Chosing the incorrect FIFO will result in deadlock. Update comments. aic79xx_inline.h Correct the offset calculation for the syncing of our qoutfifo. Update ahd_check_cmdcmpltqueues() for the larger completion entries. aic79xx_pci.c: Attach to HostRAID controllers by default. In the future I may add a sysctl to modify the behavior, but since FreeBSD does not have any HostRAID drivers, failing to attach just results in more email and bug reports for the author. MFC After: 1week
2004-02-04 16:38:38 +00:00
bmov CCSCBRAM, SCBPTR, 2;
mov CCSCBRAM, SCB_SGPTR[0];
mov CCSCBRAM, QOUTFIFO_ENTRY_VALID_TAG;
mov NONE, SDSCB_QOFF;
inc INT_COALESCING_CMDCOUNT;
aic79xx.c: Implement the SCB_SILENT flag. This is useful for hushing up the driver during DV or other operations that we expect to cause transmission errors. The messages will still print if the SHOW_MASKED_ERRORS debug option is enabled. Save and restore the NEGOADDR address when setting new transfer settings. The sequencer performs lookups in the negotiation table too and it expects NEGOADDR to remain consistent across pause/unpause sessions. Consistently use "offset" instead of "period" to determine if we are running sync or not. Add a SHOW_MESSAGES diagnostic for when we assert ATN during message processing. Print out IU, QAS, and RTI features when showing transfer options. Limit the syncrate after all option conformance changes have taken place in ahd_devlimited_syncrate. Changes in options may change the final syncrate we accept. Keep a copy of the hs_mailbox in our softc so that we can perform read/modify/write operations on the hs_mailbox without having to pause the sequencer to read the last written value. Use the ENINT_COALESS flag in the hs_mailbox to toggle interrupt coalessing. Add entrypoints for enabling interrupt coalessing and setting both a timeout (how long to wait for commands to be coalessed) and a maximum commands to coaless value. Add a statistics timer that decides when to enable or disable interrupt coalessing based on load. Add a routine, ahd_reset_cmds_pending() which is used to update the CMDS_PENDING sequencer variable whenever error recovery compeltes SCBs without notifying the sequencer. Since ahd_reset_cmds_pending is called during ahd_unpause() only if we've aborted SCBs, its call to ahd_flush_qoutfifo should not cause recursion through ahd_run_qoutfifo(). A panic has been added to ensure that this recursion does not occur. In ahd_search_qinfifo, update the CMDS_PENDING sequencer variable directly. ahd_search_qinififo can be called in situations where using ahd_reset_cmds_pending() might cause recursion. Since we can safely determine the exact number to reduce CMDS_PENDING by in this scenario without running the qoutfifo, the manual update is sufficient. Clean up diagnostics. Add ahd_flush_qoutfifo() which will run the qoutfifo as well as complete any commands sitting on the sequencer's COMPLETE_SCB lists or the good status FIFO. Use this routine in several places that did similar things in an add-hoc, but incomplete, fashion. A call to this routine was also added to ahd_abort_scbs() to close a race. In ahd_pause_and_flushwork() only return once selections are safely disabled. Flush all completed commands via ahd_flush_qoutfifo(). Remove "Now packetized" diagnostic now that this information is incorperated into the actual negotiation messages that are displayed. When forcing renegotiation, don't clober the current ppr_options. Much of the driver uses this information to determine if we are currently packetized or not. Remove some stray spaces at column 1 in ahd_set_tags. When complaining about getting a host message loop request with no pending messages, print out the SCB_CONTROL register down on the card. Modify the ahd_sent_msg() routine to handle a search for an outgoing identify message. Use this to detect a msg reject on an identify message which typically indicates that the target thought we were packetized. Force a renegotiation in this case. In ahd_search_qinfifo(), wait more effectively for SCB DMA activities to cease. We also disable SCB fetch operations since we are about to change the qinfifo and any fetch in progress will likely be invalidated. In ahd_qinfifo_count(), fix the qinfifo empty case. In ahd_dump_card_state(), print out CCSCBCTL in the correct mode. If we are a narrow controller, don't set the current width to unknown when forcing a future negotiation. This just confuses the code into attempting a wide negotiation on a narrow bus. Add support for task management function completions. Modify ahd_handle_devreset so that it can handle lun resets in addition to target resets. Use ahd_handle_devreset for lun and target reset task management functions. Handle the abort task TMF race case better. We now wait until any current selections are over and then set the TMF back to zero. This should cause the sequencer to ignore the abort TMF completion should it occur. Correct a bug in the illegal phase handler that caused us to drop down to narrow when handling the unexpected command phase case after 3rd party reset of a packetized device. Indicate the features, bugs, and flags set in the softc that are used to control firmware patch download when booting verbose. aic79xx.h: Add coalessing and HS_MAILBOX fields. Add per-softc variables for the stats "daemon". Add a debug option for interrupt coalessing activities. Add two new softc flags: o AHD_UPDATE_PEND_CMDS Run ahd_reset_cmds_pending() on the next unpause. o AHD_RUNNING_QOUTFIFO Used to catch recursion through ahd_run_qoutfifo(). aic79xx.reg: Correct register addresses related to the software timer and the DFDBCTL register. Add constants paramaterizing the software timer. Add scratch ram locations for storing interrupt coalessing tunables. Break INTMASK in SEQITNCTL out into INTMASK1 and INTMASK2. In at least the REV A, these are writable bits. We make use of that for a swtimer workaround in the sequencer. Since HS_MAILBOX autoclears, provide a sequencer variable to store its contents. Add SEQINT codes for handling task management completions. aic79xx.seq: Correct ignore wide residue processing check for a wide negotiation being in effect. We must be in the SCSI register window in order to access the negotiation table. Use the software timer and a commands completed count to implement interrupt coalessing. The command complete is deferred until either the maximum command threshold or a the expiration of a command deferral timer. If we have more SCBs to complete to the host (sitting in COMPLETE_SCB lists), always try to coaless them up to our coalessing limit. If coalessing is enabled, but we have fewer commands oustanting than the host's min coalessing limit, complete the command immediately. Add code to track the number of commands outstanding. Commands are outstanding from the time they are placed into the execution queue until the DMA to post completion is setup. Add a workaround for intvec_2 interrupts on the H2A4. In H2A4, the mode pointer is not saved for intvec2, but is restored on iret. This can lead to the restoration of a bogus mode ptr. Manually clear the intmask bits and do a normal return to compensate. We use intvec_2 to track interrupt coalessing timeouts. Since we cannot disable the swtimer's countdown, simply mask its interrupt once we no longer care about it firing. In idle_loop_cchan, update LOCAL_HS_MAILBOX everytime we are notified of an HS_MAILBOX update via the HS_MAILBOX_ACT bit in QOFF_CTLSTA. We have to use a local copy of persistant portions of the HS_MAILBOX as the mailbox auto-clears on any read. Move the test for the cfg4istat interrupt up an instruction to hopefully close a race between the next outgoing selection and our disabling of selections. Add a missing ret to the last instruction in load_overrun_buf. Add notifications to the host of task management completions as well as the completions for commands that completed successfully before their corresponding TMF could be sent. Hold a critical section during select-out processing until we have a fully identified connection. This removes a race condition with the legacy abort handler. Correct a few spelling errors in some comments. aic79xx_inline.h: Call ahd_reset_cmds_pending() in ahd_unpause if required. Update cmdcmplt interrupt statistics in our interrupt handler. Allow callers to ahd_send_scb() to set the task management function. aic79xx_pci.c: Disable SERR and pause the controller prior to performing our mmapped I/O test. The U320 controllers do not support "auto-access-pause". aic79xx_osm.c: Set the task management function now that ahd_send_scb() doesn't do it for us. We also perform a lun reset in response to BDR requests to packetized devices.
2003-01-20 20:17:35 +00:00
add CMDS_PENDING, -1;
adc CMDS_PENDING[1], -1;
cmp SCB_NEXT_COMPLETE[1], SCB_LIST_NULL je fill_qoutfifo_done;
cmp CCSCBADDR, CCSCBADDR_MAX je fill_qoutfifo_done;
test QOFF_CTLSTA, SDSCB_ROLLOVR jnz fill_qoutfifo_done;
aic79xx.c: Allow 500us between pauses in ahd_pause_and_flushwork(). The maximum we will wait is now 500ms. In the same routine, remove any attempt to clear ENSELO. Let the firmware do it once the current selection has completed. This avoids some race conditions having to do with non-packetized completions and the auto-clearing of ENSELO on packetized completions. Also avoid attempts to clear critical sections when interrups are pending. We are going to loop again anyway, so clearing critical sections is a waste of time. It also may not be possible to clear a critical section if the source of the interrupt was a SEQINT. aic79xx_pci.c: Use the Generic 9005 mask when looking for generic 7901B parts. This allows the driver to attach to 7901B parts on motherboards using a non-Adaptec subvendor ID. aic79xx_inline.h: Test for the SCBRAM_RD_BUG against the bugs field, not the flags field in the softc. aic79xx.c: Cancel pending transactions on devices that respond with a selection timeout. This decreases the duration of timeout recovery when a device disappears. aic79xx.c: Don't bother forcing renegotiation on a selection timeout now that we use the device reset handler to abort any pending commands on the target. The device reset handler already takes us down to async narrow and forces a renegotiation. In the device reset handlers, only send a BDR sent async event if the status is not CAM_SEL_TIMEOUT. This avoids sending this event in the selection timeout case aic79xx.c: Modify the Core timeout handler to verify that another command has the potential to timeout before passing off a command timeout as due to some other command. This safety measure is added in response to a timeout recovery failure on H2B where it appears that incoming reselection status was lost during a drive pull test. In that case, the recovery handler continued to wait for the command that was active on the bus indefinetly. While the root cause of the above issue is still being determined seems a prudent safeguard. aic79xx_pci.c: Add a specific probe entry for the Dell OEM 39320(B). aic79xx.c: aic79xx.h: aic79xx.reg: aic79xx.seq: Modify the aic79xx firmware to never cross a cacheline or ADB boundary when DMA'ing completion entries to the host. In PCI mode, at least in 32/33 configurations, the SCB DMA engine may lose its place in the data-stream should the target force a retry on something other than an 8byte aligned boundary. In PCI-X mode, we do this to avoid split transactions since many chipsets seem to be unable to format proper split completions to continue the data transfer. The above change allows us to drop our completion entries from 8 bytes to 4. We were using 8 byte entries to ensure that PCI retries could only occur on an 8byte aligned boundary. Now that the sequencer guarantees this by splitting up completions, we can safely drop the size to 4 bytes (2 byte tag, one byte SG_RESID, one byte pad). Both the split-completion and PCI retry problems only show up under high tag load when interrupt coalescing is being especially effective. The switch from a 2byte completion entry to an 8 byte entry to solve the PCI problem increased the chance of incurring a split in PCI-X mode when multiple transactions were completed at once. Dropping the completion size to 4 bytes also means that we can complete more commands in a single DMA (128byte FIFO -> 32 commands instead of 16). aic79xx.c: Modify the SCSIINT handler to defer clearing sequencer critical sections to the individual interrupt handlers. This allows us to immediately disable any outgoing selections in the case of an unexpected busfree so we don't inadvertantly clear ENSELO *after* a new selection has started. Doing so may cause the sequencer to miss a successful selection. In ahd_update_pending_scbs(), only clear ENSELO if the bus is currently busy and a selection is not already in progress or the sequencer has yet to handle a pending selection. While we want to ensure that the selection for the SCB at the head of the selection queue is restarted so that any change in negotiation request can take effect, we can't clobber pending selection state without confusing the sequencer into missing a selection.
2004-05-11 20:46:05 +00:00
/*
* Don't cross an ADB or Cachline boundary when DMA'ing
* completion entries. In PCI mode, at least in 32/33
* configurations, the SCB DMA engine may lose its place
* in the data-stream should the target force a retry on
* something other than an 8byte aligned boundary. In
* PCI-X mode, we do this to avoid split transactions since
* many chipsets seem to be unable to format proper split
* completions to continue the data transfer.
*/
add SINDEX, A, CCSCBADDR;
test SINDEX, CACHELINE_MASK jz fill_qoutfifo_done;
bmov SCBPTR, SCB_NEXT_COMPLETE, 2;
jmp fill_qoutfifo_loop;
fill_qoutfifo_done:
mov SCBHCNT, CCSCBADDR;
mvi CCSCBCTL, CCSCBEN|CCSCBRESET;
bmov COMPLETE_SCB_HEAD, SCB_NEXT_COMPLETE, 2;
mvi SCB_NEXT_COMPLETE[1], SCB_LIST_NULL ret;
fetch_new_scb:
bmov SCBHADDR, NEXT_QUEUED_SCB_ADDR, 4;
mvi CCARREN|CCSCBEN|CCSCBDIR|CCSCBRESET jmp dma_scb;
dma_complete_scb:
bmov SCBPTR, COMPLETE_DMA_SCB_HEAD, 2;
bmov SCBHADDR, SCB_BUSADDR, 4;
Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-05-04 00:20:07 +00:00
mvi CCARREN|CCSCBEN|CCSCBRESET jmp dma_scb;
/*
* Either post or fetch an SCB from host memory. The caller
* is responsible for polling for transfer completion.
*
* Prerequisits: Mode == M_CCHAN
* SINDEX contains CCSCBCTL flags
* SCBHADDR set to Host SCB address
* SCBPTR set to SCB src location on "push" operations
*/
SET_SRC_MODE M_CCHAN;
SET_DST_MODE M_CCHAN;
dma_scb:
mvi SCBHCNT, SCB_TRANSFER_SIZE;
Upgrade to version 1.1 of the aic79xx U320 driver. aic79xx.c: o Remove redundant ahd_update_modes() call. o Correct panic in diagnostic should state corruption cause the SCB Id to be invalid during a selection timeout. o Add workaround for missing BUSFREEREV feature in Rev A silicon. o Corect formatting nits. o Use register pretty printing in more places. o Save and restore our SCB pointer when updating the waiting queue list for an "expected" LQ-out busfree. o In ahd_clear_intstat, deal with the missing autoclear in the CLRLQO* registers. o BE fixup in a diagnostic printf. o Make sure that we are in the proper mode before disabling selections in ahd_update_pending_scbs. o Add more diagnostics. o task_attribute_nonpkt_tag -> task_attribute: we don't need a nonpkt_tag field anymore for allowing all 512 SCBs to be used in non-packetized connections. o Negotiate HOLD_MCS to U320 devices. o Add a few additional mode assertions. o Restore the chip mode after clearing out the qinfifo so that code using ahd_abort_scbs sees a consistent mode. o Simplify the DMA engine shutdown routine prior to performing a bus reset. o Perform the sequencer restart after a chip reset prior to setting up our timer to poll for the reset to be complete. On some OSes, the timer could actually pre-empt us and order is important here. o Have our "reset poller" set the expected mode since there is no guarantee of what mode will be in force when we are called from the OS timer. o Save and restore the SCB pointer in ahd_dump_card_state(). This routine must not modify card state. o Ditto for ahd_dump_scbs(). aic79xx.h: o Add a few more chip bug definitions. o Align our tag on a 32bit boundary. aic79xx.reg: aic79xx.seq: o Start work on removing workarounds for Rev B. o Use a special location in scratch from for stroring our SCBPTR during legacy FIFO allocations. This corrects problems in mixed packetized/non-packetized configurations where calling into a FIFO task corrupted our SCBPTR. o Don't rely on DMA priority to guarantee that all data in our FIFOs will flush prior to a command completion notification going out of the command channel. We've never seen this assumption fail, but better safe than sorry. o Deal with missing BUSFREEREV feature in H2A. o Simplify disconnect list code now that the list will always have only a single entry. o Implement the AHD_REG_SLOW_SETTLE_BUG workaround. o Swith to using "REG_ISR" for local mode scratch during our ISR. o Add a missing jmp to the data_group_dma_loop after our data pointers have been re-initialized by the kernel. o Correct test in the bitbucket code so that we actually wait for the bitbucket to complete before signaling the kernel of the overrun condition. o Reposition pkt_saveptrs to avoid a jmp instruction. o Update a comment to reflect that the code now waits for a FIFO to drain prior to issuing a CLRCHN. aic79xx_inline.h: o Remove unused untagged queue handling code. o Don't attempt to htole64 what could be a 32bit value. aic79xx_pci.c: o Set additional bug flags for rev A chips.
2002-09-26 22:54:00 +00:00
mov CCSCBCTL, SINDEX ret;
setjmp:
aic79xx.c: aic79xx.seq: Convert the COMPLETE_DMA_SCB list to an "stailq". This allows us to safely keep the SCB that is currently being DMA'ed back the host on the head of the list while processing completions off of the bus. The newly completed SCBs are appended to the tail of the queue. In the past, we just dequeued the SCB that was in flight from the list, but this could result in a lost completion should the host perform certain types of error recovery that must cancel all in-flight SCB DMA operations. Switch from using a 16bit completion entry, holding just the tag and the completion valid bit, to a 64bit completion entry that also contains a "status packet valid" indicator. This solves two problems: o The SCB DMA engine on at least Rev B. silicon does not properly deal with a PCI disconnect that occurs at a non-64bit aligned offset in the chips "source buffer". When the transfer is resumed, the DMA engine continues at the correct offset, but may wrap to the head of the buffer causing duplicate completions to be reported to the host. By using a completion buffer in host memory that is 64bit aligned and using 64bit completion entries, such disconnects should only occur at aligned addresses. This assumes that the host bridge will only disconnect on cache-line boundaries and that cache-lines are multpiles of 64bits. o By embedding the status information in the completion entry we can avoid an extra memory reference to the HSCB for commands that complete without error. Use the comparison of a "host freeze count" and a "sequencer freeze count" to allow the host to process most SCBs that complete with non-zero status without having to clear critical sections. Instead the host can just pause the sequencer, performs any necessary cleanup in the waiting for selection list, increments its freeze count on the controller, and unpauses. This is only possible because the sequencer defers completions of SCBs with bad status until after all pending selections have completed. The sequencer then avoids referencing any data structures the host may touch during completion of the SCB until the freeze counts match. aic79xx.c: Change the strategy for allocating our sentinal HSCB for the QINFIFO. In the past, this allocation was tacked onto the QOUTFIFO allocation. Now that the qoutfifo has grown to accomodate larger completion entries, the old approach will result in a 64byte allocation that costs an extra page of coherent memory. We now do this extra allocation via ahd_alloc_scbs() where the "unused space" can be used to allocate "normal" HSCBs. In our packetized busfree handler, use the ENSELO bit to differentiate between packetized and non-packetized unexpected busfree events that occur just after selection, but before the sequencer has had the oportunity to service the selection. When cleaning out the waiting for selection list, use the SCSI mode instead of the command channel mode. The SCB pointer in the command channel mode may be referenced by the SCB dma engine even while the sequencer is paused, whereas the SCSI mode SCB pointer is only accessed by the sequencer. Print the "complete on qfreeze" sequencer SCB completion list in ahd_dump_card_state(). This list holds all SCB completions that are deferred until a pending select-out qfreeze event has taken effect. aic79xx.h: Add definitions and structures to handle the new SCB completion scheme. Add a controller flag that indicates if the controller is in HostRAID mode. aic79xx.reg: Remove macros used for toggling from one data fifo mode to the other. They have not been in use for some time. Add scratch ram fields for our new qfreeze count scheme, converting the complete dma list into an "stailq", and providing for the "complete on qfreeze" SCB completion list. Some other fields were moved to retain proper field alignment (alignment >= field size in bytes). aic79xx.seq: Add code to our idle loop to: o Process deferred completions once a qfreeze event has taken full effect. o Thaw the queue once the sequencer and host qfreeze counts match. Generate 64bit completion entries passing the SCB_SGPTR field as the "good status" indicator. The first bit in this field is only set if we have a valid status packet to send to the host. Convert the COMPLETE_DMA_SCB list to an "stailq". When using "setjmp" to register an idle loop handler, do not combine the "ret" with the block move to pop the stack address in the same instruction. At least on the A, this results in a return to the setjmp caller, not to the new address at the top of the stack. Since we want the latter (we want the newly registered handler to only be invoked from the idle loop), we must use a separate ret instruction. Add a few missing critical sections. Close a race condition that can occur on Rev A. silicon. If both FIFOs happen to be allocated before the sequencer has a chance to service the FIFO that was allocated first, we must take special care to service the FIFO that is not active on the SCSI bus first. This guarantees that a FIFO will be freed to handle any snapshot requests for the FIFO that is still on the bus. Chosing the incorrect FIFO will result in deadlock. Update comments. aic79xx_inline.h Correct the offset calculation for the syncing of our qoutfifo. Update ahd_check_cmdcmpltqueues() for the larger completion entries. aic79xx_pci.c: Attach to HostRAID controllers by default. In the future I may add a sysctl to modify the behavior, but since FreeBSD does not have any HostRAID drivers, failing to attach just results in more email and bug reports for the author. MFC After: 1week
2004-02-04 16:38:38 +00:00
/*
* At least on the A, a return in the same
* instruction as the bmov results in a return
* to the caller, not to the new address at the
* top of the stack. Since we want the latter
* (we use setjmp to register a handler from an
* interrupt context but not invoke that handler
* until we return to our idle loop), use a
* separate ret instruction.
*/
bmov LONGJMP_ADDR, STACK, 2;
ret;
setjmp_inline:
bmov LONGJMP_ADDR, STACK, 2;
longjmp:
bmov STACK, LONGJMP_ADDR, 2 ret;
END_CRITICAL;
/*************************** Chip Bug Work Arounds ****************************/
/*
* Must disable interrupts when setting the mode pointer
* register as an interrupt occurring mid update will
* fail to store the new mode value for restoration on
* an iret.
*/
Upgrade to version 1.1 of the aic79xx U320 driver. aic79xx.c: o Remove redundant ahd_update_modes() call. o Correct panic in diagnostic should state corruption cause the SCB Id to be invalid during a selection timeout. o Add workaround for missing BUSFREEREV feature in Rev A silicon. o Corect formatting nits. o Use register pretty printing in more places. o Save and restore our SCB pointer when updating the waiting queue list for an "expected" LQ-out busfree. o In ahd_clear_intstat, deal with the missing autoclear in the CLRLQO* registers. o BE fixup in a diagnostic printf. o Make sure that we are in the proper mode before disabling selections in ahd_update_pending_scbs. o Add more diagnostics. o task_attribute_nonpkt_tag -> task_attribute: we don't need a nonpkt_tag field anymore for allowing all 512 SCBs to be used in non-packetized connections. o Negotiate HOLD_MCS to U320 devices. o Add a few additional mode assertions. o Restore the chip mode after clearing out the qinfifo so that code using ahd_abort_scbs sees a consistent mode. o Simplify the DMA engine shutdown routine prior to performing a bus reset. o Perform the sequencer restart after a chip reset prior to setting up our timer to poll for the reset to be complete. On some OSes, the timer could actually pre-empt us and order is important here. o Have our "reset poller" set the expected mode since there is no guarantee of what mode will be in force when we are called from the OS timer. o Save and restore the SCB pointer in ahd_dump_card_state(). This routine must not modify card state. o Ditto for ahd_dump_scbs(). aic79xx.h: o Add a few more chip bug definitions. o Align our tag on a 32bit boundary. aic79xx.reg: aic79xx.seq: o Start work on removing workarounds for Rev B. o Use a special location in scratch from for stroring our SCBPTR during legacy FIFO allocations. This corrects problems in mixed packetized/non-packetized configurations where calling into a FIFO task corrupted our SCBPTR. o Don't rely on DMA priority to guarantee that all data in our FIFOs will flush prior to a command completion notification going out of the command channel. We've never seen this assumption fail, but better safe than sorry. o Deal with missing BUSFREEREV feature in H2A. o Simplify disconnect list code now that the list will always have only a single entry. o Implement the AHD_REG_SLOW_SETTLE_BUG workaround. o Swith to using "REG_ISR" for local mode scratch during our ISR. o Add a missing jmp to the data_group_dma_loop after our data pointers have been re-initialized by the kernel. o Correct test in the bitbucket code so that we actually wait for the bitbucket to complete before signaling the kernel of the overrun condition. o Reposition pkt_saveptrs to avoid a jmp instruction. o Update a comment to reflect that the code now waits for a FIFO to drain prior to issuing a CLRCHN. aic79xx_inline.h: o Remove unused untagged queue handling code. o Don't attempt to htole64 what could be a 32bit value. aic79xx_pci.c: o Set additional bug flags for rev A chips.
2002-09-26 22:54:00 +00:00
if ((ahd->bugs & AHD_SET_MODE_BUG) != 0) {
set_mode_work_around:
mvi SEQINTCTL, INTVEC1DSL;
mov MODE_PTR, SINDEX;
clr SEQINTCTL ret;
Upgrade to version 1.1 of the aic79xx U320 driver. aic79xx.c: o Remove redundant ahd_update_modes() call. o Correct panic in diagnostic should state corruption cause the SCB Id to be invalid during a selection timeout. o Add workaround for missing BUSFREEREV feature in Rev A silicon. o Corect formatting nits. o Use register pretty printing in more places. o Save and restore our SCB pointer when updating the waiting queue list for an "expected" LQ-out busfree. o In ahd_clear_intstat, deal with the missing autoclear in the CLRLQO* registers. o BE fixup in a diagnostic printf. o Make sure that we are in the proper mode before disabling selections in ahd_update_pending_scbs. o Add more diagnostics. o task_attribute_nonpkt_tag -> task_attribute: we don't need a nonpkt_tag field anymore for allowing all 512 SCBs to be used in non-packetized connections. o Negotiate HOLD_MCS to U320 devices. o Add a few additional mode assertions. o Restore the chip mode after clearing out the qinfifo so that code using ahd_abort_scbs sees a consistent mode. o Simplify the DMA engine shutdown routine prior to performing a bus reset. o Perform the sequencer restart after a chip reset prior to setting up our timer to poll for the reset to be complete. On some OSes, the timer could actually pre-empt us and order is important here. o Have our "reset poller" set the expected mode since there is no guarantee of what mode will be in force when we are called from the OS timer. o Save and restore the SCB pointer in ahd_dump_card_state(). This routine must not modify card state. o Ditto for ahd_dump_scbs(). aic79xx.h: o Add a few more chip bug definitions. o Align our tag on a 32bit boundary. aic79xx.reg: aic79xx.seq: o Start work on removing workarounds for Rev B. o Use a special location in scratch from for stroring our SCBPTR during legacy FIFO allocations. This corrects problems in mixed packetized/non-packetized configurations where calling into a FIFO task corrupted our SCBPTR. o Don't rely on DMA priority to guarantee that all data in our FIFOs will flush prior to a command completion notification going out of the command channel. We've never seen this assumption fail, but better safe than sorry. o Deal with missing BUSFREEREV feature in H2A. o Simplify disconnect list code now that the list will always have only a single entry. o Implement the AHD_REG_SLOW_SETTLE_BUG workaround. o Swith to using "REG_ISR" for local mode scratch during our ISR. o Add a missing jmp to the data_group_dma_loop after our data pointers have been re-initialized by the kernel. o Correct test in the bitbucket code so that we actually wait for the bitbucket to complete before signaling the kernel of the overrun condition. o Reposition pkt_saveptrs to avoid a jmp instruction. o Update a comment to reflect that the code now waits for a FIFO to drain prior to issuing a CLRCHN. aic79xx_inline.h: o Remove unused untagged queue handling code. o Don't attempt to htole64 what could be a 32bit value. aic79xx_pci.c: o Set additional bug flags for rev A chips.
2002-09-26 22:54:00 +00:00
}
if ((ahd->bugs & AHD_INTCOLLISION_BUG) != 0) {
set_seqint_work_around:
mov SEQINTCODE, SINDEX;
mvi SEQINTCODE, NO_SEQINT ret;
}
/************************ Packetized LongJmp Routines *************************/
SET_SRC_MODE M_SCSI;
SET_DST_MODE M_SCSI;
start_selection:
BEGIN_CRITICAL;
if ((ahd->bugs & AHD_SENT_SCB_UPDATE_BUG) != 0) {
/*
* Razor #494
* Rev A hardware fails to update LAST/CURR/NEXTSCB
* correctly after a packetized selection in several
* situations:
*
* 1) If only one command existed in the queue, the
* LAST/CURR/NEXTSCB are unchanged.
*
* 2) In a non QAS, protocol allowed phase change,
* the queue is shifted 1 too far. LASTSCB is
* the last SCB that was correctly processed.
*
* 3) In the QAS case, if the full list of commands
* was successfully sent, NEXTSCB is NULL and neither
* CURRSCB nor LASTSCB can be trusted. We must
* manually walk the list counting MAXCMDCNT elements
* to find the last SCB that was sent correctly.
*
* To simplify the workaround for this bug in SELDO
* handling, we initialize LASTSCB prior to enabling
* selection so we can rely on it even for case #1 above.
*/
bmov LASTSCB, WAITING_TID_HEAD, 2;
}
bmov CURRSCB, WAITING_TID_HEAD, 2;
bmov SCBPTR, WAITING_TID_HEAD, 2;
shr SELOID, 4, SCB_SCSIID;
/*
* If we want to send a message to the device, ensure
* we are selecting with atn irregardless of our packetized
* agreement. Since SPI4 only allows target reset or PPR
* messages if this is a packetized connection, the change
* to our negotiation table entry for this selection will
* be cleared when the message is acted on.
*/
test SCB_CONTROL, MK_MESSAGE jz . + 3;
mov NEGOADDR, SELOID;
or NEGCONOPTS, ENAUTOATNO;
or SCSISEQ0, ENSELO ret;
END_CRITICAL;
/*
* Allocate a FIFO for a non-packetized transaction.
* In RevA hardware, both FIFOs must be free before we
* can allocate a FIFO for a non-packetized transaction.
*/
allocate_fifo_loop:
/*
* Do whatever work is required to free a FIFO.
*/
call idle_loop_service_fifos;
Upgrade to version 1.1 of the aic79xx U320 driver. aic79xx.c: o Remove redundant ahd_update_modes() call. o Correct panic in diagnostic should state corruption cause the SCB Id to be invalid during a selection timeout. o Add workaround for missing BUSFREEREV feature in Rev A silicon. o Corect formatting nits. o Use register pretty printing in more places. o Save and restore our SCB pointer when updating the waiting queue list for an "expected" LQ-out busfree. o In ahd_clear_intstat, deal with the missing autoclear in the CLRLQO* registers. o BE fixup in a diagnostic printf. o Make sure that we are in the proper mode before disabling selections in ahd_update_pending_scbs. o Add more diagnostics. o task_attribute_nonpkt_tag -> task_attribute: we don't need a nonpkt_tag field anymore for allowing all 512 SCBs to be used in non-packetized connections. o Negotiate HOLD_MCS to U320 devices. o Add a few additional mode assertions. o Restore the chip mode after clearing out the qinfifo so that code using ahd_abort_scbs sees a consistent mode. o Simplify the DMA engine shutdown routine prior to performing a bus reset. o Perform the sequencer restart after a chip reset prior to setting up our timer to poll for the reset to be complete. On some OSes, the timer could actually pre-empt us and order is important here. o Have our "reset poller" set the expected mode since there is no guarantee of what mode will be in force when we are called from the OS timer. o Save and restore the SCB pointer in ahd_dump_card_state(). This routine must not modify card state. o Ditto for ahd_dump_scbs(). aic79xx.h: o Add a few more chip bug definitions. o Align our tag on a 32bit boundary. aic79xx.reg: aic79xx.seq: o Start work on removing workarounds for Rev B. o Use a special location in scratch from for stroring our SCBPTR during legacy FIFO allocations. This corrects problems in mixed packetized/non-packetized configurations where calling into a FIFO task corrupted our SCBPTR. o Don't rely on DMA priority to guarantee that all data in our FIFOs will flush prior to a command completion notification going out of the command channel. We've never seen this assumption fail, but better safe than sorry. o Deal with missing BUSFREEREV feature in H2A. o Simplify disconnect list code now that the list will always have only a single entry. o Implement the AHD_REG_SLOW_SETTLE_BUG workaround. o Swith to using "REG_ISR" for local mode scratch during our ISR. o Add a missing jmp to the data_group_dma_loop after our data pointers have been re-initialized by the kernel. o Correct test in the bitbucket code so that we actually wait for the bitbucket to complete before signaling the kernel of the overrun condition. o Reposition pkt_saveptrs to avoid a jmp instruction. o Update a comment to reflect that the code now waits for a FIFO to drain prior to issuing a CLRCHN. aic79xx_inline.h: o Remove unused untagged queue handling code. o Don't attempt to htole64 what could be a 32bit value. aic79xx_pci.c: o Set additional bug flags for rev A chips.
2002-09-26 22:54:00 +00:00
SET_MODE(M_SCSI, M_SCSI)
allocate_fifo:
if ((ahd->bugs & AHD_NONPACKFIFO_BUG) != 0) {
and A, FIFO0FREE|FIFO1FREE, DFFSTAT;
cmp A, FIFO0FREE|FIFO1FREE jne allocate_fifo_loop;
} else {
test DFFSTAT, FIFO1FREE jnz allocate_fifo1;
test DFFSTAT, FIFO0FREE jz allocate_fifo_loop;
mvi DFFSTAT, B_CURRFIFO_0;
SET_MODE(M_DFF0, M_DFF0)
bmov SCBPTR, ALLOCFIFO_SCBPTR, 2 ret;
}
SET_SRC_MODE M_SCSI;
SET_DST_MODE M_SCSI;
allocate_fifo1:
mvi DFFSTAT, CURRFIFO_1;
Upgrade to version 1.1 of the aic79xx U320 driver. aic79xx.c: o Remove redundant ahd_update_modes() call. o Correct panic in diagnostic should state corruption cause the SCB Id to be invalid during a selection timeout. o Add workaround for missing BUSFREEREV feature in Rev A silicon. o Corect formatting nits. o Use register pretty printing in more places. o Save and restore our SCB pointer when updating the waiting queue list for an "expected" LQ-out busfree. o In ahd_clear_intstat, deal with the missing autoclear in the CLRLQO* registers. o BE fixup in a diagnostic printf. o Make sure that we are in the proper mode before disabling selections in ahd_update_pending_scbs. o Add more diagnostics. o task_attribute_nonpkt_tag -> task_attribute: we don't need a nonpkt_tag field anymore for allowing all 512 SCBs to be used in non-packetized connections. o Negotiate HOLD_MCS to U320 devices. o Add a few additional mode assertions. o Restore the chip mode after clearing out the qinfifo so that code using ahd_abort_scbs sees a consistent mode. o Simplify the DMA engine shutdown routine prior to performing a bus reset. o Perform the sequencer restart after a chip reset prior to setting up our timer to poll for the reset to be complete. On some OSes, the timer could actually pre-empt us and order is important here. o Have our "reset poller" set the expected mode since there is no guarantee of what mode will be in force when we are called from the OS timer. o Save and restore the SCB pointer in ahd_dump_card_state(). This routine must not modify card state. o Ditto for ahd_dump_scbs(). aic79xx.h: o Add a few more chip bug definitions. o Align our tag on a 32bit boundary. aic79xx.reg: aic79xx.seq: o Start work on removing workarounds for Rev B. o Use a special location in scratch from for stroring our SCBPTR during legacy FIFO allocations. This corrects problems in mixed packetized/non-packetized configurations where calling into a FIFO task corrupted our SCBPTR. o Don't rely on DMA priority to guarantee that all data in our FIFOs will flush prior to a command completion notification going out of the command channel. We've never seen this assumption fail, but better safe than sorry. o Deal with missing BUSFREEREV feature in H2A. o Simplify disconnect list code now that the list will always have only a single entry. o Implement the AHD_REG_SLOW_SETTLE_BUG workaround. o Swith to using "REG_ISR" for local mode scratch during our ISR. o Add a missing jmp to the data_group_dma_loop after our data pointers have been re-initialized by the kernel. o Correct test in the bitbucket code so that we actually wait for the bitbucket to complete before signaling the kernel of the overrun condition. o Reposition pkt_saveptrs to avoid a jmp instruction. o Update a comment to reflect that the code now waits for a FIFO to drain prior to issuing a CLRCHN. aic79xx_inline.h: o Remove unused untagged queue handling code. o Don't attempt to htole64 what could be a 32bit value. aic79xx_pci.c: o Set additional bug flags for rev A chips.
2002-09-26 22:54:00 +00:00
SET_MODE(M_DFF1, M_DFF1)
bmov SCBPTR, ALLOCFIFO_SCBPTR, 2 ret;
/*
* We have been reselected as an initiator
* or selected as a target.
*/
SET_SRC_MODE M_SCSI;
SET_DST_MODE M_SCSI;
select_in:
ahc_eisa.c: ahc_pci.c: ahd_pci.c: aic7xxx.c: aic79xx.c: aic_osm_lib.c: aic_osm_lib.h: Use common OSM routines from aic_osm_lib for bus dma operations, delay routines, accessing CCBs, byte swapping, etc. aic7xxx_pci.c: Provide a better description for the 2915/30LP on attach. aic7xxx.c: aic79xx.c: aic7770.c: aic79xx_pci.c: aic7xxx_pci.c: aic7xxx_93cx6.c: Move FBSDID behind an ifdef so that these core files will still compile under other OSes. aic79xx.h: aic79xx_pci.c: aic79xx.seq: To speed up non-packetized CDB delivery in Rev B, all CDB acks are "released" to the output sync as soon as the command phase starts. There is only one problem with this approach. If the target changes phase before all data are sent, we have left over acks that can go out on the bus in a data phase. Due to other chip contraints, this only happens if the target goes to data-in, but if the acks go out before we can test SDONE, we'll think that the transfer has completed successfully. Work around this by taking advantage of the 400ns or 800ns dead time between command phase and the REQ of the new phase. If the transfer has completed successfully, SCSIEN should fall *long* before we see a phase change. We thus treat any phasemiss that occurs before SCSIEN falls as an incomplete transfer. aic79xx.h: Add the AHD_FAST_CDB_DELIVERY feature. aic79xx_pci.c: Set AHD_FAST_CDB_DELIVERY for all Rev. B parts. aic79xx.seq: Test for PHASEMIS in the command phase for all AHD_FAST_CDB_DELIVERY controlelrs. ahd_pci.c: ahc_pci.c: aic7xxx.h: aic79xx.h: Move definition of controller BAR offsets to core header files. aic7xxx.c: aic79xx.c: In the softc free routine, leave removal of a softc from the global list of softcs to the OSM (the caller of this routine). This allows us to avoid holding the softc list_lock during device destruction where we may have to sleep waiting for our recovery thread to halt. ahc_pci.c: Use ahc_pci_test_register access to validate I/O mapped in addition to the tests already performed for memory mapped access. Remove unused ahc_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. ahd_pci.c: Remove reduntant definition of controller BAR offsets. These are also defined in aic79xx.h. Remove unused ahd_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. aic7xxx.c: aic79xx.c: aic79xx.h: aic7xxx.h: aic7xxx_osm.c: aic79xx_osm.c: Move timeout handling to the driver cores. In the case of the aic79xx driver, the algorithm has been enhanced to try target resets before performing a bus reset. For the aic7xxx driver, the algorithm is unchanged. Although the drivers do not currently sleep during recovery (recovery is timeout driven), the cores do expect all processing to be performed via a recovery thread. Our timeout handlers are now little stubs that wakeup the recovery thread. aic79xx.c: aic79xx.h: aic79xx_inline.h: Change shared_data allocation to use a map_node so that the sentinel hscb can use this map node in ahd_swap_with_next_hscb. This routine now swaps the hscb_map pointer in additon to the hscb contents so that any sync operations occur on the correct map. physaddr -> busaddr Pointed out by: Jason Thorpe <thorpej@wasabisystems.com> aic79xx.c: Make more use of the in/out/w/l/q macros for accessing byte registers in the chip. Correct some issues in the ahd_flush_qoutfifo() routine. o Run the qoutfifo only once the command channel DMA engine has been halted. This closes a window where we might have missed some entries. o Change ahd_run_data_fifo() to not loop to completion. If we happen to start on the wrong FIFO and the other FIFO has a snapshot savepointers, we might deadlock. This required our delay between FIFO tests to be moved to the ahd_flush_qoutfifo() routine. o Update/add comments. o Remove spurious test for COMPLETE_DMA list being empty when completing transactions from the GSFIFO with residuals. The SCB must be put on the COMPLETE_DMA scb list unconditionally. o When halting command channel DMA activity, we must disable the DMA channel in all cases but an update of the QOUTFIFO. The latter case is required so that the sequencer will update its position in the QOUTFIFO. Previously, we left the channel enabled for all "push" DMAs. This left us vulnerable to the sequencer handling an SCB push long after that SCB was already processed manually by this routine. o Correct the polarity of tests involving ahd_scb_active_in_fifo(). This routine returns non-zero for true. Return to processing bad status completions through the qoutfifo. This reduces the time that the sequencer is kept paused when handling transactions with bad status or underruns. When waiting for the controller to quiece selections, add a delay to our loop. Otherwise we may fail to wait long enough for the sequencer to comply. On H2A4 hardware, use the slow slewrate for non-paced transfers. This mirrors what the Adaptec Windows drivers do. On the Rev B. only slow down the CRC timing for older U160 devices that might need the slower timing. We define "older" as devices that do not support packetized protocol. Wait up to 5000 * 5us for the SEEPROM to become unbusy. Write ops seem to take much longer than read ops. aic79xx.seq: For controllers with the FAINT_LED bug, turn the diagnostic led feature on during selection and reselection. This covers the non-packetized case. The LED will be disabled for non-packetized transfers once we return to the top level idle loop. Add more comments about the busy LED workaround. Extend a critical section around the entire command channel idle loop process. Previously the portion of this handler that directly manipulated the linked list of completed SCBs was not protected. This is the likely cause of the recent reports of commands being completed twice by the driver. Extend critical sections across the test for, and the longjump to, longjump routines. This prevents the firmware from trying to jump to a longjmp handler that was just cleared by the host. Improve the locations of several critical section begin and end points. Typically these changes remove instructions that did not need to be inside a critical section. Close the "busfree after selection, but before busfree interrupts can be enabled" race to just a single sequencer instruction. We now test the BSY line explicitly before clearing the busfree status and enabling the busfree interrupt. Close a race condition in the processing of HS_MAILBOX updates. We now clear the "updated" status before the copy. This ensures that we don't accidentally clear the status incorrectly when the host sneaks in an update just after our last copy, but before we clear the status. This race has never been observed. Don't re-enable SCSIEN if we lose the race to disable SCSIEN in our interrupt handler's workaround for the RevA data-valid too early issue. aic79xx_inline.h: Add comments indicating that the order in which bytes are read or written in ahd_inw and ahd_outw is important. This allows us to use these inlines when accessing registers with side-effects. aic79xx_pci.c: The 29320 and the 29320B are 7902 not 7901 based products. Correct the driver banner. aic7xxx.h: Enable the use of the auto-access pause feature on the aic7870 and aic7880. It was disabled due to an oversight. aic7xxx.reg: Move TARG_IMMEDIATE_SCB to alias LAST_MSG to avoid leaving garbage in MWI_RESIDUAL. This prevents spurious overflows whn operating target mode on controllers that require the MWI_RESIDUAL work-around. aic7xxx.seq: AHC_TMODE_WIDEODD_BUG is a bug, not a softc flag. Reference the correct softc field when testing for its presence. Set the NOT_IDENTIFIED and NO_CDB_SENT bits in SEQ_FLAGS to indicate that the nexus is invalid in await busfree. aic7xxx_93cx6.c: Add support for the C56/C66 versions of the EWEN and EWDS commands. aic7xxx.c: aic7xxx_pci.c: Move test for the validity of left over BIOS data to ahc_test_register_access(). This guarantees that any left over CHIPRST value is not clobbered by our register access test and lost to the test that was in ahc_reset.
2003-12-17 00:02:10 +00:00
if ((ahd->bugs & AHD_FAINT_LED_BUG) != 0) {
/*
* On Rev A. hardware, the busy LED is only
* turned on automaically during selections
* and re-selections. Make the LED status
* more useful by forcing it to be on from
* the point of selection until our idle
* loop determines that neither of our FIFOs
* are busy. This handles the non-packetized
* case nicely as we will not return to the
* idle loop until the busfree at the end of
* each transaction.
*/
or SBLKCTL, DIAGLEDEN|DIAGLEDON;
}
Upgrade to version 1.1 of the aic79xx U320 driver. aic79xx.c: o Remove redundant ahd_update_modes() call. o Correct panic in diagnostic should state corruption cause the SCB Id to be invalid during a selection timeout. o Add workaround for missing BUSFREEREV feature in Rev A silicon. o Corect formatting nits. o Use register pretty printing in more places. o Save and restore our SCB pointer when updating the waiting queue list for an "expected" LQ-out busfree. o In ahd_clear_intstat, deal with the missing autoclear in the CLRLQO* registers. o BE fixup in a diagnostic printf. o Make sure that we are in the proper mode before disabling selections in ahd_update_pending_scbs. o Add more diagnostics. o task_attribute_nonpkt_tag -> task_attribute: we don't need a nonpkt_tag field anymore for allowing all 512 SCBs to be used in non-packetized connections. o Negotiate HOLD_MCS to U320 devices. o Add a few additional mode assertions. o Restore the chip mode after clearing out the qinfifo so that code using ahd_abort_scbs sees a consistent mode. o Simplify the DMA engine shutdown routine prior to performing a bus reset. o Perform the sequencer restart after a chip reset prior to setting up our timer to poll for the reset to be complete. On some OSes, the timer could actually pre-empt us and order is important here. o Have our "reset poller" set the expected mode since there is no guarantee of what mode will be in force when we are called from the OS timer. o Save and restore the SCB pointer in ahd_dump_card_state(). This routine must not modify card state. o Ditto for ahd_dump_scbs(). aic79xx.h: o Add a few more chip bug definitions. o Align our tag on a 32bit boundary. aic79xx.reg: aic79xx.seq: o Start work on removing workarounds for Rev B. o Use a special location in scratch from for stroring our SCBPTR during legacy FIFO allocations. This corrects problems in mixed packetized/non-packetized configurations where calling into a FIFO task corrupted our SCBPTR. o Don't rely on DMA priority to guarantee that all data in our FIFOs will flush prior to a command completion notification going out of the command channel. We've never seen this assumption fail, but better safe than sorry. o Deal with missing BUSFREEREV feature in H2A. o Simplify disconnect list code now that the list will always have only a single entry. o Implement the AHD_REG_SLOW_SETTLE_BUG workaround. o Swith to using "REG_ISR" for local mode scratch during our ISR. o Add a missing jmp to the data_group_dma_loop after our data pointers have been re-initialized by the kernel. o Correct test in the bitbucket code so that we actually wait for the bitbucket to complete before signaling the kernel of the overrun condition. o Reposition pkt_saveptrs to avoid a jmp instruction. o Update a comment to reflect that the code now waits for a FIFO to drain prior to issuing a CLRCHN. aic79xx_inline.h: o Remove unused untagged queue handling code. o Don't attempt to htole64 what could be a 32bit value. aic79xx_pci.c: o Set additional bug flags for rev A chips.
2002-09-26 22:54:00 +00:00
if ((ahd->bugs & AHD_BUSFREEREV_BUG) != 0) {
/*
ahc_eisa.c: ahc_pci.c: ahd_pci.c: aic7xxx.c: aic79xx.c: aic_osm_lib.c: aic_osm_lib.h: Use common OSM routines from aic_osm_lib for bus dma operations, delay routines, accessing CCBs, byte swapping, etc. aic7xxx_pci.c: Provide a better description for the 2915/30LP on attach. aic7xxx.c: aic79xx.c: aic7770.c: aic79xx_pci.c: aic7xxx_pci.c: aic7xxx_93cx6.c: Move FBSDID behind an ifdef so that these core files will still compile under other OSes. aic79xx.h: aic79xx_pci.c: aic79xx.seq: To speed up non-packetized CDB delivery in Rev B, all CDB acks are "released" to the output sync as soon as the command phase starts. There is only one problem with this approach. If the target changes phase before all data are sent, we have left over acks that can go out on the bus in a data phase. Due to other chip contraints, this only happens if the target goes to data-in, but if the acks go out before we can test SDONE, we'll think that the transfer has completed successfully. Work around this by taking advantage of the 400ns or 800ns dead time between command phase and the REQ of the new phase. If the transfer has completed successfully, SCSIEN should fall *long* before we see a phase change. We thus treat any phasemiss that occurs before SCSIEN falls as an incomplete transfer. aic79xx.h: Add the AHD_FAST_CDB_DELIVERY feature. aic79xx_pci.c: Set AHD_FAST_CDB_DELIVERY for all Rev. B parts. aic79xx.seq: Test for PHASEMIS in the command phase for all AHD_FAST_CDB_DELIVERY controlelrs. ahd_pci.c: ahc_pci.c: aic7xxx.h: aic79xx.h: Move definition of controller BAR offsets to core header files. aic7xxx.c: aic79xx.c: In the softc free routine, leave removal of a softc from the global list of softcs to the OSM (the caller of this routine). This allows us to avoid holding the softc list_lock during device destruction where we may have to sleep waiting for our recovery thread to halt. ahc_pci.c: Use ahc_pci_test_register access to validate I/O mapped in addition to the tests already performed for memory mapped access. Remove unused ahc_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. ahd_pci.c: Remove reduntant definition of controller BAR offsets. These are also defined in aic79xx.h. Remove unused ahd_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. aic7xxx.c: aic79xx.c: aic79xx.h: aic7xxx.h: aic7xxx_osm.c: aic79xx_osm.c: Move timeout handling to the driver cores. In the case of the aic79xx driver, the algorithm has been enhanced to try target resets before performing a bus reset. For the aic7xxx driver, the algorithm is unchanged. Although the drivers do not currently sleep during recovery (recovery is timeout driven), the cores do expect all processing to be performed via a recovery thread. Our timeout handlers are now little stubs that wakeup the recovery thread. aic79xx.c: aic79xx.h: aic79xx_inline.h: Change shared_data allocation to use a map_node so that the sentinel hscb can use this map node in ahd_swap_with_next_hscb. This routine now swaps the hscb_map pointer in additon to the hscb contents so that any sync operations occur on the correct map. physaddr -> busaddr Pointed out by: Jason Thorpe <thorpej@wasabisystems.com> aic79xx.c: Make more use of the in/out/w/l/q macros for accessing byte registers in the chip. Correct some issues in the ahd_flush_qoutfifo() routine. o Run the qoutfifo only once the command channel DMA engine has been halted. This closes a window where we might have missed some entries. o Change ahd_run_data_fifo() to not loop to completion. If we happen to start on the wrong FIFO and the other FIFO has a snapshot savepointers, we might deadlock. This required our delay between FIFO tests to be moved to the ahd_flush_qoutfifo() routine. o Update/add comments. o Remove spurious test for COMPLETE_DMA list being empty when completing transactions from the GSFIFO with residuals. The SCB must be put on the COMPLETE_DMA scb list unconditionally. o When halting command channel DMA activity, we must disable the DMA channel in all cases but an update of the QOUTFIFO. The latter case is required so that the sequencer will update its position in the QOUTFIFO. Previously, we left the channel enabled for all "push" DMAs. This left us vulnerable to the sequencer handling an SCB push long after that SCB was already processed manually by this routine. o Correct the polarity of tests involving ahd_scb_active_in_fifo(). This routine returns non-zero for true. Return to processing bad status completions through the qoutfifo. This reduces the time that the sequencer is kept paused when handling transactions with bad status or underruns. When waiting for the controller to quiece selections, add a delay to our loop. Otherwise we may fail to wait long enough for the sequencer to comply. On H2A4 hardware, use the slow slewrate for non-paced transfers. This mirrors what the Adaptec Windows drivers do. On the Rev B. only slow down the CRC timing for older U160 devices that might need the slower timing. We define "older" as devices that do not support packetized protocol. Wait up to 5000 * 5us for the SEEPROM to become unbusy. Write ops seem to take much longer than read ops. aic79xx.seq: For controllers with the FAINT_LED bug, turn the diagnostic led feature on during selection and reselection. This covers the non-packetized case. The LED will be disabled for non-packetized transfers once we return to the top level idle loop. Add more comments about the busy LED workaround. Extend a critical section around the entire command channel idle loop process. Previously the portion of this handler that directly manipulated the linked list of completed SCBs was not protected. This is the likely cause of the recent reports of commands being completed twice by the driver. Extend critical sections across the test for, and the longjump to, longjump routines. This prevents the firmware from trying to jump to a longjmp handler that was just cleared by the host. Improve the locations of several critical section begin and end points. Typically these changes remove instructions that did not need to be inside a critical section. Close the "busfree after selection, but before busfree interrupts can be enabled" race to just a single sequencer instruction. We now test the BSY line explicitly before clearing the busfree status and enabling the busfree interrupt. Close a race condition in the processing of HS_MAILBOX updates. We now clear the "updated" status before the copy. This ensures that we don't accidentally clear the status incorrectly when the host sneaks in an update just after our last copy, but before we clear the status. This race has never been observed. Don't re-enable SCSIEN if we lose the race to disable SCSIEN in our interrupt handler's workaround for the RevA data-valid too early issue. aic79xx_inline.h: Add comments indicating that the order in which bytes are read or written in ahd_inw and ahd_outw is important. This allows us to use these inlines when accessing registers with side-effects. aic79xx_pci.c: The 29320 and the 29320B are 7902 not 7901 based products. Correct the driver banner. aic7xxx.h: Enable the use of the auto-access pause feature on the aic7870 and aic7880. It was disabled due to an oversight. aic7xxx.reg: Move TARG_IMMEDIATE_SCB to alias LAST_MSG to avoid leaving garbage in MWI_RESIDUAL. This prevents spurious overflows whn operating target mode on controllers that require the MWI_RESIDUAL work-around. aic7xxx.seq: AHC_TMODE_WIDEODD_BUG is a bug, not a softc flag. Reference the correct softc field when testing for its presence. Set the NOT_IDENTIFIED and NO_CDB_SENT bits in SEQ_FLAGS to indicate that the nexus is invalid in await busfree. aic7xxx_93cx6.c: Add support for the C56/C66 versions of the EWEN and EWDS commands. aic7xxx.c: aic7xxx_pci.c: Move test for the validity of left over BIOS data to ahc_test_register_access(). This guarantees that any left over CHIPRST value is not clobbered by our register access test and lost to the test that was in ahc_reset.
2003-12-17 00:02:10 +00:00
* Test to ensure that the bus has not
* already gone free prior to clearing
* any stale busfree status. This avoids
* a window whereby a busfree just after
* a selection could be missed.
Upgrade to version 1.1 of the aic79xx U320 driver. aic79xx.c: o Remove redundant ahd_update_modes() call. o Correct panic in diagnostic should state corruption cause the SCB Id to be invalid during a selection timeout. o Add workaround for missing BUSFREEREV feature in Rev A silicon. o Corect formatting nits. o Use register pretty printing in more places. o Save and restore our SCB pointer when updating the waiting queue list for an "expected" LQ-out busfree. o In ahd_clear_intstat, deal with the missing autoclear in the CLRLQO* registers. o BE fixup in a diagnostic printf. o Make sure that we are in the proper mode before disabling selections in ahd_update_pending_scbs. o Add more diagnostics. o task_attribute_nonpkt_tag -> task_attribute: we don't need a nonpkt_tag field anymore for allowing all 512 SCBs to be used in non-packetized connections. o Negotiate HOLD_MCS to U320 devices. o Add a few additional mode assertions. o Restore the chip mode after clearing out the qinfifo so that code using ahd_abort_scbs sees a consistent mode. o Simplify the DMA engine shutdown routine prior to performing a bus reset. o Perform the sequencer restart after a chip reset prior to setting up our timer to poll for the reset to be complete. On some OSes, the timer could actually pre-empt us and order is important here. o Have our "reset poller" set the expected mode since there is no guarantee of what mode will be in force when we are called from the OS timer. o Save and restore the SCB pointer in ahd_dump_card_state(). This routine must not modify card state. o Ditto for ahd_dump_scbs(). aic79xx.h: o Add a few more chip bug definitions. o Align our tag on a 32bit boundary. aic79xx.reg: aic79xx.seq: o Start work on removing workarounds for Rev B. o Use a special location in scratch from for stroring our SCBPTR during legacy FIFO allocations. This corrects problems in mixed packetized/non-packetized configurations where calling into a FIFO task corrupted our SCBPTR. o Don't rely on DMA priority to guarantee that all data in our FIFOs will flush prior to a command completion notification going out of the command channel. We've never seen this assumption fail, but better safe than sorry. o Deal with missing BUSFREEREV feature in H2A. o Simplify disconnect list code now that the list will always have only a single entry. o Implement the AHD_REG_SLOW_SETTLE_BUG workaround. o Swith to using "REG_ISR" for local mode scratch during our ISR. o Add a missing jmp to the data_group_dma_loop after our data pointers have been re-initialized by the kernel. o Correct test in the bitbucket code so that we actually wait for the bitbucket to complete before signaling the kernel of the overrun condition. o Reposition pkt_saveptrs to avoid a jmp instruction. o Update a comment to reflect that the code now waits for a FIFO to drain prior to issuing a CLRCHN. aic79xx_inline.h: o Remove unused untagged queue handling code. o Don't attempt to htole64 what could be a 32bit value. aic79xx_pci.c: o Set additional bug flags for rev A chips.
2002-09-26 22:54:00 +00:00
*/
ahc_eisa.c: ahc_pci.c: ahd_pci.c: aic7xxx.c: aic79xx.c: aic_osm_lib.c: aic_osm_lib.h: Use common OSM routines from aic_osm_lib for bus dma operations, delay routines, accessing CCBs, byte swapping, etc. aic7xxx_pci.c: Provide a better description for the 2915/30LP on attach. aic7xxx.c: aic79xx.c: aic7770.c: aic79xx_pci.c: aic7xxx_pci.c: aic7xxx_93cx6.c: Move FBSDID behind an ifdef so that these core files will still compile under other OSes. aic79xx.h: aic79xx_pci.c: aic79xx.seq: To speed up non-packetized CDB delivery in Rev B, all CDB acks are "released" to the output sync as soon as the command phase starts. There is only one problem with this approach. If the target changes phase before all data are sent, we have left over acks that can go out on the bus in a data phase. Due to other chip contraints, this only happens if the target goes to data-in, but if the acks go out before we can test SDONE, we'll think that the transfer has completed successfully. Work around this by taking advantage of the 400ns or 800ns dead time between command phase and the REQ of the new phase. If the transfer has completed successfully, SCSIEN should fall *long* before we see a phase change. We thus treat any phasemiss that occurs before SCSIEN falls as an incomplete transfer. aic79xx.h: Add the AHD_FAST_CDB_DELIVERY feature. aic79xx_pci.c: Set AHD_FAST_CDB_DELIVERY for all Rev. B parts. aic79xx.seq: Test for PHASEMIS in the command phase for all AHD_FAST_CDB_DELIVERY controlelrs. ahd_pci.c: ahc_pci.c: aic7xxx.h: aic79xx.h: Move definition of controller BAR offsets to core header files. aic7xxx.c: aic79xx.c: In the softc free routine, leave removal of a softc from the global list of softcs to the OSM (the caller of this routine). This allows us to avoid holding the softc list_lock during device destruction where we may have to sleep waiting for our recovery thread to halt. ahc_pci.c: Use ahc_pci_test_register access to validate I/O mapped in addition to the tests already performed for memory mapped access. Remove unused ahc_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. ahd_pci.c: Remove reduntant definition of controller BAR offsets. These are also defined in aic79xx.h. Remove unused ahd_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. aic7xxx.c: aic79xx.c: aic79xx.h: aic7xxx.h: aic7xxx_osm.c: aic79xx_osm.c: Move timeout handling to the driver cores. In the case of the aic79xx driver, the algorithm has been enhanced to try target resets before performing a bus reset. For the aic7xxx driver, the algorithm is unchanged. Although the drivers do not currently sleep during recovery (recovery is timeout driven), the cores do expect all processing to be performed via a recovery thread. Our timeout handlers are now little stubs that wakeup the recovery thread. aic79xx.c: aic79xx.h: aic79xx_inline.h: Change shared_data allocation to use a map_node so that the sentinel hscb can use this map node in ahd_swap_with_next_hscb. This routine now swaps the hscb_map pointer in additon to the hscb contents so that any sync operations occur on the correct map. physaddr -> busaddr Pointed out by: Jason Thorpe <thorpej@wasabisystems.com> aic79xx.c: Make more use of the in/out/w/l/q macros for accessing byte registers in the chip. Correct some issues in the ahd_flush_qoutfifo() routine. o Run the qoutfifo only once the command channel DMA engine has been halted. This closes a window where we might have missed some entries. o Change ahd_run_data_fifo() to not loop to completion. If we happen to start on the wrong FIFO and the other FIFO has a snapshot savepointers, we might deadlock. This required our delay between FIFO tests to be moved to the ahd_flush_qoutfifo() routine. o Update/add comments. o Remove spurious test for COMPLETE_DMA list being empty when completing transactions from the GSFIFO with residuals. The SCB must be put on the COMPLETE_DMA scb list unconditionally. o When halting command channel DMA activity, we must disable the DMA channel in all cases but an update of the QOUTFIFO. The latter case is required so that the sequencer will update its position in the QOUTFIFO. Previously, we left the channel enabled for all "push" DMAs. This left us vulnerable to the sequencer handling an SCB push long after that SCB was already processed manually by this routine. o Correct the polarity of tests involving ahd_scb_active_in_fifo(). This routine returns non-zero for true. Return to processing bad status completions through the qoutfifo. This reduces the time that the sequencer is kept paused when handling transactions with bad status or underruns. When waiting for the controller to quiece selections, add a delay to our loop. Otherwise we may fail to wait long enough for the sequencer to comply. On H2A4 hardware, use the slow slewrate for non-paced transfers. This mirrors what the Adaptec Windows drivers do. On the Rev B. only slow down the CRC timing for older U160 devices that might need the slower timing. We define "older" as devices that do not support packetized protocol. Wait up to 5000 * 5us for the SEEPROM to become unbusy. Write ops seem to take much longer than read ops. aic79xx.seq: For controllers with the FAINT_LED bug, turn the diagnostic led feature on during selection and reselection. This covers the non-packetized case. The LED will be disabled for non-packetized transfers once we return to the top level idle loop. Add more comments about the busy LED workaround. Extend a critical section around the entire command channel idle loop process. Previously the portion of this handler that directly manipulated the linked list of completed SCBs was not protected. This is the likely cause of the recent reports of commands being completed twice by the driver. Extend critical sections across the test for, and the longjump to, longjump routines. This prevents the firmware from trying to jump to a longjmp handler that was just cleared by the host. Improve the locations of several critical section begin and end points. Typically these changes remove instructions that did not need to be inside a critical section. Close the "busfree after selection, but before busfree interrupts can be enabled" race to just a single sequencer instruction. We now test the BSY line explicitly before clearing the busfree status and enabling the busfree interrupt. Close a race condition in the processing of HS_MAILBOX updates. We now clear the "updated" status before the copy. This ensures that we don't accidentally clear the status incorrectly when the host sneaks in an update just after our last copy, but before we clear the status. This race has never been observed. Don't re-enable SCSIEN if we lose the race to disable SCSIEN in our interrupt handler's workaround for the RevA data-valid too early issue. aic79xx_inline.h: Add comments indicating that the order in which bytes are read or written in ahd_inw and ahd_outw is important. This allows us to use these inlines when accessing registers with side-effects. aic79xx_pci.c: The 29320 and the 29320B are 7902 not 7901 based products. Correct the driver banner. aic7xxx.h: Enable the use of the auto-access pause feature on the aic7870 and aic7880. It was disabled due to an oversight. aic7xxx.reg: Move TARG_IMMEDIATE_SCB to alias LAST_MSG to avoid leaving garbage in MWI_RESIDUAL. This prevents spurious overflows whn operating target mode on controllers that require the MWI_RESIDUAL work-around. aic7xxx.seq: AHC_TMODE_WIDEODD_BUG is a bug, not a softc flag. Reference the correct softc field when testing for its presence. Set the NOT_IDENTIFIED and NO_CDB_SENT bits in SEQ_FLAGS to indicate that the nexus is invalid in await busfree. aic7xxx_93cx6.c: Add support for the C56/C66 versions of the EWEN and EWDS commands. aic7xxx.c: aic7xxx_pci.c: Move test for the validity of left over BIOS data to ahc_test_register_access(). This guarantees that any left over CHIPRST value is not clobbered by our register access test and lost to the test that was in ahc_reset.
2003-12-17 00:02:10 +00:00
test SCSISIGI, BSYI jz . + 2;
Upgrade to version 1.1 of the aic79xx U320 driver. aic79xx.c: o Remove redundant ahd_update_modes() call. o Correct panic in diagnostic should state corruption cause the SCB Id to be invalid during a selection timeout. o Add workaround for missing BUSFREEREV feature in Rev A silicon. o Corect formatting nits. o Use register pretty printing in more places. o Save and restore our SCB pointer when updating the waiting queue list for an "expected" LQ-out busfree. o In ahd_clear_intstat, deal with the missing autoclear in the CLRLQO* registers. o BE fixup in a diagnostic printf. o Make sure that we are in the proper mode before disabling selections in ahd_update_pending_scbs. o Add more diagnostics. o task_attribute_nonpkt_tag -> task_attribute: we don't need a nonpkt_tag field anymore for allowing all 512 SCBs to be used in non-packetized connections. o Negotiate HOLD_MCS to U320 devices. o Add a few additional mode assertions. o Restore the chip mode after clearing out the qinfifo so that code using ahd_abort_scbs sees a consistent mode. o Simplify the DMA engine shutdown routine prior to performing a bus reset. o Perform the sequencer restart after a chip reset prior to setting up our timer to poll for the reset to be complete. On some OSes, the timer could actually pre-empt us and order is important here. o Have our "reset poller" set the expected mode since there is no guarantee of what mode will be in force when we are called from the OS timer. o Save and restore the SCB pointer in ahd_dump_card_state(). This routine must not modify card state. o Ditto for ahd_dump_scbs(). aic79xx.h: o Add a few more chip bug definitions. o Align our tag on a 32bit boundary. aic79xx.reg: aic79xx.seq: o Start work on removing workarounds for Rev B. o Use a special location in scratch from for stroring our SCBPTR during legacy FIFO allocations. This corrects problems in mixed packetized/non-packetized configurations where calling into a FIFO task corrupted our SCBPTR. o Don't rely on DMA priority to guarantee that all data in our FIFOs will flush prior to a command completion notification going out of the command channel. We've never seen this assumption fail, but better safe than sorry. o Deal with missing BUSFREEREV feature in H2A. o Simplify disconnect list code now that the list will always have only a single entry. o Implement the AHD_REG_SLOW_SETTLE_BUG workaround. o Swith to using "REG_ISR" for local mode scratch during our ISR. o Add a missing jmp to the data_group_dma_loop after our data pointers have been re-initialized by the kernel. o Correct test in the bitbucket code so that we actually wait for the bitbucket to complete before signaling the kernel of the overrun condition. o Reposition pkt_saveptrs to avoid a jmp instruction. o Update a comment to reflect that the code now waits for a FIFO to drain prior to issuing a CLRCHN. aic79xx_inline.h: o Remove unused untagged queue handling code. o Don't attempt to htole64 what could be a 32bit value. aic79xx_pci.c: o Set additional bug flags for rev A chips.
2002-09-26 22:54:00 +00:00
mvi CLRSINT1,CLRBUSFREE;
or SIMODE1, ENBUSFREE;
}
or SXFRCTL0, SPIOEN;
and SAVED_SCSIID, SELID_MASK, SELID;
and A, OID, IOWNID;
or SAVED_SCSIID, A;
mvi CLRSINT0, CLRSELDI;
jmp ITloop;
/*
* We have successfully selected out.
*
* Clear SELDO.
* Dequeue all SCBs sent from the waiting queue
* Requeue all SCBs *not* sent to the tail of the waiting queue
* Take Razor #494 into account for above.
*
* In Packetized Mode:
* Return to the idle loop. Our interrupt handler will take
* care of any incoming L_Qs.
*
* In Non-Packetize Mode:
* Continue to our normal state machine.
*/
SET_SRC_MODE M_SCSI;
SET_DST_MODE M_SCSI;
select_out:
BEGIN_CRITICAL;
ahc_eisa.c: ahc_pci.c: ahd_pci.c: aic7xxx.c: aic79xx.c: aic_osm_lib.c: aic_osm_lib.h: Use common OSM routines from aic_osm_lib for bus dma operations, delay routines, accessing CCBs, byte swapping, etc. aic7xxx_pci.c: Provide a better description for the 2915/30LP on attach. aic7xxx.c: aic79xx.c: aic7770.c: aic79xx_pci.c: aic7xxx_pci.c: aic7xxx_93cx6.c: Move FBSDID behind an ifdef so that these core files will still compile under other OSes. aic79xx.h: aic79xx_pci.c: aic79xx.seq: To speed up non-packetized CDB delivery in Rev B, all CDB acks are "released" to the output sync as soon as the command phase starts. There is only one problem with this approach. If the target changes phase before all data are sent, we have left over acks that can go out on the bus in a data phase. Due to other chip contraints, this only happens if the target goes to data-in, but if the acks go out before we can test SDONE, we'll think that the transfer has completed successfully. Work around this by taking advantage of the 400ns or 800ns dead time between command phase and the REQ of the new phase. If the transfer has completed successfully, SCSIEN should fall *long* before we see a phase change. We thus treat any phasemiss that occurs before SCSIEN falls as an incomplete transfer. aic79xx.h: Add the AHD_FAST_CDB_DELIVERY feature. aic79xx_pci.c: Set AHD_FAST_CDB_DELIVERY for all Rev. B parts. aic79xx.seq: Test for PHASEMIS in the command phase for all AHD_FAST_CDB_DELIVERY controlelrs. ahd_pci.c: ahc_pci.c: aic7xxx.h: aic79xx.h: Move definition of controller BAR offsets to core header files. aic7xxx.c: aic79xx.c: In the softc free routine, leave removal of a softc from the global list of softcs to the OSM (the caller of this routine). This allows us to avoid holding the softc list_lock during device destruction where we may have to sleep waiting for our recovery thread to halt. ahc_pci.c: Use ahc_pci_test_register access to validate I/O mapped in addition to the tests already performed for memory mapped access. Remove unused ahc_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. ahd_pci.c: Remove reduntant definition of controller BAR offsets. These are also defined in aic79xx.h. Remove unused ahd_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. aic7xxx.c: aic79xx.c: aic79xx.h: aic7xxx.h: aic7xxx_osm.c: aic79xx_osm.c: Move timeout handling to the driver cores. In the case of the aic79xx driver, the algorithm has been enhanced to try target resets before performing a bus reset. For the aic7xxx driver, the algorithm is unchanged. Although the drivers do not currently sleep during recovery (recovery is timeout driven), the cores do expect all processing to be performed via a recovery thread. Our timeout handlers are now little stubs that wakeup the recovery thread. aic79xx.c: aic79xx.h: aic79xx_inline.h: Change shared_data allocation to use a map_node so that the sentinel hscb can use this map node in ahd_swap_with_next_hscb. This routine now swaps the hscb_map pointer in additon to the hscb contents so that any sync operations occur on the correct map. physaddr -> busaddr Pointed out by: Jason Thorpe <thorpej@wasabisystems.com> aic79xx.c: Make more use of the in/out/w/l/q macros for accessing byte registers in the chip. Correct some issues in the ahd_flush_qoutfifo() routine. o Run the qoutfifo only once the command channel DMA engine has been halted. This closes a window where we might have missed some entries. o Change ahd_run_data_fifo() to not loop to completion. If we happen to start on the wrong FIFO and the other FIFO has a snapshot savepointers, we might deadlock. This required our delay between FIFO tests to be moved to the ahd_flush_qoutfifo() routine. o Update/add comments. o Remove spurious test for COMPLETE_DMA list being empty when completing transactions from the GSFIFO with residuals. The SCB must be put on the COMPLETE_DMA scb list unconditionally. o When halting command channel DMA activity, we must disable the DMA channel in all cases but an update of the QOUTFIFO. The latter case is required so that the sequencer will update its position in the QOUTFIFO. Previously, we left the channel enabled for all "push" DMAs. This left us vulnerable to the sequencer handling an SCB push long after that SCB was already processed manually by this routine. o Correct the polarity of tests involving ahd_scb_active_in_fifo(). This routine returns non-zero for true. Return to processing bad status completions through the qoutfifo. This reduces the time that the sequencer is kept paused when handling transactions with bad status or underruns. When waiting for the controller to quiece selections, add a delay to our loop. Otherwise we may fail to wait long enough for the sequencer to comply. On H2A4 hardware, use the slow slewrate for non-paced transfers. This mirrors what the Adaptec Windows drivers do. On the Rev B. only slow down the CRC timing for older U160 devices that might need the slower timing. We define "older" as devices that do not support packetized protocol. Wait up to 5000 * 5us for the SEEPROM to become unbusy. Write ops seem to take much longer than read ops. aic79xx.seq: For controllers with the FAINT_LED bug, turn the diagnostic led feature on during selection and reselection. This covers the non-packetized case. The LED will be disabled for non-packetized transfers once we return to the top level idle loop. Add more comments about the busy LED workaround. Extend a critical section around the entire command channel idle loop process. Previously the portion of this handler that directly manipulated the linked list of completed SCBs was not protected. This is the likely cause of the recent reports of commands being completed twice by the driver. Extend critical sections across the test for, and the longjump to, longjump routines. This prevents the firmware from trying to jump to a longjmp handler that was just cleared by the host. Improve the locations of several critical section begin and end points. Typically these changes remove instructions that did not need to be inside a critical section. Close the "busfree after selection, but before busfree interrupts can be enabled" race to just a single sequencer instruction. We now test the BSY line explicitly before clearing the busfree status and enabling the busfree interrupt. Close a race condition in the processing of HS_MAILBOX updates. We now clear the "updated" status before the copy. This ensures that we don't accidentally clear the status incorrectly when the host sneaks in an update just after our last copy, but before we clear the status. This race has never been observed. Don't re-enable SCSIEN if we lose the race to disable SCSIEN in our interrupt handler's workaround for the RevA data-valid too early issue. aic79xx_inline.h: Add comments indicating that the order in which bytes are read or written in ahd_inw and ahd_outw is important. This allows us to use these inlines when accessing registers with side-effects. aic79xx_pci.c: The 29320 and the 29320B are 7902 not 7901 based products. Correct the driver banner. aic7xxx.h: Enable the use of the auto-access pause feature on the aic7870 and aic7880. It was disabled due to an oversight. aic7xxx.reg: Move TARG_IMMEDIATE_SCB to alias LAST_MSG to avoid leaving garbage in MWI_RESIDUAL. This prevents spurious overflows whn operating target mode on controllers that require the MWI_RESIDUAL work-around. aic7xxx.seq: AHC_TMODE_WIDEODD_BUG is a bug, not a softc flag. Reference the correct softc field when testing for its presence. Set the NOT_IDENTIFIED and NO_CDB_SENT bits in SEQ_FLAGS to indicate that the nexus is invalid in await busfree. aic7xxx_93cx6.c: Add support for the C56/C66 versions of the EWEN and EWDS commands. aic7xxx.c: aic7xxx_pci.c: Move test for the validity of left over BIOS data to ahc_test_register_access(). This guarantees that any left over CHIPRST value is not clobbered by our register access test and lost to the test that was in ahc_reset.
2003-12-17 00:02:10 +00:00
if ((ahd->bugs & AHD_FAINT_LED_BUG) != 0) {
/*
* On Rev A. hardware, the busy LED is only
* turned on automaically during selections
* and re-selections. Make the LED status
* more useful by forcing it to be on from
* the point of re-selection until our idle
* loop determines that neither of our FIFOs
* are busy. This handles the non-packetized
* case nicely as we will not return to the
* idle loop until the busfree at the end of
* each transaction.
*/
or SBLKCTL, DIAGLEDEN|DIAGLEDON;
}
/* Clear out all SCBs that have been successfully sent. */
if ((ahd->bugs & AHD_SENT_SCB_UPDATE_BUG) != 0) {
/*
* For packetized, the LQO manager clears ENSELO on
* the assertion of SELDO. If we are non-packetized,
aic79xx.c: Clear the LQICRC_NLQ status should it pop up after we have already handled the SCSIPERR. During some streaming operations this status can be delayed until the stream ends. Without this change, the driver would complain about a "Missing case in ahd_handle_scsiint". In the LQOBUSFREE handler... Don't return the LQOMGR back to the idle state until after we have cleaned up ENSELO and any status related to this selection. The last thing we need is the LQO manager starting another select-out before we have updated the execution queue. It is not clear whether the LQOMGR would, or would not start a new selection early. Make sure ENSELO is off prior to clearing SELDO by flushing device writes. Move assignment of the next target SCB pointer inside of an if to make the code clearer. The effect is the same. Dump card state in both "Unexpected PKT busfree" paths. In ahd_reset(), set the chip to SCSI mode before reading SXFRCTL1. That register only exists in the SCSI mode. Also set the mode explicitly to the SCSI mode after chip reset due to paranoia. Re-arrange code so that SXFRCTL1 is restored as quickly after the chip reset as possible. S/G structurs must be 8byte aligned. Make this official by saying so in our DMA tag. Disable CIO bus stretch on MDFFSTAT if SHVALID is about to come true. This can cause a CIO bus lockup if a PCI or PCI-X error occurs while the stretch is occurring - the host cannot service the PCI-X error since the CIO bus is locked out and SHVALID will never resolve. The stretch was added in the Rev B to simplify the wait for SHVALID to resolve, but the code to do this in the open source sequencer is so simple it was never removed. Consistently use MAX_OFFSET for the user max syncrate set from non-volatile storage. This ensures that the offset does not conflict with AH?_OFFSET_UNKNOWN. Have ahd_pause_and_flushwork set the mode to ensure that it has access to the registers it checks. Also modify the checking of intstat so that the check against 0xFF can actually succeed if the INT_PEND mask is something other than 0xFF. Although there are no cardbus U320 controllers, this check may be needed to recover from a hot-plug PCI removal that occurs without informing the driver. Fix a typo. sg_prefetch_cnt -> sg_prefetch_align. This fixes an infinite loop at card initialization if the cacheline size is 0. aic79xx.h: Add AHD_EARLY_REQ_BUG bug flag. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. Change AH?_OFFSET_UNKNOWN to 0xFF. This is a value that the curr->offset can never be, unlike '0' which we previously used. This fixes code that only checks for a non-zero offset to determine if a sync negotiation is required since it will fire in the unknown case even if the goal is async. aic79xx.reg: Add comments for LQISTAT bits indicating their names in the 7902 data book. We use slightly different and more descriptive names in the firmware. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. aic79xx.seq: Update comments regarding rundown of the GSFIFO to reflect reality. Fix spelling errors. Since we use an 8byte address and 1 byte length, shorten the size of a block move for the legacy DMA'ed CDB case from 11 to 9 bytes. Remove code that, assuming the abort pending feature worked, would set MK_MESSAGE in the SCB's control byte on completion to catch invalid reselections. Since we don't see interrupts for completed selections, this status update could occur prior to us noticing the SELDO. The "select-out" queue logic will get confused by the MK_MESSAGE bit being set as this is used to catch packatized connections where we select-out with ATN. Since the abort pending feature doesn't work on any released controllers yet, this code was never executed. Add support for the AHD_EARLY_REQ_BUG. Don't ignore persistent REQ assertions just because they were asserted within the bus settle delay window. This allows us to tolerate devices like the GEM318 that violate the SCSI spec. Remove unintentional settnig of SG_CACHE_AVAIL. Writing this bit should have no effect, but who knows... On the Rev A, we must wait for HDMAENACK before loading additional segments to avoid clobbering the address of the first segment in the S/G FIFO. This resolves data-corruption issues with certain IBM (now Hitachi) and Fujitsu U320 drives. Rearrange calc_residual to avoid an extra jmp instruction. On RevA Silicon, if the target returns us to data-out after we have already trained for data-out, it is possible for us to transition the free running clock to data-valid before the required 100ns P1 setup time (8 P1 assertions in fast-160 mode). This will only happen if this L-Q is a continuation of a data transfer for which we have already prefetched data into our FIFO (LQ/Data followed by LQ/Data for the same write transaction). This can cause some target implementations to miss the first few data transfers on the bus. We detect this situation by noticing that this is the first data transfer after an LQ (LQIWORKONLQ true), that the data transfer is a continuation of a transfer already setup in our FIFO (SAVEPTRS interrupt), and that the transaction is a write (DIRECTION set in DFCNTRL). The delay is performed by disabling SCSIEN until we see the first REQ from the target. Only compile in snapshot savepointers handler for RevA silicon where it is enabled. Handle the cfg4icmd packetized interrupt. We just need to load the address and count, start the DMA, and CLRCHN once the transfer is complete. Fix an oversight in the overrun handler for packetized status operations. We need to wait for either CTXTDONE or an overrun when checking for an overrun. The previous code did not wait and thus could decide that no overrun had occurred even though an overrun will occur on the next data-valid req. Add some comment to this section for clarity. Use LAST_SEG_DONE instead of LASTSDONE for testing transfer completion in the packetized status case. LASTSDONE may come up more quickly since it only records completion on the SCSI side, but since LAST_SEG_DONE is used everywhere else (and needs to be), this is less confusing. Add a missing invalidation of the longjmp address in the non-pack handler. This code needs additional review. aic79xx_inline.h: Fix spelling error. aic79xx_osm.c: Set the cdb length for CDBs dma'ed from host memory. Add a comment indicating that, should CAM start supporting cdbs larger than 16bytes, the driver could store the CDB in the status buffer. aic79xx_pci.c: Add a table entry for the 39320A. Added a missing comma to an error string table. Fix spelling errors.
2003-02-27 23:23:16 +00:00
* LASTSCB and CURRSCB are accurate.
*/
test SCSISEQ0, ENSELO jnz use_lastscb;
/*
* The update is correct for LQOSTAT1 errors. All
* but LQOBUSFREE are handled by kernel interrupts.
* If we see LQOBUSFREE, return to the idle loop.
* Once we are out of the select_out critical section,
* the kernel will cleanup the LQOBUSFREE and we will
* eventually restart the selection if appropriate.
*/
test LQOSTAT1, LQOBUSFREE jnz idle_loop;
/*
* On a phase change oustside of packet boundaries,
* LASTSCB points to the currently active SCB context
* on the bus.
*/
test LQOSTAT2, LQOPHACHGOUTPKT jnz use_lastscb;
/*
* If the hardware has traversed the whole list, NEXTSCB
* will be NULL, CURRSCB and LASTSCB cannot be trusted,
* but MAXCMDCNT is accurate. If we stop part way through
* the list or only had one command to issue, NEXTSCB[1] is
* not NULL and LASTSCB is the last command to go out.
*/
cmp NEXTSCB[1], SCB_LIST_NULL jne use_lastscb;
/*
* Brute force walk.
*/
bmov SCBPTR, WAITING_TID_HEAD, 2;
mvi SEQINTCTL, INTVEC1DSL;
mvi MODE_PTR, MK_MODE(M_CFG, M_CFG);
mov A, MAXCMDCNT;
mvi MODE_PTR, MK_MODE(M_SCSI, M_SCSI);
clr SEQINTCTL;
find_lastscb_loop:
dec A;
test A, 0xFF jz found_last_sent_scb;
bmov SCBPTR, SCB_NEXT, 2;
jmp find_lastscb_loop;
use_lastscb:
bmov SCBPTR, LASTSCB, 2;
found_last_sent_scb:
bmov CURRSCB, SCBPTR, 2;
curscb_ww_done:
} else {
bmov SCBPTR, CURRSCB, 2;
}
/*
Correct a very rare case where command ordering could be compromised by a transaction performing a driver handled message sequence (an scb with the MK_MESSAGE flag set). SCBs that perform host managed messaging must always be at the head of their per-target selection queue so that the firmware knows to manually assert ATN if the current negotiation agreement is packetized. In the past we guaranteed this by queuing these SCBs separarately in the execution queue. This exposes the system to potential command reordering in two cases: 1) Another SCB for the same ITL nexus is queued that does not have the MK_MESSAGE flag set. This SCB will be queued to the per-target list which can be serviced before the MK_MESSAGE scb that preceeded it. 2) If the target cannot accept all of the commands in the per-target selection queue in one selection, the remainder is queued to the tail of the selection queues so as to effect round-robin scheduling. This could allow the MK_MESSAGE scb to be sent to the target before the requeued commands. This commit changes the firmware policy to defer queuing MK_MESSAGE SCBs into the selection queues until this can be done without affecting order. This means that the target's selection queue is either empty, or the last SCB on the execution queue is also a MK_MESSAGE SCB. During any wait, the firmware halts the download of new SCBs so only a single "holding location" is required. Luckily, MK_MESSAGE SCBs are rare and typically occur only during CAM's bus probe where only one command is outstanding at a time. However, during some recovery scenarios, the reordering *could* occur. aic79xx.c: Update ahd_search_qinfifo() and helper routines to search for pending MK_MESSAGE scbs and properly restitch the execution queue if either the MK_MESSAGE SCB is being aborted, or the MK_MESSAGE SCB can be queued due to the execution queue draining due to aborts. Enable LQOBUSFREE status to assert an interrupt. This should be redundant since a BUSFREE interrupt should always occur along with an LQOBUSFREE event, but on the Rev A, this doesn't seem to be guaranteed. When a PPR request is rejected when a previously existing packetized agreement is in place, assume that the target has been reset without our knowledge and revert to async/narrow transfers. This corrects two issues: the stale ENATNO setting that was used to send the PPR is cleared so the firmware is not confused by a future packetized selection with ATN asserted but no MK_MESSAGE flag in the SCB and it speeds up recovery by aborting any pending packetized transactions that by definition are now dead. When re-queueing SCBs after a failed negotiation attempt, ensure command ordering by freezing the device queue first. Traverse the list of pending SCBs rather than the whole SCB array on the controller when pushing MK_MESSAGE flag changes out to the controller. The original code was optimized for the aic7xxx controllers where there are fewer controller slots then pending SCBs and the firmware picks SCB slots. For the U320 controller, the hope is that we have fewer pending SCBs then the 512 slots on the controller. Enhance some diagnostics. Factor out some common code. aic79xx.h: Add prototype for new ahd_done_with_status() that is used to factor out some commone code. aic79xx.reg: Add definisions for the pending MK_MESSAGE SCB. aic79xx.seq: Defer MK_MESSAGE SCB queing to the execution queue so as to preserve command ordering. Re-arrange some of the selection processing code so the above change had no performance impact on the common code path. Close a few critical section holes. When entering a non-packetized phase, manually enable busfree interrupts, since the controller hardware does not do this automatically. aic79xx_inline.h: Enhance logging for queued SCBs. aic79xx_osm.c: Add new a new DDB ahd command, ahd_dump, which invokes the ahd_dump_card_state() routine on the unit specified with the ahd_sunit DDB command. aic79xx_pci.c: Turn on the BUSFREEREV bug for the Rev B. controller. This is required to close the busfree during non-packetized phase hole.
2004-08-04 17:55:34 +00:00
* The whole list made it. Clear our tail pointer to indicate
* that the per-target selection queue is now empty.
*/
Correct a very rare case where command ordering could be compromised by a transaction performing a driver handled message sequence (an scb with the MK_MESSAGE flag set). SCBs that perform host managed messaging must always be at the head of their per-target selection queue so that the firmware knows to manually assert ATN if the current negotiation agreement is packetized. In the past we guaranteed this by queuing these SCBs separarately in the execution queue. This exposes the system to potential command reordering in two cases: 1) Another SCB for the same ITL nexus is queued that does not have the MK_MESSAGE flag set. This SCB will be queued to the per-target list which can be serviced before the MK_MESSAGE scb that preceeded it. 2) If the target cannot accept all of the commands in the per-target selection queue in one selection, the remainder is queued to the tail of the selection queues so as to effect round-robin scheduling. This could allow the MK_MESSAGE scb to be sent to the target before the requeued commands. This commit changes the firmware policy to defer queuing MK_MESSAGE SCBs into the selection queues until this can be done without affecting order. This means that the target's selection queue is either empty, or the last SCB on the execution queue is also a MK_MESSAGE SCB. During any wait, the firmware halts the download of new SCBs so only a single "holding location" is required. Luckily, MK_MESSAGE SCBs are rare and typically occur only during CAM's bus probe where only one command is outstanding at a time. However, during some recovery scenarios, the reordering *could* occur. aic79xx.c: Update ahd_search_qinfifo() and helper routines to search for pending MK_MESSAGE scbs and properly restitch the execution queue if either the MK_MESSAGE SCB is being aborted, or the MK_MESSAGE SCB can be queued due to the execution queue draining due to aborts. Enable LQOBUSFREE status to assert an interrupt. This should be redundant since a BUSFREE interrupt should always occur along with an LQOBUSFREE event, but on the Rev A, this doesn't seem to be guaranteed. When a PPR request is rejected when a previously existing packetized agreement is in place, assume that the target has been reset without our knowledge and revert to async/narrow transfers. This corrects two issues: the stale ENATNO setting that was used to send the PPR is cleared so the firmware is not confused by a future packetized selection with ATN asserted but no MK_MESSAGE flag in the SCB and it speeds up recovery by aborting any pending packetized transactions that by definition are now dead. When re-queueing SCBs after a failed negotiation attempt, ensure command ordering by freezing the device queue first. Traverse the list of pending SCBs rather than the whole SCB array on the controller when pushing MK_MESSAGE flag changes out to the controller. The original code was optimized for the aic7xxx controllers where there are fewer controller slots then pending SCBs and the firmware picks SCB slots. For the U320 controller, the hope is that we have fewer pending SCBs then the 512 slots on the controller. Enhance some diagnostics. Factor out some common code. aic79xx.h: Add prototype for new ahd_done_with_status() that is used to factor out some commone code. aic79xx.reg: Add definisions for the pending MK_MESSAGE SCB. aic79xx.seq: Defer MK_MESSAGE SCB queing to the execution queue so as to preserve command ordering. Re-arrange some of the selection processing code so the above change had no performance impact on the common code path. Close a few critical section holes. When entering a non-packetized phase, manually enable busfree interrupts, since the controller hardware does not do this automatically. aic79xx_inline.h: Enhance logging for queued SCBs. aic79xx_osm.c: Add new a new DDB ahd command, ahd_dump, which invokes the ahd_dump_card_state() routine on the unit specified with the ahd_sunit DDB command. aic79xx_pci.c: Turn on the BUSFREEREV bug for the Rev B. controller. This is required to close the busfree during non-packetized phase hole.
2004-08-04 17:55:34 +00:00
cmp SCB_NEXT[1], SCB_LIST_NULL je select_out_clear_tail;
/*
Correct a very rare case where command ordering could be compromised by a transaction performing a driver handled message sequence (an scb with the MK_MESSAGE flag set). SCBs that perform host managed messaging must always be at the head of their per-target selection queue so that the firmware knows to manually assert ATN if the current negotiation agreement is packetized. In the past we guaranteed this by queuing these SCBs separarately in the execution queue. This exposes the system to potential command reordering in two cases: 1) Another SCB for the same ITL nexus is queued that does not have the MK_MESSAGE flag set. This SCB will be queued to the per-target list which can be serviced before the MK_MESSAGE scb that preceeded it. 2) If the target cannot accept all of the commands in the per-target selection queue in one selection, the remainder is queued to the tail of the selection queues so as to effect round-robin scheduling. This could allow the MK_MESSAGE scb to be sent to the target before the requeued commands. This commit changes the firmware policy to defer queuing MK_MESSAGE SCBs into the selection queues until this can be done without affecting order. This means that the target's selection queue is either empty, or the last SCB on the execution queue is also a MK_MESSAGE SCB. During any wait, the firmware halts the download of new SCBs so only a single "holding location" is required. Luckily, MK_MESSAGE SCBs are rare and typically occur only during CAM's bus probe where only one command is outstanding at a time. However, during some recovery scenarios, the reordering *could* occur. aic79xx.c: Update ahd_search_qinfifo() and helper routines to search for pending MK_MESSAGE scbs and properly restitch the execution queue if either the MK_MESSAGE SCB is being aborted, or the MK_MESSAGE SCB can be queued due to the execution queue draining due to aborts. Enable LQOBUSFREE status to assert an interrupt. This should be redundant since a BUSFREE interrupt should always occur along with an LQOBUSFREE event, but on the Rev A, this doesn't seem to be guaranteed. When a PPR request is rejected when a previously existing packetized agreement is in place, assume that the target has been reset without our knowledge and revert to async/narrow transfers. This corrects two issues: the stale ENATNO setting that was used to send the PPR is cleared so the firmware is not confused by a future packetized selection with ATN asserted but no MK_MESSAGE flag in the SCB and it speeds up recovery by aborting any pending packetized transactions that by definition are now dead. When re-queueing SCBs after a failed negotiation attempt, ensure command ordering by freezing the device queue first. Traverse the list of pending SCBs rather than the whole SCB array on the controller when pushing MK_MESSAGE flag changes out to the controller. The original code was optimized for the aic7xxx controllers where there are fewer controller slots then pending SCBs and the firmware picks SCB slots. For the U320 controller, the hope is that we have fewer pending SCBs then the 512 slots on the controller. Enhance some diagnostics. Factor out some common code. aic79xx.h: Add prototype for new ahd_done_with_status() that is used to factor out some commone code. aic79xx.reg: Add definisions for the pending MK_MESSAGE SCB. aic79xx.seq: Defer MK_MESSAGE SCB queing to the execution queue so as to preserve command ordering. Re-arrange some of the selection processing code so the above change had no performance impact on the common code path. Close a few critical section holes. When entering a non-packetized phase, manually enable busfree interrupts, since the controller hardware does not do this automatically. aic79xx_inline.h: Enhance logging for queued SCBs. aic79xx_osm.c: Add new a new DDB ahd command, ahd_dump, which invokes the ahd_dump_card_state() routine on the unit specified with the ahd_sunit DDB command. aic79xx_pci.c: Turn on the BUSFREEREV bug for the Rev B. controller. This is required to close the busfree during non-packetized phase hole.
2004-08-04 17:55:34 +00:00
* Requeue any SCBs not sent, to the tail of the waiting Q.
* We know that neither the per-TID list nor the list of
Correct a very rare case where command ordering could be compromised by a transaction performing a driver handled message sequence (an scb with the MK_MESSAGE flag set). SCBs that perform host managed messaging must always be at the head of their per-target selection queue so that the firmware knows to manually assert ATN if the current negotiation agreement is packetized. In the past we guaranteed this by queuing these SCBs separarately in the execution queue. This exposes the system to potential command reordering in two cases: 1) Another SCB for the same ITL nexus is queued that does not have the MK_MESSAGE flag set. This SCB will be queued to the per-target list which can be serviced before the MK_MESSAGE scb that preceeded it. 2) If the target cannot accept all of the commands in the per-target selection queue in one selection, the remainder is queued to the tail of the selection queues so as to effect round-robin scheduling. This could allow the MK_MESSAGE scb to be sent to the target before the requeued commands. This commit changes the firmware policy to defer queuing MK_MESSAGE SCBs into the selection queues until this can be done without affecting order. This means that the target's selection queue is either empty, or the last SCB on the execution queue is also a MK_MESSAGE SCB. During any wait, the firmware halts the download of new SCBs so only a single "holding location" is required. Luckily, MK_MESSAGE SCBs are rare and typically occur only during CAM's bus probe where only one command is outstanding at a time. However, during some recovery scenarios, the reordering *could* occur. aic79xx.c: Update ahd_search_qinfifo() and helper routines to search for pending MK_MESSAGE scbs and properly restitch the execution queue if either the MK_MESSAGE SCB is being aborted, or the MK_MESSAGE SCB can be queued due to the execution queue draining due to aborts. Enable LQOBUSFREE status to assert an interrupt. This should be redundant since a BUSFREE interrupt should always occur along with an LQOBUSFREE event, but on the Rev A, this doesn't seem to be guaranteed. When a PPR request is rejected when a previously existing packetized agreement is in place, assume that the target has been reset without our knowledge and revert to async/narrow transfers. This corrects two issues: the stale ENATNO setting that was used to send the PPR is cleared so the firmware is not confused by a future packetized selection with ATN asserted but no MK_MESSAGE flag in the SCB and it speeds up recovery by aborting any pending packetized transactions that by definition are now dead. When re-queueing SCBs after a failed negotiation attempt, ensure command ordering by freezing the device queue first. Traverse the list of pending SCBs rather than the whole SCB array on the controller when pushing MK_MESSAGE flag changes out to the controller. The original code was optimized for the aic7xxx controllers where there are fewer controller slots then pending SCBs and the firmware picks SCB slots. For the U320 controller, the hope is that we have fewer pending SCBs then the 512 slots on the controller. Enhance some diagnostics. Factor out some common code. aic79xx.h: Add prototype for new ahd_done_with_status() that is used to factor out some commone code. aic79xx.reg: Add definisions for the pending MK_MESSAGE SCB. aic79xx.seq: Defer MK_MESSAGE SCB queing to the execution queue so as to preserve command ordering. Re-arrange some of the selection processing code so the above change had no performance impact on the common code path. Close a few critical section holes. When entering a non-packetized phase, manually enable busfree interrupts, since the controller hardware does not do this automatically. aic79xx_inline.h: Enhance logging for queued SCBs. aic79xx_osm.c: Add new a new DDB ahd command, ahd_dump, which invokes the ahd_dump_card_state() routine on the unit specified with the ahd_sunit DDB command. aic79xx_pci.c: Turn on the BUSFREEREV bug for the Rev B. controller. This is required to close the busfree during non-packetized phase hole.
2004-08-04 17:55:34 +00:00
* TIDs is empty. Use this knowledge to our advantage and
* queue the remainder to the tail of the global execution
* queue.
*/
bmov REG0, SCB_NEXT, 2;
Correct a very rare case where command ordering could be compromised by a transaction performing a driver handled message sequence (an scb with the MK_MESSAGE flag set). SCBs that perform host managed messaging must always be at the head of their per-target selection queue so that the firmware knows to manually assert ATN if the current negotiation agreement is packetized. In the past we guaranteed this by queuing these SCBs separarately in the execution queue. This exposes the system to potential command reordering in two cases: 1) Another SCB for the same ITL nexus is queued that does not have the MK_MESSAGE flag set. This SCB will be queued to the per-target list which can be serviced before the MK_MESSAGE scb that preceeded it. 2) If the target cannot accept all of the commands in the per-target selection queue in one selection, the remainder is queued to the tail of the selection queues so as to effect round-robin scheduling. This could allow the MK_MESSAGE scb to be sent to the target before the requeued commands. This commit changes the firmware policy to defer queuing MK_MESSAGE SCBs into the selection queues until this can be done without affecting order. This means that the target's selection queue is either empty, or the last SCB on the execution queue is also a MK_MESSAGE SCB. During any wait, the firmware halts the download of new SCBs so only a single "holding location" is required. Luckily, MK_MESSAGE SCBs are rare and typically occur only during CAM's bus probe where only one command is outstanding at a time. However, during some recovery scenarios, the reordering *could* occur. aic79xx.c: Update ahd_search_qinfifo() and helper routines to search for pending MK_MESSAGE scbs and properly restitch the execution queue if either the MK_MESSAGE SCB is being aborted, or the MK_MESSAGE SCB can be queued due to the execution queue draining due to aborts. Enable LQOBUSFREE status to assert an interrupt. This should be redundant since a BUSFREE interrupt should always occur along with an LQOBUSFREE event, but on the Rev A, this doesn't seem to be guaranteed. When a PPR request is rejected when a previously existing packetized agreement is in place, assume that the target has been reset without our knowledge and revert to async/narrow transfers. This corrects two issues: the stale ENATNO setting that was used to send the PPR is cleared so the firmware is not confused by a future packetized selection with ATN asserted but no MK_MESSAGE flag in the SCB and it speeds up recovery by aborting any pending packetized transactions that by definition are now dead. When re-queueing SCBs after a failed negotiation attempt, ensure command ordering by freezing the device queue first. Traverse the list of pending SCBs rather than the whole SCB array on the controller when pushing MK_MESSAGE flag changes out to the controller. The original code was optimized for the aic7xxx controllers where there are fewer controller slots then pending SCBs and the firmware picks SCB slots. For the U320 controller, the hope is that we have fewer pending SCBs then the 512 slots on the controller. Enhance some diagnostics. Factor out some common code. aic79xx.h: Add prototype for new ahd_done_with_status() that is used to factor out some commone code. aic79xx.reg: Add definisions for the pending MK_MESSAGE SCB. aic79xx.seq: Defer MK_MESSAGE SCB queing to the execution queue so as to preserve command ordering. Re-arrange some of the selection processing code so the above change had no performance impact on the common code path. Close a few critical section holes. When entering a non-packetized phase, manually enable busfree interrupts, since the controller hardware does not do this automatically. aic79xx_inline.h: Enhance logging for queued SCBs. aic79xx_osm.c: Add new a new DDB ahd command, ahd_dump, which invokes the ahd_dump_card_state() routine on the unit specified with the ahd_sunit DDB command. aic79xx_pci.c: Turn on the BUSFREEREV bug for the Rev B. controller. This is required to close the busfree during non-packetized phase hole.
2004-08-04 17:55:34 +00:00
select_out_queue_remainder:
bmov SCBPTR, WAITING_TID_TAIL, 2;
bmov SCB_NEXT2, REG0, 2;
bmov WAITING_TID_TAIL, REG0, 2;
jmp select_out_inc_tid_q;
Correct a very rare case where command ordering could be compromised by a transaction performing a driver handled message sequence (an scb with the MK_MESSAGE flag set). SCBs that perform host managed messaging must always be at the head of their per-target selection queue so that the firmware knows to manually assert ATN if the current negotiation agreement is packetized. In the past we guaranteed this by queuing these SCBs separarately in the execution queue. This exposes the system to potential command reordering in two cases: 1) Another SCB for the same ITL nexus is queued that does not have the MK_MESSAGE flag set. This SCB will be queued to the per-target list which can be serviced before the MK_MESSAGE scb that preceeded it. 2) If the target cannot accept all of the commands in the per-target selection queue in one selection, the remainder is queued to the tail of the selection queues so as to effect round-robin scheduling. This could allow the MK_MESSAGE scb to be sent to the target before the requeued commands. This commit changes the firmware policy to defer queuing MK_MESSAGE SCBs into the selection queues until this can be done without affecting order. This means that the target's selection queue is either empty, or the last SCB on the execution queue is also a MK_MESSAGE SCB. During any wait, the firmware halts the download of new SCBs so only a single "holding location" is required. Luckily, MK_MESSAGE SCBs are rare and typically occur only during CAM's bus probe where only one command is outstanding at a time. However, during some recovery scenarios, the reordering *could* occur. aic79xx.c: Update ahd_search_qinfifo() and helper routines to search for pending MK_MESSAGE scbs and properly restitch the execution queue if either the MK_MESSAGE SCB is being aborted, or the MK_MESSAGE SCB can be queued due to the execution queue draining due to aborts. Enable LQOBUSFREE status to assert an interrupt. This should be redundant since a BUSFREE interrupt should always occur along with an LQOBUSFREE event, but on the Rev A, this doesn't seem to be guaranteed. When a PPR request is rejected when a previously existing packetized agreement is in place, assume that the target has been reset without our knowledge and revert to async/narrow transfers. This corrects two issues: the stale ENATNO setting that was used to send the PPR is cleared so the firmware is not confused by a future packetized selection with ATN asserted but no MK_MESSAGE flag in the SCB and it speeds up recovery by aborting any pending packetized transactions that by definition are now dead. When re-queueing SCBs after a failed negotiation attempt, ensure command ordering by freezing the device queue first. Traverse the list of pending SCBs rather than the whole SCB array on the controller when pushing MK_MESSAGE flag changes out to the controller. The original code was optimized for the aic7xxx controllers where there are fewer controller slots then pending SCBs and the firmware picks SCB slots. For the U320 controller, the hope is that we have fewer pending SCBs then the 512 slots on the controller. Enhance some diagnostics. Factor out some common code. aic79xx.h: Add prototype for new ahd_done_with_status() that is used to factor out some commone code. aic79xx.reg: Add definisions for the pending MK_MESSAGE SCB. aic79xx.seq: Defer MK_MESSAGE SCB queing to the execution queue so as to preserve command ordering. Re-arrange some of the selection processing code so the above change had no performance impact on the common code path. Close a few critical section holes. When entering a non-packetized phase, manually enable busfree interrupts, since the controller hardware does not do this automatically. aic79xx_inline.h: Enhance logging for queued SCBs. aic79xx_osm.c: Add new a new DDB ahd command, ahd_dump, which invokes the ahd_dump_card_state() routine on the unit specified with the ahd_sunit DDB command. aic79xx_pci.c: Turn on the BUSFREEREV bug for the Rev B. controller. This is required to close the busfree during non-packetized phase hole.
2004-08-04 17:55:34 +00:00
select_out_clear_tail:
/*
* Queue any pending MK_MESSAGE SCB for this target now
* that the queue is empty.
*/
test SEQ_FLAGS2, PENDING_MK_MESSAGE jz select_out_no_mk_message_scb;
mov A, MK_MESSAGE_SCSIID;
cmp SCB_SCSIID, A jne select_out_no_mk_message_scb;
and SEQ_FLAGS2, ~PENDING_MK_MESSAGE;
bmov REG0, MK_MESSAGE_SCB, 2;
jmp select_out_queue_remainder;
select_out_no_mk_message_scb:
/*
Correct a very rare case where command ordering could be compromised by a transaction performing a driver handled message sequence (an scb with the MK_MESSAGE flag set). SCBs that perform host managed messaging must always be at the head of their per-target selection queue so that the firmware knows to manually assert ATN if the current negotiation agreement is packetized. In the past we guaranteed this by queuing these SCBs separarately in the execution queue. This exposes the system to potential command reordering in two cases: 1) Another SCB for the same ITL nexus is queued that does not have the MK_MESSAGE flag set. This SCB will be queued to the per-target list which can be serviced before the MK_MESSAGE scb that preceeded it. 2) If the target cannot accept all of the commands in the per-target selection queue in one selection, the remainder is queued to the tail of the selection queues so as to effect round-robin scheduling. This could allow the MK_MESSAGE scb to be sent to the target before the requeued commands. This commit changes the firmware policy to defer queuing MK_MESSAGE SCBs into the selection queues until this can be done without affecting order. This means that the target's selection queue is either empty, or the last SCB on the execution queue is also a MK_MESSAGE SCB. During any wait, the firmware halts the download of new SCBs so only a single "holding location" is required. Luckily, MK_MESSAGE SCBs are rare and typically occur only during CAM's bus probe where only one command is outstanding at a time. However, during some recovery scenarios, the reordering *could* occur. aic79xx.c: Update ahd_search_qinfifo() and helper routines to search for pending MK_MESSAGE scbs and properly restitch the execution queue if either the MK_MESSAGE SCB is being aborted, or the MK_MESSAGE SCB can be queued due to the execution queue draining due to aborts. Enable LQOBUSFREE status to assert an interrupt. This should be redundant since a BUSFREE interrupt should always occur along with an LQOBUSFREE event, but on the Rev A, this doesn't seem to be guaranteed. When a PPR request is rejected when a previously existing packetized agreement is in place, assume that the target has been reset without our knowledge and revert to async/narrow transfers. This corrects two issues: the stale ENATNO setting that was used to send the PPR is cleared so the firmware is not confused by a future packetized selection with ATN asserted but no MK_MESSAGE flag in the SCB and it speeds up recovery by aborting any pending packetized transactions that by definition are now dead. When re-queueing SCBs after a failed negotiation attempt, ensure command ordering by freezing the device queue first. Traverse the list of pending SCBs rather than the whole SCB array on the controller when pushing MK_MESSAGE flag changes out to the controller. The original code was optimized for the aic7xxx controllers where there are fewer controller slots then pending SCBs and the firmware picks SCB slots. For the U320 controller, the hope is that we have fewer pending SCBs then the 512 slots on the controller. Enhance some diagnostics. Factor out some common code. aic79xx.h: Add prototype for new ahd_done_with_status() that is used to factor out some commone code. aic79xx.reg: Add definisions for the pending MK_MESSAGE SCB. aic79xx.seq: Defer MK_MESSAGE SCB queing to the execution queue so as to preserve command ordering. Re-arrange some of the selection processing code so the above change had no performance impact on the common code path. Close a few critical section holes. When entering a non-packetized phase, manually enable busfree interrupts, since the controller hardware does not do this automatically. aic79xx_inline.h: Enhance logging for queued SCBs. aic79xx_osm.c: Add new a new DDB ahd command, ahd_dump, which invokes the ahd_dump_card_state() routine on the unit specified with the ahd_sunit DDB command. aic79xx_pci.c: Turn on the BUSFREEREV bug for the Rev B. controller. This is required to close the busfree during non-packetized phase hole.
2004-08-04 17:55:34 +00:00
* Clear this target's execution tail and increment the queue.
*/
shr DINDEX, 3, SCB_SCSIID;
or DINDEX, 1; /* Want only the second byte */
mvi DINDEX[1], ((WAITING_SCB_TAILS) >> 8);
mvi DINDIR, SCB_LIST_NULL;
select_out_inc_tid_q:
bmov SCBPTR, WAITING_TID_HEAD, 2;
bmov WAITING_TID_HEAD, SCB_NEXT2, 2;
cmp WAITING_TID_HEAD[1], SCB_LIST_NULL jne . + 2;
mvi WAITING_TID_TAIL[1], SCB_LIST_NULL;
bmov SCBPTR, CURRSCB, 2;
mvi CLRSINT0, CLRSELDO;
Correct a very rare case where command ordering could be compromised by a transaction performing a driver handled message sequence (an scb with the MK_MESSAGE flag set). SCBs that perform host managed messaging must always be at the head of their per-target selection queue so that the firmware knows to manually assert ATN if the current negotiation agreement is packetized. In the past we guaranteed this by queuing these SCBs separarately in the execution queue. This exposes the system to potential command reordering in two cases: 1) Another SCB for the same ITL nexus is queued that does not have the MK_MESSAGE flag set. This SCB will be queued to the per-target list which can be serviced before the MK_MESSAGE scb that preceeded it. 2) If the target cannot accept all of the commands in the per-target selection queue in one selection, the remainder is queued to the tail of the selection queues so as to effect round-robin scheduling. This could allow the MK_MESSAGE scb to be sent to the target before the requeued commands. This commit changes the firmware policy to defer queuing MK_MESSAGE SCBs into the selection queues until this can be done without affecting order. This means that the target's selection queue is either empty, or the last SCB on the execution queue is also a MK_MESSAGE SCB. During any wait, the firmware halts the download of new SCBs so only a single "holding location" is required. Luckily, MK_MESSAGE SCBs are rare and typically occur only during CAM's bus probe where only one command is outstanding at a time. However, during some recovery scenarios, the reordering *could* occur. aic79xx.c: Update ahd_search_qinfifo() and helper routines to search for pending MK_MESSAGE scbs and properly restitch the execution queue if either the MK_MESSAGE SCB is being aborted, or the MK_MESSAGE SCB can be queued due to the execution queue draining due to aborts. Enable LQOBUSFREE status to assert an interrupt. This should be redundant since a BUSFREE interrupt should always occur along with an LQOBUSFREE event, but on the Rev A, this doesn't seem to be guaranteed. When a PPR request is rejected when a previously existing packetized agreement is in place, assume that the target has been reset without our knowledge and revert to async/narrow transfers. This corrects two issues: the stale ENATNO setting that was used to send the PPR is cleared so the firmware is not confused by a future packetized selection with ATN asserted but no MK_MESSAGE flag in the SCB and it speeds up recovery by aborting any pending packetized transactions that by definition are now dead. When re-queueing SCBs after a failed negotiation attempt, ensure command ordering by freezing the device queue first. Traverse the list of pending SCBs rather than the whole SCB array on the controller when pushing MK_MESSAGE flag changes out to the controller. The original code was optimized for the aic7xxx controllers where there are fewer controller slots then pending SCBs and the firmware picks SCB slots. For the U320 controller, the hope is that we have fewer pending SCBs then the 512 slots on the controller. Enhance some diagnostics. Factor out some common code. aic79xx.h: Add prototype for new ahd_done_with_status() that is used to factor out some commone code. aic79xx.reg: Add definisions for the pending MK_MESSAGE SCB. aic79xx.seq: Defer MK_MESSAGE SCB queing to the execution queue so as to preserve command ordering. Re-arrange some of the selection processing code so the above change had no performance impact on the common code path. Close a few critical section holes. When entering a non-packetized phase, manually enable busfree interrupts, since the controller hardware does not do this automatically. aic79xx_inline.h: Enhance logging for queued SCBs. aic79xx_osm.c: Add new a new DDB ahd command, ahd_dump, which invokes the ahd_dump_card_state() routine on the unit specified with the ahd_sunit DDB command. aic79xx_pci.c: Turn on the BUSFREEREV bug for the Rev B. controller. This is required to close the busfree during non-packetized phase hole.
2004-08-04 17:55:34 +00:00
test LQOSTAT2, LQOPHACHGOUTPKT jnz unexpected_nonpkt_mode_cleared;
test LQOSTAT1, LQOPHACHGINPKT jnz unexpected_nonpkt_mode_cleared;
/*
* If this is a packetized connection, return to our
* idle_loop and let our interrupt handler deal with
* any connection setup/teardown issues. The only
aic79xx.c: Implement the SCB_SILENT flag. This is useful for hushing up the driver during DV or other operations that we expect to cause transmission errors. The messages will still print if the SHOW_MASKED_ERRORS debug option is enabled. Save and restore the NEGOADDR address when setting new transfer settings. The sequencer performs lookups in the negotiation table too and it expects NEGOADDR to remain consistent across pause/unpause sessions. Consistently use "offset" instead of "period" to determine if we are running sync or not. Add a SHOW_MESSAGES diagnostic for when we assert ATN during message processing. Print out IU, QAS, and RTI features when showing transfer options. Limit the syncrate after all option conformance changes have taken place in ahd_devlimited_syncrate. Changes in options may change the final syncrate we accept. Keep a copy of the hs_mailbox in our softc so that we can perform read/modify/write operations on the hs_mailbox without having to pause the sequencer to read the last written value. Use the ENINT_COALESS flag in the hs_mailbox to toggle interrupt coalessing. Add entrypoints for enabling interrupt coalessing and setting both a timeout (how long to wait for commands to be coalessed) and a maximum commands to coaless value. Add a statistics timer that decides when to enable or disable interrupt coalessing based on load. Add a routine, ahd_reset_cmds_pending() which is used to update the CMDS_PENDING sequencer variable whenever error recovery compeltes SCBs without notifying the sequencer. Since ahd_reset_cmds_pending is called during ahd_unpause() only if we've aborted SCBs, its call to ahd_flush_qoutfifo should not cause recursion through ahd_run_qoutfifo(). A panic has been added to ensure that this recursion does not occur. In ahd_search_qinfifo, update the CMDS_PENDING sequencer variable directly. ahd_search_qinififo can be called in situations where using ahd_reset_cmds_pending() might cause recursion. Since we can safely determine the exact number to reduce CMDS_PENDING by in this scenario without running the qoutfifo, the manual update is sufficient. Clean up diagnostics. Add ahd_flush_qoutfifo() which will run the qoutfifo as well as complete any commands sitting on the sequencer's COMPLETE_SCB lists or the good status FIFO. Use this routine in several places that did similar things in an add-hoc, but incomplete, fashion. A call to this routine was also added to ahd_abort_scbs() to close a race. In ahd_pause_and_flushwork() only return once selections are safely disabled. Flush all completed commands via ahd_flush_qoutfifo(). Remove "Now packetized" diagnostic now that this information is incorperated into the actual negotiation messages that are displayed. When forcing renegotiation, don't clober the current ppr_options. Much of the driver uses this information to determine if we are currently packetized or not. Remove some stray spaces at column 1 in ahd_set_tags. When complaining about getting a host message loop request with no pending messages, print out the SCB_CONTROL register down on the card. Modify the ahd_sent_msg() routine to handle a search for an outgoing identify message. Use this to detect a msg reject on an identify message which typically indicates that the target thought we were packetized. Force a renegotiation in this case. In ahd_search_qinfifo(), wait more effectively for SCB DMA activities to cease. We also disable SCB fetch operations since we are about to change the qinfifo and any fetch in progress will likely be invalidated. In ahd_qinfifo_count(), fix the qinfifo empty case. In ahd_dump_card_state(), print out CCSCBCTL in the correct mode. If we are a narrow controller, don't set the current width to unknown when forcing a future negotiation. This just confuses the code into attempting a wide negotiation on a narrow bus. Add support for task management function completions. Modify ahd_handle_devreset so that it can handle lun resets in addition to target resets. Use ahd_handle_devreset for lun and target reset task management functions. Handle the abort task TMF race case better. We now wait until any current selections are over and then set the TMF back to zero. This should cause the sequencer to ignore the abort TMF completion should it occur. Correct a bug in the illegal phase handler that caused us to drop down to narrow when handling the unexpected command phase case after 3rd party reset of a packetized device. Indicate the features, bugs, and flags set in the softc that are used to control firmware patch download when booting verbose. aic79xx.h: Add coalessing and HS_MAILBOX fields. Add per-softc variables for the stats "daemon". Add a debug option for interrupt coalessing activities. Add two new softc flags: o AHD_UPDATE_PEND_CMDS Run ahd_reset_cmds_pending() on the next unpause. o AHD_RUNNING_QOUTFIFO Used to catch recursion through ahd_run_qoutfifo(). aic79xx.reg: Correct register addresses related to the software timer and the DFDBCTL register. Add constants paramaterizing the software timer. Add scratch ram locations for storing interrupt coalessing tunables. Break INTMASK in SEQITNCTL out into INTMASK1 and INTMASK2. In at least the REV A, these are writable bits. We make use of that for a swtimer workaround in the sequencer. Since HS_MAILBOX autoclears, provide a sequencer variable to store its contents. Add SEQINT codes for handling task management completions. aic79xx.seq: Correct ignore wide residue processing check for a wide negotiation being in effect. We must be in the SCSI register window in order to access the negotiation table. Use the software timer and a commands completed count to implement interrupt coalessing. The command complete is deferred until either the maximum command threshold or a the expiration of a command deferral timer. If we have more SCBs to complete to the host (sitting in COMPLETE_SCB lists), always try to coaless them up to our coalessing limit. If coalessing is enabled, but we have fewer commands oustanting than the host's min coalessing limit, complete the command immediately. Add code to track the number of commands outstanding. Commands are outstanding from the time they are placed into the execution queue until the DMA to post completion is setup. Add a workaround for intvec_2 interrupts on the H2A4. In H2A4, the mode pointer is not saved for intvec2, but is restored on iret. This can lead to the restoration of a bogus mode ptr. Manually clear the intmask bits and do a normal return to compensate. We use intvec_2 to track interrupt coalessing timeouts. Since we cannot disable the swtimer's countdown, simply mask its interrupt once we no longer care about it firing. In idle_loop_cchan, update LOCAL_HS_MAILBOX everytime we are notified of an HS_MAILBOX update via the HS_MAILBOX_ACT bit in QOFF_CTLSTA. We have to use a local copy of persistant portions of the HS_MAILBOX as the mailbox auto-clears on any read. Move the test for the cfg4istat interrupt up an instruction to hopefully close a race between the next outgoing selection and our disabling of selections. Add a missing ret to the last instruction in load_overrun_buf. Add notifications to the host of task management completions as well as the completions for commands that completed successfully before their corresponding TMF could be sent. Hold a critical section during select-out processing until we have a fully identified connection. This removes a race condition with the legacy abort handler. Correct a few spelling errors in some comments. aic79xx_inline.h: Call ahd_reset_cmds_pending() in ahd_unpause if required. Update cmdcmplt interrupt statistics in our interrupt handler. Allow callers to ahd_send_scb() to set the task management function. aic79xx_pci.c: Disable SERR and pause the controller prior to performing our mmapped I/O test. The U320 controllers do not support "auto-access-pause". aic79xx_osm.c: Set the task management function now that ahd_send_scb() doesn't do it for us. We also perform a lun reset in response to BDR requests to packetized devices.
2003-01-20 20:17:35 +00:00
* exceptions are the case of MK_MESSAGE and task management
* SCBs.
*/
if ((ahd->bugs & AHD_LQO_ATNO_BUG) != 0) {
aic79xx.c: Implement the SCB_SILENT flag. This is useful for hushing up the driver during DV or other operations that we expect to cause transmission errors. The messages will still print if the SHOW_MASKED_ERRORS debug option is enabled. Save and restore the NEGOADDR address when setting new transfer settings. The sequencer performs lookups in the negotiation table too and it expects NEGOADDR to remain consistent across pause/unpause sessions. Consistently use "offset" instead of "period" to determine if we are running sync or not. Add a SHOW_MESSAGES diagnostic for when we assert ATN during message processing. Print out IU, QAS, and RTI features when showing transfer options. Limit the syncrate after all option conformance changes have taken place in ahd_devlimited_syncrate. Changes in options may change the final syncrate we accept. Keep a copy of the hs_mailbox in our softc so that we can perform read/modify/write operations on the hs_mailbox without having to pause the sequencer to read the last written value. Use the ENINT_COALESS flag in the hs_mailbox to toggle interrupt coalessing. Add entrypoints for enabling interrupt coalessing and setting both a timeout (how long to wait for commands to be coalessed) and a maximum commands to coaless value. Add a statistics timer that decides when to enable or disable interrupt coalessing based on load. Add a routine, ahd_reset_cmds_pending() which is used to update the CMDS_PENDING sequencer variable whenever error recovery compeltes SCBs without notifying the sequencer. Since ahd_reset_cmds_pending is called during ahd_unpause() only if we've aborted SCBs, its call to ahd_flush_qoutfifo should not cause recursion through ahd_run_qoutfifo(). A panic has been added to ensure that this recursion does not occur. In ahd_search_qinfifo, update the CMDS_PENDING sequencer variable directly. ahd_search_qinififo can be called in situations where using ahd_reset_cmds_pending() might cause recursion. Since we can safely determine the exact number to reduce CMDS_PENDING by in this scenario without running the qoutfifo, the manual update is sufficient. Clean up diagnostics. Add ahd_flush_qoutfifo() which will run the qoutfifo as well as complete any commands sitting on the sequencer's COMPLETE_SCB lists or the good status FIFO. Use this routine in several places that did similar things in an add-hoc, but incomplete, fashion. A call to this routine was also added to ahd_abort_scbs() to close a race. In ahd_pause_and_flushwork() only return once selections are safely disabled. Flush all completed commands via ahd_flush_qoutfifo(). Remove "Now packetized" diagnostic now that this information is incorperated into the actual negotiation messages that are displayed. When forcing renegotiation, don't clober the current ppr_options. Much of the driver uses this information to determine if we are currently packetized or not. Remove some stray spaces at column 1 in ahd_set_tags. When complaining about getting a host message loop request with no pending messages, print out the SCB_CONTROL register down on the card. Modify the ahd_sent_msg() routine to handle a search for an outgoing identify message. Use this to detect a msg reject on an identify message which typically indicates that the target thought we were packetized. Force a renegotiation in this case. In ahd_search_qinfifo(), wait more effectively for SCB DMA activities to cease. We also disable SCB fetch operations since we are about to change the qinfifo and any fetch in progress will likely be invalidated. In ahd_qinfifo_count(), fix the qinfifo empty case. In ahd_dump_card_state(), print out CCSCBCTL in the correct mode. If we are a narrow controller, don't set the current width to unknown when forcing a future negotiation. This just confuses the code into attempting a wide negotiation on a narrow bus. Add support for task management function completions. Modify ahd_handle_devreset so that it can handle lun resets in addition to target resets. Use ahd_handle_devreset for lun and target reset task management functions. Handle the abort task TMF race case better. We now wait until any current selections are over and then set the TMF back to zero. This should cause the sequencer to ignore the abort TMF completion should it occur. Correct a bug in the illegal phase handler that caused us to drop down to narrow when handling the unexpected command phase case after 3rd party reset of a packetized device. Indicate the features, bugs, and flags set in the softc that are used to control firmware patch download when booting verbose. aic79xx.h: Add coalessing and HS_MAILBOX fields. Add per-softc variables for the stats "daemon". Add a debug option for interrupt coalessing activities. Add two new softc flags: o AHD_UPDATE_PEND_CMDS Run ahd_reset_cmds_pending() on the next unpause. o AHD_RUNNING_QOUTFIFO Used to catch recursion through ahd_run_qoutfifo(). aic79xx.reg: Correct register addresses related to the software timer and the DFDBCTL register. Add constants paramaterizing the software timer. Add scratch ram locations for storing interrupt coalessing tunables. Break INTMASK in SEQITNCTL out into INTMASK1 and INTMASK2. In at least the REV A, these are writable bits. We make use of that for a swtimer workaround in the sequencer. Since HS_MAILBOX autoclears, provide a sequencer variable to store its contents. Add SEQINT codes for handling task management completions. aic79xx.seq: Correct ignore wide residue processing check for a wide negotiation being in effect. We must be in the SCSI register window in order to access the negotiation table. Use the software timer and a commands completed count to implement interrupt coalessing. The command complete is deferred until either the maximum command threshold or a the expiration of a command deferral timer. If we have more SCBs to complete to the host (sitting in COMPLETE_SCB lists), always try to coaless them up to our coalessing limit. If coalessing is enabled, but we have fewer commands oustanting than the host's min coalessing limit, complete the command immediately. Add code to track the number of commands outstanding. Commands are outstanding from the time they are placed into the execution queue until the DMA to post completion is setup. Add a workaround for intvec_2 interrupts on the H2A4. In H2A4, the mode pointer is not saved for intvec2, but is restored on iret. This can lead to the restoration of a bogus mode ptr. Manually clear the intmask bits and do a normal return to compensate. We use intvec_2 to track interrupt coalessing timeouts. Since we cannot disable the swtimer's countdown, simply mask its interrupt once we no longer care about it firing. In idle_loop_cchan, update LOCAL_HS_MAILBOX everytime we are notified of an HS_MAILBOX update via the HS_MAILBOX_ACT bit in QOFF_CTLSTA. We have to use a local copy of persistant portions of the HS_MAILBOX as the mailbox auto-clears on any read. Move the test for the cfg4istat interrupt up an instruction to hopefully close a race between the next outgoing selection and our disabling of selections. Add a missing ret to the last instruction in load_overrun_buf. Add notifications to the host of task management completions as well as the completions for commands that completed successfully before their corresponding TMF could be sent. Hold a critical section during select-out processing until we have a fully identified connection. This removes a race condition with the legacy abort handler. Correct a few spelling errors in some comments. aic79xx_inline.h: Call ahd_reset_cmds_pending() in ahd_unpause if required. Update cmdcmplt interrupt statistics in our interrupt handler. Allow callers to ahd_send_scb() to set the task management function. aic79xx_pci.c: Disable SERR and pause the controller prior to performing our mmapped I/O test. The U320 controllers do not support "auto-access-pause". aic79xx_osm.c: Set the task management function now that ahd_send_scb() doesn't do it for us. We also perform a lun reset in response to BDR requests to packetized devices.
2003-01-20 20:17:35 +00:00
/*
* In the A, the LQO manager transitions to LQOSTOP0 even if
* we have selected out with ATN asserted and the target
* REQs in a non-packet phase.
*/
test SCB_CONTROL, MK_MESSAGE jz select_out_no_message;
test SCSISIGO, ATNO jnz select_out_non_packetized;
select_out_no_message:
}
aic79xx.c: Implement the SCB_SILENT flag. This is useful for hushing up the driver during DV or other operations that we expect to cause transmission errors. The messages will still print if the SHOW_MASKED_ERRORS debug option is enabled. Save and restore the NEGOADDR address when setting new transfer settings. The sequencer performs lookups in the negotiation table too and it expects NEGOADDR to remain consistent across pause/unpause sessions. Consistently use "offset" instead of "period" to determine if we are running sync or not. Add a SHOW_MESSAGES diagnostic for when we assert ATN during message processing. Print out IU, QAS, and RTI features when showing transfer options. Limit the syncrate after all option conformance changes have taken place in ahd_devlimited_syncrate. Changes in options may change the final syncrate we accept. Keep a copy of the hs_mailbox in our softc so that we can perform read/modify/write operations on the hs_mailbox without having to pause the sequencer to read the last written value. Use the ENINT_COALESS flag in the hs_mailbox to toggle interrupt coalessing. Add entrypoints for enabling interrupt coalessing and setting both a timeout (how long to wait for commands to be coalessed) and a maximum commands to coaless value. Add a statistics timer that decides when to enable or disable interrupt coalessing based on load. Add a routine, ahd_reset_cmds_pending() which is used to update the CMDS_PENDING sequencer variable whenever error recovery compeltes SCBs without notifying the sequencer. Since ahd_reset_cmds_pending is called during ahd_unpause() only if we've aborted SCBs, its call to ahd_flush_qoutfifo should not cause recursion through ahd_run_qoutfifo(). A panic has been added to ensure that this recursion does not occur. In ahd_search_qinfifo, update the CMDS_PENDING sequencer variable directly. ahd_search_qinififo can be called in situations where using ahd_reset_cmds_pending() might cause recursion. Since we can safely determine the exact number to reduce CMDS_PENDING by in this scenario without running the qoutfifo, the manual update is sufficient. Clean up diagnostics. Add ahd_flush_qoutfifo() which will run the qoutfifo as well as complete any commands sitting on the sequencer's COMPLETE_SCB lists or the good status FIFO. Use this routine in several places that did similar things in an add-hoc, but incomplete, fashion. A call to this routine was also added to ahd_abort_scbs() to close a race. In ahd_pause_and_flushwork() only return once selections are safely disabled. Flush all completed commands via ahd_flush_qoutfifo(). Remove "Now packetized" diagnostic now that this information is incorperated into the actual negotiation messages that are displayed. When forcing renegotiation, don't clober the current ppr_options. Much of the driver uses this information to determine if we are currently packetized or not. Remove some stray spaces at column 1 in ahd_set_tags. When complaining about getting a host message loop request with no pending messages, print out the SCB_CONTROL register down on the card. Modify the ahd_sent_msg() routine to handle a search for an outgoing identify message. Use this to detect a msg reject on an identify message which typically indicates that the target thought we were packetized. Force a renegotiation in this case. In ahd_search_qinfifo(), wait more effectively for SCB DMA activities to cease. We also disable SCB fetch operations since we are about to change the qinfifo and any fetch in progress will likely be invalidated. In ahd_qinfifo_count(), fix the qinfifo empty case. In ahd_dump_card_state(), print out CCSCBCTL in the correct mode. If we are a narrow controller, don't set the current width to unknown when forcing a future negotiation. This just confuses the code into attempting a wide negotiation on a narrow bus. Add support for task management function completions. Modify ahd_handle_devreset so that it can handle lun resets in addition to target resets. Use ahd_handle_devreset for lun and target reset task management functions. Handle the abort task TMF race case better. We now wait until any current selections are over and then set the TMF back to zero. This should cause the sequencer to ignore the abort TMF completion should it occur. Correct a bug in the illegal phase handler that caused us to drop down to narrow when handling the unexpected command phase case after 3rd party reset of a packetized device. Indicate the features, bugs, and flags set in the softc that are used to control firmware patch download when booting verbose. aic79xx.h: Add coalessing and HS_MAILBOX fields. Add per-softc variables for the stats "daemon". Add a debug option for interrupt coalessing activities. Add two new softc flags: o AHD_UPDATE_PEND_CMDS Run ahd_reset_cmds_pending() on the next unpause. o AHD_RUNNING_QOUTFIFO Used to catch recursion through ahd_run_qoutfifo(). aic79xx.reg: Correct register addresses related to the software timer and the DFDBCTL register. Add constants paramaterizing the software timer. Add scratch ram locations for storing interrupt coalessing tunables. Break INTMASK in SEQITNCTL out into INTMASK1 and INTMASK2. In at least the REV A, these are writable bits. We make use of that for a swtimer workaround in the sequencer. Since HS_MAILBOX autoclears, provide a sequencer variable to store its contents. Add SEQINT codes for handling task management completions. aic79xx.seq: Correct ignore wide residue processing check for a wide negotiation being in effect. We must be in the SCSI register window in order to access the negotiation table. Use the software timer and a commands completed count to implement interrupt coalessing. The command complete is deferred until either the maximum command threshold or a the expiration of a command deferral timer. If we have more SCBs to complete to the host (sitting in COMPLETE_SCB lists), always try to coaless them up to our coalessing limit. If coalessing is enabled, but we have fewer commands oustanting than the host's min coalessing limit, complete the command immediately. Add code to track the number of commands outstanding. Commands are outstanding from the time they are placed into the execution queue until the DMA to post completion is setup. Add a workaround for intvec_2 interrupts on the H2A4. In H2A4, the mode pointer is not saved for intvec2, but is restored on iret. This can lead to the restoration of a bogus mode ptr. Manually clear the intmask bits and do a normal return to compensate. We use intvec_2 to track interrupt coalessing timeouts. Since we cannot disable the swtimer's countdown, simply mask its interrupt once we no longer care about it firing. In idle_loop_cchan, update LOCAL_HS_MAILBOX everytime we are notified of an HS_MAILBOX update via the HS_MAILBOX_ACT bit in QOFF_CTLSTA. We have to use a local copy of persistant portions of the HS_MAILBOX as the mailbox auto-clears on any read. Move the test for the cfg4istat interrupt up an instruction to hopefully close a race between the next outgoing selection and our disabling of selections. Add a missing ret to the last instruction in load_overrun_buf. Add notifications to the host of task management completions as well as the completions for commands that completed successfully before their corresponding TMF could be sent. Hold a critical section during select-out processing until we have a fully identified connection. This removes a race condition with the legacy abort handler. Correct a few spelling errors in some comments. aic79xx_inline.h: Call ahd_reset_cmds_pending() in ahd_unpause if required. Update cmdcmplt interrupt statistics in our interrupt handler. Allow callers to ahd_send_scb() to set the task management function. aic79xx_pci.c: Disable SERR and pause the controller prior to performing our mmapped I/O test. The U320 controllers do not support "auto-access-pause". aic79xx_osm.c: Set the task management function now that ahd_send_scb() doesn't do it for us. We also perform a lun reset in response to BDR requests to packetized devices.
2003-01-20 20:17:35 +00:00
test LQOSTAT2, LQOSTOP0 jz select_out_non_packetized;
test SCB_TASK_MANAGEMENT, 0xFF jz idle_loop;
SET_SEQINTCODE(TASKMGMT_FUNC_COMPLETE)
jmp idle_loop;
select_out_non_packetized:
/* Non packetized request. */
and SCSISEQ0, ~ENSELO;
Upgrade to version 1.1 of the aic79xx U320 driver. aic79xx.c: o Remove redundant ahd_update_modes() call. o Correct panic in diagnostic should state corruption cause the SCB Id to be invalid during a selection timeout. o Add workaround for missing BUSFREEREV feature in Rev A silicon. o Corect formatting nits. o Use register pretty printing in more places. o Save and restore our SCB pointer when updating the waiting queue list for an "expected" LQ-out busfree. o In ahd_clear_intstat, deal with the missing autoclear in the CLRLQO* registers. o BE fixup in a diagnostic printf. o Make sure that we are in the proper mode before disabling selections in ahd_update_pending_scbs. o Add more diagnostics. o task_attribute_nonpkt_tag -> task_attribute: we don't need a nonpkt_tag field anymore for allowing all 512 SCBs to be used in non-packetized connections. o Negotiate HOLD_MCS to U320 devices. o Add a few additional mode assertions. o Restore the chip mode after clearing out the qinfifo so that code using ahd_abort_scbs sees a consistent mode. o Simplify the DMA engine shutdown routine prior to performing a bus reset. o Perform the sequencer restart after a chip reset prior to setting up our timer to poll for the reset to be complete. On some OSes, the timer could actually pre-empt us and order is important here. o Have our "reset poller" set the expected mode since there is no guarantee of what mode will be in force when we are called from the OS timer. o Save and restore the SCB pointer in ahd_dump_card_state(). This routine must not modify card state. o Ditto for ahd_dump_scbs(). aic79xx.h: o Add a few more chip bug definitions. o Align our tag on a 32bit boundary. aic79xx.reg: aic79xx.seq: o Start work on removing workarounds for Rev B. o Use a special location in scratch from for stroring our SCBPTR during legacy FIFO allocations. This corrects problems in mixed packetized/non-packetized configurations where calling into a FIFO task corrupted our SCBPTR. o Don't rely on DMA priority to guarantee that all data in our FIFOs will flush prior to a command completion notification going out of the command channel. We've never seen this assumption fail, but better safe than sorry. o Deal with missing BUSFREEREV feature in H2A. o Simplify disconnect list code now that the list will always have only a single entry. o Implement the AHD_REG_SLOW_SETTLE_BUG workaround. o Swith to using "REG_ISR" for local mode scratch during our ISR. o Add a missing jmp to the data_group_dma_loop after our data pointers have been re-initialized by the kernel. o Correct test in the bitbucket code so that we actually wait for the bitbucket to complete before signaling the kernel of the overrun condition. o Reposition pkt_saveptrs to avoid a jmp instruction. o Update a comment to reflect that the code now waits for a FIFO to drain prior to issuing a CLRCHN. aic79xx_inline.h: o Remove unused untagged queue handling code. o Don't attempt to htole64 what could be a 32bit value. aic79xx_pci.c: o Set additional bug flags for rev A chips.
2002-09-26 22:54:00 +00:00
if ((ahd->bugs & AHD_BUSFREEREV_BUG) != 0) {
/*
ahc_eisa.c: ahc_pci.c: ahd_pci.c: aic7xxx.c: aic79xx.c: aic_osm_lib.c: aic_osm_lib.h: Use common OSM routines from aic_osm_lib for bus dma operations, delay routines, accessing CCBs, byte swapping, etc. aic7xxx_pci.c: Provide a better description for the 2915/30LP on attach. aic7xxx.c: aic79xx.c: aic7770.c: aic79xx_pci.c: aic7xxx_pci.c: aic7xxx_93cx6.c: Move FBSDID behind an ifdef so that these core files will still compile under other OSes. aic79xx.h: aic79xx_pci.c: aic79xx.seq: To speed up non-packetized CDB delivery in Rev B, all CDB acks are "released" to the output sync as soon as the command phase starts. There is only one problem with this approach. If the target changes phase before all data are sent, we have left over acks that can go out on the bus in a data phase. Due to other chip contraints, this only happens if the target goes to data-in, but if the acks go out before we can test SDONE, we'll think that the transfer has completed successfully. Work around this by taking advantage of the 400ns or 800ns dead time between command phase and the REQ of the new phase. If the transfer has completed successfully, SCSIEN should fall *long* before we see a phase change. We thus treat any phasemiss that occurs before SCSIEN falls as an incomplete transfer. aic79xx.h: Add the AHD_FAST_CDB_DELIVERY feature. aic79xx_pci.c: Set AHD_FAST_CDB_DELIVERY for all Rev. B parts. aic79xx.seq: Test for PHASEMIS in the command phase for all AHD_FAST_CDB_DELIVERY controlelrs. ahd_pci.c: ahc_pci.c: aic7xxx.h: aic79xx.h: Move definition of controller BAR offsets to core header files. aic7xxx.c: aic79xx.c: In the softc free routine, leave removal of a softc from the global list of softcs to the OSM (the caller of this routine). This allows us to avoid holding the softc list_lock during device destruction where we may have to sleep waiting for our recovery thread to halt. ahc_pci.c: Use ahc_pci_test_register access to validate I/O mapped in addition to the tests already performed for memory mapped access. Remove unused ahc_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. ahd_pci.c: Remove reduntant definition of controller BAR offsets. These are also defined in aic79xx.h. Remove unused ahd_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. aic7xxx.c: aic79xx.c: aic79xx.h: aic7xxx.h: aic7xxx_osm.c: aic79xx_osm.c: Move timeout handling to the driver cores. In the case of the aic79xx driver, the algorithm has been enhanced to try target resets before performing a bus reset. For the aic7xxx driver, the algorithm is unchanged. Although the drivers do not currently sleep during recovery (recovery is timeout driven), the cores do expect all processing to be performed via a recovery thread. Our timeout handlers are now little stubs that wakeup the recovery thread. aic79xx.c: aic79xx.h: aic79xx_inline.h: Change shared_data allocation to use a map_node so that the sentinel hscb can use this map node in ahd_swap_with_next_hscb. This routine now swaps the hscb_map pointer in additon to the hscb contents so that any sync operations occur on the correct map. physaddr -> busaddr Pointed out by: Jason Thorpe <thorpej@wasabisystems.com> aic79xx.c: Make more use of the in/out/w/l/q macros for accessing byte registers in the chip. Correct some issues in the ahd_flush_qoutfifo() routine. o Run the qoutfifo only once the command channel DMA engine has been halted. This closes a window where we might have missed some entries. o Change ahd_run_data_fifo() to not loop to completion. If we happen to start on the wrong FIFO and the other FIFO has a snapshot savepointers, we might deadlock. This required our delay between FIFO tests to be moved to the ahd_flush_qoutfifo() routine. o Update/add comments. o Remove spurious test for COMPLETE_DMA list being empty when completing transactions from the GSFIFO with residuals. The SCB must be put on the COMPLETE_DMA scb list unconditionally. o When halting command channel DMA activity, we must disable the DMA channel in all cases but an update of the QOUTFIFO. The latter case is required so that the sequencer will update its position in the QOUTFIFO. Previously, we left the channel enabled for all "push" DMAs. This left us vulnerable to the sequencer handling an SCB push long after that SCB was already processed manually by this routine. o Correct the polarity of tests involving ahd_scb_active_in_fifo(). This routine returns non-zero for true. Return to processing bad status completions through the qoutfifo. This reduces the time that the sequencer is kept paused when handling transactions with bad status or underruns. When waiting for the controller to quiece selections, add a delay to our loop. Otherwise we may fail to wait long enough for the sequencer to comply. On H2A4 hardware, use the slow slewrate for non-paced transfers. This mirrors what the Adaptec Windows drivers do. On the Rev B. only slow down the CRC timing for older U160 devices that might need the slower timing. We define "older" as devices that do not support packetized protocol. Wait up to 5000 * 5us for the SEEPROM to become unbusy. Write ops seem to take much longer than read ops. aic79xx.seq: For controllers with the FAINT_LED bug, turn the diagnostic led feature on during selection and reselection. This covers the non-packetized case. The LED will be disabled for non-packetized transfers once we return to the top level idle loop. Add more comments about the busy LED workaround. Extend a critical section around the entire command channel idle loop process. Previously the portion of this handler that directly manipulated the linked list of completed SCBs was not protected. This is the likely cause of the recent reports of commands being completed twice by the driver. Extend critical sections across the test for, and the longjump to, longjump routines. This prevents the firmware from trying to jump to a longjmp handler that was just cleared by the host. Improve the locations of several critical section begin and end points. Typically these changes remove instructions that did not need to be inside a critical section. Close the "busfree after selection, but before busfree interrupts can be enabled" race to just a single sequencer instruction. We now test the BSY line explicitly before clearing the busfree status and enabling the busfree interrupt. Close a race condition in the processing of HS_MAILBOX updates. We now clear the "updated" status before the copy. This ensures that we don't accidentally clear the status incorrectly when the host sneaks in an update just after our last copy, but before we clear the status. This race has never been observed. Don't re-enable SCSIEN if we lose the race to disable SCSIEN in our interrupt handler's workaround for the RevA data-valid too early issue. aic79xx_inline.h: Add comments indicating that the order in which bytes are read or written in ahd_inw and ahd_outw is important. This allows us to use these inlines when accessing registers with side-effects. aic79xx_pci.c: The 29320 and the 29320B are 7902 not 7901 based products. Correct the driver banner. aic7xxx.h: Enable the use of the auto-access pause feature on the aic7870 and aic7880. It was disabled due to an oversight. aic7xxx.reg: Move TARG_IMMEDIATE_SCB to alias LAST_MSG to avoid leaving garbage in MWI_RESIDUAL. This prevents spurious overflows whn operating target mode on controllers that require the MWI_RESIDUAL work-around. aic7xxx.seq: AHC_TMODE_WIDEODD_BUG is a bug, not a softc flag. Reference the correct softc field when testing for its presence. Set the NOT_IDENTIFIED and NO_CDB_SENT bits in SEQ_FLAGS to indicate that the nexus is invalid in await busfree. aic7xxx_93cx6.c: Add support for the C56/C66 versions of the EWEN and EWDS commands. aic7xxx.c: aic7xxx_pci.c: Move test for the validity of left over BIOS data to ahc_test_register_access(). This guarantees that any left over CHIPRST value is not clobbered by our register access test and lost to the test that was in ahc_reset.
2003-12-17 00:02:10 +00:00
* Test to ensure that the bus has not
* already gone free prior to clearing
* any stale busfree status. This avoids
* a window whereby a busfree just after
* a selection could be missed.
Upgrade to version 1.1 of the aic79xx U320 driver. aic79xx.c: o Remove redundant ahd_update_modes() call. o Correct panic in diagnostic should state corruption cause the SCB Id to be invalid during a selection timeout. o Add workaround for missing BUSFREEREV feature in Rev A silicon. o Corect formatting nits. o Use register pretty printing in more places. o Save and restore our SCB pointer when updating the waiting queue list for an "expected" LQ-out busfree. o In ahd_clear_intstat, deal with the missing autoclear in the CLRLQO* registers. o BE fixup in a diagnostic printf. o Make sure that we are in the proper mode before disabling selections in ahd_update_pending_scbs. o Add more diagnostics. o task_attribute_nonpkt_tag -> task_attribute: we don't need a nonpkt_tag field anymore for allowing all 512 SCBs to be used in non-packetized connections. o Negotiate HOLD_MCS to U320 devices. o Add a few additional mode assertions. o Restore the chip mode after clearing out the qinfifo so that code using ahd_abort_scbs sees a consistent mode. o Simplify the DMA engine shutdown routine prior to performing a bus reset. o Perform the sequencer restart after a chip reset prior to setting up our timer to poll for the reset to be complete. On some OSes, the timer could actually pre-empt us and order is important here. o Have our "reset poller" set the expected mode since there is no guarantee of what mode will be in force when we are called from the OS timer. o Save and restore the SCB pointer in ahd_dump_card_state(). This routine must not modify card state. o Ditto for ahd_dump_scbs(). aic79xx.h: o Add a few more chip bug definitions. o Align our tag on a 32bit boundary. aic79xx.reg: aic79xx.seq: o Start work on removing workarounds for Rev B. o Use a special location in scratch from for stroring our SCBPTR during legacy FIFO allocations. This corrects problems in mixed packetized/non-packetized configurations where calling into a FIFO task corrupted our SCBPTR. o Don't rely on DMA priority to guarantee that all data in our FIFOs will flush prior to a command completion notification going out of the command channel. We've never seen this assumption fail, but better safe than sorry. o Deal with missing BUSFREEREV feature in H2A. o Simplify disconnect list code now that the list will always have only a single entry. o Implement the AHD_REG_SLOW_SETTLE_BUG workaround. o Swith to using "REG_ISR" for local mode scratch during our ISR. o Add a missing jmp to the data_group_dma_loop after our data pointers have been re-initialized by the kernel. o Correct test in the bitbucket code so that we actually wait for the bitbucket to complete before signaling the kernel of the overrun condition. o Reposition pkt_saveptrs to avoid a jmp instruction. o Update a comment to reflect that the code now waits for a FIFO to drain prior to issuing a CLRCHN. aic79xx_inline.h: o Remove unused untagged queue handling code. o Don't attempt to htole64 what could be a 32bit value. aic79xx_pci.c: o Set additional bug flags for rev A chips.
2002-09-26 22:54:00 +00:00
*/
ahc_eisa.c: ahc_pci.c: ahd_pci.c: aic7xxx.c: aic79xx.c: aic_osm_lib.c: aic_osm_lib.h: Use common OSM routines from aic_osm_lib for bus dma operations, delay routines, accessing CCBs, byte swapping, etc. aic7xxx_pci.c: Provide a better description for the 2915/30LP on attach. aic7xxx.c: aic79xx.c: aic7770.c: aic79xx_pci.c: aic7xxx_pci.c: aic7xxx_93cx6.c: Move FBSDID behind an ifdef so that these core files will still compile under other OSes. aic79xx.h: aic79xx_pci.c: aic79xx.seq: To speed up non-packetized CDB delivery in Rev B, all CDB acks are "released" to the output sync as soon as the command phase starts. There is only one problem with this approach. If the target changes phase before all data are sent, we have left over acks that can go out on the bus in a data phase. Due to other chip contraints, this only happens if the target goes to data-in, but if the acks go out before we can test SDONE, we'll think that the transfer has completed successfully. Work around this by taking advantage of the 400ns or 800ns dead time between command phase and the REQ of the new phase. If the transfer has completed successfully, SCSIEN should fall *long* before we see a phase change. We thus treat any phasemiss that occurs before SCSIEN falls as an incomplete transfer. aic79xx.h: Add the AHD_FAST_CDB_DELIVERY feature. aic79xx_pci.c: Set AHD_FAST_CDB_DELIVERY for all Rev. B parts. aic79xx.seq: Test for PHASEMIS in the command phase for all AHD_FAST_CDB_DELIVERY controlelrs. ahd_pci.c: ahc_pci.c: aic7xxx.h: aic79xx.h: Move definition of controller BAR offsets to core header files. aic7xxx.c: aic79xx.c: In the softc free routine, leave removal of a softc from the global list of softcs to the OSM (the caller of this routine). This allows us to avoid holding the softc list_lock during device destruction where we may have to sleep waiting for our recovery thread to halt. ahc_pci.c: Use ahc_pci_test_register access to validate I/O mapped in addition to the tests already performed for memory mapped access. Remove unused ahc_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. ahd_pci.c: Remove reduntant definition of controller BAR offsets. These are also defined in aic79xx.h. Remove unused ahd_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. aic7xxx.c: aic79xx.c: aic79xx.h: aic7xxx.h: aic7xxx_osm.c: aic79xx_osm.c: Move timeout handling to the driver cores. In the case of the aic79xx driver, the algorithm has been enhanced to try target resets before performing a bus reset. For the aic7xxx driver, the algorithm is unchanged. Although the drivers do not currently sleep during recovery (recovery is timeout driven), the cores do expect all processing to be performed via a recovery thread. Our timeout handlers are now little stubs that wakeup the recovery thread. aic79xx.c: aic79xx.h: aic79xx_inline.h: Change shared_data allocation to use a map_node so that the sentinel hscb can use this map node in ahd_swap_with_next_hscb. This routine now swaps the hscb_map pointer in additon to the hscb contents so that any sync operations occur on the correct map. physaddr -> busaddr Pointed out by: Jason Thorpe <thorpej@wasabisystems.com> aic79xx.c: Make more use of the in/out/w/l/q macros for accessing byte registers in the chip. Correct some issues in the ahd_flush_qoutfifo() routine. o Run the qoutfifo only once the command channel DMA engine has been halted. This closes a window where we might have missed some entries. o Change ahd_run_data_fifo() to not loop to completion. If we happen to start on the wrong FIFO and the other FIFO has a snapshot savepointers, we might deadlock. This required our delay between FIFO tests to be moved to the ahd_flush_qoutfifo() routine. o Update/add comments. o Remove spurious test for COMPLETE_DMA list being empty when completing transactions from the GSFIFO with residuals. The SCB must be put on the COMPLETE_DMA scb list unconditionally. o When halting command channel DMA activity, we must disable the DMA channel in all cases but an update of the QOUTFIFO. The latter case is required so that the sequencer will update its position in the QOUTFIFO. Previously, we left the channel enabled for all "push" DMAs. This left us vulnerable to the sequencer handling an SCB push long after that SCB was already processed manually by this routine. o Correct the polarity of tests involving ahd_scb_active_in_fifo(). This routine returns non-zero for true. Return to processing bad status completions through the qoutfifo. This reduces the time that the sequencer is kept paused when handling transactions with bad status or underruns. When waiting for the controller to quiece selections, add a delay to our loop. Otherwise we may fail to wait long enough for the sequencer to comply. On H2A4 hardware, use the slow slewrate for non-paced transfers. This mirrors what the Adaptec Windows drivers do. On the Rev B. only slow down the CRC timing for older U160 devices that might need the slower timing. We define "older" as devices that do not support packetized protocol. Wait up to 5000 * 5us for the SEEPROM to become unbusy. Write ops seem to take much longer than read ops. aic79xx.seq: For controllers with the FAINT_LED bug, turn the diagnostic led feature on during selection and reselection. This covers the non-packetized case. The LED will be disabled for non-packetized transfers once we return to the top level idle loop. Add more comments about the busy LED workaround. Extend a critical section around the entire command channel idle loop process. Previously the portion of this handler that directly manipulated the linked list of completed SCBs was not protected. This is the likely cause of the recent reports of commands being completed twice by the driver. Extend critical sections across the test for, and the longjump to, longjump routines. This prevents the firmware from trying to jump to a longjmp handler that was just cleared by the host. Improve the locations of several critical section begin and end points. Typically these changes remove instructions that did not need to be inside a critical section. Close the "busfree after selection, but before busfree interrupts can be enabled" race to just a single sequencer instruction. We now test the BSY line explicitly before clearing the busfree status and enabling the busfree interrupt. Close a race condition in the processing of HS_MAILBOX updates. We now clear the "updated" status before the copy. This ensures that we don't accidentally clear the status incorrectly when the host sneaks in an update just after our last copy, but before we clear the status. This race has never been observed. Don't re-enable SCSIEN if we lose the race to disable SCSIEN in our interrupt handler's workaround for the RevA data-valid too early issue. aic79xx_inline.h: Add comments indicating that the order in which bytes are read or written in ahd_inw and ahd_outw is important. This allows us to use these inlines when accessing registers with side-effects. aic79xx_pci.c: The 29320 and the 29320B are 7902 not 7901 based products. Correct the driver banner. aic7xxx.h: Enable the use of the auto-access pause feature on the aic7870 and aic7880. It was disabled due to an oversight. aic7xxx.reg: Move TARG_IMMEDIATE_SCB to alias LAST_MSG to avoid leaving garbage in MWI_RESIDUAL. This prevents spurious overflows whn operating target mode on controllers that require the MWI_RESIDUAL work-around. aic7xxx.seq: AHC_TMODE_WIDEODD_BUG is a bug, not a softc flag. Reference the correct softc field when testing for its presence. Set the NOT_IDENTIFIED and NO_CDB_SENT bits in SEQ_FLAGS to indicate that the nexus is invalid in await busfree. aic7xxx_93cx6.c: Add support for the C56/C66 versions of the EWEN and EWDS commands. aic7xxx.c: aic7xxx_pci.c: Move test for the validity of left over BIOS data to ahc_test_register_access(). This guarantees that any left over CHIPRST value is not clobbered by our register access test and lost to the test that was in ahc_reset.
2003-12-17 00:02:10 +00:00
test SCSISIGI, BSYI jz . + 2;
Upgrade to version 1.1 of the aic79xx U320 driver. aic79xx.c: o Remove redundant ahd_update_modes() call. o Correct panic in diagnostic should state corruption cause the SCB Id to be invalid during a selection timeout. o Add workaround for missing BUSFREEREV feature in Rev A silicon. o Corect formatting nits. o Use register pretty printing in more places. o Save and restore our SCB pointer when updating the waiting queue list for an "expected" LQ-out busfree. o In ahd_clear_intstat, deal with the missing autoclear in the CLRLQO* registers. o BE fixup in a diagnostic printf. o Make sure that we are in the proper mode before disabling selections in ahd_update_pending_scbs. o Add more diagnostics. o task_attribute_nonpkt_tag -> task_attribute: we don't need a nonpkt_tag field anymore for allowing all 512 SCBs to be used in non-packetized connections. o Negotiate HOLD_MCS to U320 devices. o Add a few additional mode assertions. o Restore the chip mode after clearing out the qinfifo so that code using ahd_abort_scbs sees a consistent mode. o Simplify the DMA engine shutdown routine prior to performing a bus reset. o Perform the sequencer restart after a chip reset prior to setting up our timer to poll for the reset to be complete. On some OSes, the timer could actually pre-empt us and order is important here. o Have our "reset poller" set the expected mode since there is no guarantee of what mode will be in force when we are called from the OS timer. o Save and restore the SCB pointer in ahd_dump_card_state(). This routine must not modify card state. o Ditto for ahd_dump_scbs(). aic79xx.h: o Add a few more chip bug definitions. o Align our tag on a 32bit boundary. aic79xx.reg: aic79xx.seq: o Start work on removing workarounds for Rev B. o Use a special location in scratch from for stroring our SCBPTR during legacy FIFO allocations. This corrects problems in mixed packetized/non-packetized configurations where calling into a FIFO task corrupted our SCBPTR. o Don't rely on DMA priority to guarantee that all data in our FIFOs will flush prior to a command completion notification going out of the command channel. We've never seen this assumption fail, but better safe than sorry. o Deal with missing BUSFREEREV feature in H2A. o Simplify disconnect list code now that the list will always have only a single entry. o Implement the AHD_REG_SLOW_SETTLE_BUG workaround. o Swith to using "REG_ISR" for local mode scratch during our ISR. o Add a missing jmp to the data_group_dma_loop after our data pointers have been re-initialized by the kernel. o Correct test in the bitbucket code so that we actually wait for the bitbucket to complete before signaling the kernel of the overrun condition. o Reposition pkt_saveptrs to avoid a jmp instruction. o Update a comment to reflect that the code now waits for a FIFO to drain prior to issuing a CLRCHN. aic79xx_inline.h: o Remove unused untagged queue handling code. o Don't attempt to htole64 what could be a 32bit value. aic79xx_pci.c: o Set additional bug flags for rev A chips.
2002-09-26 22:54:00 +00:00
mvi CLRSINT1,CLRBUSFREE;
or SIMODE1, ENBUSFREE;
}
mov SAVED_SCSIID, SCB_SCSIID;
mov SAVED_LUN, SCB_LUN;
aic79xx.c: Implement the SCB_SILENT flag. This is useful for hushing up the driver during DV or other operations that we expect to cause transmission errors. The messages will still print if the SHOW_MASKED_ERRORS debug option is enabled. Save and restore the NEGOADDR address when setting new transfer settings. The sequencer performs lookups in the negotiation table too and it expects NEGOADDR to remain consistent across pause/unpause sessions. Consistently use "offset" instead of "period" to determine if we are running sync or not. Add a SHOW_MESSAGES diagnostic for when we assert ATN during message processing. Print out IU, QAS, and RTI features when showing transfer options. Limit the syncrate after all option conformance changes have taken place in ahd_devlimited_syncrate. Changes in options may change the final syncrate we accept. Keep a copy of the hs_mailbox in our softc so that we can perform read/modify/write operations on the hs_mailbox without having to pause the sequencer to read the last written value. Use the ENINT_COALESS flag in the hs_mailbox to toggle interrupt coalessing. Add entrypoints for enabling interrupt coalessing and setting both a timeout (how long to wait for commands to be coalessed) and a maximum commands to coaless value. Add a statistics timer that decides when to enable or disable interrupt coalessing based on load. Add a routine, ahd_reset_cmds_pending() which is used to update the CMDS_PENDING sequencer variable whenever error recovery compeltes SCBs without notifying the sequencer. Since ahd_reset_cmds_pending is called during ahd_unpause() only if we've aborted SCBs, its call to ahd_flush_qoutfifo should not cause recursion through ahd_run_qoutfifo(). A panic has been added to ensure that this recursion does not occur. In ahd_search_qinfifo, update the CMDS_PENDING sequencer variable directly. ahd_search_qinififo can be called in situations where using ahd_reset_cmds_pending() might cause recursion. Since we can safely determine the exact number to reduce CMDS_PENDING by in this scenario without running the qoutfifo, the manual update is sufficient. Clean up diagnostics. Add ahd_flush_qoutfifo() which will run the qoutfifo as well as complete any commands sitting on the sequencer's COMPLETE_SCB lists or the good status FIFO. Use this routine in several places that did similar things in an add-hoc, but incomplete, fashion. A call to this routine was also added to ahd_abort_scbs() to close a race. In ahd_pause_and_flushwork() only return once selections are safely disabled. Flush all completed commands via ahd_flush_qoutfifo(). Remove "Now packetized" diagnostic now that this information is incorperated into the actual negotiation messages that are displayed. When forcing renegotiation, don't clober the current ppr_options. Much of the driver uses this information to determine if we are currently packetized or not. Remove some stray spaces at column 1 in ahd_set_tags. When complaining about getting a host message loop request with no pending messages, print out the SCB_CONTROL register down on the card. Modify the ahd_sent_msg() routine to handle a search for an outgoing identify message. Use this to detect a msg reject on an identify message which typically indicates that the target thought we were packetized. Force a renegotiation in this case. In ahd_search_qinfifo(), wait more effectively for SCB DMA activities to cease. We also disable SCB fetch operations since we are about to change the qinfifo and any fetch in progress will likely be invalidated. In ahd_qinfifo_count(), fix the qinfifo empty case. In ahd_dump_card_state(), print out CCSCBCTL in the correct mode. If we are a narrow controller, don't set the current width to unknown when forcing a future negotiation. This just confuses the code into attempting a wide negotiation on a narrow bus. Add support for task management function completions. Modify ahd_handle_devreset so that it can handle lun resets in addition to target resets. Use ahd_handle_devreset for lun and target reset task management functions. Handle the abort task TMF race case better. We now wait until any current selections are over and then set the TMF back to zero. This should cause the sequencer to ignore the abort TMF completion should it occur. Correct a bug in the illegal phase handler that caused us to drop down to narrow when handling the unexpected command phase case after 3rd party reset of a packetized device. Indicate the features, bugs, and flags set in the softc that are used to control firmware patch download when booting verbose. aic79xx.h: Add coalessing and HS_MAILBOX fields. Add per-softc variables for the stats "daemon". Add a debug option for interrupt coalessing activities. Add two new softc flags: o AHD_UPDATE_PEND_CMDS Run ahd_reset_cmds_pending() on the next unpause. o AHD_RUNNING_QOUTFIFO Used to catch recursion through ahd_run_qoutfifo(). aic79xx.reg: Correct register addresses related to the software timer and the DFDBCTL register. Add constants paramaterizing the software timer. Add scratch ram locations for storing interrupt coalessing tunables. Break INTMASK in SEQITNCTL out into INTMASK1 and INTMASK2. In at least the REV A, these are writable bits. We make use of that for a swtimer workaround in the sequencer. Since HS_MAILBOX autoclears, provide a sequencer variable to store its contents. Add SEQINT codes for handling task management completions. aic79xx.seq: Correct ignore wide residue processing check for a wide negotiation being in effect. We must be in the SCSI register window in order to access the negotiation table. Use the software timer and a commands completed count to implement interrupt coalessing. The command complete is deferred until either the maximum command threshold or a the expiration of a command deferral timer. If we have more SCBs to complete to the host (sitting in COMPLETE_SCB lists), always try to coaless them up to our coalessing limit. If coalessing is enabled, but we have fewer commands oustanting than the host's min coalessing limit, complete the command immediately. Add code to track the number of commands outstanding. Commands are outstanding from the time they are placed into the execution queue until the DMA to post completion is setup. Add a workaround for intvec_2 interrupts on the H2A4. In H2A4, the mode pointer is not saved for intvec2, but is restored on iret. This can lead to the restoration of a bogus mode ptr. Manually clear the intmask bits and do a normal return to compensate. We use intvec_2 to track interrupt coalessing timeouts. Since we cannot disable the swtimer's countdown, simply mask its interrupt once we no longer care about it firing. In idle_loop_cchan, update LOCAL_HS_MAILBOX everytime we are notified of an HS_MAILBOX update via the HS_MAILBOX_ACT bit in QOFF_CTLSTA. We have to use a local copy of persistant portions of the HS_MAILBOX as the mailbox auto-clears on any read. Move the test for the cfg4istat interrupt up an instruction to hopefully close a race between the next outgoing selection and our disabling of selections. Add a missing ret to the last instruction in load_overrun_buf. Add notifications to the host of task management completions as well as the completions for commands that completed successfully before their corresponding TMF could be sent. Hold a critical section during select-out processing until we have a fully identified connection. This removes a race condition with the legacy abort handler. Correct a few spelling errors in some comments. aic79xx_inline.h: Call ahd_reset_cmds_pending() in ahd_unpause if required. Update cmdcmplt interrupt statistics in our interrupt handler. Allow callers to ahd_send_scb() to set the task management function. aic79xx_pci.c: Disable SERR and pause the controller prior to performing our mmapped I/O test. The U320 controllers do not support "auto-access-pause". aic79xx_osm.c: Set the task management function now that ahd_send_scb() doesn't do it for us. We also perform a lun reset in response to BDR requests to packetized devices.
2003-01-20 20:17:35 +00:00
mvi SEQ_FLAGS, NO_CDB_SENT;
END_CRITICAL;
or SXFRCTL0, SPIOEN;
/*
* As soon as we get a successful selection, the target
* should go into the message out phase since we have ATN
* asserted.
*/
mvi MSG_OUT, MSG_IDENTIFYFLAG;
/*
* Main loop for information transfer phases. Wait for the
* target to assert REQ before checking MSG, C/D and I/O for
* the bus phase.
*/
mesgin_phasemis:
ITloop:
call phase_lock;
mov A, LASTPHASE;
test A, ~P_DATAIN_DT jz p_data;
cmp A,P_COMMAND je p_command;
cmp A,P_MESGOUT je p_mesgout;
cmp A,P_STATUS je p_status;
cmp A,P_MESGIN je p_mesgin;
SET_SEQINTCODE(BAD_PHASE)
jmp ITloop; /* Try reading the bus again. */
/*
* Command phase. Set up the DMA registers and let 'er rip.
*/
p_command:
test SEQ_FLAGS, NOT_IDENTIFIED jz p_command_okay;
SET_SEQINTCODE(PROTO_VIOLATION)
p_command_okay:
test MODE_PTR, ~(MK_MODE(M_DFF1, M_DFF1))
jnz p_command_allocate_fifo;
/*
* Command retry. Free our current FIFO and
* re-allocate a FIFO so transfer state is
* reset.
*/
SET_SRC_MODE M_DFF1;
SET_DST_MODE M_DFF1;
mvi DFFSXFRCTL, RSTCHN|CLRSHCNT;
SET_MODE(M_SCSI, M_SCSI)
p_command_allocate_fifo:
Upgrade to version 1.1 of the aic79xx U320 driver. aic79xx.c: o Remove redundant ahd_update_modes() call. o Correct panic in diagnostic should state corruption cause the SCB Id to be invalid during a selection timeout. o Add workaround for missing BUSFREEREV feature in Rev A silicon. o Corect formatting nits. o Use register pretty printing in more places. o Save and restore our SCB pointer when updating the waiting queue list for an "expected" LQ-out busfree. o In ahd_clear_intstat, deal with the missing autoclear in the CLRLQO* registers. o BE fixup in a diagnostic printf. o Make sure that we are in the proper mode before disabling selections in ahd_update_pending_scbs. o Add more diagnostics. o task_attribute_nonpkt_tag -> task_attribute: we don't need a nonpkt_tag field anymore for allowing all 512 SCBs to be used in non-packetized connections. o Negotiate HOLD_MCS to U320 devices. o Add a few additional mode assertions. o Restore the chip mode after clearing out the qinfifo so that code using ahd_abort_scbs sees a consistent mode. o Simplify the DMA engine shutdown routine prior to performing a bus reset. o Perform the sequencer restart after a chip reset prior to setting up our timer to poll for the reset to be complete. On some OSes, the timer could actually pre-empt us and order is important here. o Have our "reset poller" set the expected mode since there is no guarantee of what mode will be in force when we are called from the OS timer. o Save and restore the SCB pointer in ahd_dump_card_state(). This routine must not modify card state. o Ditto for ahd_dump_scbs(). aic79xx.h: o Add a few more chip bug definitions. o Align our tag on a 32bit boundary. aic79xx.reg: aic79xx.seq: o Start work on removing workarounds for Rev B. o Use a special location in scratch from for stroring our SCBPTR during legacy FIFO allocations. This corrects problems in mixed packetized/non-packetized configurations where calling into a FIFO task corrupted our SCBPTR. o Don't rely on DMA priority to guarantee that all data in our FIFOs will flush prior to a command completion notification going out of the command channel. We've never seen this assumption fail, but better safe than sorry. o Deal with missing BUSFREEREV feature in H2A. o Simplify disconnect list code now that the list will always have only a single entry. o Implement the AHD_REG_SLOW_SETTLE_BUG workaround. o Swith to using "REG_ISR" for local mode scratch during our ISR. o Add a missing jmp to the data_group_dma_loop after our data pointers have been re-initialized by the kernel. o Correct test in the bitbucket code so that we actually wait for the bitbucket to complete before signaling the kernel of the overrun condition. o Reposition pkt_saveptrs to avoid a jmp instruction. o Update a comment to reflect that the code now waits for a FIFO to drain prior to issuing a CLRCHN. aic79xx_inline.h: o Remove unused untagged queue handling code. o Don't attempt to htole64 what could be a 32bit value. aic79xx_pci.c: o Set additional bug flags for rev A chips.
2002-09-26 22:54:00 +00:00
bmov ALLOCFIFO_SCBPTR, SCBPTR, 2;
call allocate_fifo;
SET_SRC_MODE M_DFF1;
SET_DST_MODE M_DFF1;
add NONE, -17, SCB_CDB_LEN;
jnc p_command_embedded;
p_command_from_host:
aic79xx.c: Clear the LQICRC_NLQ status should it pop up after we have already handled the SCSIPERR. During some streaming operations this status can be delayed until the stream ends. Without this change, the driver would complain about a "Missing case in ahd_handle_scsiint". In the LQOBUSFREE handler... Don't return the LQOMGR back to the idle state until after we have cleaned up ENSELO and any status related to this selection. The last thing we need is the LQO manager starting another select-out before we have updated the execution queue. It is not clear whether the LQOMGR would, or would not start a new selection early. Make sure ENSELO is off prior to clearing SELDO by flushing device writes. Move assignment of the next target SCB pointer inside of an if to make the code clearer. The effect is the same. Dump card state in both "Unexpected PKT busfree" paths. In ahd_reset(), set the chip to SCSI mode before reading SXFRCTL1. That register only exists in the SCSI mode. Also set the mode explicitly to the SCSI mode after chip reset due to paranoia. Re-arrange code so that SXFRCTL1 is restored as quickly after the chip reset as possible. S/G structurs must be 8byte aligned. Make this official by saying so in our DMA tag. Disable CIO bus stretch on MDFFSTAT if SHVALID is about to come true. This can cause a CIO bus lockup if a PCI or PCI-X error occurs while the stretch is occurring - the host cannot service the PCI-X error since the CIO bus is locked out and SHVALID will never resolve. The stretch was added in the Rev B to simplify the wait for SHVALID to resolve, but the code to do this in the open source sequencer is so simple it was never removed. Consistently use MAX_OFFSET for the user max syncrate set from non-volatile storage. This ensures that the offset does not conflict with AH?_OFFSET_UNKNOWN. Have ahd_pause_and_flushwork set the mode to ensure that it has access to the registers it checks. Also modify the checking of intstat so that the check against 0xFF can actually succeed if the INT_PEND mask is something other than 0xFF. Although there are no cardbus U320 controllers, this check may be needed to recover from a hot-plug PCI removal that occurs without informing the driver. Fix a typo. sg_prefetch_cnt -> sg_prefetch_align. This fixes an infinite loop at card initialization if the cacheline size is 0. aic79xx.h: Add AHD_EARLY_REQ_BUG bug flag. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. Change AH?_OFFSET_UNKNOWN to 0xFF. This is a value that the curr->offset can never be, unlike '0' which we previously used. This fixes code that only checks for a non-zero offset to determine if a sync negotiation is required since it will fire in the unknown case even if the goal is async. aic79xx.reg: Add comments for LQISTAT bits indicating their names in the 7902 data book. We use slightly different and more descriptive names in the firmware. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. aic79xx.seq: Update comments regarding rundown of the GSFIFO to reflect reality. Fix spelling errors. Since we use an 8byte address and 1 byte length, shorten the size of a block move for the legacy DMA'ed CDB case from 11 to 9 bytes. Remove code that, assuming the abort pending feature worked, would set MK_MESSAGE in the SCB's control byte on completion to catch invalid reselections. Since we don't see interrupts for completed selections, this status update could occur prior to us noticing the SELDO. The "select-out" queue logic will get confused by the MK_MESSAGE bit being set as this is used to catch packatized connections where we select-out with ATN. Since the abort pending feature doesn't work on any released controllers yet, this code was never executed. Add support for the AHD_EARLY_REQ_BUG. Don't ignore persistent REQ assertions just because they were asserted within the bus settle delay window. This allows us to tolerate devices like the GEM318 that violate the SCSI spec. Remove unintentional settnig of SG_CACHE_AVAIL. Writing this bit should have no effect, but who knows... On the Rev A, we must wait for HDMAENACK before loading additional segments to avoid clobbering the address of the first segment in the S/G FIFO. This resolves data-corruption issues with certain IBM (now Hitachi) and Fujitsu U320 drives. Rearrange calc_residual to avoid an extra jmp instruction. On RevA Silicon, if the target returns us to data-out after we have already trained for data-out, it is possible for us to transition the free running clock to data-valid before the required 100ns P1 setup time (8 P1 assertions in fast-160 mode). This will only happen if this L-Q is a continuation of a data transfer for which we have already prefetched data into our FIFO (LQ/Data followed by LQ/Data for the same write transaction). This can cause some target implementations to miss the first few data transfers on the bus. We detect this situation by noticing that this is the first data transfer after an LQ (LQIWORKONLQ true), that the data transfer is a continuation of a transfer already setup in our FIFO (SAVEPTRS interrupt), and that the transaction is a write (DIRECTION set in DFCNTRL). The delay is performed by disabling SCSIEN until we see the first REQ from the target. Only compile in snapshot savepointers handler for RevA silicon where it is enabled. Handle the cfg4icmd packetized interrupt. We just need to load the address and count, start the DMA, and CLRCHN once the transfer is complete. Fix an oversight in the overrun handler for packetized status operations. We need to wait for either CTXTDONE or an overrun when checking for an overrun. The previous code did not wait and thus could decide that no overrun had occurred even though an overrun will occur on the next data-valid req. Add some comment to this section for clarity. Use LAST_SEG_DONE instead of LASTSDONE for testing transfer completion in the packetized status case. LASTSDONE may come up more quickly since it only records completion on the SCSI side, but since LAST_SEG_DONE is used everywhere else (and needs to be), this is less confusing. Add a missing invalidation of the longjmp address in the non-pack handler. This code needs additional review. aic79xx_inline.h: Fix spelling error. aic79xx_osm.c: Set the cdb length for CDBs dma'ed from host memory. Add a comment indicating that, should CAM start supporting cdbs larger than 16bytes, the driver could store the CDB in the status buffer. aic79xx_pci.c: Add a table entry for the 39320A. Added a missing comma to an error string table. Fix spelling errors.
2003-02-27 23:23:16 +00:00
bmov HADDR[0], SCB_HOST_CDB_PTR, 9;
mvi SG_CACHE_PRE, LAST_SEG;
mvi DFCNTRL, (PRELOADEN|SCSIEN|HDMAEN);
jmp p_command_xfer;
p_command_embedded:
bmov SHCNT[0], SCB_CDB_LEN, 1;
bmov DFDAT, SCB_CDB_STORE, 16;
mvi DFCNTRL, SCSIEN;
p_command_xfer:
and SEQ_FLAGS, ~NO_CDB_SENT;
ahc_eisa.c: ahc_pci.c: ahd_pci.c: aic7xxx.c: aic79xx.c: aic_osm_lib.c: aic_osm_lib.h: Use common OSM routines from aic_osm_lib for bus dma operations, delay routines, accessing CCBs, byte swapping, etc. aic7xxx_pci.c: Provide a better description for the 2915/30LP on attach. aic7xxx.c: aic79xx.c: aic7770.c: aic79xx_pci.c: aic7xxx_pci.c: aic7xxx_93cx6.c: Move FBSDID behind an ifdef so that these core files will still compile under other OSes. aic79xx.h: aic79xx_pci.c: aic79xx.seq: To speed up non-packetized CDB delivery in Rev B, all CDB acks are "released" to the output sync as soon as the command phase starts. There is only one problem with this approach. If the target changes phase before all data are sent, we have left over acks that can go out on the bus in a data phase. Due to other chip contraints, this only happens if the target goes to data-in, but if the acks go out before we can test SDONE, we'll think that the transfer has completed successfully. Work around this by taking advantage of the 400ns or 800ns dead time between command phase and the REQ of the new phase. If the transfer has completed successfully, SCSIEN should fall *long* before we see a phase change. We thus treat any phasemiss that occurs before SCSIEN falls as an incomplete transfer. aic79xx.h: Add the AHD_FAST_CDB_DELIVERY feature. aic79xx_pci.c: Set AHD_FAST_CDB_DELIVERY for all Rev. B parts. aic79xx.seq: Test for PHASEMIS in the command phase for all AHD_FAST_CDB_DELIVERY controlelrs. ahd_pci.c: ahc_pci.c: aic7xxx.h: aic79xx.h: Move definition of controller BAR offsets to core header files. aic7xxx.c: aic79xx.c: In the softc free routine, leave removal of a softc from the global list of softcs to the OSM (the caller of this routine). This allows us to avoid holding the softc list_lock during device destruction where we may have to sleep waiting for our recovery thread to halt. ahc_pci.c: Use ahc_pci_test_register access to validate I/O mapped in addition to the tests already performed for memory mapped access. Remove unused ahc_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. ahd_pci.c: Remove reduntant definition of controller BAR offsets. These are also defined in aic79xx.h. Remove unused ahd_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. aic7xxx.c: aic79xx.c: aic79xx.h: aic7xxx.h: aic7xxx_osm.c: aic79xx_osm.c: Move timeout handling to the driver cores. In the case of the aic79xx driver, the algorithm has been enhanced to try target resets before performing a bus reset. For the aic7xxx driver, the algorithm is unchanged. Although the drivers do not currently sleep during recovery (recovery is timeout driven), the cores do expect all processing to be performed via a recovery thread. Our timeout handlers are now little stubs that wakeup the recovery thread. aic79xx.c: aic79xx.h: aic79xx_inline.h: Change shared_data allocation to use a map_node so that the sentinel hscb can use this map node in ahd_swap_with_next_hscb. This routine now swaps the hscb_map pointer in additon to the hscb contents so that any sync operations occur on the correct map. physaddr -> busaddr Pointed out by: Jason Thorpe <thorpej@wasabisystems.com> aic79xx.c: Make more use of the in/out/w/l/q macros for accessing byte registers in the chip. Correct some issues in the ahd_flush_qoutfifo() routine. o Run the qoutfifo only once the command channel DMA engine has been halted. This closes a window where we might have missed some entries. o Change ahd_run_data_fifo() to not loop to completion. If we happen to start on the wrong FIFO and the other FIFO has a snapshot savepointers, we might deadlock. This required our delay between FIFO tests to be moved to the ahd_flush_qoutfifo() routine. o Update/add comments. o Remove spurious test for COMPLETE_DMA list being empty when completing transactions from the GSFIFO with residuals. The SCB must be put on the COMPLETE_DMA scb list unconditionally. o When halting command channel DMA activity, we must disable the DMA channel in all cases but an update of the QOUTFIFO. The latter case is required so that the sequencer will update its position in the QOUTFIFO. Previously, we left the channel enabled for all "push" DMAs. This left us vulnerable to the sequencer handling an SCB push long after that SCB was already processed manually by this routine. o Correct the polarity of tests involving ahd_scb_active_in_fifo(). This routine returns non-zero for true. Return to processing bad status completions through the qoutfifo. This reduces the time that the sequencer is kept paused when handling transactions with bad status or underruns. When waiting for the controller to quiece selections, add a delay to our loop. Otherwise we may fail to wait long enough for the sequencer to comply. On H2A4 hardware, use the slow slewrate for non-paced transfers. This mirrors what the Adaptec Windows drivers do. On the Rev B. only slow down the CRC timing for older U160 devices that might need the slower timing. We define "older" as devices that do not support packetized protocol. Wait up to 5000 * 5us for the SEEPROM to become unbusy. Write ops seem to take much longer than read ops. aic79xx.seq: For controllers with the FAINT_LED bug, turn the diagnostic led feature on during selection and reselection. This covers the non-packetized case. The LED will be disabled for non-packetized transfers once we return to the top level idle loop. Add more comments about the busy LED workaround. Extend a critical section around the entire command channel idle loop process. Previously the portion of this handler that directly manipulated the linked list of completed SCBs was not protected. This is the likely cause of the recent reports of commands being completed twice by the driver. Extend critical sections across the test for, and the longjump to, longjump routines. This prevents the firmware from trying to jump to a longjmp handler that was just cleared by the host. Improve the locations of several critical section begin and end points. Typically these changes remove instructions that did not need to be inside a critical section. Close the "busfree after selection, but before busfree interrupts can be enabled" race to just a single sequencer instruction. We now test the BSY line explicitly before clearing the busfree status and enabling the busfree interrupt. Close a race condition in the processing of HS_MAILBOX updates. We now clear the "updated" status before the copy. This ensures that we don't accidentally clear the status incorrectly when the host sneaks in an update just after our last copy, but before we clear the status. This race has never been observed. Don't re-enable SCSIEN if we lose the race to disable SCSIEN in our interrupt handler's workaround for the RevA data-valid too early issue. aic79xx_inline.h: Add comments indicating that the order in which bytes are read or written in ahd_inw and ahd_outw is important. This allows us to use these inlines when accessing registers with side-effects. aic79xx_pci.c: The 29320 and the 29320B are 7902 not 7901 based products. Correct the driver banner. aic7xxx.h: Enable the use of the auto-access pause feature on the aic7870 and aic7880. It was disabled due to an oversight. aic7xxx.reg: Move TARG_IMMEDIATE_SCB to alias LAST_MSG to avoid leaving garbage in MWI_RESIDUAL. This prevents spurious overflows whn operating target mode on controllers that require the MWI_RESIDUAL work-around. aic7xxx.seq: AHC_TMODE_WIDEODD_BUG is a bug, not a softc flag. Reference the correct softc field when testing for its presence. Set the NOT_IDENTIFIED and NO_CDB_SENT bits in SEQ_FLAGS to indicate that the nexus is invalid in await busfree. aic7xxx_93cx6.c: Add support for the C56/C66 versions of the EWEN and EWDS commands. aic7xxx.c: aic7xxx_pci.c: Move test for the validity of left over BIOS data to ahc_test_register_access(). This guarantees that any left over CHIPRST value is not clobbered by our register access test and lost to the test that was in ahc_reset.
2003-12-17 00:02:10 +00:00
if ((ahd->features & AHD_FAST_CDB_DELIVERY) != 0) {
/*
* To speed up CDB delivery in Rev B, all CDB acks
* are "released" to the output sync as soon as the
* command phase starts. There is only one problem
* with this approach. If the target changes phase
* before all data are sent, we have left over acks
* that can go out on the bus in a data phase. Due
* to other chip contraints, this only happens if
* the target goes to data-in, but if the acks go
* out before we can test SDONE, we'll think that
* the transfer has completed successfully. Work
* around this by taking advantage of the 400ns or
* 800ns dead time between command phase and the REQ
* of the new phase. If the transfer has completed
* successfully, SCSIEN should fall *long* before we
* see a phase change. We thus treat any phasemiss
* that occurs before SCSIEN falls as an incomplete
* transfer.
*/
test SSTAT1, PHASEMIS jnz p_command_xfer_failed;
test DFCNTRL, SCSIEN jnz . - 1;
} else {
test DFCNTRL, SCSIEN jnz .;
}
/*
* DMA Channel automatically disabled.
* Don't allow a data phase if the command
* was not fully transferred.
*/
test SSTAT2, SDONE jnz ITloop;
ahc_eisa.c: ahc_pci.c: ahd_pci.c: aic7xxx.c: aic79xx.c: aic_osm_lib.c: aic_osm_lib.h: Use common OSM routines from aic_osm_lib for bus dma operations, delay routines, accessing CCBs, byte swapping, etc. aic7xxx_pci.c: Provide a better description for the 2915/30LP on attach. aic7xxx.c: aic79xx.c: aic7770.c: aic79xx_pci.c: aic7xxx_pci.c: aic7xxx_93cx6.c: Move FBSDID behind an ifdef so that these core files will still compile under other OSes. aic79xx.h: aic79xx_pci.c: aic79xx.seq: To speed up non-packetized CDB delivery in Rev B, all CDB acks are "released" to the output sync as soon as the command phase starts. There is only one problem with this approach. If the target changes phase before all data are sent, we have left over acks that can go out on the bus in a data phase. Due to other chip contraints, this only happens if the target goes to data-in, but if the acks go out before we can test SDONE, we'll think that the transfer has completed successfully. Work around this by taking advantage of the 400ns or 800ns dead time between command phase and the REQ of the new phase. If the transfer has completed successfully, SCSIEN should fall *long* before we see a phase change. We thus treat any phasemiss that occurs before SCSIEN falls as an incomplete transfer. aic79xx.h: Add the AHD_FAST_CDB_DELIVERY feature. aic79xx_pci.c: Set AHD_FAST_CDB_DELIVERY for all Rev. B parts. aic79xx.seq: Test for PHASEMIS in the command phase for all AHD_FAST_CDB_DELIVERY controlelrs. ahd_pci.c: ahc_pci.c: aic7xxx.h: aic79xx.h: Move definition of controller BAR offsets to core header files. aic7xxx.c: aic79xx.c: In the softc free routine, leave removal of a softc from the global list of softcs to the OSM (the caller of this routine). This allows us to avoid holding the softc list_lock during device destruction where we may have to sleep waiting for our recovery thread to halt. ahc_pci.c: Use ahc_pci_test_register access to validate I/O mapped in addition to the tests already performed for memory mapped access. Remove unused ahc_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. ahd_pci.c: Remove reduntant definition of controller BAR offsets. These are also defined in aic79xx.h. Remove unused ahd_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. aic7xxx.c: aic79xx.c: aic79xx.h: aic7xxx.h: aic7xxx_osm.c: aic79xx_osm.c: Move timeout handling to the driver cores. In the case of the aic79xx driver, the algorithm has been enhanced to try target resets before performing a bus reset. For the aic7xxx driver, the algorithm is unchanged. Although the drivers do not currently sleep during recovery (recovery is timeout driven), the cores do expect all processing to be performed via a recovery thread. Our timeout handlers are now little stubs that wakeup the recovery thread. aic79xx.c: aic79xx.h: aic79xx_inline.h: Change shared_data allocation to use a map_node so that the sentinel hscb can use this map node in ahd_swap_with_next_hscb. This routine now swaps the hscb_map pointer in additon to the hscb contents so that any sync operations occur on the correct map. physaddr -> busaddr Pointed out by: Jason Thorpe <thorpej@wasabisystems.com> aic79xx.c: Make more use of the in/out/w/l/q macros for accessing byte registers in the chip. Correct some issues in the ahd_flush_qoutfifo() routine. o Run the qoutfifo only once the command channel DMA engine has been halted. This closes a window where we might have missed some entries. o Change ahd_run_data_fifo() to not loop to completion. If we happen to start on the wrong FIFO and the other FIFO has a snapshot savepointers, we might deadlock. This required our delay between FIFO tests to be moved to the ahd_flush_qoutfifo() routine. o Update/add comments. o Remove spurious test for COMPLETE_DMA list being empty when completing transactions from the GSFIFO with residuals. The SCB must be put on the COMPLETE_DMA scb list unconditionally. o When halting command channel DMA activity, we must disable the DMA channel in all cases but an update of the QOUTFIFO. The latter case is required so that the sequencer will update its position in the QOUTFIFO. Previously, we left the channel enabled for all "push" DMAs. This left us vulnerable to the sequencer handling an SCB push long after that SCB was already processed manually by this routine. o Correct the polarity of tests involving ahd_scb_active_in_fifo(). This routine returns non-zero for true. Return to processing bad status completions through the qoutfifo. This reduces the time that the sequencer is kept paused when handling transactions with bad status or underruns. When waiting for the controller to quiece selections, add a delay to our loop. Otherwise we may fail to wait long enough for the sequencer to comply. On H2A4 hardware, use the slow slewrate for non-paced transfers. This mirrors what the Adaptec Windows drivers do. On the Rev B. only slow down the CRC timing for older U160 devices that might need the slower timing. We define "older" as devices that do not support packetized protocol. Wait up to 5000 * 5us for the SEEPROM to become unbusy. Write ops seem to take much longer than read ops. aic79xx.seq: For controllers with the FAINT_LED bug, turn the diagnostic led feature on during selection and reselection. This covers the non-packetized case. The LED will be disabled for non-packetized transfers once we return to the top level idle loop. Add more comments about the busy LED workaround. Extend a critical section around the entire command channel idle loop process. Previously the portion of this handler that directly manipulated the linked list of completed SCBs was not protected. This is the likely cause of the recent reports of commands being completed twice by the driver. Extend critical sections across the test for, and the longjump to, longjump routines. This prevents the firmware from trying to jump to a longjmp handler that was just cleared by the host. Improve the locations of several critical section begin and end points. Typically these changes remove instructions that did not need to be inside a critical section. Close the "busfree after selection, but before busfree interrupts can be enabled" race to just a single sequencer instruction. We now test the BSY line explicitly before clearing the busfree status and enabling the busfree interrupt. Close a race condition in the processing of HS_MAILBOX updates. We now clear the "updated" status before the copy. This ensures that we don't accidentally clear the status incorrectly when the host sneaks in an update just after our last copy, but before we clear the status. This race has never been observed. Don't re-enable SCSIEN if we lose the race to disable SCSIEN in our interrupt handler's workaround for the RevA data-valid too early issue. aic79xx_inline.h: Add comments indicating that the order in which bytes are read or written in ahd_inw and ahd_outw is important. This allows us to use these inlines when accessing registers with side-effects. aic79xx_pci.c: The 29320 and the 29320B are 7902 not 7901 based products. Correct the driver banner. aic7xxx.h: Enable the use of the auto-access pause feature on the aic7870 and aic7880. It was disabled due to an oversight. aic7xxx.reg: Move TARG_IMMEDIATE_SCB to alias LAST_MSG to avoid leaving garbage in MWI_RESIDUAL. This prevents spurious overflows whn operating target mode on controllers that require the MWI_RESIDUAL work-around. aic7xxx.seq: AHC_TMODE_WIDEODD_BUG is a bug, not a softc flag. Reference the correct softc field when testing for its presence. Set the NOT_IDENTIFIED and NO_CDB_SENT bits in SEQ_FLAGS to indicate that the nexus is invalid in await busfree. aic7xxx_93cx6.c: Add support for the C56/C66 versions of the EWEN and EWDS commands. aic7xxx.c: aic7xxx_pci.c: Move test for the validity of left over BIOS data to ahc_test_register_access(). This guarantees that any left over CHIPRST value is not clobbered by our register access test and lost to the test that was in ahc_reset.
2003-12-17 00:02:10 +00:00
p_command_xfer_failed:
or SEQ_FLAGS, NO_CDB_SENT;
jmp ITloop;
/*
* Status phase. Wait for the data byte to appear, then read it
* and store it into the SCB.
*/
SET_SRC_MODE M_SCSI;
SET_DST_MODE M_SCSI;
p_status:
test SEQ_FLAGS,NOT_IDENTIFIED jnz mesgin_proto_violation;
p_status_okay:
mov SCB_SCSI_STATUS, SCSIDAT;
or SCB_CONTROL, STATUS_RCVD;
jmp ITloop;
/*
* Message out phase. If MSG_OUT is MSG_IDENTIFYFLAG, build a full
* indentify message sequence and send it to the target. The host may
* override this behavior by setting the MK_MESSAGE bit in the SCB
* control byte. This will cause us to interrupt the host and allow
* it to handle the message phase completely on its own. If the bit
* associated with this target is set, we will also interrupt the host,
* thereby allowing it to send a message on the next selection regardless
* of the transaction being sent.
*
* If MSG_OUT is == HOST_MSG, also interrupt the host and take a message.
* This is done to allow the host to send messages outside of an identify
* sequence while protecting the seqencer from testing the MK_MESSAGE bit
* on an SCB that might not be for the current nexus. (For example, a
* BDR message in responce to a bad reselection would leave us pointed to
* an SCB that doesn't have anything to do with the current target).
*
* Otherwise, treat MSG_OUT as a 1 byte message to send (abort, abort tag,
* bus device reset).
*
* When there are no messages to send, MSG_OUT should be set to MSG_NOOP,
* in case the target decides to put us in this phase for some strange
* reason.
*/
p_mesgout_retry:
/* Turn on ATN for the retry */
mvi SCSISIGO, ATNO;
p_mesgout:
mov SINDEX, MSG_OUT;
cmp SINDEX, MSG_IDENTIFYFLAG jne p_mesgout_from_host;
test SCB_CONTROL,MK_MESSAGE jnz host_message_loop;
p_mesgout_identify:
or SINDEX, MSG_IDENTIFYFLAG|DISCENB, SCB_LUN;
test SCB_CONTROL, DISCENB jnz . + 2;
and SINDEX, ~DISCENB;
/*
* Send a tag message if TAG_ENB is set in the SCB control block.
* Use SCB_NONPACKET_TAG as the tag value.
*/
p_mesgout_tag:
test SCB_CONTROL,TAG_ENB jz p_mesgout_onebyte;
mov SCSIDAT, SINDEX; /* Send the identify message */
call phase_lock;
cmp LASTPHASE, P_MESGOUT jne p_mesgout_done;
and SCSIDAT,TAG_ENB|SCB_TAG_TYPE,SCB_CONTROL;
call phase_lock;
cmp LASTPHASE, P_MESGOUT jne p_mesgout_done;
mov SCBPTR jmp p_mesgout_onebyte;
/*
* Interrupt the driver, and allow it to handle this message
* phase and any required retries.
*/
p_mesgout_from_host:
cmp SINDEX, HOST_MSG jne p_mesgout_onebyte;
jmp host_message_loop;
p_mesgout_onebyte:
mvi CLRSINT1, CLRATNO;
mov SCSIDAT, SINDEX;
/*
* If the next bus phase after ATN drops is message out, it means
* that the target is requesting that the last message(s) be resent.
*/
call phase_lock;
cmp LASTPHASE, P_MESGOUT je p_mesgout_retry;
p_mesgout_done:
mvi CLRSINT1,CLRATNO; /* Be sure to turn ATNO off */
mov LAST_MSG, MSG_OUT;
mvi MSG_OUT, MSG_NOOP; /* No message left */
jmp ITloop;
/*
* Message in phase. Bytes are read using Automatic PIO mode.
*/
p_mesgin:
/* read the 1st message byte */
mvi ACCUM call inb_first;
test A,MSG_IDENTIFYFLAG jnz mesgin_identify;
cmp A,MSG_DISCONNECT je mesgin_disconnect;
cmp A,MSG_SAVEDATAPOINTER je mesgin_sdptrs;
cmp ALLZEROS,A je mesgin_complete;
cmp A,MSG_RESTOREPOINTERS je mesgin_rdptrs;
cmp A,MSG_IGN_WIDE_RESIDUE je mesgin_ign_wide_residue;
cmp A,MSG_NOOP je mesgin_done;
/*
* Pushed message loop to allow the kernel to
* run it's own message state engine. To avoid an
* extra nop instruction after signaling the kernel,
* we perform the phase_lock before checking to see
* if we should exit the loop and skip the phase_lock
* in the ITloop. Performing back to back phase_locks
* shouldn't hurt, but why do it twice...
*/
host_message_loop:
call phase_lock; /* Benign the first time through. */
SET_SEQINTCODE(HOST_MSG_LOOP)
cmp RETURN_1, EXIT_MSG_LOOP je ITloop;
cmp RETURN_1, CONT_MSG_LOOP_WRITE jne . + 3;
mov SCSIDAT, RETURN_2;
jmp host_message_loop;
/* Must be CONT_MSG_LOOP_READ */
mov NONE, SCSIDAT; /* ACK Byte */
jmp host_message_loop;
mesgin_ign_wide_residue:
aic79xx.c: Implement the SCB_SILENT flag. This is useful for hushing up the driver during DV or other operations that we expect to cause transmission errors. The messages will still print if the SHOW_MASKED_ERRORS debug option is enabled. Save and restore the NEGOADDR address when setting new transfer settings. The sequencer performs lookups in the negotiation table too and it expects NEGOADDR to remain consistent across pause/unpause sessions. Consistently use "offset" instead of "period" to determine if we are running sync or not. Add a SHOW_MESSAGES diagnostic for when we assert ATN during message processing. Print out IU, QAS, and RTI features when showing transfer options. Limit the syncrate after all option conformance changes have taken place in ahd_devlimited_syncrate. Changes in options may change the final syncrate we accept. Keep a copy of the hs_mailbox in our softc so that we can perform read/modify/write operations on the hs_mailbox without having to pause the sequencer to read the last written value. Use the ENINT_COALESS flag in the hs_mailbox to toggle interrupt coalessing. Add entrypoints for enabling interrupt coalessing and setting both a timeout (how long to wait for commands to be coalessed) and a maximum commands to coaless value. Add a statistics timer that decides when to enable or disable interrupt coalessing based on load. Add a routine, ahd_reset_cmds_pending() which is used to update the CMDS_PENDING sequencer variable whenever error recovery compeltes SCBs without notifying the sequencer. Since ahd_reset_cmds_pending is called during ahd_unpause() only if we've aborted SCBs, its call to ahd_flush_qoutfifo should not cause recursion through ahd_run_qoutfifo(). A panic has been added to ensure that this recursion does not occur. In ahd_search_qinfifo, update the CMDS_PENDING sequencer variable directly. ahd_search_qinififo can be called in situations where using ahd_reset_cmds_pending() might cause recursion. Since we can safely determine the exact number to reduce CMDS_PENDING by in this scenario without running the qoutfifo, the manual update is sufficient. Clean up diagnostics. Add ahd_flush_qoutfifo() which will run the qoutfifo as well as complete any commands sitting on the sequencer's COMPLETE_SCB lists or the good status FIFO. Use this routine in several places that did similar things in an add-hoc, but incomplete, fashion. A call to this routine was also added to ahd_abort_scbs() to close a race. In ahd_pause_and_flushwork() only return once selections are safely disabled. Flush all completed commands via ahd_flush_qoutfifo(). Remove "Now packetized" diagnostic now that this information is incorperated into the actual negotiation messages that are displayed. When forcing renegotiation, don't clober the current ppr_options. Much of the driver uses this information to determine if we are currently packetized or not. Remove some stray spaces at column 1 in ahd_set_tags. When complaining about getting a host message loop request with no pending messages, print out the SCB_CONTROL register down on the card. Modify the ahd_sent_msg() routine to handle a search for an outgoing identify message. Use this to detect a msg reject on an identify message which typically indicates that the target thought we were packetized. Force a renegotiation in this case. In ahd_search_qinfifo(), wait more effectively for SCB DMA activities to cease. We also disable SCB fetch operations since we are about to change the qinfifo and any fetch in progress will likely be invalidated. In ahd_qinfifo_count(), fix the qinfifo empty case. In ahd_dump_card_state(), print out CCSCBCTL in the correct mode. If we are a narrow controller, don't set the current width to unknown when forcing a future negotiation. This just confuses the code into attempting a wide negotiation on a narrow bus. Add support for task management function completions. Modify ahd_handle_devreset so that it can handle lun resets in addition to target resets. Use ahd_handle_devreset for lun and target reset task management functions. Handle the abort task TMF race case better. We now wait until any current selections are over and then set the TMF back to zero. This should cause the sequencer to ignore the abort TMF completion should it occur. Correct a bug in the illegal phase handler that caused us to drop down to narrow when handling the unexpected command phase case after 3rd party reset of a packetized device. Indicate the features, bugs, and flags set in the softc that are used to control firmware patch download when booting verbose. aic79xx.h: Add coalessing and HS_MAILBOX fields. Add per-softc variables for the stats "daemon". Add a debug option for interrupt coalessing activities. Add two new softc flags: o AHD_UPDATE_PEND_CMDS Run ahd_reset_cmds_pending() on the next unpause. o AHD_RUNNING_QOUTFIFO Used to catch recursion through ahd_run_qoutfifo(). aic79xx.reg: Correct register addresses related to the software timer and the DFDBCTL register. Add constants paramaterizing the software timer. Add scratch ram locations for storing interrupt coalessing tunables. Break INTMASK in SEQITNCTL out into INTMASK1 and INTMASK2. In at least the REV A, these are writable bits. We make use of that for a swtimer workaround in the sequencer. Since HS_MAILBOX autoclears, provide a sequencer variable to store its contents. Add SEQINT codes for handling task management completions. aic79xx.seq: Correct ignore wide residue processing check for a wide negotiation being in effect. We must be in the SCSI register window in order to access the negotiation table. Use the software timer and a commands completed count to implement interrupt coalessing. The command complete is deferred until either the maximum command threshold or a the expiration of a command deferral timer. If we have more SCBs to complete to the host (sitting in COMPLETE_SCB lists), always try to coaless them up to our coalessing limit. If coalessing is enabled, but we have fewer commands oustanting than the host's min coalessing limit, complete the command immediately. Add code to track the number of commands outstanding. Commands are outstanding from the time they are placed into the execution queue until the DMA to post completion is setup. Add a workaround for intvec_2 interrupts on the H2A4. In H2A4, the mode pointer is not saved for intvec2, but is restored on iret. This can lead to the restoration of a bogus mode ptr. Manually clear the intmask bits and do a normal return to compensate. We use intvec_2 to track interrupt coalessing timeouts. Since we cannot disable the swtimer's countdown, simply mask its interrupt once we no longer care about it firing. In idle_loop_cchan, update LOCAL_HS_MAILBOX everytime we are notified of an HS_MAILBOX update via the HS_MAILBOX_ACT bit in QOFF_CTLSTA. We have to use a local copy of persistant portions of the HS_MAILBOX as the mailbox auto-clears on any read. Move the test for the cfg4istat interrupt up an instruction to hopefully close a race between the next outgoing selection and our disabling of selections. Add a missing ret to the last instruction in load_overrun_buf. Add notifications to the host of task management completions as well as the completions for commands that completed successfully before their corresponding TMF could be sent. Hold a critical section during select-out processing until we have a fully identified connection. This removes a race condition with the legacy abort handler. Correct a few spelling errors in some comments. aic79xx_inline.h: Call ahd_reset_cmds_pending() in ahd_unpause if required. Update cmdcmplt interrupt statistics in our interrupt handler. Allow callers to ahd_send_scb() to set the task management function. aic79xx_pci.c: Disable SERR and pause the controller prior to performing our mmapped I/O test. The U320 controllers do not support "auto-access-pause". aic79xx_osm.c: Set the task management function now that ahd_send_scb() doesn't do it for us. We also perform a lun reset in response to BDR requests to packetized devices.
2003-01-20 20:17:35 +00:00
mov SAVED_MODE, MODE_PTR;
SET_MODE(M_SCSI, M_SCSI)
shr NEGOADDR, 4, SAVED_SCSIID;
aic79xx.c: Implement the SCB_SILENT flag. This is useful for hushing up the driver during DV or other operations that we expect to cause transmission errors. The messages will still print if the SHOW_MASKED_ERRORS debug option is enabled. Save and restore the NEGOADDR address when setting new transfer settings. The sequencer performs lookups in the negotiation table too and it expects NEGOADDR to remain consistent across pause/unpause sessions. Consistently use "offset" instead of "period" to determine if we are running sync or not. Add a SHOW_MESSAGES diagnostic for when we assert ATN during message processing. Print out IU, QAS, and RTI features when showing transfer options. Limit the syncrate after all option conformance changes have taken place in ahd_devlimited_syncrate. Changes in options may change the final syncrate we accept. Keep a copy of the hs_mailbox in our softc so that we can perform read/modify/write operations on the hs_mailbox without having to pause the sequencer to read the last written value. Use the ENINT_COALESS flag in the hs_mailbox to toggle interrupt coalessing. Add entrypoints for enabling interrupt coalessing and setting both a timeout (how long to wait for commands to be coalessed) and a maximum commands to coaless value. Add a statistics timer that decides when to enable or disable interrupt coalessing based on load. Add a routine, ahd_reset_cmds_pending() which is used to update the CMDS_PENDING sequencer variable whenever error recovery compeltes SCBs without notifying the sequencer. Since ahd_reset_cmds_pending is called during ahd_unpause() only if we've aborted SCBs, its call to ahd_flush_qoutfifo should not cause recursion through ahd_run_qoutfifo(). A panic has been added to ensure that this recursion does not occur. In ahd_search_qinfifo, update the CMDS_PENDING sequencer variable directly. ahd_search_qinififo can be called in situations where using ahd_reset_cmds_pending() might cause recursion. Since we can safely determine the exact number to reduce CMDS_PENDING by in this scenario without running the qoutfifo, the manual update is sufficient. Clean up diagnostics. Add ahd_flush_qoutfifo() which will run the qoutfifo as well as complete any commands sitting on the sequencer's COMPLETE_SCB lists or the good status FIFO. Use this routine in several places that did similar things in an add-hoc, but incomplete, fashion. A call to this routine was also added to ahd_abort_scbs() to close a race. In ahd_pause_and_flushwork() only return once selections are safely disabled. Flush all completed commands via ahd_flush_qoutfifo(). Remove "Now packetized" diagnostic now that this information is incorperated into the actual negotiation messages that are displayed. When forcing renegotiation, don't clober the current ppr_options. Much of the driver uses this information to determine if we are currently packetized or not. Remove some stray spaces at column 1 in ahd_set_tags. When complaining about getting a host message loop request with no pending messages, print out the SCB_CONTROL register down on the card. Modify the ahd_sent_msg() routine to handle a search for an outgoing identify message. Use this to detect a msg reject on an identify message which typically indicates that the target thought we were packetized. Force a renegotiation in this case. In ahd_search_qinfifo(), wait more effectively for SCB DMA activities to cease. We also disable SCB fetch operations since we are about to change the qinfifo and any fetch in progress will likely be invalidated. In ahd_qinfifo_count(), fix the qinfifo empty case. In ahd_dump_card_state(), print out CCSCBCTL in the correct mode. If we are a narrow controller, don't set the current width to unknown when forcing a future negotiation. This just confuses the code into attempting a wide negotiation on a narrow bus. Add support for task management function completions. Modify ahd_handle_devreset so that it can handle lun resets in addition to target resets. Use ahd_handle_devreset for lun and target reset task management functions. Handle the abort task TMF race case better. We now wait until any current selections are over and then set the TMF back to zero. This should cause the sequencer to ignore the abort TMF completion should it occur. Correct a bug in the illegal phase handler that caused us to drop down to narrow when handling the unexpected command phase case after 3rd party reset of a packetized device. Indicate the features, bugs, and flags set in the softc that are used to control firmware patch download when booting verbose. aic79xx.h: Add coalessing and HS_MAILBOX fields. Add per-softc variables for the stats "daemon". Add a debug option for interrupt coalessing activities. Add two new softc flags: o AHD_UPDATE_PEND_CMDS Run ahd_reset_cmds_pending() on the next unpause. o AHD_RUNNING_QOUTFIFO Used to catch recursion through ahd_run_qoutfifo(). aic79xx.reg: Correct register addresses related to the software timer and the DFDBCTL register. Add constants paramaterizing the software timer. Add scratch ram locations for storing interrupt coalessing tunables. Break INTMASK in SEQITNCTL out into INTMASK1 and INTMASK2. In at least the REV A, these are writable bits. We make use of that for a swtimer workaround in the sequencer. Since HS_MAILBOX autoclears, provide a sequencer variable to store its contents. Add SEQINT codes for handling task management completions. aic79xx.seq: Correct ignore wide residue processing check for a wide negotiation being in effect. We must be in the SCSI register window in order to access the negotiation table. Use the software timer and a commands completed count to implement interrupt coalessing. The command complete is deferred until either the maximum command threshold or a the expiration of a command deferral timer. If we have more SCBs to complete to the host (sitting in COMPLETE_SCB lists), always try to coaless them up to our coalessing limit. If coalessing is enabled, but we have fewer commands oustanting than the host's min coalessing limit, complete the command immediately. Add code to track the number of commands outstanding. Commands are outstanding from the time they are placed into the execution queue until the DMA to post completion is setup. Add a workaround for intvec_2 interrupts on the H2A4. In H2A4, the mode pointer is not saved for intvec2, but is restored on iret. This can lead to the restoration of a bogus mode ptr. Manually clear the intmask bits and do a normal return to compensate. We use intvec_2 to track interrupt coalessing timeouts. Since we cannot disable the swtimer's countdown, simply mask its interrupt once we no longer care about it firing. In idle_loop_cchan, update LOCAL_HS_MAILBOX everytime we are notified of an HS_MAILBOX update via the HS_MAILBOX_ACT bit in QOFF_CTLSTA. We have to use a local copy of persistant portions of the HS_MAILBOX as the mailbox auto-clears on any read. Move the test for the cfg4istat interrupt up an instruction to hopefully close a race between the next outgoing selection and our disabling of selections. Add a missing ret to the last instruction in load_overrun_buf. Add notifications to the host of task management completions as well as the completions for commands that completed successfully before their corresponding TMF could be sent. Hold a critical section during select-out processing until we have a fully identified connection. This removes a race condition with the legacy abort handler. Correct a few spelling errors in some comments. aic79xx_inline.h: Call ahd_reset_cmds_pending() in ahd_unpause if required. Update cmdcmplt interrupt statistics in our interrupt handler. Allow callers to ahd_send_scb() to set the task management function. aic79xx_pci.c: Disable SERR and pause the controller prior to performing our mmapped I/O test. The U320 controllers do not support "auto-access-pause". aic79xx_osm.c: Set the task management function now that ahd_send_scb() doesn't do it for us. We also perform a lun reset in response to BDR requests to packetized devices.
2003-01-20 20:17:35 +00:00
mov A, NEGCONOPTS;
RESTORE_MODE(SAVED_MODE)
test A, WIDEXFER jz mesgin_reject;
/* Pull the residue byte */
mvi REG0 call inb_next;
cmp REG0, 0x01 jne mesgin_reject;
test SCB_RESIDUAL_SGPTR[0], SG_LIST_NULL jz . + 2;
test SCB_TASK_ATTRIBUTE, SCB_XFERLEN_ODD jnz mesgin_done;
SET_SEQINTCODE(IGN_WIDE_RES)
jmp mesgin_done;
mesgin_proto_violation:
SET_SEQINTCODE(PROTO_VIOLATION)
jmp mesgin_done;
mesgin_reject:
mvi MSG_MESSAGE_REJECT call mk_mesg;
mesgin_done:
mov NONE,SCSIDAT; /*dummy read from latch to ACK*/
jmp ITloop;
Upgrade to version 1.1 of the aic79xx U320 driver. aic79xx.c: o Remove redundant ahd_update_modes() call. o Correct panic in diagnostic should state corruption cause the SCB Id to be invalid during a selection timeout. o Add workaround for missing BUSFREEREV feature in Rev A silicon. o Corect formatting nits. o Use register pretty printing in more places. o Save and restore our SCB pointer when updating the waiting queue list for an "expected" LQ-out busfree. o In ahd_clear_intstat, deal with the missing autoclear in the CLRLQO* registers. o BE fixup in a diagnostic printf. o Make sure that we are in the proper mode before disabling selections in ahd_update_pending_scbs. o Add more diagnostics. o task_attribute_nonpkt_tag -> task_attribute: we don't need a nonpkt_tag field anymore for allowing all 512 SCBs to be used in non-packetized connections. o Negotiate HOLD_MCS to U320 devices. o Add a few additional mode assertions. o Restore the chip mode after clearing out the qinfifo so that code using ahd_abort_scbs sees a consistent mode. o Simplify the DMA engine shutdown routine prior to performing a bus reset. o Perform the sequencer restart after a chip reset prior to setting up our timer to poll for the reset to be complete. On some OSes, the timer could actually pre-empt us and order is important here. o Have our "reset poller" set the expected mode since there is no guarantee of what mode will be in force when we are called from the OS timer. o Save and restore the SCB pointer in ahd_dump_card_state(). This routine must not modify card state. o Ditto for ahd_dump_scbs(). aic79xx.h: o Add a few more chip bug definitions. o Align our tag on a 32bit boundary. aic79xx.reg: aic79xx.seq: o Start work on removing workarounds for Rev B. o Use a special location in scratch from for stroring our SCBPTR during legacy FIFO allocations. This corrects problems in mixed packetized/non-packetized configurations where calling into a FIFO task corrupted our SCBPTR. o Don't rely on DMA priority to guarantee that all data in our FIFOs will flush prior to a command completion notification going out of the command channel. We've never seen this assumption fail, but better safe than sorry. o Deal with missing BUSFREEREV feature in H2A. o Simplify disconnect list code now that the list will always have only a single entry. o Implement the AHD_REG_SLOW_SETTLE_BUG workaround. o Swith to using "REG_ISR" for local mode scratch during our ISR. o Add a missing jmp to the data_group_dma_loop after our data pointers have been re-initialized by the kernel. o Correct test in the bitbucket code so that we actually wait for the bitbucket to complete before signaling the kernel of the overrun condition. o Reposition pkt_saveptrs to avoid a jmp instruction. o Update a comment to reflect that the code now waits for a FIFO to drain prior to issuing a CLRCHN. aic79xx_inline.h: o Remove unused untagged queue handling code. o Don't attempt to htole64 what could be a 32bit value. aic79xx_pci.c: o Set additional bug flags for rev A chips.
2002-09-26 22:54:00 +00:00
#define INDEX_DISC_LIST(scsiid, lun) \
and A, 0xC0, scsiid; \
or SCBPTR, A, lun; \
Upgrade to version 1.1 of the aic79xx U320 driver. aic79xx.c: o Remove redundant ahd_update_modes() call. o Correct panic in diagnostic should state corruption cause the SCB Id to be invalid during a selection timeout. o Add workaround for missing BUSFREEREV feature in Rev A silicon. o Corect formatting nits. o Use register pretty printing in more places. o Save and restore our SCB pointer when updating the waiting queue list for an "expected" LQ-out busfree. o In ahd_clear_intstat, deal with the missing autoclear in the CLRLQO* registers. o BE fixup in a diagnostic printf. o Make sure that we are in the proper mode before disabling selections in ahd_update_pending_scbs. o Add more diagnostics. o task_attribute_nonpkt_tag -> task_attribute: we don't need a nonpkt_tag field anymore for allowing all 512 SCBs to be used in non-packetized connections. o Negotiate HOLD_MCS to U320 devices. o Add a few additional mode assertions. o Restore the chip mode after clearing out the qinfifo so that code using ahd_abort_scbs sees a consistent mode. o Simplify the DMA engine shutdown routine prior to performing a bus reset. o Perform the sequencer restart after a chip reset prior to setting up our timer to poll for the reset to be complete. On some OSes, the timer could actually pre-empt us and order is important here. o Have our "reset poller" set the expected mode since there is no guarantee of what mode will be in force when we are called from the OS timer. o Save and restore the SCB pointer in ahd_dump_card_state(). This routine must not modify card state. o Ditto for ahd_dump_scbs(). aic79xx.h: o Add a few more chip bug definitions. o Align our tag on a 32bit boundary. aic79xx.reg: aic79xx.seq: o Start work on removing workarounds for Rev B. o Use a special location in scratch from for stroring our SCBPTR during legacy FIFO allocations. This corrects problems in mixed packetized/non-packetized configurations where calling into a FIFO task corrupted our SCBPTR. o Don't rely on DMA priority to guarantee that all data in our FIFOs will flush prior to a command completion notification going out of the command channel. We've never seen this assumption fail, but better safe than sorry. o Deal with missing BUSFREEREV feature in H2A. o Simplify disconnect list code now that the list will always have only a single entry. o Implement the AHD_REG_SLOW_SETTLE_BUG workaround. o Swith to using "REG_ISR" for local mode scratch during our ISR. o Add a missing jmp to the data_group_dma_loop after our data pointers have been re-initialized by the kernel. o Correct test in the bitbucket code so that we actually wait for the bitbucket to complete before signaling the kernel of the overrun condition. o Reposition pkt_saveptrs to avoid a jmp instruction. o Update a comment to reflect that the code now waits for a FIFO to drain prior to issuing a CLRCHN. aic79xx_inline.h: o Remove unused untagged queue handling code. o Don't attempt to htole64 what could be a 32bit value. aic79xx_pci.c: o Set additional bug flags for rev A chips.
2002-09-26 22:54:00 +00:00
clr SCBPTR[1]; \
and SINDEX, 0x30, scsiid; \
shr SINDEX, 3; /* Multiply by 2 */ \
add SINDEX, (SCB_DISCONNECTED_LISTS & 0xFF); \
mvi SINDEX[1], ((SCB_DISCONNECTED_LISTS >> 8) & 0xFF)
mesgin_identify:
/*
* Determine whether a target is using tagged or non-tagged
* transactions by first looking at the transaction stored in
* the per-device, disconnected array. If there is no untagged
* transaction for this target, this must be a tagged transaction.
*/
and SAVED_LUN, MSG_IDENTIFY_LUNMASK, A;
INDEX_DISC_LIST(SAVED_SCSIID, SAVED_LUN);
bmov DINDEX, SINDEX, 2;
bmov REG0, SINDIR, 2;
cmp REG0[1], SCB_LIST_NULL je snoop_tag;
/* Untagged. Clear the busy table entry and setup the SCB. */
bmov DINDIR, ALLONES, 2;
bmov SCBPTR, REG0, 2;
jmp setup_SCB;
/*
* Here we "snoop" the bus looking for a SIMPLE QUEUE TAG message.
* If we get one, we use the tag returned to find the proper
* SCB. After receiving the tag, look for the SCB at SCB locations tag and
* tag + 256.
*/
snoop_tag:
if ((ahd->flags & AHD_SEQUENCER_DEBUG) != 0) {
or SEQ_FLAGS, 0x80;
}
mov NONE, SCSIDAT; /* ACK Identify MSG */
call phase_lock;
if ((ahd->flags & AHD_SEQUENCER_DEBUG) != 0) {
or SEQ_FLAGS, 0x1;
}
cmp LASTPHASE, P_MESGIN jne not_found_ITloop;
if ((ahd->flags & AHD_SEQUENCER_DEBUG) != 0) {
or SEQ_FLAGS, 0x2;
}
cmp SCSIBUS, MSG_SIMPLE_Q_TAG jne not_found;
get_tag:
clr SCBPTR[1];
mvi SCBPTR call inb_next; /* tag value */
verify_scb:
test SCB_CONTROL,DISCONNECTED jz verify_other_scb;
mov A, SAVED_SCSIID;
cmp SCB_SCSIID, A jne verify_other_scb;
mov A, SAVED_LUN;
cmp SCB_LUN, A je setup_SCB_disconnected;
verify_other_scb:
xor SCBPTR[1], 1;
test SCBPTR[1], 0xFF jnz verify_scb;
jmp not_found;
/*
* Ensure that the SCB the tag points to is for
* an SCB transaction to the reconnecting target.
*/
setup_SCB:
if ((ahd->flags & AHD_SEQUENCER_DEBUG) != 0) {
or SEQ_FLAGS, 0x10;
}
test SCB_CONTROL,DISCONNECTED jz not_found;
setup_SCB_disconnected:
and SCB_CONTROL,~DISCONNECTED;
clr SEQ_FLAGS; /* make note of IDENTIFY */
Upgrade to version 1.1 of the aic79xx U320 driver. aic79xx.c: o Remove redundant ahd_update_modes() call. o Correct panic in diagnostic should state corruption cause the SCB Id to be invalid during a selection timeout. o Add workaround for missing BUSFREEREV feature in Rev A silicon. o Corect formatting nits. o Use register pretty printing in more places. o Save and restore our SCB pointer when updating the waiting queue list for an "expected" LQ-out busfree. o In ahd_clear_intstat, deal with the missing autoclear in the CLRLQO* registers. o BE fixup in a diagnostic printf. o Make sure that we are in the proper mode before disabling selections in ahd_update_pending_scbs. o Add more diagnostics. o task_attribute_nonpkt_tag -> task_attribute: we don't need a nonpkt_tag field anymore for allowing all 512 SCBs to be used in non-packetized connections. o Negotiate HOLD_MCS to U320 devices. o Add a few additional mode assertions. o Restore the chip mode after clearing out the qinfifo so that code using ahd_abort_scbs sees a consistent mode. o Simplify the DMA engine shutdown routine prior to performing a bus reset. o Perform the sequencer restart after a chip reset prior to setting up our timer to poll for the reset to be complete. On some OSes, the timer could actually pre-empt us and order is important here. o Have our "reset poller" set the expected mode since there is no guarantee of what mode will be in force when we are called from the OS timer. o Save and restore the SCB pointer in ahd_dump_card_state(). This routine must not modify card state. o Ditto for ahd_dump_scbs(). aic79xx.h: o Add a few more chip bug definitions. o Align our tag on a 32bit boundary. aic79xx.reg: aic79xx.seq: o Start work on removing workarounds for Rev B. o Use a special location in scratch from for stroring our SCBPTR during legacy FIFO allocations. This corrects problems in mixed packetized/non-packetized configurations where calling into a FIFO task corrupted our SCBPTR. o Don't rely on DMA priority to guarantee that all data in our FIFOs will flush prior to a command completion notification going out of the command channel. We've never seen this assumption fail, but better safe than sorry. o Deal with missing BUSFREEREV feature in H2A. o Simplify disconnect list code now that the list will always have only a single entry. o Implement the AHD_REG_SLOW_SETTLE_BUG workaround. o Swith to using "REG_ISR" for local mode scratch during our ISR. o Add a missing jmp to the data_group_dma_loop after our data pointers have been re-initialized by the kernel. o Correct test in the bitbucket code so that we actually wait for the bitbucket to complete before signaling the kernel of the overrun condition. o Reposition pkt_saveptrs to avoid a jmp instruction. o Update a comment to reflect that the code now waits for a FIFO to drain prior to issuing a CLRCHN. aic79xx_inline.h: o Remove unused untagged queue handling code. o Don't attempt to htole64 what could be a 32bit value. aic79xx_pci.c: o Set additional bug flags for rev A chips.
2002-09-26 22:54:00 +00:00
test SCB_SGPTR, SG_LIST_NULL jnz . + 3;
bmov ALLOCFIFO_SCBPTR, SCBPTR, 2;
call allocate_fifo;
/* See if the host wants to send a message upon reconnection */
test SCB_CONTROL, MK_MESSAGE jz mesgin_done;
mvi HOST_MSG call mk_mesg;
jmp mesgin_done;
not_found:
SET_SEQINTCODE(NO_MATCH)
jmp mesgin_done;
not_found_ITloop:
SET_SEQINTCODE(NO_MATCH)
jmp ITloop;
/*
* We received a "command complete" message. Put the SCB on the complete
* queue and trigger a completion interrupt via the idle loop. Before doing
aic79xx.c: aic79xx.seq: Convert the COMPLETE_DMA_SCB list to an "stailq". This allows us to safely keep the SCB that is currently being DMA'ed back the host on the head of the list while processing completions off of the bus. The newly completed SCBs are appended to the tail of the queue. In the past, we just dequeued the SCB that was in flight from the list, but this could result in a lost completion should the host perform certain types of error recovery that must cancel all in-flight SCB DMA operations. Switch from using a 16bit completion entry, holding just the tag and the completion valid bit, to a 64bit completion entry that also contains a "status packet valid" indicator. This solves two problems: o The SCB DMA engine on at least Rev B. silicon does not properly deal with a PCI disconnect that occurs at a non-64bit aligned offset in the chips "source buffer". When the transfer is resumed, the DMA engine continues at the correct offset, but may wrap to the head of the buffer causing duplicate completions to be reported to the host. By using a completion buffer in host memory that is 64bit aligned and using 64bit completion entries, such disconnects should only occur at aligned addresses. This assumes that the host bridge will only disconnect on cache-line boundaries and that cache-lines are multpiles of 64bits. o By embedding the status information in the completion entry we can avoid an extra memory reference to the HSCB for commands that complete without error. Use the comparison of a "host freeze count" and a "sequencer freeze count" to allow the host to process most SCBs that complete with non-zero status without having to clear critical sections. Instead the host can just pause the sequencer, performs any necessary cleanup in the waiting for selection list, increments its freeze count on the controller, and unpauses. This is only possible because the sequencer defers completions of SCBs with bad status until after all pending selections have completed. The sequencer then avoids referencing any data structures the host may touch during completion of the SCB until the freeze counts match. aic79xx.c: Change the strategy for allocating our sentinal HSCB for the QINFIFO. In the past, this allocation was tacked onto the QOUTFIFO allocation. Now that the qoutfifo has grown to accomodate larger completion entries, the old approach will result in a 64byte allocation that costs an extra page of coherent memory. We now do this extra allocation via ahd_alloc_scbs() where the "unused space" can be used to allocate "normal" HSCBs. In our packetized busfree handler, use the ENSELO bit to differentiate between packetized and non-packetized unexpected busfree events that occur just after selection, but before the sequencer has had the oportunity to service the selection. When cleaning out the waiting for selection list, use the SCSI mode instead of the command channel mode. The SCB pointer in the command channel mode may be referenced by the SCB dma engine even while the sequencer is paused, whereas the SCSI mode SCB pointer is only accessed by the sequencer. Print the "complete on qfreeze" sequencer SCB completion list in ahd_dump_card_state(). This list holds all SCB completions that are deferred until a pending select-out qfreeze event has taken effect. aic79xx.h: Add definitions and structures to handle the new SCB completion scheme. Add a controller flag that indicates if the controller is in HostRAID mode. aic79xx.reg: Remove macros used for toggling from one data fifo mode to the other. They have not been in use for some time. Add scratch ram fields for our new qfreeze count scheme, converting the complete dma list into an "stailq", and providing for the "complete on qfreeze" SCB completion list. Some other fields were moved to retain proper field alignment (alignment >= field size in bytes). aic79xx.seq: Add code to our idle loop to: o Process deferred completions once a qfreeze event has taken full effect. o Thaw the queue once the sequencer and host qfreeze counts match. Generate 64bit completion entries passing the SCB_SGPTR field as the "good status" indicator. The first bit in this field is only set if we have a valid status packet to send to the host. Convert the COMPLETE_DMA_SCB list to an "stailq". When using "setjmp" to register an idle loop handler, do not combine the "ret" with the block move to pop the stack address in the same instruction. At least on the A, this results in a return to the setjmp caller, not to the new address at the top of the stack. Since we want the latter (we want the newly registered handler to only be invoked from the idle loop), we must use a separate ret instruction. Add a few missing critical sections. Close a race condition that can occur on Rev A. silicon. If both FIFOs happen to be allocated before the sequencer has a chance to service the FIFO that was allocated first, we must take special care to service the FIFO that is not active on the SCSI bus first. This guarantees that a FIFO will be freed to handle any snapshot requests for the FIFO that is still on the bus. Chosing the incorrect FIFO will result in deadlock. Update comments. aic79xx_inline.h Correct the offset calculation for the syncing of our qoutfifo. Update ahd_check_cmdcmpltqueues() for the larger completion entries. aic79xx_pci.c: Attach to HostRAID controllers by default. In the future I may add a sysctl to modify the behavior, but since FreeBSD does not have any HostRAID drivers, failing to attach just results in more email and bug reports for the author. MFC After: 1week
2004-02-04 16:38:38 +00:00
* so, check to see if there is a residual or the status byte is something
* other than STATUS_GOOD (0). In either of these conditions, we upload the
* SCB back to the host so it can process this information.
*/
mesgin_complete:
/*
* If ATN is raised, we still want to give the target a message.
* Perhaps there was a parity error on this last message byte.
* Either way, the target should take us to message out phase
* and then attempt to complete the command again. We should use a
* critical section here to guard against a timeout triggering
* for this command and setting ATN while we are still processing
* the completion.
test SCSISIGI, ATNI jnz mesgin_done;
*/
/*
* If we are identified and have successfully sent the CDB,
* any status will do. Optimize this fast path.
*/
test SCB_CONTROL, STATUS_RCVD jz mesgin_proto_violation;
test SEQ_FLAGS, NOT_IDENTIFIED|NO_CDB_SENT jz complete_accepted;
/*
* If the target never sent an identify message but instead went
* to mesgin to give an invalid message, let the host abort us.
*/
test SEQ_FLAGS, NOT_IDENTIFIED jnz mesgin_proto_violation;
/*
* If we recevied good status but never successfully sent the
* cdb, abort the command.
*/
test SCB_SCSI_STATUS,0xff jnz complete_accepted;
test SEQ_FLAGS, NO_CDB_SENT jnz mesgin_proto_violation;
complete_accepted:
/*
* See if we attempted to deliver a message but the target ingnored us.
*/
test SCB_CONTROL, MK_MESSAGE jz complete_nomsg;
SET_SEQINTCODE(MKMSG_FAILED)
complete_nomsg:
call queue_scb_completion;
jmp await_busfree;
aic79xx.c: aic79xx.seq: Convert the COMPLETE_DMA_SCB list to an "stailq". This allows us to safely keep the SCB that is currently being DMA'ed back the host on the head of the list while processing completions off of the bus. The newly completed SCBs are appended to the tail of the queue. In the past, we just dequeued the SCB that was in flight from the list, but this could result in a lost completion should the host perform certain types of error recovery that must cancel all in-flight SCB DMA operations. Switch from using a 16bit completion entry, holding just the tag and the completion valid bit, to a 64bit completion entry that also contains a "status packet valid" indicator. This solves two problems: o The SCB DMA engine on at least Rev B. silicon does not properly deal with a PCI disconnect that occurs at a non-64bit aligned offset in the chips "source buffer". When the transfer is resumed, the DMA engine continues at the correct offset, but may wrap to the head of the buffer causing duplicate completions to be reported to the host. By using a completion buffer in host memory that is 64bit aligned and using 64bit completion entries, such disconnects should only occur at aligned addresses. This assumes that the host bridge will only disconnect on cache-line boundaries and that cache-lines are multpiles of 64bits. o By embedding the status information in the completion entry we can avoid an extra memory reference to the HSCB for commands that complete without error. Use the comparison of a "host freeze count" and a "sequencer freeze count" to allow the host to process most SCBs that complete with non-zero status without having to clear critical sections. Instead the host can just pause the sequencer, performs any necessary cleanup in the waiting for selection list, increments its freeze count on the controller, and unpauses. This is only possible because the sequencer defers completions of SCBs with bad status until after all pending selections have completed. The sequencer then avoids referencing any data structures the host may touch during completion of the SCB until the freeze counts match. aic79xx.c: Change the strategy for allocating our sentinal HSCB for the QINFIFO. In the past, this allocation was tacked onto the QOUTFIFO allocation. Now that the qoutfifo has grown to accomodate larger completion entries, the old approach will result in a 64byte allocation that costs an extra page of coherent memory. We now do this extra allocation via ahd_alloc_scbs() where the "unused space" can be used to allocate "normal" HSCBs. In our packetized busfree handler, use the ENSELO bit to differentiate between packetized and non-packetized unexpected busfree events that occur just after selection, but before the sequencer has had the oportunity to service the selection. When cleaning out the waiting for selection list, use the SCSI mode instead of the command channel mode. The SCB pointer in the command channel mode may be referenced by the SCB dma engine even while the sequencer is paused, whereas the SCSI mode SCB pointer is only accessed by the sequencer. Print the "complete on qfreeze" sequencer SCB completion list in ahd_dump_card_state(). This list holds all SCB completions that are deferred until a pending select-out qfreeze event has taken effect. aic79xx.h: Add definitions and structures to handle the new SCB completion scheme. Add a controller flag that indicates if the controller is in HostRAID mode. aic79xx.reg: Remove macros used for toggling from one data fifo mode to the other. They have not been in use for some time. Add scratch ram fields for our new qfreeze count scheme, converting the complete dma list into an "stailq", and providing for the "complete on qfreeze" SCB completion list. Some other fields were moved to retain proper field alignment (alignment >= field size in bytes). aic79xx.seq: Add code to our idle loop to: o Process deferred completions once a qfreeze event has taken full effect. o Thaw the queue once the sequencer and host qfreeze counts match. Generate 64bit completion entries passing the SCB_SGPTR field as the "good status" indicator. The first bit in this field is only set if we have a valid status packet to send to the host. Convert the COMPLETE_DMA_SCB list to an "stailq". When using "setjmp" to register an idle loop handler, do not combine the "ret" with the block move to pop the stack address in the same instruction. At least on the A, this results in a return to the setjmp caller, not to the new address at the top of the stack. Since we want the latter (we want the newly registered handler to only be invoked from the idle loop), we must use a separate ret instruction. Add a few missing critical sections. Close a race condition that can occur on Rev A. silicon. If both FIFOs happen to be allocated before the sequencer has a chance to service the FIFO that was allocated first, we must take special care to service the FIFO that is not active on the SCSI bus first. This guarantees that a FIFO will be freed to handle any snapshot requests for the FIFO that is still on the bus. Chosing the incorrect FIFO will result in deadlock. Update comments. aic79xx_inline.h Correct the offset calculation for the syncing of our qoutfifo. Update ahd_check_cmdcmpltqueues() for the larger completion entries. aic79xx_pci.c: Attach to HostRAID controllers by default. In the future I may add a sysctl to modify the behavior, but since FreeBSD does not have any HostRAID drivers, failing to attach just results in more email and bug reports for the author. MFC After: 1week
2004-02-04 16:38:38 +00:00
BEGIN_CRITICAL;
freeze_queue:
/* Cancel any pending select-out. */
aic79xx.c: Implement the SCB_SILENT flag. This is useful for hushing up the driver during DV or other operations that we expect to cause transmission errors. The messages will still print if the SHOW_MASKED_ERRORS debug option is enabled. Save and restore the NEGOADDR address when setting new transfer settings. The sequencer performs lookups in the negotiation table too and it expects NEGOADDR to remain consistent across pause/unpause sessions. Consistently use "offset" instead of "period" to determine if we are running sync or not. Add a SHOW_MESSAGES diagnostic for when we assert ATN during message processing. Print out IU, QAS, and RTI features when showing transfer options. Limit the syncrate after all option conformance changes have taken place in ahd_devlimited_syncrate. Changes in options may change the final syncrate we accept. Keep a copy of the hs_mailbox in our softc so that we can perform read/modify/write operations on the hs_mailbox without having to pause the sequencer to read the last written value. Use the ENINT_COALESS flag in the hs_mailbox to toggle interrupt coalessing. Add entrypoints for enabling interrupt coalessing and setting both a timeout (how long to wait for commands to be coalessed) and a maximum commands to coaless value. Add a statistics timer that decides when to enable or disable interrupt coalessing based on load. Add a routine, ahd_reset_cmds_pending() which is used to update the CMDS_PENDING sequencer variable whenever error recovery compeltes SCBs without notifying the sequencer. Since ahd_reset_cmds_pending is called during ahd_unpause() only if we've aborted SCBs, its call to ahd_flush_qoutfifo should not cause recursion through ahd_run_qoutfifo(). A panic has been added to ensure that this recursion does not occur. In ahd_search_qinfifo, update the CMDS_PENDING sequencer variable directly. ahd_search_qinififo can be called in situations where using ahd_reset_cmds_pending() might cause recursion. Since we can safely determine the exact number to reduce CMDS_PENDING by in this scenario without running the qoutfifo, the manual update is sufficient. Clean up diagnostics. Add ahd_flush_qoutfifo() which will run the qoutfifo as well as complete any commands sitting on the sequencer's COMPLETE_SCB lists or the good status FIFO. Use this routine in several places that did similar things in an add-hoc, but incomplete, fashion. A call to this routine was also added to ahd_abort_scbs() to close a race. In ahd_pause_and_flushwork() only return once selections are safely disabled. Flush all completed commands via ahd_flush_qoutfifo(). Remove "Now packetized" diagnostic now that this information is incorperated into the actual negotiation messages that are displayed. When forcing renegotiation, don't clober the current ppr_options. Much of the driver uses this information to determine if we are currently packetized or not. Remove some stray spaces at column 1 in ahd_set_tags. When complaining about getting a host message loop request with no pending messages, print out the SCB_CONTROL register down on the card. Modify the ahd_sent_msg() routine to handle a search for an outgoing identify message. Use this to detect a msg reject on an identify message which typically indicates that the target thought we were packetized. Force a renegotiation in this case. In ahd_search_qinfifo(), wait more effectively for SCB DMA activities to cease. We also disable SCB fetch operations since we are about to change the qinfifo and any fetch in progress will likely be invalidated. In ahd_qinfifo_count(), fix the qinfifo empty case. In ahd_dump_card_state(), print out CCSCBCTL in the correct mode. If we are a narrow controller, don't set the current width to unknown when forcing a future negotiation. This just confuses the code into attempting a wide negotiation on a narrow bus. Add support for task management function completions. Modify ahd_handle_devreset so that it can handle lun resets in addition to target resets. Use ahd_handle_devreset for lun and target reset task management functions. Handle the abort task TMF race case better. We now wait until any current selections are over and then set the TMF back to zero. This should cause the sequencer to ignore the abort TMF completion should it occur. Correct a bug in the illegal phase handler that caused us to drop down to narrow when handling the unexpected command phase case after 3rd party reset of a packetized device. Indicate the features, bugs, and flags set in the softc that are used to control firmware patch download when booting verbose. aic79xx.h: Add coalessing and HS_MAILBOX fields. Add per-softc variables for the stats "daemon". Add a debug option for interrupt coalessing activities. Add two new softc flags: o AHD_UPDATE_PEND_CMDS Run ahd_reset_cmds_pending() on the next unpause. o AHD_RUNNING_QOUTFIFO Used to catch recursion through ahd_run_qoutfifo(). aic79xx.reg: Correct register addresses related to the software timer and the DFDBCTL register. Add constants paramaterizing the software timer. Add scratch ram locations for storing interrupt coalessing tunables. Break INTMASK in SEQITNCTL out into INTMASK1 and INTMASK2. In at least the REV A, these are writable bits. We make use of that for a swtimer workaround in the sequencer. Since HS_MAILBOX autoclears, provide a sequencer variable to store its contents. Add SEQINT codes for handling task management completions. aic79xx.seq: Correct ignore wide residue processing check for a wide negotiation being in effect. We must be in the SCSI register window in order to access the negotiation table. Use the software timer and a commands completed count to implement interrupt coalessing. The command complete is deferred until either the maximum command threshold or a the expiration of a command deferral timer. If we have more SCBs to complete to the host (sitting in COMPLETE_SCB lists), always try to coaless them up to our coalessing limit. If coalessing is enabled, but we have fewer commands oustanting than the host's min coalessing limit, complete the command immediately. Add code to track the number of commands outstanding. Commands are outstanding from the time they are placed into the execution queue until the DMA to post completion is setup. Add a workaround for intvec_2 interrupts on the H2A4. In H2A4, the mode pointer is not saved for intvec2, but is restored on iret. This can lead to the restoration of a bogus mode ptr. Manually clear the intmask bits and do a normal return to compensate. We use intvec_2 to track interrupt coalessing timeouts. Since we cannot disable the swtimer's countdown, simply mask its interrupt once we no longer care about it firing. In idle_loop_cchan, update LOCAL_HS_MAILBOX everytime we are notified of an HS_MAILBOX update via the HS_MAILBOX_ACT bit in QOFF_CTLSTA. We have to use a local copy of persistant portions of the HS_MAILBOX as the mailbox auto-clears on any read. Move the test for the cfg4istat interrupt up an instruction to hopefully close a race between the next outgoing selection and our disabling of selections. Add a missing ret to the last instruction in load_overrun_buf. Add notifications to the host of task management completions as well as the completions for commands that completed successfully before their corresponding TMF could be sent. Hold a critical section during select-out processing until we have a fully identified connection. This removes a race condition with the legacy abort handler. Correct a few spelling errors in some comments. aic79xx_inline.h: Call ahd_reset_cmds_pending() in ahd_unpause if required. Update cmdcmplt interrupt statistics in our interrupt handler. Allow callers to ahd_send_scb() to set the task management function. aic79xx_pci.c: Disable SERR and pause the controller prior to performing our mmapped I/O test. The U320 controllers do not support "auto-access-pause". aic79xx_osm.c: Set the task management function now that ahd_send_scb() doesn't do it for us. We also perform a lun reset in response to BDR requests to packetized devices.
2003-01-20 20:17:35 +00:00
test SSTAT0, SELDO|SELINGO jnz . + 2;
and SCSISEQ0, ~ENSELO;
mov ACCUM_SAVE, A;
clr A;
add QFREEZE_COUNT, 1;
adc QFREEZE_COUNT[1], A;
or SEQ_FLAGS2, SELECTOUT_QFROZEN;
mov A, ACCUM_SAVE ret;
aic79xx.c: aic79xx.seq: Convert the COMPLETE_DMA_SCB list to an "stailq". This allows us to safely keep the SCB that is currently being DMA'ed back the host on the head of the list while processing completions off of the bus. The newly completed SCBs are appended to the tail of the queue. In the past, we just dequeued the SCB that was in flight from the list, but this could result in a lost completion should the host perform certain types of error recovery that must cancel all in-flight SCB DMA operations. Switch from using a 16bit completion entry, holding just the tag and the completion valid bit, to a 64bit completion entry that also contains a "status packet valid" indicator. This solves two problems: o The SCB DMA engine on at least Rev B. silicon does not properly deal with a PCI disconnect that occurs at a non-64bit aligned offset in the chips "source buffer". When the transfer is resumed, the DMA engine continues at the correct offset, but may wrap to the head of the buffer causing duplicate completions to be reported to the host. By using a completion buffer in host memory that is 64bit aligned and using 64bit completion entries, such disconnects should only occur at aligned addresses. This assumes that the host bridge will only disconnect on cache-line boundaries and that cache-lines are multpiles of 64bits. o By embedding the status information in the completion entry we can avoid an extra memory reference to the HSCB for commands that complete without error. Use the comparison of a "host freeze count" and a "sequencer freeze count" to allow the host to process most SCBs that complete with non-zero status without having to clear critical sections. Instead the host can just pause the sequencer, performs any necessary cleanup in the waiting for selection list, increments its freeze count on the controller, and unpauses. This is only possible because the sequencer defers completions of SCBs with bad status until after all pending selections have completed. The sequencer then avoids referencing any data structures the host may touch during completion of the SCB until the freeze counts match. aic79xx.c: Change the strategy for allocating our sentinal HSCB for the QINFIFO. In the past, this allocation was tacked onto the QOUTFIFO allocation. Now that the qoutfifo has grown to accomodate larger completion entries, the old approach will result in a 64byte allocation that costs an extra page of coherent memory. We now do this extra allocation via ahd_alloc_scbs() where the "unused space" can be used to allocate "normal" HSCBs. In our packetized busfree handler, use the ENSELO bit to differentiate between packetized and non-packetized unexpected busfree events that occur just after selection, but before the sequencer has had the oportunity to service the selection. When cleaning out the waiting for selection list, use the SCSI mode instead of the command channel mode. The SCB pointer in the command channel mode may be referenced by the SCB dma engine even while the sequencer is paused, whereas the SCSI mode SCB pointer is only accessed by the sequencer. Print the "complete on qfreeze" sequencer SCB completion list in ahd_dump_card_state(). This list holds all SCB completions that are deferred until a pending select-out qfreeze event has taken effect. aic79xx.h: Add definitions and structures to handle the new SCB completion scheme. Add a controller flag that indicates if the controller is in HostRAID mode. aic79xx.reg: Remove macros used for toggling from one data fifo mode to the other. They have not been in use for some time. Add scratch ram fields for our new qfreeze count scheme, converting the complete dma list into an "stailq", and providing for the "complete on qfreeze" SCB completion list. Some other fields were moved to retain proper field alignment (alignment >= field size in bytes). aic79xx.seq: Add code to our idle loop to: o Process deferred completions once a qfreeze event has taken full effect. o Thaw the queue once the sequencer and host qfreeze counts match. Generate 64bit completion entries passing the SCB_SGPTR field as the "good status" indicator. The first bit in this field is only set if we have a valid status packet to send to the host. Convert the COMPLETE_DMA_SCB list to an "stailq". When using "setjmp" to register an idle loop handler, do not combine the "ret" with the block move to pop the stack address in the same instruction. At least on the A, this results in a return to the setjmp caller, not to the new address at the top of the stack. Since we want the latter (we want the newly registered handler to only be invoked from the idle loop), we must use a separate ret instruction. Add a few missing critical sections. Close a race condition that can occur on Rev A. silicon. If both FIFOs happen to be allocated before the sequencer has a chance to service the FIFO that was allocated first, we must take special care to service the FIFO that is not active on the SCSI bus first. This guarantees that a FIFO will be freed to handle any snapshot requests for the FIFO that is still on the bus. Chosing the incorrect FIFO will result in deadlock. Update comments. aic79xx_inline.h Correct the offset calculation for the syncing of our qoutfifo. Update ahd_check_cmdcmpltqueues() for the larger completion entries. aic79xx_pci.c: Attach to HostRAID controllers by default. In the future I may add a sysctl to modify the behavior, but since FreeBSD does not have any HostRAID drivers, failing to attach just results in more email and bug reports for the author. MFC After: 1week
2004-02-04 16:38:38 +00:00
END_CRITICAL;
Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-05-04 00:20:07 +00:00
/*
* Complete the current FIFO's SCB if data for this same
* SCB is not transferring in the other FIFO.
*/
SET_SRC_MODE M_DFF1;
SET_DST_MODE M_DFF1;
pkt_complete_scb_if_fifos_idle:
bmov ARG_1, SCBPTR, 2;
mvi DFFSXFRCTL, CLRCHN;
Upgrade to version 1.1 of the aic79xx U320 driver. aic79xx.c: o Remove redundant ahd_update_modes() call. o Correct panic in diagnostic should state corruption cause the SCB Id to be invalid during a selection timeout. o Add workaround for missing BUSFREEREV feature in Rev A silicon. o Corect formatting nits. o Use register pretty printing in more places. o Save and restore our SCB pointer when updating the waiting queue list for an "expected" LQ-out busfree. o In ahd_clear_intstat, deal with the missing autoclear in the CLRLQO* registers. o BE fixup in a diagnostic printf. o Make sure that we are in the proper mode before disabling selections in ahd_update_pending_scbs. o Add more diagnostics. o task_attribute_nonpkt_tag -> task_attribute: we don't need a nonpkt_tag field anymore for allowing all 512 SCBs to be used in non-packetized connections. o Negotiate HOLD_MCS to U320 devices. o Add a few additional mode assertions. o Restore the chip mode after clearing out the qinfifo so that code using ahd_abort_scbs sees a consistent mode. o Simplify the DMA engine shutdown routine prior to performing a bus reset. o Perform the sequencer restart after a chip reset prior to setting up our timer to poll for the reset to be complete. On some OSes, the timer could actually pre-empt us and order is important here. o Have our "reset poller" set the expected mode since there is no guarantee of what mode will be in force when we are called from the OS timer. o Save and restore the SCB pointer in ahd_dump_card_state(). This routine must not modify card state. o Ditto for ahd_dump_scbs(). aic79xx.h: o Add a few more chip bug definitions. o Align our tag on a 32bit boundary. aic79xx.reg: aic79xx.seq: o Start work on removing workarounds for Rev B. o Use a special location in scratch from for stroring our SCBPTR during legacy FIFO allocations. This corrects problems in mixed packetized/non-packetized configurations where calling into a FIFO task corrupted our SCBPTR. o Don't rely on DMA priority to guarantee that all data in our FIFOs will flush prior to a command completion notification going out of the command channel. We've never seen this assumption fail, but better safe than sorry. o Deal with missing BUSFREEREV feature in H2A. o Simplify disconnect list code now that the list will always have only a single entry. o Implement the AHD_REG_SLOW_SETTLE_BUG workaround. o Swith to using "REG_ISR" for local mode scratch during our ISR. o Add a missing jmp to the data_group_dma_loop after our data pointers have been re-initialized by the kernel. o Correct test in the bitbucket code so that we actually wait for the bitbucket to complete before signaling the kernel of the overrun condition. o Reposition pkt_saveptrs to avoid a jmp instruction. o Update a comment to reflect that the code now waits for a FIFO to drain prior to issuing a CLRCHN. aic79xx_inline.h: o Remove unused untagged queue handling code. o Don't attempt to htole64 what could be a 32bit value. aic79xx_pci.c: o Set additional bug flags for rev A chips.
2002-09-26 22:54:00 +00:00
SET_MODE(M_SCSI, M_SCSI)
bmov SCBPTR, ARG_1, 2;
Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-05-04 00:20:07 +00:00
test SCB_FIFO_USE_COUNT, 0xFF jnz return;
queue_scb_completion:
test SCB_SCSI_STATUS,0xff jnz bad_status;
/*
* Check for residuals
*/
test SCB_SGPTR, SG_LIST_NULL jnz complete; /* No xfer */
test SCB_SGPTR, SG_FULL_RESID jnz upload_scb;/* Never xfered */
test SCB_RESIDUAL_SGPTR, SG_LIST_NULL jz upload_scb;
complete:
ahc_eisa.c: ahc_pci.c: ahd_pci.c: aic7xxx.c: aic79xx.c: aic_osm_lib.c: aic_osm_lib.h: Use common OSM routines from aic_osm_lib for bus dma operations, delay routines, accessing CCBs, byte swapping, etc. aic7xxx_pci.c: Provide a better description for the 2915/30LP on attach. aic7xxx.c: aic79xx.c: aic7770.c: aic79xx_pci.c: aic7xxx_pci.c: aic7xxx_93cx6.c: Move FBSDID behind an ifdef so that these core files will still compile under other OSes. aic79xx.h: aic79xx_pci.c: aic79xx.seq: To speed up non-packetized CDB delivery in Rev B, all CDB acks are "released" to the output sync as soon as the command phase starts. There is only one problem with this approach. If the target changes phase before all data are sent, we have left over acks that can go out on the bus in a data phase. Due to other chip contraints, this only happens if the target goes to data-in, but if the acks go out before we can test SDONE, we'll think that the transfer has completed successfully. Work around this by taking advantage of the 400ns or 800ns dead time between command phase and the REQ of the new phase. If the transfer has completed successfully, SCSIEN should fall *long* before we see a phase change. We thus treat any phasemiss that occurs before SCSIEN falls as an incomplete transfer. aic79xx.h: Add the AHD_FAST_CDB_DELIVERY feature. aic79xx_pci.c: Set AHD_FAST_CDB_DELIVERY for all Rev. B parts. aic79xx.seq: Test for PHASEMIS in the command phase for all AHD_FAST_CDB_DELIVERY controlelrs. ahd_pci.c: ahc_pci.c: aic7xxx.h: aic79xx.h: Move definition of controller BAR offsets to core header files. aic7xxx.c: aic79xx.c: In the softc free routine, leave removal of a softc from the global list of softcs to the OSM (the caller of this routine). This allows us to avoid holding the softc list_lock during device destruction where we may have to sleep waiting for our recovery thread to halt. ahc_pci.c: Use ahc_pci_test_register access to validate I/O mapped in addition to the tests already performed for memory mapped access. Remove unused ahc_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. ahd_pci.c: Remove reduntant definition of controller BAR offsets. These are also defined in aic79xx.h. Remove unused ahd_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. aic7xxx.c: aic79xx.c: aic79xx.h: aic7xxx.h: aic7xxx_osm.c: aic79xx_osm.c: Move timeout handling to the driver cores. In the case of the aic79xx driver, the algorithm has been enhanced to try target resets before performing a bus reset. For the aic7xxx driver, the algorithm is unchanged. Although the drivers do not currently sleep during recovery (recovery is timeout driven), the cores do expect all processing to be performed via a recovery thread. Our timeout handlers are now little stubs that wakeup the recovery thread. aic79xx.c: aic79xx.h: aic79xx_inline.h: Change shared_data allocation to use a map_node so that the sentinel hscb can use this map node in ahd_swap_with_next_hscb. This routine now swaps the hscb_map pointer in additon to the hscb contents so that any sync operations occur on the correct map. physaddr -> busaddr Pointed out by: Jason Thorpe <thorpej@wasabisystems.com> aic79xx.c: Make more use of the in/out/w/l/q macros for accessing byte registers in the chip. Correct some issues in the ahd_flush_qoutfifo() routine. o Run the qoutfifo only once the command channel DMA engine has been halted. This closes a window where we might have missed some entries. o Change ahd_run_data_fifo() to not loop to completion. If we happen to start on the wrong FIFO and the other FIFO has a snapshot savepointers, we might deadlock. This required our delay between FIFO tests to be moved to the ahd_flush_qoutfifo() routine. o Update/add comments. o Remove spurious test for COMPLETE_DMA list being empty when completing transactions from the GSFIFO with residuals. The SCB must be put on the COMPLETE_DMA scb list unconditionally. o When halting command channel DMA activity, we must disable the DMA channel in all cases but an update of the QOUTFIFO. The latter case is required so that the sequencer will update its position in the QOUTFIFO. Previously, we left the channel enabled for all "push" DMAs. This left us vulnerable to the sequencer handling an SCB push long after that SCB was already processed manually by this routine. o Correct the polarity of tests involving ahd_scb_active_in_fifo(). This routine returns non-zero for true. Return to processing bad status completions through the qoutfifo. This reduces the time that the sequencer is kept paused when handling transactions with bad status or underruns. When waiting for the controller to quiece selections, add a delay to our loop. Otherwise we may fail to wait long enough for the sequencer to comply. On H2A4 hardware, use the slow slewrate for non-paced transfers. This mirrors what the Adaptec Windows drivers do. On the Rev B. only slow down the CRC timing for older U160 devices that might need the slower timing. We define "older" as devices that do not support packetized protocol. Wait up to 5000 * 5us for the SEEPROM to become unbusy. Write ops seem to take much longer than read ops. aic79xx.seq: For controllers with the FAINT_LED bug, turn the diagnostic led feature on during selection and reselection. This covers the non-packetized case. The LED will be disabled for non-packetized transfers once we return to the top level idle loop. Add more comments about the busy LED workaround. Extend a critical section around the entire command channel idle loop process. Previously the portion of this handler that directly manipulated the linked list of completed SCBs was not protected. This is the likely cause of the recent reports of commands being completed twice by the driver. Extend critical sections across the test for, and the longjump to, longjump routines. This prevents the firmware from trying to jump to a longjmp handler that was just cleared by the host. Improve the locations of several critical section begin and end points. Typically these changes remove instructions that did not need to be inside a critical section. Close the "busfree after selection, but before busfree interrupts can be enabled" race to just a single sequencer instruction. We now test the BSY line explicitly before clearing the busfree status and enabling the busfree interrupt. Close a race condition in the processing of HS_MAILBOX updates. We now clear the "updated" status before the copy. This ensures that we don't accidentally clear the status incorrectly when the host sneaks in an update just after our last copy, but before we clear the status. This race has never been observed. Don't re-enable SCSIEN if we lose the race to disable SCSIEN in our interrupt handler's workaround for the RevA data-valid too early issue. aic79xx_inline.h: Add comments indicating that the order in which bytes are read or written in ahd_inw and ahd_outw is important. This allows us to use these inlines when accessing registers with side-effects. aic79xx_pci.c: The 29320 and the 29320B are 7902 not 7901 based products. Correct the driver banner. aic7xxx.h: Enable the use of the auto-access pause feature on the aic7870 and aic7880. It was disabled due to an oversight. aic7xxx.reg: Move TARG_IMMEDIATE_SCB to alias LAST_MSG to avoid leaving garbage in MWI_RESIDUAL. This prevents spurious overflows whn operating target mode on controllers that require the MWI_RESIDUAL work-around. aic7xxx.seq: AHC_TMODE_WIDEODD_BUG is a bug, not a softc flag. Reference the correct softc field when testing for its presence. Set the NOT_IDENTIFIED and NO_CDB_SENT bits in SEQ_FLAGS to indicate that the nexus is invalid in await busfree. aic7xxx_93cx6.c: Add support for the C56/C66 versions of the EWEN and EWDS commands. aic7xxx.c: aic7xxx_pci.c: Move test for the validity of left over BIOS data to ahc_test_register_access(). This guarantees that any left over CHIPRST value is not clobbered by our register access test and lost to the test that was in ahc_reset.
2003-12-17 00:02:10 +00:00
BEGIN_CRITICAL;
bmov SCB_NEXT_COMPLETE, COMPLETE_SCB_HEAD, 2;
bmov COMPLETE_SCB_HEAD, SCBPTR, 2 ret;
ahc_eisa.c: ahc_pci.c: ahd_pci.c: aic7xxx.c: aic79xx.c: aic_osm_lib.c: aic_osm_lib.h: Use common OSM routines from aic_osm_lib for bus dma operations, delay routines, accessing CCBs, byte swapping, etc. aic7xxx_pci.c: Provide a better description for the 2915/30LP on attach. aic7xxx.c: aic79xx.c: aic7770.c: aic79xx_pci.c: aic7xxx_pci.c: aic7xxx_93cx6.c: Move FBSDID behind an ifdef so that these core files will still compile under other OSes. aic79xx.h: aic79xx_pci.c: aic79xx.seq: To speed up non-packetized CDB delivery in Rev B, all CDB acks are "released" to the output sync as soon as the command phase starts. There is only one problem with this approach. If the target changes phase before all data are sent, we have left over acks that can go out on the bus in a data phase. Due to other chip contraints, this only happens if the target goes to data-in, but if the acks go out before we can test SDONE, we'll think that the transfer has completed successfully. Work around this by taking advantage of the 400ns or 800ns dead time between command phase and the REQ of the new phase. If the transfer has completed successfully, SCSIEN should fall *long* before we see a phase change. We thus treat any phasemiss that occurs before SCSIEN falls as an incomplete transfer. aic79xx.h: Add the AHD_FAST_CDB_DELIVERY feature. aic79xx_pci.c: Set AHD_FAST_CDB_DELIVERY for all Rev. B parts. aic79xx.seq: Test for PHASEMIS in the command phase for all AHD_FAST_CDB_DELIVERY controlelrs. ahd_pci.c: ahc_pci.c: aic7xxx.h: aic79xx.h: Move definition of controller BAR offsets to core header files. aic7xxx.c: aic79xx.c: In the softc free routine, leave removal of a softc from the global list of softcs to the OSM (the caller of this routine). This allows us to avoid holding the softc list_lock during device destruction where we may have to sleep waiting for our recovery thread to halt. ahc_pci.c: Use ahc_pci_test_register access to validate I/O mapped in addition to the tests already performed for memory mapped access. Remove unused ahc_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. ahd_pci.c: Remove reduntant definition of controller BAR offsets. These are also defined in aic79xx.h. Remove unused ahd_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. aic7xxx.c: aic79xx.c: aic79xx.h: aic7xxx.h: aic7xxx_osm.c: aic79xx_osm.c: Move timeout handling to the driver cores. In the case of the aic79xx driver, the algorithm has been enhanced to try target resets before performing a bus reset. For the aic7xxx driver, the algorithm is unchanged. Although the drivers do not currently sleep during recovery (recovery is timeout driven), the cores do expect all processing to be performed via a recovery thread. Our timeout handlers are now little stubs that wakeup the recovery thread. aic79xx.c: aic79xx.h: aic79xx_inline.h: Change shared_data allocation to use a map_node so that the sentinel hscb can use this map node in ahd_swap_with_next_hscb. This routine now swaps the hscb_map pointer in additon to the hscb contents so that any sync operations occur on the correct map. physaddr -> busaddr Pointed out by: Jason Thorpe <thorpej@wasabisystems.com> aic79xx.c: Make more use of the in/out/w/l/q macros for accessing byte registers in the chip. Correct some issues in the ahd_flush_qoutfifo() routine. o Run the qoutfifo only once the command channel DMA engine has been halted. This closes a window where we might have missed some entries. o Change ahd_run_data_fifo() to not loop to completion. If we happen to start on the wrong FIFO and the other FIFO has a snapshot savepointers, we might deadlock. This required our delay between FIFO tests to be moved to the ahd_flush_qoutfifo() routine. o Update/add comments. o Remove spurious test for COMPLETE_DMA list being empty when completing transactions from the GSFIFO with residuals. The SCB must be put on the COMPLETE_DMA scb list unconditionally. o When halting command channel DMA activity, we must disable the DMA channel in all cases but an update of the QOUTFIFO. The latter case is required so that the sequencer will update its position in the QOUTFIFO. Previously, we left the channel enabled for all "push" DMAs. This left us vulnerable to the sequencer handling an SCB push long after that SCB was already processed manually by this routine. o Correct the polarity of tests involving ahd_scb_active_in_fifo(). This routine returns non-zero for true. Return to processing bad status completions through the qoutfifo. This reduces the time that the sequencer is kept paused when handling transactions with bad status or underruns. When waiting for the controller to quiece selections, add a delay to our loop. Otherwise we may fail to wait long enough for the sequencer to comply. On H2A4 hardware, use the slow slewrate for non-paced transfers. This mirrors what the Adaptec Windows drivers do. On the Rev B. only slow down the CRC timing for older U160 devices that might need the slower timing. We define "older" as devices that do not support packetized protocol. Wait up to 5000 * 5us for the SEEPROM to become unbusy. Write ops seem to take much longer than read ops. aic79xx.seq: For controllers with the FAINT_LED bug, turn the diagnostic led feature on during selection and reselection. This covers the non-packetized case. The LED will be disabled for non-packetized transfers once we return to the top level idle loop. Add more comments about the busy LED workaround. Extend a critical section around the entire command channel idle loop process. Previously the portion of this handler that directly manipulated the linked list of completed SCBs was not protected. This is the likely cause of the recent reports of commands being completed twice by the driver. Extend critical sections across the test for, and the longjump to, longjump routines. This prevents the firmware from trying to jump to a longjmp handler that was just cleared by the host. Improve the locations of several critical section begin and end points. Typically these changes remove instructions that did not need to be inside a critical section. Close the "busfree after selection, but before busfree interrupts can be enabled" race to just a single sequencer instruction. We now test the BSY line explicitly before clearing the busfree status and enabling the busfree interrupt. Close a race condition in the processing of HS_MAILBOX updates. We now clear the "updated" status before the copy. This ensures that we don't accidentally clear the status incorrectly when the host sneaks in an update just after our last copy, but before we clear the status. This race has never been observed. Don't re-enable SCSIEN if we lose the race to disable SCSIEN in our interrupt handler's workaround for the RevA data-valid too early issue. aic79xx_inline.h: Add comments indicating that the order in which bytes are read or written in ahd_inw and ahd_outw is important. This allows us to use these inlines when accessing registers with side-effects. aic79xx_pci.c: The 29320 and the 29320B are 7902 not 7901 based products. Correct the driver banner. aic7xxx.h: Enable the use of the auto-access pause feature on the aic7870 and aic7880. It was disabled due to an oversight. aic7xxx.reg: Move TARG_IMMEDIATE_SCB to alias LAST_MSG to avoid leaving garbage in MWI_RESIDUAL. This prevents spurious overflows whn operating target mode on controllers that require the MWI_RESIDUAL work-around. aic7xxx.seq: AHC_TMODE_WIDEODD_BUG is a bug, not a softc flag. Reference the correct softc field when testing for its presence. Set the NOT_IDENTIFIED and NO_CDB_SENT bits in SEQ_FLAGS to indicate that the nexus is invalid in await busfree. aic7xxx_93cx6.c: Add support for the C56/C66 versions of the EWEN and EWDS commands. aic7xxx.c: aic7xxx_pci.c: Move test for the validity of left over BIOS data to ahc_test_register_access(). This guarantees that any left over CHIPRST value is not clobbered by our register access test and lost to the test that was in ahc_reset.
2003-12-17 00:02:10 +00:00
END_CRITICAL;
bad_status:
cmp SCB_SCSI_STATUS, STATUS_PKT_SENSE je upload_scb;
call freeze_queue;
upload_scb:
Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-05-04 00:20:07 +00:00
/*
* Restore SCB TAG since we reuse this field
* in the sequencer. We don't want to corrupt
* it on the host.
*/
bmov SCB_TAG, SCBPTR, 2;
ahc_eisa.c: ahc_pci.c: ahd_pci.c: aic7xxx.c: aic79xx.c: aic_osm_lib.c: aic_osm_lib.h: Use common OSM routines from aic_osm_lib for bus dma operations, delay routines, accessing CCBs, byte swapping, etc. aic7xxx_pci.c: Provide a better description for the 2915/30LP on attach. aic7xxx.c: aic79xx.c: aic7770.c: aic79xx_pci.c: aic7xxx_pci.c: aic7xxx_93cx6.c: Move FBSDID behind an ifdef so that these core files will still compile under other OSes. aic79xx.h: aic79xx_pci.c: aic79xx.seq: To speed up non-packetized CDB delivery in Rev B, all CDB acks are "released" to the output sync as soon as the command phase starts. There is only one problem with this approach. If the target changes phase before all data are sent, we have left over acks that can go out on the bus in a data phase. Due to other chip contraints, this only happens if the target goes to data-in, but if the acks go out before we can test SDONE, we'll think that the transfer has completed successfully. Work around this by taking advantage of the 400ns or 800ns dead time between command phase and the REQ of the new phase. If the transfer has completed successfully, SCSIEN should fall *long* before we see a phase change. We thus treat any phasemiss that occurs before SCSIEN falls as an incomplete transfer. aic79xx.h: Add the AHD_FAST_CDB_DELIVERY feature. aic79xx_pci.c: Set AHD_FAST_CDB_DELIVERY for all Rev. B parts. aic79xx.seq: Test for PHASEMIS in the command phase for all AHD_FAST_CDB_DELIVERY controlelrs. ahd_pci.c: ahc_pci.c: aic7xxx.h: aic79xx.h: Move definition of controller BAR offsets to core header files. aic7xxx.c: aic79xx.c: In the softc free routine, leave removal of a softc from the global list of softcs to the OSM (the caller of this routine). This allows us to avoid holding the softc list_lock during device destruction where we may have to sleep waiting for our recovery thread to halt. ahc_pci.c: Use ahc_pci_test_register access to validate I/O mapped in addition to the tests already performed for memory mapped access. Remove unused ahc_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. ahd_pci.c: Remove reduntant definition of controller BAR offsets. These are also defined in aic79xx.h. Remove unused ahd_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. aic7xxx.c: aic79xx.c: aic79xx.h: aic7xxx.h: aic7xxx_osm.c: aic79xx_osm.c: Move timeout handling to the driver cores. In the case of the aic79xx driver, the algorithm has been enhanced to try target resets before performing a bus reset. For the aic7xxx driver, the algorithm is unchanged. Although the drivers do not currently sleep during recovery (recovery is timeout driven), the cores do expect all processing to be performed via a recovery thread. Our timeout handlers are now little stubs that wakeup the recovery thread. aic79xx.c: aic79xx.h: aic79xx_inline.h: Change shared_data allocation to use a map_node so that the sentinel hscb can use this map node in ahd_swap_with_next_hscb. This routine now swaps the hscb_map pointer in additon to the hscb contents so that any sync operations occur on the correct map. physaddr -> busaddr Pointed out by: Jason Thorpe <thorpej@wasabisystems.com> aic79xx.c: Make more use of the in/out/w/l/q macros for accessing byte registers in the chip. Correct some issues in the ahd_flush_qoutfifo() routine. o Run the qoutfifo only once the command channel DMA engine has been halted. This closes a window where we might have missed some entries. o Change ahd_run_data_fifo() to not loop to completion. If we happen to start on the wrong FIFO and the other FIFO has a snapshot savepointers, we might deadlock. This required our delay between FIFO tests to be moved to the ahd_flush_qoutfifo() routine. o Update/add comments. o Remove spurious test for COMPLETE_DMA list being empty when completing transactions from the GSFIFO with residuals. The SCB must be put on the COMPLETE_DMA scb list unconditionally. o When halting command channel DMA activity, we must disable the DMA channel in all cases but an update of the QOUTFIFO. The latter case is required so that the sequencer will update its position in the QOUTFIFO. Previously, we left the channel enabled for all "push" DMAs. This left us vulnerable to the sequencer handling an SCB push long after that SCB was already processed manually by this routine. o Correct the polarity of tests involving ahd_scb_active_in_fifo(). This routine returns non-zero for true. Return to processing bad status completions through the qoutfifo. This reduces the time that the sequencer is kept paused when handling transactions with bad status or underruns. When waiting for the controller to quiece selections, add a delay to our loop. Otherwise we may fail to wait long enough for the sequencer to comply. On H2A4 hardware, use the slow slewrate for non-paced transfers. This mirrors what the Adaptec Windows drivers do. On the Rev B. only slow down the CRC timing for older U160 devices that might need the slower timing. We define "older" as devices that do not support packetized protocol. Wait up to 5000 * 5us for the SEEPROM to become unbusy. Write ops seem to take much longer than read ops. aic79xx.seq: For controllers with the FAINT_LED bug, turn the diagnostic led feature on during selection and reselection. This covers the non-packetized case. The LED will be disabled for non-packetized transfers once we return to the top level idle loop. Add more comments about the busy LED workaround. Extend a critical section around the entire command channel idle loop process. Previously the portion of this handler that directly manipulated the linked list of completed SCBs was not protected. This is the likely cause of the recent reports of commands being completed twice by the driver. Extend critical sections across the test for, and the longjump to, longjump routines. This prevents the firmware from trying to jump to a longjmp handler that was just cleared by the host. Improve the locations of several critical section begin and end points. Typically these changes remove instructions that did not need to be inside a critical section. Close the "busfree after selection, but before busfree interrupts can be enabled" race to just a single sequencer instruction. We now test the BSY line explicitly before clearing the busfree status and enabling the busfree interrupt. Close a race condition in the processing of HS_MAILBOX updates. We now clear the "updated" status before the copy. This ensures that we don't accidentally clear the status incorrectly when the host sneaks in an update just after our last copy, but before we clear the status. This race has never been observed. Don't re-enable SCSIEN if we lose the race to disable SCSIEN in our interrupt handler's workaround for the RevA data-valid too early issue. aic79xx_inline.h: Add comments indicating that the order in which bytes are read or written in ahd_inw and ahd_outw is important. This allows us to use these inlines when accessing registers with side-effects. aic79xx_pci.c: The 29320 and the 29320B are 7902 not 7901 based products. Correct the driver banner. aic7xxx.h: Enable the use of the auto-access pause feature on the aic7870 and aic7880. It was disabled due to an oversight. aic7xxx.reg: Move TARG_IMMEDIATE_SCB to alias LAST_MSG to avoid leaving garbage in MWI_RESIDUAL. This prevents spurious overflows whn operating target mode on controllers that require the MWI_RESIDUAL work-around. aic7xxx.seq: AHC_TMODE_WIDEODD_BUG is a bug, not a softc flag. Reference the correct softc field when testing for its presence. Set the NOT_IDENTIFIED and NO_CDB_SENT bits in SEQ_FLAGS to indicate that the nexus is invalid in await busfree. aic7xxx_93cx6.c: Add support for the C56/C66 versions of the EWEN and EWDS commands. aic7xxx.c: aic7xxx_pci.c: Move test for the validity of left over BIOS data to ahc_test_register_access(). This guarantees that any left over CHIPRST value is not clobbered by our register access test and lost to the test that was in ahc_reset.
2003-12-17 00:02:10 +00:00
BEGIN_CRITICAL;
aic79xx.c: aic79xx.seq: Convert the COMPLETE_DMA_SCB list to an "stailq". This allows us to safely keep the SCB that is currently being DMA'ed back the host on the head of the list while processing completions off of the bus. The newly completed SCBs are appended to the tail of the queue. In the past, we just dequeued the SCB that was in flight from the list, but this could result in a lost completion should the host perform certain types of error recovery that must cancel all in-flight SCB DMA operations. Switch from using a 16bit completion entry, holding just the tag and the completion valid bit, to a 64bit completion entry that also contains a "status packet valid" indicator. This solves two problems: o The SCB DMA engine on at least Rev B. silicon does not properly deal with a PCI disconnect that occurs at a non-64bit aligned offset in the chips "source buffer". When the transfer is resumed, the DMA engine continues at the correct offset, but may wrap to the head of the buffer causing duplicate completions to be reported to the host. By using a completion buffer in host memory that is 64bit aligned and using 64bit completion entries, such disconnects should only occur at aligned addresses. This assumes that the host bridge will only disconnect on cache-line boundaries and that cache-lines are multpiles of 64bits. o By embedding the status information in the completion entry we can avoid an extra memory reference to the HSCB for commands that complete without error. Use the comparison of a "host freeze count" and a "sequencer freeze count" to allow the host to process most SCBs that complete with non-zero status without having to clear critical sections. Instead the host can just pause the sequencer, performs any necessary cleanup in the waiting for selection list, increments its freeze count on the controller, and unpauses. This is only possible because the sequencer defers completions of SCBs with bad status until after all pending selections have completed. The sequencer then avoids referencing any data structures the host may touch during completion of the SCB until the freeze counts match. aic79xx.c: Change the strategy for allocating our sentinal HSCB for the QINFIFO. In the past, this allocation was tacked onto the QOUTFIFO allocation. Now that the qoutfifo has grown to accomodate larger completion entries, the old approach will result in a 64byte allocation that costs an extra page of coherent memory. We now do this extra allocation via ahd_alloc_scbs() where the "unused space" can be used to allocate "normal" HSCBs. In our packetized busfree handler, use the ENSELO bit to differentiate between packetized and non-packetized unexpected busfree events that occur just after selection, but before the sequencer has had the oportunity to service the selection. When cleaning out the waiting for selection list, use the SCSI mode instead of the command channel mode. The SCB pointer in the command channel mode may be referenced by the SCB dma engine even while the sequencer is paused, whereas the SCSI mode SCB pointer is only accessed by the sequencer. Print the "complete on qfreeze" sequencer SCB completion list in ahd_dump_card_state(). This list holds all SCB completions that are deferred until a pending select-out qfreeze event has taken effect. aic79xx.h: Add definitions and structures to handle the new SCB completion scheme. Add a controller flag that indicates if the controller is in HostRAID mode. aic79xx.reg: Remove macros used for toggling from one data fifo mode to the other. They have not been in use for some time. Add scratch ram fields for our new qfreeze count scheme, converting the complete dma list into an "stailq", and providing for the "complete on qfreeze" SCB completion list. Some other fields were moved to retain proper field alignment (alignment >= field size in bytes). aic79xx.seq: Add code to our idle loop to: o Process deferred completions once a qfreeze event has taken full effect. o Thaw the queue once the sequencer and host qfreeze counts match. Generate 64bit completion entries passing the SCB_SGPTR field as the "good status" indicator. The first bit in this field is only set if we have a valid status packet to send to the host. Convert the COMPLETE_DMA_SCB list to an "stailq". When using "setjmp" to register an idle loop handler, do not combine the "ret" with the block move to pop the stack address in the same instruction. At least on the A, this results in a return to the setjmp caller, not to the new address at the top of the stack. Since we want the latter (we want the newly registered handler to only be invoked from the idle loop), we must use a separate ret instruction. Add a few missing critical sections. Close a race condition that can occur on Rev A. silicon. If both FIFOs happen to be allocated before the sequencer has a chance to service the FIFO that was allocated first, we must take special care to service the FIFO that is not active on the SCSI bus first. This guarantees that a FIFO will be freed to handle any snapshot requests for the FIFO that is still on the bus. Chosing the incorrect FIFO will result in deadlock. Update comments. aic79xx_inline.h Correct the offset calculation for the syncing of our qoutfifo. Update ahd_check_cmdcmpltqueues() for the larger completion entries. aic79xx_pci.c: Attach to HostRAID controllers by default. In the future I may add a sysctl to modify the behavior, but since FreeBSD does not have any HostRAID drivers, failing to attach just results in more email and bug reports for the author. MFC After: 1week
2004-02-04 16:38:38 +00:00
or SCB_SGPTR, SG_STATUS_VALID;
mvi SCB_NEXT_COMPLETE[1], SCB_LIST_NULL;
cmp COMPLETE_DMA_SCB_HEAD[1], SCB_LIST_NULL jne add_dma_scb_tail;
bmov COMPLETE_DMA_SCB_HEAD, SCBPTR, 2;
aic79xx.c: aic79xx.seq: Convert the COMPLETE_DMA_SCB list to an "stailq". This allows us to safely keep the SCB that is currently being DMA'ed back the host on the head of the list while processing completions off of the bus. The newly completed SCBs are appended to the tail of the queue. In the past, we just dequeued the SCB that was in flight from the list, but this could result in a lost completion should the host perform certain types of error recovery that must cancel all in-flight SCB DMA operations. Switch from using a 16bit completion entry, holding just the tag and the completion valid bit, to a 64bit completion entry that also contains a "status packet valid" indicator. This solves two problems: o The SCB DMA engine on at least Rev B. silicon does not properly deal with a PCI disconnect that occurs at a non-64bit aligned offset in the chips "source buffer". When the transfer is resumed, the DMA engine continues at the correct offset, but may wrap to the head of the buffer causing duplicate completions to be reported to the host. By using a completion buffer in host memory that is 64bit aligned and using 64bit completion entries, such disconnects should only occur at aligned addresses. This assumes that the host bridge will only disconnect on cache-line boundaries and that cache-lines are multpiles of 64bits. o By embedding the status information in the completion entry we can avoid an extra memory reference to the HSCB for commands that complete without error. Use the comparison of a "host freeze count" and a "sequencer freeze count" to allow the host to process most SCBs that complete with non-zero status without having to clear critical sections. Instead the host can just pause the sequencer, performs any necessary cleanup in the waiting for selection list, increments its freeze count on the controller, and unpauses. This is only possible because the sequencer defers completions of SCBs with bad status until after all pending selections have completed. The sequencer then avoids referencing any data structures the host may touch during completion of the SCB until the freeze counts match. aic79xx.c: Change the strategy for allocating our sentinal HSCB for the QINFIFO. In the past, this allocation was tacked onto the QOUTFIFO allocation. Now that the qoutfifo has grown to accomodate larger completion entries, the old approach will result in a 64byte allocation that costs an extra page of coherent memory. We now do this extra allocation via ahd_alloc_scbs() where the "unused space" can be used to allocate "normal" HSCBs. In our packetized busfree handler, use the ENSELO bit to differentiate between packetized and non-packetized unexpected busfree events that occur just after selection, but before the sequencer has had the oportunity to service the selection. When cleaning out the waiting for selection list, use the SCSI mode instead of the command channel mode. The SCB pointer in the command channel mode may be referenced by the SCB dma engine even while the sequencer is paused, whereas the SCSI mode SCB pointer is only accessed by the sequencer. Print the "complete on qfreeze" sequencer SCB completion list in ahd_dump_card_state(). This list holds all SCB completions that are deferred until a pending select-out qfreeze event has taken effect. aic79xx.h: Add definitions and structures to handle the new SCB completion scheme. Add a controller flag that indicates if the controller is in HostRAID mode. aic79xx.reg: Remove macros used for toggling from one data fifo mode to the other. They have not been in use for some time. Add scratch ram fields for our new qfreeze count scheme, converting the complete dma list into an "stailq", and providing for the "complete on qfreeze" SCB completion list. Some other fields were moved to retain proper field alignment (alignment >= field size in bytes). aic79xx.seq: Add code to our idle loop to: o Process deferred completions once a qfreeze event has taken full effect. o Thaw the queue once the sequencer and host qfreeze counts match. Generate 64bit completion entries passing the SCB_SGPTR field as the "good status" indicator. The first bit in this field is only set if we have a valid status packet to send to the host. Convert the COMPLETE_DMA_SCB list to an "stailq". When using "setjmp" to register an idle loop handler, do not combine the "ret" with the block move to pop the stack address in the same instruction. At least on the A, this results in a return to the setjmp caller, not to the new address at the top of the stack. Since we want the latter (we want the newly registered handler to only be invoked from the idle loop), we must use a separate ret instruction. Add a few missing critical sections. Close a race condition that can occur on Rev A. silicon. If both FIFOs happen to be allocated before the sequencer has a chance to service the FIFO that was allocated first, we must take special care to service the FIFO that is not active on the SCSI bus first. This guarantees that a FIFO will be freed to handle any snapshot requests for the FIFO that is still on the bus. Chosing the incorrect FIFO will result in deadlock. Update comments. aic79xx_inline.h Correct the offset calculation for the syncing of our qoutfifo. Update ahd_check_cmdcmpltqueues() for the larger completion entries. aic79xx_pci.c: Attach to HostRAID controllers by default. In the future I may add a sysctl to modify the behavior, but since FreeBSD does not have any HostRAID drivers, failing to attach just results in more email and bug reports for the author. MFC After: 1week
2004-02-04 16:38:38 +00:00
bmov COMPLETE_DMA_SCB_TAIL, SCBPTR, 2 ret;
add_dma_scb_tail:
bmov REG0, SCBPTR, 2;
bmov SCBPTR, COMPLETE_DMA_SCB_TAIL, 2;
bmov SCB_NEXT_COMPLETE, REG0, 2;
bmov COMPLETE_DMA_SCB_TAIL, REG0, 2 ret;
ahc_eisa.c: ahc_pci.c: ahd_pci.c: aic7xxx.c: aic79xx.c: aic_osm_lib.c: aic_osm_lib.h: Use common OSM routines from aic_osm_lib for bus dma operations, delay routines, accessing CCBs, byte swapping, etc. aic7xxx_pci.c: Provide a better description for the 2915/30LP on attach. aic7xxx.c: aic79xx.c: aic7770.c: aic79xx_pci.c: aic7xxx_pci.c: aic7xxx_93cx6.c: Move FBSDID behind an ifdef so that these core files will still compile under other OSes. aic79xx.h: aic79xx_pci.c: aic79xx.seq: To speed up non-packetized CDB delivery in Rev B, all CDB acks are "released" to the output sync as soon as the command phase starts. There is only one problem with this approach. If the target changes phase before all data are sent, we have left over acks that can go out on the bus in a data phase. Due to other chip contraints, this only happens if the target goes to data-in, but if the acks go out before we can test SDONE, we'll think that the transfer has completed successfully. Work around this by taking advantage of the 400ns or 800ns dead time between command phase and the REQ of the new phase. If the transfer has completed successfully, SCSIEN should fall *long* before we see a phase change. We thus treat any phasemiss that occurs before SCSIEN falls as an incomplete transfer. aic79xx.h: Add the AHD_FAST_CDB_DELIVERY feature. aic79xx_pci.c: Set AHD_FAST_CDB_DELIVERY for all Rev. B parts. aic79xx.seq: Test for PHASEMIS in the command phase for all AHD_FAST_CDB_DELIVERY controlelrs. ahd_pci.c: ahc_pci.c: aic7xxx.h: aic79xx.h: Move definition of controller BAR offsets to core header files. aic7xxx.c: aic79xx.c: In the softc free routine, leave removal of a softc from the global list of softcs to the OSM (the caller of this routine). This allows us to avoid holding the softc list_lock during device destruction where we may have to sleep waiting for our recovery thread to halt. ahc_pci.c: Use ahc_pci_test_register access to validate I/O mapped in addition to the tests already performed for memory mapped access. Remove unused ahc_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. ahd_pci.c: Remove reduntant definition of controller BAR offsets. These are also defined in aic79xx.h. Remove unused ahd_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. aic7xxx.c: aic79xx.c: aic79xx.h: aic7xxx.h: aic7xxx_osm.c: aic79xx_osm.c: Move timeout handling to the driver cores. In the case of the aic79xx driver, the algorithm has been enhanced to try target resets before performing a bus reset. For the aic7xxx driver, the algorithm is unchanged. Although the drivers do not currently sleep during recovery (recovery is timeout driven), the cores do expect all processing to be performed via a recovery thread. Our timeout handlers are now little stubs that wakeup the recovery thread. aic79xx.c: aic79xx.h: aic79xx_inline.h: Change shared_data allocation to use a map_node so that the sentinel hscb can use this map node in ahd_swap_with_next_hscb. This routine now swaps the hscb_map pointer in additon to the hscb contents so that any sync operations occur on the correct map. physaddr -> busaddr Pointed out by: Jason Thorpe <thorpej@wasabisystems.com> aic79xx.c: Make more use of the in/out/w/l/q macros for accessing byte registers in the chip. Correct some issues in the ahd_flush_qoutfifo() routine. o Run the qoutfifo only once the command channel DMA engine has been halted. This closes a window where we might have missed some entries. o Change ahd_run_data_fifo() to not loop to completion. If we happen to start on the wrong FIFO and the other FIFO has a snapshot savepointers, we might deadlock. This required our delay between FIFO tests to be moved to the ahd_flush_qoutfifo() routine. o Update/add comments. o Remove spurious test for COMPLETE_DMA list being empty when completing transactions from the GSFIFO with residuals. The SCB must be put on the COMPLETE_DMA scb list unconditionally. o When halting command channel DMA activity, we must disable the DMA channel in all cases but an update of the QOUTFIFO. The latter case is required so that the sequencer will update its position in the QOUTFIFO. Previously, we left the channel enabled for all "push" DMAs. This left us vulnerable to the sequencer handling an SCB push long after that SCB was already processed manually by this routine. o Correct the polarity of tests involving ahd_scb_active_in_fifo(). This routine returns non-zero for true. Return to processing bad status completions through the qoutfifo. This reduces the time that the sequencer is kept paused when handling transactions with bad status or underruns. When waiting for the controller to quiece selections, add a delay to our loop. Otherwise we may fail to wait long enough for the sequencer to comply. On H2A4 hardware, use the slow slewrate for non-paced transfers. This mirrors what the Adaptec Windows drivers do. On the Rev B. only slow down the CRC timing for older U160 devices that might need the slower timing. We define "older" as devices that do not support packetized protocol. Wait up to 5000 * 5us for the SEEPROM to become unbusy. Write ops seem to take much longer than read ops. aic79xx.seq: For controllers with the FAINT_LED bug, turn the diagnostic led feature on during selection and reselection. This covers the non-packetized case. The LED will be disabled for non-packetized transfers once we return to the top level idle loop. Add more comments about the busy LED workaround. Extend a critical section around the entire command channel idle loop process. Previously the portion of this handler that directly manipulated the linked list of completed SCBs was not protected. This is the likely cause of the recent reports of commands being completed twice by the driver. Extend critical sections across the test for, and the longjump to, longjump routines. This prevents the firmware from trying to jump to a longjmp handler that was just cleared by the host. Improve the locations of several critical section begin and end points. Typically these changes remove instructions that did not need to be inside a critical section. Close the "busfree after selection, but before busfree interrupts can be enabled" race to just a single sequencer instruction. We now test the BSY line explicitly before clearing the busfree status and enabling the busfree interrupt. Close a race condition in the processing of HS_MAILBOX updates. We now clear the "updated" status before the copy. This ensures that we don't accidentally clear the status incorrectly when the host sneaks in an update just after our last copy, but before we clear the status. This race has never been observed. Don't re-enable SCSIEN if we lose the race to disable SCSIEN in our interrupt handler's workaround for the RevA data-valid too early issue. aic79xx_inline.h: Add comments indicating that the order in which bytes are read or written in ahd_inw and ahd_outw is important. This allows us to use these inlines when accessing registers with side-effects. aic79xx_pci.c: The 29320 and the 29320B are 7902 not 7901 based products. Correct the driver banner. aic7xxx.h: Enable the use of the auto-access pause feature on the aic7870 and aic7880. It was disabled due to an oversight. aic7xxx.reg: Move TARG_IMMEDIATE_SCB to alias LAST_MSG to avoid leaving garbage in MWI_RESIDUAL. This prevents spurious overflows whn operating target mode on controllers that require the MWI_RESIDUAL work-around. aic7xxx.seq: AHC_TMODE_WIDEODD_BUG is a bug, not a softc flag. Reference the correct softc field when testing for its presence. Set the NOT_IDENTIFIED and NO_CDB_SENT bits in SEQ_FLAGS to indicate that the nexus is invalid in await busfree. aic7xxx_93cx6.c: Add support for the C56/C66 versions of the EWEN and EWDS commands. aic7xxx.c: aic7xxx_pci.c: Move test for the validity of left over BIOS data to ahc_test_register_access(). This guarantees that any left over CHIPRST value is not clobbered by our register access test and lost to the test that was in ahc_reset.
2003-12-17 00:02:10 +00:00
END_CRITICAL;
/*
* Is it a disconnect message? Set a flag in the SCB to remind us
* and await the bus going free. If this is an untagged transaction
* store the SCB id for it in our untagged target table for lookup on
* a reselction.
*/
mesgin_disconnect:
/*
* If ATN is raised, we still want to give the target a message.
* Perhaps there was a parity error on this last message byte
* or we want to abort this command. Either way, the target
* should take us to message out phase and then attempt to
* disconnect again.
* XXX - Wait for more testing.
test SCSISIGI, ATNI jnz mesgin_done;
*/
test SEQ_FLAGS, NOT_IDENTIFIED|NO_CDB_SENT
jnz mesgin_proto_violation;
or SCB_CONTROL,DISCONNECTED;
test SCB_CONTROL, TAG_ENB jnz await_busfree;
queue_disc_scb:
bmov REG0, SCBPTR, 2;
INDEX_DISC_LIST(SAVED_SCSIID, SAVED_LUN);
bmov DINDEX, SINDEX, 2;
bmov DINDIR, REG0, 2;
bmov SCBPTR, REG0, 2;
/* FALLTHROUGH */
await_busfree:
and SIMODE1, ~ENBUSFREE;
Upgrade to version 1.1 of the aic79xx U320 driver. aic79xx.c: o Remove redundant ahd_update_modes() call. o Correct panic in diagnostic should state corruption cause the SCB Id to be invalid during a selection timeout. o Add workaround for missing BUSFREEREV feature in Rev A silicon. o Corect formatting nits. o Use register pretty printing in more places. o Save and restore our SCB pointer when updating the waiting queue list for an "expected" LQ-out busfree. o In ahd_clear_intstat, deal with the missing autoclear in the CLRLQO* registers. o BE fixup in a diagnostic printf. o Make sure that we are in the proper mode before disabling selections in ahd_update_pending_scbs. o Add more diagnostics. o task_attribute_nonpkt_tag -> task_attribute: we don't need a nonpkt_tag field anymore for allowing all 512 SCBs to be used in non-packetized connections. o Negotiate HOLD_MCS to U320 devices. o Add a few additional mode assertions. o Restore the chip mode after clearing out the qinfifo so that code using ahd_abort_scbs sees a consistent mode. o Simplify the DMA engine shutdown routine prior to performing a bus reset. o Perform the sequencer restart after a chip reset prior to setting up our timer to poll for the reset to be complete. On some OSes, the timer could actually pre-empt us and order is important here. o Have our "reset poller" set the expected mode since there is no guarantee of what mode will be in force when we are called from the OS timer. o Save and restore the SCB pointer in ahd_dump_card_state(). This routine must not modify card state. o Ditto for ahd_dump_scbs(). aic79xx.h: o Add a few more chip bug definitions. o Align our tag on a 32bit boundary. aic79xx.reg: aic79xx.seq: o Start work on removing workarounds for Rev B. o Use a special location in scratch from for stroring our SCBPTR during legacy FIFO allocations. This corrects problems in mixed packetized/non-packetized configurations where calling into a FIFO task corrupted our SCBPTR. o Don't rely on DMA priority to guarantee that all data in our FIFOs will flush prior to a command completion notification going out of the command channel. We've never seen this assumption fail, but better safe than sorry. o Deal with missing BUSFREEREV feature in H2A. o Simplify disconnect list code now that the list will always have only a single entry. o Implement the AHD_REG_SLOW_SETTLE_BUG workaround. o Swith to using "REG_ISR" for local mode scratch during our ISR. o Add a missing jmp to the data_group_dma_loop after our data pointers have been re-initialized by the kernel. o Correct test in the bitbucket code so that we actually wait for the bitbucket to complete before signaling the kernel of the overrun condition. o Reposition pkt_saveptrs to avoid a jmp instruction. o Update a comment to reflect that the code now waits for a FIFO to drain prior to issuing a CLRCHN. aic79xx_inline.h: o Remove unused untagged queue handling code. o Don't attempt to htole64 what could be a 32bit value. aic79xx_pci.c: o Set additional bug flags for rev A chips.
2002-09-26 22:54:00 +00:00
if ((ahd->bugs & AHD_BUSFREEREV_BUG) == 0) {
/*
* In the BUSFREEREV_BUG case, the
* busfree status was cleared at the
* beginning of the connection.
*/
mvi CLRSINT1,CLRBUSFREE;
}
mov NONE, SCSIDAT; /* Ack the last byte */
test MODE_PTR, ~(MK_MODE(M_DFF1, M_DFF1))
jnz await_busfree_not_m_dff;
SET_SRC_MODE M_DFF1;
SET_DST_MODE M_DFF1;
await_busfree_clrchn:
mvi DFFSXFRCTL, CLRCHN;
await_busfree_not_m_dff:
ahc_eisa.c: ahc_pci.c: ahd_pci.c: aic7xxx.c: aic79xx.c: aic_osm_lib.c: aic_osm_lib.h: Use common OSM routines from aic_osm_lib for bus dma operations, delay routines, accessing CCBs, byte swapping, etc. aic7xxx_pci.c: Provide a better description for the 2915/30LP on attach. aic7xxx.c: aic79xx.c: aic7770.c: aic79xx_pci.c: aic7xxx_pci.c: aic7xxx_93cx6.c: Move FBSDID behind an ifdef so that these core files will still compile under other OSes. aic79xx.h: aic79xx_pci.c: aic79xx.seq: To speed up non-packetized CDB delivery in Rev B, all CDB acks are "released" to the output sync as soon as the command phase starts. There is only one problem with this approach. If the target changes phase before all data are sent, we have left over acks that can go out on the bus in a data phase. Due to other chip contraints, this only happens if the target goes to data-in, but if the acks go out before we can test SDONE, we'll think that the transfer has completed successfully. Work around this by taking advantage of the 400ns or 800ns dead time between command phase and the REQ of the new phase. If the transfer has completed successfully, SCSIEN should fall *long* before we see a phase change. We thus treat any phasemiss that occurs before SCSIEN falls as an incomplete transfer. aic79xx.h: Add the AHD_FAST_CDB_DELIVERY feature. aic79xx_pci.c: Set AHD_FAST_CDB_DELIVERY for all Rev. B parts. aic79xx.seq: Test for PHASEMIS in the command phase for all AHD_FAST_CDB_DELIVERY controlelrs. ahd_pci.c: ahc_pci.c: aic7xxx.h: aic79xx.h: Move definition of controller BAR offsets to core header files. aic7xxx.c: aic79xx.c: In the softc free routine, leave removal of a softc from the global list of softcs to the OSM (the caller of this routine). This allows us to avoid holding the softc list_lock during device destruction where we may have to sleep waiting for our recovery thread to halt. ahc_pci.c: Use ahc_pci_test_register access to validate I/O mapped in addition to the tests already performed for memory mapped access. Remove unused ahc_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. ahd_pci.c: Remove reduntant definition of controller BAR offsets. These are also defined in aic79xx.h. Remove unused ahd_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. aic7xxx.c: aic79xx.c: aic79xx.h: aic7xxx.h: aic7xxx_osm.c: aic79xx_osm.c: Move timeout handling to the driver cores. In the case of the aic79xx driver, the algorithm has been enhanced to try target resets before performing a bus reset. For the aic7xxx driver, the algorithm is unchanged. Although the drivers do not currently sleep during recovery (recovery is timeout driven), the cores do expect all processing to be performed via a recovery thread. Our timeout handlers are now little stubs that wakeup the recovery thread. aic79xx.c: aic79xx.h: aic79xx_inline.h: Change shared_data allocation to use a map_node so that the sentinel hscb can use this map node in ahd_swap_with_next_hscb. This routine now swaps the hscb_map pointer in additon to the hscb contents so that any sync operations occur on the correct map. physaddr -> busaddr Pointed out by: Jason Thorpe <thorpej@wasabisystems.com> aic79xx.c: Make more use of the in/out/w/l/q macros for accessing byte registers in the chip. Correct some issues in the ahd_flush_qoutfifo() routine. o Run the qoutfifo only once the command channel DMA engine has been halted. This closes a window where we might have missed some entries. o Change ahd_run_data_fifo() to not loop to completion. If we happen to start on the wrong FIFO and the other FIFO has a snapshot savepointers, we might deadlock. This required our delay between FIFO tests to be moved to the ahd_flush_qoutfifo() routine. o Update/add comments. o Remove spurious test for COMPLETE_DMA list being empty when completing transactions from the GSFIFO with residuals. The SCB must be put on the COMPLETE_DMA scb list unconditionally. o When halting command channel DMA activity, we must disable the DMA channel in all cases but an update of the QOUTFIFO. The latter case is required so that the sequencer will update its position in the QOUTFIFO. Previously, we left the channel enabled for all "push" DMAs. This left us vulnerable to the sequencer handling an SCB push long after that SCB was already processed manually by this routine. o Correct the polarity of tests involving ahd_scb_active_in_fifo(). This routine returns non-zero for true. Return to processing bad status completions through the qoutfifo. This reduces the time that the sequencer is kept paused when handling transactions with bad status or underruns. When waiting for the controller to quiece selections, add a delay to our loop. Otherwise we may fail to wait long enough for the sequencer to comply. On H2A4 hardware, use the slow slewrate for non-paced transfers. This mirrors what the Adaptec Windows drivers do. On the Rev B. only slow down the CRC timing for older U160 devices that might need the slower timing. We define "older" as devices that do not support packetized protocol. Wait up to 5000 * 5us for the SEEPROM to become unbusy. Write ops seem to take much longer than read ops. aic79xx.seq: For controllers with the FAINT_LED bug, turn the diagnostic led feature on during selection and reselection. This covers the non-packetized case. The LED will be disabled for non-packetized transfers once we return to the top level idle loop. Add more comments about the busy LED workaround. Extend a critical section around the entire command channel idle loop process. Previously the portion of this handler that directly manipulated the linked list of completed SCBs was not protected. This is the likely cause of the recent reports of commands being completed twice by the driver. Extend critical sections across the test for, and the longjump to, longjump routines. This prevents the firmware from trying to jump to a longjmp handler that was just cleared by the host. Improve the locations of several critical section begin and end points. Typically these changes remove instructions that did not need to be inside a critical section. Close the "busfree after selection, but before busfree interrupts can be enabled" race to just a single sequencer instruction. We now test the BSY line explicitly before clearing the busfree status and enabling the busfree interrupt. Close a race condition in the processing of HS_MAILBOX updates. We now clear the "updated" status before the copy. This ensures that we don't accidentally clear the status incorrectly when the host sneaks in an update just after our last copy, but before we clear the status. This race has never been observed. Don't re-enable SCSIEN if we lose the race to disable SCSIEN in our interrupt handler's workaround for the RevA data-valid too early issue. aic79xx_inline.h: Add comments indicating that the order in which bytes are read or written in ahd_inw and ahd_outw is important. This allows us to use these inlines when accessing registers with side-effects. aic79xx_pci.c: The 29320 and the 29320B are 7902 not 7901 based products. Correct the driver banner. aic7xxx.h: Enable the use of the auto-access pause feature on the aic7870 and aic7880. It was disabled due to an oversight. aic7xxx.reg: Move TARG_IMMEDIATE_SCB to alias LAST_MSG to avoid leaving garbage in MWI_RESIDUAL. This prevents spurious overflows whn operating target mode on controllers that require the MWI_RESIDUAL work-around. aic7xxx.seq: AHC_TMODE_WIDEODD_BUG is a bug, not a softc flag. Reference the correct softc field when testing for its presence. Set the NOT_IDENTIFIED and NO_CDB_SENT bits in SEQ_FLAGS to indicate that the nexus is invalid in await busfree. aic7xxx_93cx6.c: Add support for the C56/C66 versions of the EWEN and EWDS commands. aic7xxx.c: aic7xxx_pci.c: Move test for the validity of left over BIOS data to ahc_test_register_access(). This guarantees that any left over CHIPRST value is not clobbered by our register access test and lost to the test that was in ahc_reset.
2003-12-17 00:02:10 +00:00
/* clear target specific flags */
mvi SEQ_FLAGS, NOT_IDENTIFIED|NO_CDB_SENT;
test SSTAT1,REQINIT|BUSFREE jz .;
ahc_eisa.c: ahc_pci.c: ahd_pci.c: aic7xxx.c: aic79xx.c: aic_osm_lib.c: aic_osm_lib.h: Use common OSM routines from aic_osm_lib for bus dma operations, delay routines, accessing CCBs, byte swapping, etc. aic7xxx_pci.c: Provide a better description for the 2915/30LP on attach. aic7xxx.c: aic79xx.c: aic7770.c: aic79xx_pci.c: aic7xxx_pci.c: aic7xxx_93cx6.c: Move FBSDID behind an ifdef so that these core files will still compile under other OSes. aic79xx.h: aic79xx_pci.c: aic79xx.seq: To speed up non-packetized CDB delivery in Rev B, all CDB acks are "released" to the output sync as soon as the command phase starts. There is only one problem with this approach. If the target changes phase before all data are sent, we have left over acks that can go out on the bus in a data phase. Due to other chip contraints, this only happens if the target goes to data-in, but if the acks go out before we can test SDONE, we'll think that the transfer has completed successfully. Work around this by taking advantage of the 400ns or 800ns dead time between command phase and the REQ of the new phase. If the transfer has completed successfully, SCSIEN should fall *long* before we see a phase change. We thus treat any phasemiss that occurs before SCSIEN falls as an incomplete transfer. aic79xx.h: Add the AHD_FAST_CDB_DELIVERY feature. aic79xx_pci.c: Set AHD_FAST_CDB_DELIVERY for all Rev. B parts. aic79xx.seq: Test for PHASEMIS in the command phase for all AHD_FAST_CDB_DELIVERY controlelrs. ahd_pci.c: ahc_pci.c: aic7xxx.h: aic79xx.h: Move definition of controller BAR offsets to core header files. aic7xxx.c: aic79xx.c: In the softc free routine, leave removal of a softc from the global list of softcs to the OSM (the caller of this routine). This allows us to avoid holding the softc list_lock during device destruction where we may have to sleep waiting for our recovery thread to halt. ahc_pci.c: Use ahc_pci_test_register access to validate I/O mapped in addition to the tests already performed for memory mapped access. Remove unused ahc_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. ahd_pci.c: Remove reduntant definition of controller BAR offsets. These are also defined in aic79xx.h. Remove unused ahd_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. aic7xxx.c: aic79xx.c: aic79xx.h: aic7xxx.h: aic7xxx_osm.c: aic79xx_osm.c: Move timeout handling to the driver cores. In the case of the aic79xx driver, the algorithm has been enhanced to try target resets before performing a bus reset. For the aic7xxx driver, the algorithm is unchanged. Although the drivers do not currently sleep during recovery (recovery is timeout driven), the cores do expect all processing to be performed via a recovery thread. Our timeout handlers are now little stubs that wakeup the recovery thread. aic79xx.c: aic79xx.h: aic79xx_inline.h: Change shared_data allocation to use a map_node so that the sentinel hscb can use this map node in ahd_swap_with_next_hscb. This routine now swaps the hscb_map pointer in additon to the hscb contents so that any sync operations occur on the correct map. physaddr -> busaddr Pointed out by: Jason Thorpe <thorpej@wasabisystems.com> aic79xx.c: Make more use of the in/out/w/l/q macros for accessing byte registers in the chip. Correct some issues in the ahd_flush_qoutfifo() routine. o Run the qoutfifo only once the command channel DMA engine has been halted. This closes a window where we might have missed some entries. o Change ahd_run_data_fifo() to not loop to completion. If we happen to start on the wrong FIFO and the other FIFO has a snapshot savepointers, we might deadlock. This required our delay between FIFO tests to be moved to the ahd_flush_qoutfifo() routine. o Update/add comments. o Remove spurious test for COMPLETE_DMA list being empty when completing transactions from the GSFIFO with residuals. The SCB must be put on the COMPLETE_DMA scb list unconditionally. o When halting command channel DMA activity, we must disable the DMA channel in all cases but an update of the QOUTFIFO. The latter case is required so that the sequencer will update its position in the QOUTFIFO. Previously, we left the channel enabled for all "push" DMAs. This left us vulnerable to the sequencer handling an SCB push long after that SCB was already processed manually by this routine. o Correct the polarity of tests involving ahd_scb_active_in_fifo(). This routine returns non-zero for true. Return to processing bad status completions through the qoutfifo. This reduces the time that the sequencer is kept paused when handling transactions with bad status or underruns. When waiting for the controller to quiece selections, add a delay to our loop. Otherwise we may fail to wait long enough for the sequencer to comply. On H2A4 hardware, use the slow slewrate for non-paced transfers. This mirrors what the Adaptec Windows drivers do. On the Rev B. only slow down the CRC timing for older U160 devices that might need the slower timing. We define "older" as devices that do not support packetized protocol. Wait up to 5000 * 5us for the SEEPROM to become unbusy. Write ops seem to take much longer than read ops. aic79xx.seq: For controllers with the FAINT_LED bug, turn the diagnostic led feature on during selection and reselection. This covers the non-packetized case. The LED will be disabled for non-packetized transfers once we return to the top level idle loop. Add more comments about the busy LED workaround. Extend a critical section around the entire command channel idle loop process. Previously the portion of this handler that directly manipulated the linked list of completed SCBs was not protected. This is the likely cause of the recent reports of commands being completed twice by the driver. Extend critical sections across the test for, and the longjump to, longjump routines. This prevents the firmware from trying to jump to a longjmp handler that was just cleared by the host. Improve the locations of several critical section begin and end points. Typically these changes remove instructions that did not need to be inside a critical section. Close the "busfree after selection, but before busfree interrupts can be enabled" race to just a single sequencer instruction. We now test the BSY line explicitly before clearing the busfree status and enabling the busfree interrupt. Close a race condition in the processing of HS_MAILBOX updates. We now clear the "updated" status before the copy. This ensures that we don't accidentally clear the status incorrectly when the host sneaks in an update just after our last copy, but before we clear the status. This race has never been observed. Don't re-enable SCSIEN if we lose the race to disable SCSIEN in our interrupt handler's workaround for the RevA data-valid too early issue. aic79xx_inline.h: Add comments indicating that the order in which bytes are read or written in ahd_inw and ahd_outw is important. This allows us to use these inlines when accessing registers with side-effects. aic79xx_pci.c: The 29320 and the 29320B are 7902 not 7901 based products. Correct the driver banner. aic7xxx.h: Enable the use of the auto-access pause feature on the aic7870 and aic7880. It was disabled due to an oversight. aic7xxx.reg: Move TARG_IMMEDIATE_SCB to alias LAST_MSG to avoid leaving garbage in MWI_RESIDUAL. This prevents spurious overflows whn operating target mode on controllers that require the MWI_RESIDUAL work-around. aic7xxx.seq: AHC_TMODE_WIDEODD_BUG is a bug, not a softc flag. Reference the correct softc field when testing for its presence. Set the NOT_IDENTIFIED and NO_CDB_SENT bits in SEQ_FLAGS to indicate that the nexus is invalid in await busfree. aic7xxx_93cx6.c: Add support for the C56/C66 versions of the EWEN and EWDS commands. aic7xxx.c: aic7xxx_pci.c: Move test for the validity of left over BIOS data to ahc_test_register_access(). This guarantees that any left over CHIPRST value is not clobbered by our register access test and lost to the test that was in ahc_reset.
2003-12-17 00:02:10 +00:00
/*
* We only set BUSFREE status once either a new
* phase has been detected or we are really
* BUSFREE. This allows the driver to know
* that we are active on the bus even though
* no identified transaction exists should a
* timeout occur while awaiting busfree.
*/
mvi LASTPHASE, P_BUSFREE;
test SSTAT1, BUSFREE jnz idle_loop;
SET_SEQINTCODE(MISSED_BUSFREE)
/*
* Save data pointers message:
* Copying RAM values back to SCB, for Save Data Pointers message, but
* only if we've actually been into a data phase to change them. This
* protects against bogus data in scratch ram and the residual counts
* since they are only initialized when we go into data_in or data_out.
* Ack the message as soon as possible.
*/
SET_SRC_MODE M_DFF1;
SET_DST_MODE M_DFF1;
mesgin_sdptrs:
mov NONE,SCSIDAT; /*dummy read from latch to ACK*/
test SEQ_FLAGS, DPHASE jz ITloop;
call save_pointers;
jmp ITloop;
save_pointers:
/*
* If we are asked to save our position at the end of the
* transfer, just mark us at the end rather than perform a
* full save.
*/
test SCB_RESIDUAL_SGPTR[0], SG_LIST_NULL jz save_pointers_full;
or SCB_SGPTR, SG_LIST_NULL ret;
save_pointers_full:
/*
* The SCB_DATAPTR becomes the current SHADDR.
* All other information comes directly from our residual
* state.
*/
bmov SCB_DATAPTR, SHADDR, 8;
bmov SCB_DATACNT, SCB_RESIDUAL_DATACNT, 8 ret;
/*
* Restore pointers message? Data pointers are recopied from the
* SCB anytime we enter a data phase for the first time, so all
* we need to do is clear the DPHASE flag and let the data phase
* code do the rest. We also reset/reallocate the FIFO to make
* sure we have a clean start for the next data or command phase.
*/
mesgin_rdptrs:
and SEQ_FLAGS, ~DPHASE;
test MODE_PTR, ~(MK_MODE(M_DFF1, M_DFF1)) jnz msgin_rdptrs_get_fifo;
mvi DFFSXFRCTL, RSTCHN|CLRSHCNT;
SET_MODE(M_SCSI, M_SCSI)
msgin_rdptrs_get_fifo:
call allocate_fifo;
jmp mesgin_done;
phase_lock:
aic79xx.c: Clear the LQICRC_NLQ status should it pop up after we have already handled the SCSIPERR. During some streaming operations this status can be delayed until the stream ends. Without this change, the driver would complain about a "Missing case in ahd_handle_scsiint". In the LQOBUSFREE handler... Don't return the LQOMGR back to the idle state until after we have cleaned up ENSELO and any status related to this selection. The last thing we need is the LQO manager starting another select-out before we have updated the execution queue. It is not clear whether the LQOMGR would, or would not start a new selection early. Make sure ENSELO is off prior to clearing SELDO by flushing device writes. Move assignment of the next target SCB pointer inside of an if to make the code clearer. The effect is the same. Dump card state in both "Unexpected PKT busfree" paths. In ahd_reset(), set the chip to SCSI mode before reading SXFRCTL1. That register only exists in the SCSI mode. Also set the mode explicitly to the SCSI mode after chip reset due to paranoia. Re-arrange code so that SXFRCTL1 is restored as quickly after the chip reset as possible. S/G structurs must be 8byte aligned. Make this official by saying so in our DMA tag. Disable CIO bus stretch on MDFFSTAT if SHVALID is about to come true. This can cause a CIO bus lockup if a PCI or PCI-X error occurs while the stretch is occurring - the host cannot service the PCI-X error since the CIO bus is locked out and SHVALID will never resolve. The stretch was added in the Rev B to simplify the wait for SHVALID to resolve, but the code to do this in the open source sequencer is so simple it was never removed. Consistently use MAX_OFFSET for the user max syncrate set from non-volatile storage. This ensures that the offset does not conflict with AH?_OFFSET_UNKNOWN. Have ahd_pause_and_flushwork set the mode to ensure that it has access to the registers it checks. Also modify the checking of intstat so that the check against 0xFF can actually succeed if the INT_PEND mask is something other than 0xFF. Although there are no cardbus U320 controllers, this check may be needed to recover from a hot-plug PCI removal that occurs without informing the driver. Fix a typo. sg_prefetch_cnt -> sg_prefetch_align. This fixes an infinite loop at card initialization if the cacheline size is 0. aic79xx.h: Add AHD_EARLY_REQ_BUG bug flag. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. Change AH?_OFFSET_UNKNOWN to 0xFF. This is a value that the curr->offset can never be, unlike '0' which we previously used. This fixes code that only checks for a non-zero offset to determine if a sync negotiation is required since it will fire in the unknown case even if the goal is async. aic79xx.reg: Add comments for LQISTAT bits indicating their names in the 7902 data book. We use slightly different and more descriptive names in the firmware. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. aic79xx.seq: Update comments regarding rundown of the GSFIFO to reflect reality. Fix spelling errors. Since we use an 8byte address and 1 byte length, shorten the size of a block move for the legacy DMA'ed CDB case from 11 to 9 bytes. Remove code that, assuming the abort pending feature worked, would set MK_MESSAGE in the SCB's control byte on completion to catch invalid reselections. Since we don't see interrupts for completed selections, this status update could occur prior to us noticing the SELDO. The "select-out" queue logic will get confused by the MK_MESSAGE bit being set as this is used to catch packatized connections where we select-out with ATN. Since the abort pending feature doesn't work on any released controllers yet, this code was never executed. Add support for the AHD_EARLY_REQ_BUG. Don't ignore persistent REQ assertions just because they were asserted within the bus settle delay window. This allows us to tolerate devices like the GEM318 that violate the SCSI spec. Remove unintentional settnig of SG_CACHE_AVAIL. Writing this bit should have no effect, but who knows... On the Rev A, we must wait for HDMAENACK before loading additional segments to avoid clobbering the address of the first segment in the S/G FIFO. This resolves data-corruption issues with certain IBM (now Hitachi) and Fujitsu U320 drives. Rearrange calc_residual to avoid an extra jmp instruction. On RevA Silicon, if the target returns us to data-out after we have already trained for data-out, it is possible for us to transition the free running clock to data-valid before the required 100ns P1 setup time (8 P1 assertions in fast-160 mode). This will only happen if this L-Q is a continuation of a data transfer for which we have already prefetched data into our FIFO (LQ/Data followed by LQ/Data for the same write transaction). This can cause some target implementations to miss the first few data transfers on the bus. We detect this situation by noticing that this is the first data transfer after an LQ (LQIWORKONLQ true), that the data transfer is a continuation of a transfer already setup in our FIFO (SAVEPTRS interrupt), and that the transaction is a write (DIRECTION set in DFCNTRL). The delay is performed by disabling SCSIEN until we see the first REQ from the target. Only compile in snapshot savepointers handler for RevA silicon where it is enabled. Handle the cfg4icmd packetized interrupt. We just need to load the address and count, start the DMA, and CLRCHN once the transfer is complete. Fix an oversight in the overrun handler for packetized status operations. We need to wait for either CTXTDONE or an overrun when checking for an overrun. The previous code did not wait and thus could decide that no overrun had occurred even though an overrun will occur on the next data-valid req. Add some comment to this section for clarity. Use LAST_SEG_DONE instead of LASTSDONE for testing transfer completion in the packetized status case. LASTSDONE may come up more quickly since it only records completion on the SCSI side, but since LAST_SEG_DONE is used everywhere else (and needs to be), this is less confusing. Add a missing invalidation of the longjmp address in the non-pack handler. This code needs additional review. aic79xx_inline.h: Fix spelling error. aic79xx_osm.c: Set the cdb length for CDBs dma'ed from host memory. Add a comment indicating that, should CAM start supporting cdbs larger than 16bytes, the driver could store the CDB in the status buffer. aic79xx_pci.c: Add a table entry for the 39320A. Added a missing comma to an error string table. Fix spelling errors.
2003-02-27 23:23:16 +00:00
if ((ahd->bugs & AHD_EARLY_REQ_BUG) != 0) {
/*
* Don't ignore persistent REQ assertions just because
* they were asserted within the bus settle delay window.
* This allows us to tolerate devices like the GEM318
* that violate the SCSI spec. We are careful not to
* count REQ while we are waiting for it to fall during
* an async phase due to our asserted ACK. Each
* sequencer instruction takes ~25ns, so the REQ must
* last at least 100ns in order to be counted as a true
* REQ.
*/
test SCSIPHASE, 0xFF jnz phase_locked;
test SCSISIGI, ACKI jnz phase_lock;
test SCSISIGI, REQI jz phase_lock;
test SCSIPHASE, 0xFF jnz phase_locked;
test SCSISIGI, ACKI jnz phase_lock;
test SCSISIGI, REQI jz phase_lock;
phase_locked:
} else {
test SCSIPHASE, 0xFF jz .;
}
test SSTAT1, SCSIPERR jnz phase_lock;
phase_lock_latch_phase:
and LASTPHASE, PHASE_MASK, SCSISIGI ret;
/*
* Functions to read data in Automatic PIO mode.
*
* An ACK is not sent on input from the target until SCSIDATL is read from.
* So we wait until SCSIDATL is latched (the usual way), then read the data
* byte directly off the bus using SCSIBUSL. When we have pulled the ATN
* line, or we just want to acknowledge the byte, then we do a dummy read
* from SCISDATL. The SCSI spec guarantees that the target will hold the
* data byte on the bus until we send our ACK.
*
* The assumption here is that these are called in a particular sequence,
* and that REQ is already set when inb_first is called. inb_{first,next}
* use the same calling convention as inb.
*/
inb_next:
mov NONE,SCSIDAT; /*dummy read from latch to ACK*/
inb_next_wait:
/*
* If there is a parity error, wait for the kernel to
* see the interrupt and prepare our message response
* before continuing.
*/
test SCSIPHASE, 0xFF jz .;
test SSTAT1, SCSIPERR jnz inb_next_wait;
inb_next_check_phase:
and LASTPHASE, PHASE_MASK, SCSISIGI;
cmp LASTPHASE, P_MESGIN jne mesgin_phasemis;
inb_first:
clr DINDEX[1];
mov DINDEX,SINDEX;
mov DINDIR,SCSIBUS ret; /*read byte directly from bus*/
inb_last:
mov NONE,SCSIDAT ret; /*dummy read from latch to ACK*/
mk_mesg:
mvi SCSISIGO, ATNO;
mov MSG_OUT,SINDEX ret;
SET_SRC_MODE M_DFF1;
SET_DST_MODE M_DFF1;
disable_ccsgen:
test SG_STATE, FETCH_INPROG jz disable_ccsgen_fetch_done;
clr CCSGCTL;
disable_ccsgen_fetch_done:
clr SG_STATE ret;
service_fifo:
/*
* Do we have any prefetch left???
*/
test SG_STATE, SEGS_AVAIL jnz idle_sg_avail;
/*
* Can this FIFO have access to the S/G cache yet?
*/
test CCSGCTL, SG_CACHE_AVAIL jz return;
/* Did we just finish fetching segs? */
test CCSGCTL, CCSGDONE jnz idle_sgfetch_complete;
/* Are we actively fetching segments? */
test CCSGCTL, CCSGENACK jnz return;
aic79xx.c: aic79xx.seq: Convert the COMPLETE_DMA_SCB list to an "stailq". This allows us to safely keep the SCB that is currently being DMA'ed back the host on the head of the list while processing completions off of the bus. The newly completed SCBs are appended to the tail of the queue. In the past, we just dequeued the SCB that was in flight from the list, but this could result in a lost completion should the host perform certain types of error recovery that must cancel all in-flight SCB DMA operations. Switch from using a 16bit completion entry, holding just the tag and the completion valid bit, to a 64bit completion entry that also contains a "status packet valid" indicator. This solves two problems: o The SCB DMA engine on at least Rev B. silicon does not properly deal with a PCI disconnect that occurs at a non-64bit aligned offset in the chips "source buffer". When the transfer is resumed, the DMA engine continues at the correct offset, but may wrap to the head of the buffer causing duplicate completions to be reported to the host. By using a completion buffer in host memory that is 64bit aligned and using 64bit completion entries, such disconnects should only occur at aligned addresses. This assumes that the host bridge will only disconnect on cache-line boundaries and that cache-lines are multpiles of 64bits. o By embedding the status information in the completion entry we can avoid an extra memory reference to the HSCB for commands that complete without error. Use the comparison of a "host freeze count" and a "sequencer freeze count" to allow the host to process most SCBs that complete with non-zero status without having to clear critical sections. Instead the host can just pause the sequencer, performs any necessary cleanup in the waiting for selection list, increments its freeze count on the controller, and unpauses. This is only possible because the sequencer defers completions of SCBs with bad status until after all pending selections have completed. The sequencer then avoids referencing any data structures the host may touch during completion of the SCB until the freeze counts match. aic79xx.c: Change the strategy for allocating our sentinal HSCB for the QINFIFO. In the past, this allocation was tacked onto the QOUTFIFO allocation. Now that the qoutfifo has grown to accomodate larger completion entries, the old approach will result in a 64byte allocation that costs an extra page of coherent memory. We now do this extra allocation via ahd_alloc_scbs() where the "unused space" can be used to allocate "normal" HSCBs. In our packetized busfree handler, use the ENSELO bit to differentiate between packetized and non-packetized unexpected busfree events that occur just after selection, but before the sequencer has had the oportunity to service the selection. When cleaning out the waiting for selection list, use the SCSI mode instead of the command channel mode. The SCB pointer in the command channel mode may be referenced by the SCB dma engine even while the sequencer is paused, whereas the SCSI mode SCB pointer is only accessed by the sequencer. Print the "complete on qfreeze" sequencer SCB completion list in ahd_dump_card_state(). This list holds all SCB completions that are deferred until a pending select-out qfreeze event has taken effect. aic79xx.h: Add definitions and structures to handle the new SCB completion scheme. Add a controller flag that indicates if the controller is in HostRAID mode. aic79xx.reg: Remove macros used for toggling from one data fifo mode to the other. They have not been in use for some time. Add scratch ram fields for our new qfreeze count scheme, converting the complete dma list into an "stailq", and providing for the "complete on qfreeze" SCB completion list. Some other fields were moved to retain proper field alignment (alignment >= field size in bytes). aic79xx.seq: Add code to our idle loop to: o Process deferred completions once a qfreeze event has taken full effect. o Thaw the queue once the sequencer and host qfreeze counts match. Generate 64bit completion entries passing the SCB_SGPTR field as the "good status" indicator. The first bit in this field is only set if we have a valid status packet to send to the host. Convert the COMPLETE_DMA_SCB list to an "stailq". When using "setjmp" to register an idle loop handler, do not combine the "ret" with the block move to pop the stack address in the same instruction. At least on the A, this results in a return to the setjmp caller, not to the new address at the top of the stack. Since we want the latter (we want the newly registered handler to only be invoked from the idle loop), we must use a separate ret instruction. Add a few missing critical sections. Close a race condition that can occur on Rev A. silicon. If both FIFOs happen to be allocated before the sequencer has a chance to service the FIFO that was allocated first, we must take special care to service the FIFO that is not active on the SCSI bus first. This guarantees that a FIFO will be freed to handle any snapshot requests for the FIFO that is still on the bus. Chosing the incorrect FIFO will result in deadlock. Update comments. aic79xx_inline.h Correct the offset calculation for the syncing of our qoutfifo. Update ahd_check_cmdcmpltqueues() for the larger completion entries. aic79xx_pci.c: Attach to HostRAID controllers by default. In the future I may add a sysctl to modify the behavior, but since FreeBSD does not have any HostRAID drivers, failing to attach just results in more email and bug reports for the author. MFC After: 1week
2004-02-04 16:38:38 +00:00
/*
* Should the other FIFO get the S/G cache first? If
* both FIFOs have been allocated since we last checked
* any FIFO, it is important that we service a FIFO
* that is not actively on the bus first. This guarantees
* that a FIFO will be freed to handle snapshot requests for
* any FIFO that is still on the bus. Chips with RTI do not
* perform snapshots, so don't bother with this test there.
*/
if ((ahd->features & AHD_RTI) == 0) {
/*
* If we're not still receiving SCSI data,
* it is safe to allocate the S/G cache to
* this FIFO.
*/
test DFCNTRL, SCSIEN jz idle_sgfetch_start;
/*
* Switch to the other FIFO. Non-RTI chips
* also have the "set mode" bug, so we must
* disable interrupts during the switch.
*/
mvi SEQINTCTL, INTVEC1DSL;
xor MODE_PTR, MK_MODE(M_DFF1, M_DFF1);
/*
* If the other FIFO needs loading, then it
* must not have claimed the S/G cache yet
* (SG_CACHE_AVAIL would have been cleared in
* the orginal FIFO mode and we test this above).
* Return to the idle loop so we can process the
* FIFO not currently on the bus first.
*/
test SG_STATE, LOADING_NEEDED jz idle_sgfetch_okay;
clr SEQINTCTL ret;
idle_sgfetch_okay:
xor MODE_PTR, MK_MODE(M_DFF1, M_DFF1);
clr SEQINTCTL;
}
idle_sgfetch_start:
/*
* We fetch a "cacheline aligned" and sized amount of data
* so we don't end up referencing a non-existant page.
* Cacheline aligned is in quotes because the kernel will
* set the prefetch amount to a reasonable level if the
* cacheline size is unknown.
*/
bmov SGHADDR, SCB_RESIDUAL_SGPTR, 4;
Upgrade to version 1.1 of the aic79xx U320 driver. aic79xx.c: o Remove redundant ahd_update_modes() call. o Correct panic in diagnostic should state corruption cause the SCB Id to be invalid during a selection timeout. o Add workaround for missing BUSFREEREV feature in Rev A silicon. o Corect formatting nits. o Use register pretty printing in more places. o Save and restore our SCB pointer when updating the waiting queue list for an "expected" LQ-out busfree. o In ahd_clear_intstat, deal with the missing autoclear in the CLRLQO* registers. o BE fixup in a diagnostic printf. o Make sure that we are in the proper mode before disabling selections in ahd_update_pending_scbs. o Add more diagnostics. o task_attribute_nonpkt_tag -> task_attribute: we don't need a nonpkt_tag field anymore for allowing all 512 SCBs to be used in non-packetized connections. o Negotiate HOLD_MCS to U320 devices. o Add a few additional mode assertions. o Restore the chip mode after clearing out the qinfifo so that code using ahd_abort_scbs sees a consistent mode. o Simplify the DMA engine shutdown routine prior to performing a bus reset. o Perform the sequencer restart after a chip reset prior to setting up our timer to poll for the reset to be complete. On some OSes, the timer could actually pre-empt us and order is important here. o Have our "reset poller" set the expected mode since there is no guarantee of what mode will be in force when we are called from the OS timer. o Save and restore the SCB pointer in ahd_dump_card_state(). This routine must not modify card state. o Ditto for ahd_dump_scbs(). aic79xx.h: o Add a few more chip bug definitions. o Align our tag on a 32bit boundary. aic79xx.reg: aic79xx.seq: o Start work on removing workarounds for Rev B. o Use a special location in scratch from for stroring our SCBPTR during legacy FIFO allocations. This corrects problems in mixed packetized/non-packetized configurations where calling into a FIFO task corrupted our SCBPTR. o Don't rely on DMA priority to guarantee that all data in our FIFOs will flush prior to a command completion notification going out of the command channel. We've never seen this assumption fail, but better safe than sorry. o Deal with missing BUSFREEREV feature in H2A. o Simplify disconnect list code now that the list will always have only a single entry. o Implement the AHD_REG_SLOW_SETTLE_BUG workaround. o Swith to using "REG_ISR" for local mode scratch during our ISR. o Add a missing jmp to the data_group_dma_loop after our data pointers have been re-initialized by the kernel. o Correct test in the bitbucket code so that we actually wait for the bitbucket to complete before signaling the kernel of the overrun condition. o Reposition pkt_saveptrs to avoid a jmp instruction. o Update a comment to reflect that the code now waits for a FIFO to drain prior to issuing a CLRCHN. aic79xx_inline.h: o Remove unused untagged queue handling code. o Don't attempt to htole64 what could be a 32bit value. aic79xx_pci.c: o Set additional bug flags for rev A chips.
2002-09-26 22:54:00 +00:00
mvi SGHCNT, SG_PREFETCH_CNT;
if ((ahd->bugs & AHD_REG_SLOW_SETTLE_BUG) != 0) {
/*
aic79xx.c: aic79xx.seq: Convert the COMPLETE_DMA_SCB list to an "stailq". This allows us to safely keep the SCB that is currently being DMA'ed back the host on the head of the list while processing completions off of the bus. The newly completed SCBs are appended to the tail of the queue. In the past, we just dequeued the SCB that was in flight from the list, but this could result in a lost completion should the host perform certain types of error recovery that must cancel all in-flight SCB DMA operations. Switch from using a 16bit completion entry, holding just the tag and the completion valid bit, to a 64bit completion entry that also contains a "status packet valid" indicator. This solves two problems: o The SCB DMA engine on at least Rev B. silicon does not properly deal with a PCI disconnect that occurs at a non-64bit aligned offset in the chips "source buffer". When the transfer is resumed, the DMA engine continues at the correct offset, but may wrap to the head of the buffer causing duplicate completions to be reported to the host. By using a completion buffer in host memory that is 64bit aligned and using 64bit completion entries, such disconnects should only occur at aligned addresses. This assumes that the host bridge will only disconnect on cache-line boundaries and that cache-lines are multpiles of 64bits. o By embedding the status information in the completion entry we can avoid an extra memory reference to the HSCB for commands that complete without error. Use the comparison of a "host freeze count" and a "sequencer freeze count" to allow the host to process most SCBs that complete with non-zero status without having to clear critical sections. Instead the host can just pause the sequencer, performs any necessary cleanup in the waiting for selection list, increments its freeze count on the controller, and unpauses. This is only possible because the sequencer defers completions of SCBs with bad status until after all pending selections have completed. The sequencer then avoids referencing any data structures the host may touch during completion of the SCB until the freeze counts match. aic79xx.c: Change the strategy for allocating our sentinal HSCB for the QINFIFO. In the past, this allocation was tacked onto the QOUTFIFO allocation. Now that the qoutfifo has grown to accomodate larger completion entries, the old approach will result in a 64byte allocation that costs an extra page of coherent memory. We now do this extra allocation via ahd_alloc_scbs() where the "unused space" can be used to allocate "normal" HSCBs. In our packetized busfree handler, use the ENSELO bit to differentiate between packetized and non-packetized unexpected busfree events that occur just after selection, but before the sequencer has had the oportunity to service the selection. When cleaning out the waiting for selection list, use the SCSI mode instead of the command channel mode. The SCB pointer in the command channel mode may be referenced by the SCB dma engine even while the sequencer is paused, whereas the SCSI mode SCB pointer is only accessed by the sequencer. Print the "complete on qfreeze" sequencer SCB completion list in ahd_dump_card_state(). This list holds all SCB completions that are deferred until a pending select-out qfreeze event has taken effect. aic79xx.h: Add definitions and structures to handle the new SCB completion scheme. Add a controller flag that indicates if the controller is in HostRAID mode. aic79xx.reg: Remove macros used for toggling from one data fifo mode to the other. They have not been in use for some time. Add scratch ram fields for our new qfreeze count scheme, converting the complete dma list into an "stailq", and providing for the "complete on qfreeze" SCB completion list. Some other fields were moved to retain proper field alignment (alignment >= field size in bytes). aic79xx.seq: Add code to our idle loop to: o Process deferred completions once a qfreeze event has taken full effect. o Thaw the queue once the sequencer and host qfreeze counts match. Generate 64bit completion entries passing the SCB_SGPTR field as the "good status" indicator. The first bit in this field is only set if we have a valid status packet to send to the host. Convert the COMPLETE_DMA_SCB list to an "stailq". When using "setjmp" to register an idle loop handler, do not combine the "ret" with the block move to pop the stack address in the same instruction. At least on the A, this results in a return to the setjmp caller, not to the new address at the top of the stack. Since we want the latter (we want the newly registered handler to only be invoked from the idle loop), we must use a separate ret instruction. Add a few missing critical sections. Close a race condition that can occur on Rev A. silicon. If both FIFOs happen to be allocated before the sequencer has a chance to service the FIFO that was allocated first, we must take special care to service the FIFO that is not active on the SCSI bus first. This guarantees that a FIFO will be freed to handle any snapshot requests for the FIFO that is still on the bus. Chosing the incorrect FIFO will result in deadlock. Update comments. aic79xx_inline.h Correct the offset calculation for the syncing of our qoutfifo. Update ahd_check_cmdcmpltqueues() for the larger completion entries. aic79xx_pci.c: Attach to HostRAID controllers by default. In the future I may add a sysctl to modify the behavior, but since FreeBSD does not have any HostRAID drivers, failing to attach just results in more email and bug reports for the author. MFC After: 1week
2004-02-04 16:38:38 +00:00
* Need two instructions between "touches" of SGHADDR.
Upgrade to version 1.1 of the aic79xx U320 driver. aic79xx.c: o Remove redundant ahd_update_modes() call. o Correct panic in diagnostic should state corruption cause the SCB Id to be invalid during a selection timeout. o Add workaround for missing BUSFREEREV feature in Rev A silicon. o Corect formatting nits. o Use register pretty printing in more places. o Save and restore our SCB pointer when updating the waiting queue list for an "expected" LQ-out busfree. o In ahd_clear_intstat, deal with the missing autoclear in the CLRLQO* registers. o BE fixup in a diagnostic printf. o Make sure that we are in the proper mode before disabling selections in ahd_update_pending_scbs. o Add more diagnostics. o task_attribute_nonpkt_tag -> task_attribute: we don't need a nonpkt_tag field anymore for allowing all 512 SCBs to be used in non-packetized connections. o Negotiate HOLD_MCS to U320 devices. o Add a few additional mode assertions. o Restore the chip mode after clearing out the qinfifo so that code using ahd_abort_scbs sees a consistent mode. o Simplify the DMA engine shutdown routine prior to performing a bus reset. o Perform the sequencer restart after a chip reset prior to setting up our timer to poll for the reset to be complete. On some OSes, the timer could actually pre-empt us and order is important here. o Have our "reset poller" set the expected mode since there is no guarantee of what mode will be in force when we are called from the OS timer. o Save and restore the SCB pointer in ahd_dump_card_state(). This routine must not modify card state. o Ditto for ahd_dump_scbs(). aic79xx.h: o Add a few more chip bug definitions. o Align our tag on a 32bit boundary. aic79xx.reg: aic79xx.seq: o Start work on removing workarounds for Rev B. o Use a special location in scratch from for stroring our SCBPTR during legacy FIFO allocations. This corrects problems in mixed packetized/non-packetized configurations where calling into a FIFO task corrupted our SCBPTR. o Don't rely on DMA priority to guarantee that all data in our FIFOs will flush prior to a command completion notification going out of the command channel. We've never seen this assumption fail, but better safe than sorry. o Deal with missing BUSFREEREV feature in H2A. o Simplify disconnect list code now that the list will always have only a single entry. o Implement the AHD_REG_SLOW_SETTLE_BUG workaround. o Swith to using "REG_ISR" for local mode scratch during our ISR. o Add a missing jmp to the data_group_dma_loop after our data pointers have been re-initialized by the kernel. o Correct test in the bitbucket code so that we actually wait for the bitbucket to complete before signaling the kernel of the overrun condition. o Reposition pkt_saveptrs to avoid a jmp instruction. o Update a comment to reflect that the code now waits for a FIFO to drain prior to issuing a CLRCHN. aic79xx_inline.h: o Remove unused untagged queue handling code. o Don't attempt to htole64 what could be a 32bit value. aic79xx_pci.c: o Set additional bug flags for rev A chips.
2002-09-26 22:54:00 +00:00
*/
nop;
}
and SGHADDR[0], SG_PREFETCH_ALIGN_MASK, SCB_RESIDUAL_SGPTR;
aic79xx.c: Clear the LQICRC_NLQ status should it pop up after we have already handled the SCSIPERR. During some streaming operations this status can be delayed until the stream ends. Without this change, the driver would complain about a "Missing case in ahd_handle_scsiint". In the LQOBUSFREE handler... Don't return the LQOMGR back to the idle state until after we have cleaned up ENSELO and any status related to this selection. The last thing we need is the LQO manager starting another select-out before we have updated the execution queue. It is not clear whether the LQOMGR would, or would not start a new selection early. Make sure ENSELO is off prior to clearing SELDO by flushing device writes. Move assignment of the next target SCB pointer inside of an if to make the code clearer. The effect is the same. Dump card state in both "Unexpected PKT busfree" paths. In ahd_reset(), set the chip to SCSI mode before reading SXFRCTL1. That register only exists in the SCSI mode. Also set the mode explicitly to the SCSI mode after chip reset due to paranoia. Re-arrange code so that SXFRCTL1 is restored as quickly after the chip reset as possible. S/G structurs must be 8byte aligned. Make this official by saying so in our DMA tag. Disable CIO bus stretch on MDFFSTAT if SHVALID is about to come true. This can cause a CIO bus lockup if a PCI or PCI-X error occurs while the stretch is occurring - the host cannot service the PCI-X error since the CIO bus is locked out and SHVALID will never resolve. The stretch was added in the Rev B to simplify the wait for SHVALID to resolve, but the code to do this in the open source sequencer is so simple it was never removed. Consistently use MAX_OFFSET for the user max syncrate set from non-volatile storage. This ensures that the offset does not conflict with AH?_OFFSET_UNKNOWN. Have ahd_pause_and_flushwork set the mode to ensure that it has access to the registers it checks. Also modify the checking of intstat so that the check against 0xFF can actually succeed if the INT_PEND mask is something other than 0xFF. Although there are no cardbus U320 controllers, this check may be needed to recover from a hot-plug PCI removal that occurs without informing the driver. Fix a typo. sg_prefetch_cnt -> sg_prefetch_align. This fixes an infinite loop at card initialization if the cacheline size is 0. aic79xx.h: Add AHD_EARLY_REQ_BUG bug flag. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. Change AH?_OFFSET_UNKNOWN to 0xFF. This is a value that the curr->offset can never be, unlike '0' which we previously used. This fixes code that only checks for a non-zero offset to determine if a sync negotiation is required since it will fire in the unknown case even if the goal is async. aic79xx.reg: Add comments for LQISTAT bits indicating their names in the 7902 data book. We use slightly different and more descriptive names in the firmware. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. aic79xx.seq: Update comments regarding rundown of the GSFIFO to reflect reality. Fix spelling errors. Since we use an 8byte address and 1 byte length, shorten the size of a block move for the legacy DMA'ed CDB case from 11 to 9 bytes. Remove code that, assuming the abort pending feature worked, would set MK_MESSAGE in the SCB's control byte on completion to catch invalid reselections. Since we don't see interrupts for completed selections, this status update could occur prior to us noticing the SELDO. The "select-out" queue logic will get confused by the MK_MESSAGE bit being set as this is used to catch packatized connections where we select-out with ATN. Since the abort pending feature doesn't work on any released controllers yet, this code was never executed. Add support for the AHD_EARLY_REQ_BUG. Don't ignore persistent REQ assertions just because they were asserted within the bus settle delay window. This allows us to tolerate devices like the GEM318 that violate the SCSI spec. Remove unintentional settnig of SG_CACHE_AVAIL. Writing this bit should have no effect, but who knows... On the Rev A, we must wait for HDMAENACK before loading additional segments to avoid clobbering the address of the first segment in the S/G FIFO. This resolves data-corruption issues with certain IBM (now Hitachi) and Fujitsu U320 drives. Rearrange calc_residual to avoid an extra jmp instruction. On RevA Silicon, if the target returns us to data-out after we have already trained for data-out, it is possible for us to transition the free running clock to data-valid before the required 100ns P1 setup time (8 P1 assertions in fast-160 mode). This will only happen if this L-Q is a continuation of a data transfer for which we have already prefetched data into our FIFO (LQ/Data followed by LQ/Data for the same write transaction). This can cause some target implementations to miss the first few data transfers on the bus. We detect this situation by noticing that this is the first data transfer after an LQ (LQIWORKONLQ true), that the data transfer is a continuation of a transfer already setup in our FIFO (SAVEPTRS interrupt), and that the transaction is a write (DIRECTION set in DFCNTRL). The delay is performed by disabling SCSIEN until we see the first REQ from the target. Only compile in snapshot savepointers handler for RevA silicon where it is enabled. Handle the cfg4icmd packetized interrupt. We just need to load the address and count, start the DMA, and CLRCHN once the transfer is complete. Fix an oversight in the overrun handler for packetized status operations. We need to wait for either CTXTDONE or an overrun when checking for an overrun. The previous code did not wait and thus could decide that no overrun had occurred even though an overrun will occur on the next data-valid req. Add some comment to this section for clarity. Use LAST_SEG_DONE instead of LASTSDONE for testing transfer completion in the packetized status case. LASTSDONE may come up more quickly since it only records completion on the SCSI side, but since LAST_SEG_DONE is used everywhere else (and needs to be), this is less confusing. Add a missing invalidation of the longjmp address in the non-pack handler. This code needs additional review. aic79xx_inline.h: Fix spelling error. aic79xx_osm.c: Set the cdb length for CDBs dma'ed from host memory. Add a comment indicating that, should CAM start supporting cdbs larger than 16bytes, the driver could store the CDB in the status buffer. aic79xx_pci.c: Add a table entry for the 39320A. Added a missing comma to an error string table. Fix spelling errors.
2003-02-27 23:23:16 +00:00
mvi CCSGCTL, CCSGEN|CCSGRESET;
or SG_STATE, FETCH_INPROG ret;
idle_sgfetch_complete:
/*
* Guard against SG_CACHE_AVAIL activating during sg fetch
* request in the other FIFO.
*/
test SG_STATE, FETCH_INPROG jz return;
clr CCSGCTL;
and CCSGADDR, SG_PREFETCH_ADDR_MASK, SCB_RESIDUAL_SGPTR;
mvi SG_STATE, SEGS_AVAIL|LOADING_NEEDED;
idle_sg_avail:
/* Does the hardware have space for another SG entry? */
test DFSTATUS, PRELOAD_AVAIL jz return;
aic79xx.c: Clear the LQICRC_NLQ status should it pop up after we have already handled the SCSIPERR. During some streaming operations this status can be delayed until the stream ends. Without this change, the driver would complain about a "Missing case in ahd_handle_scsiint". In the LQOBUSFREE handler... Don't return the LQOMGR back to the idle state until after we have cleaned up ENSELO and any status related to this selection. The last thing we need is the LQO manager starting another select-out before we have updated the execution queue. It is not clear whether the LQOMGR would, or would not start a new selection early. Make sure ENSELO is off prior to clearing SELDO by flushing device writes. Move assignment of the next target SCB pointer inside of an if to make the code clearer. The effect is the same. Dump card state in both "Unexpected PKT busfree" paths. In ahd_reset(), set the chip to SCSI mode before reading SXFRCTL1. That register only exists in the SCSI mode. Also set the mode explicitly to the SCSI mode after chip reset due to paranoia. Re-arrange code so that SXFRCTL1 is restored as quickly after the chip reset as possible. S/G structurs must be 8byte aligned. Make this official by saying so in our DMA tag. Disable CIO bus stretch on MDFFSTAT if SHVALID is about to come true. This can cause a CIO bus lockup if a PCI or PCI-X error occurs while the stretch is occurring - the host cannot service the PCI-X error since the CIO bus is locked out and SHVALID will never resolve. The stretch was added in the Rev B to simplify the wait for SHVALID to resolve, but the code to do this in the open source sequencer is so simple it was never removed. Consistently use MAX_OFFSET for the user max syncrate set from non-volatile storage. This ensures that the offset does not conflict with AH?_OFFSET_UNKNOWN. Have ahd_pause_and_flushwork set the mode to ensure that it has access to the registers it checks. Also modify the checking of intstat so that the check against 0xFF can actually succeed if the INT_PEND mask is something other than 0xFF. Although there are no cardbus U320 controllers, this check may be needed to recover from a hot-plug PCI removal that occurs without informing the driver. Fix a typo. sg_prefetch_cnt -> sg_prefetch_align. This fixes an infinite loop at card initialization if the cacheline size is 0. aic79xx.h: Add AHD_EARLY_REQ_BUG bug flag. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. Change AH?_OFFSET_UNKNOWN to 0xFF. This is a value that the curr->offset can never be, unlike '0' which we previously used. This fixes code that only checks for a non-zero offset to determine if a sync negotiation is required since it will fire in the unknown case even if the goal is async. aic79xx.reg: Add comments for LQISTAT bits indicating their names in the 7902 data book. We use slightly different and more descriptive names in the firmware. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. aic79xx.seq: Update comments regarding rundown of the GSFIFO to reflect reality. Fix spelling errors. Since we use an 8byte address and 1 byte length, shorten the size of a block move for the legacy DMA'ed CDB case from 11 to 9 bytes. Remove code that, assuming the abort pending feature worked, would set MK_MESSAGE in the SCB's control byte on completion to catch invalid reselections. Since we don't see interrupts for completed selections, this status update could occur prior to us noticing the SELDO. The "select-out" queue logic will get confused by the MK_MESSAGE bit being set as this is used to catch packatized connections where we select-out with ATN. Since the abort pending feature doesn't work on any released controllers yet, this code was never executed. Add support for the AHD_EARLY_REQ_BUG. Don't ignore persistent REQ assertions just because they were asserted within the bus settle delay window. This allows us to tolerate devices like the GEM318 that violate the SCSI spec. Remove unintentional settnig of SG_CACHE_AVAIL. Writing this bit should have no effect, but who knows... On the Rev A, we must wait for HDMAENACK before loading additional segments to avoid clobbering the address of the first segment in the S/G FIFO. This resolves data-corruption issues with certain IBM (now Hitachi) and Fujitsu U320 drives. Rearrange calc_residual to avoid an extra jmp instruction. On RevA Silicon, if the target returns us to data-out after we have already trained for data-out, it is possible for us to transition the free running clock to data-valid before the required 100ns P1 setup time (8 P1 assertions in fast-160 mode). This will only happen if this L-Q is a continuation of a data transfer for which we have already prefetched data into our FIFO (LQ/Data followed by LQ/Data for the same write transaction). This can cause some target implementations to miss the first few data transfers on the bus. We detect this situation by noticing that this is the first data transfer after an LQ (LQIWORKONLQ true), that the data transfer is a continuation of a transfer already setup in our FIFO (SAVEPTRS interrupt), and that the transaction is a write (DIRECTION set in DFCNTRL). The delay is performed by disabling SCSIEN until we see the first REQ from the target. Only compile in snapshot savepointers handler for RevA silicon where it is enabled. Handle the cfg4icmd packetized interrupt. We just need to load the address and count, start the DMA, and CLRCHN once the transfer is complete. Fix an oversight in the overrun handler for packetized status operations. We need to wait for either CTXTDONE or an overrun when checking for an overrun. The previous code did not wait and thus could decide that no overrun had occurred even though an overrun will occur on the next data-valid req. Add some comment to this section for clarity. Use LAST_SEG_DONE instead of LASTSDONE for testing transfer completion in the packetized status case. LASTSDONE may come up more quickly since it only records completion on the SCSI side, but since LAST_SEG_DONE is used everywhere else (and needs to be), this is less confusing. Add a missing invalidation of the longjmp address in the non-pack handler. This code needs additional review. aic79xx_inline.h: Fix spelling error. aic79xx_osm.c: Set the cdb length for CDBs dma'ed from host memory. Add a comment indicating that, should CAM start supporting cdbs larger than 16bytes, the driver could store the CDB in the status buffer. aic79xx_pci.c: Add a table entry for the 39320A. Added a missing comma to an error string table. Fix spelling errors.
2003-02-27 23:23:16 +00:00
/*
* On the A, preloading a segment before HDMAENACK
* comes true can clobber the shaddow address of the
* first segment in the S/G FIFO. Wait until it is
* safe to proceed.
*/
if ((ahd->features & AHD_NEW_DFCNTRL_OPTS) == 0) {
test DFCNTRL, HDMAENACK jz return;
}
if ((ahd->flags & AHD_64BIT_ADDRESSING) != 0) {
bmov HADDR, CCSGRAM, 8;
} else {
bmov HADDR, CCSGRAM, 4;
}
bmov HCNT, CCSGRAM, 3;
bmov SCB_RESIDUAL_DATACNT[3], CCSGRAM, 1;
if ((ahd->flags & AHD_39BIT_ADDRESSING) != 0) {
and HADDR[4], SG_HIGH_ADDR_BITS, SCB_RESIDUAL_DATACNT[3];
}
if ((ahd->flags & AHD_64BIT_ADDRESSING) != 0) {
/* Skip 4 bytes of pad. */
add CCSGADDR, 4;
}
sg_advance:
clr A; /* add sizeof(struct scatter) */
add SCB_RESIDUAL_SGPTR[0],SG_SIZEOF;
adc SCB_RESIDUAL_SGPTR[1],A;
adc SCB_RESIDUAL_SGPTR[2],A;
adc SCB_RESIDUAL_SGPTR[3],A;
mov SINDEX, SCB_RESIDUAL_SGPTR[0];
test SCB_RESIDUAL_DATACNT[3], SG_LAST_SEG jz . + 3;
or SINDEX, LAST_SEG;
clr SG_STATE;
mov SG_CACHE_PRE, SINDEX;
if ((ahd->features & AHD_NEW_DFCNTRL_OPTS) != 0) {
/*
* Use SCSIENWRDIS so that SCSIEN is never
* modified by this operation.
*/
aic79xx.c: Clear the LQICRC_NLQ status should it pop up after we have already handled the SCSIPERR. During some streaming operations this status can be delayed until the stream ends. Without this change, the driver would complain about a "Missing case in ahd_handle_scsiint". In the LQOBUSFREE handler... Don't return the LQOMGR back to the idle state until after we have cleaned up ENSELO and any status related to this selection. The last thing we need is the LQO manager starting another select-out before we have updated the execution queue. It is not clear whether the LQOMGR would, or would not start a new selection early. Make sure ENSELO is off prior to clearing SELDO by flushing device writes. Move assignment of the next target SCB pointer inside of an if to make the code clearer. The effect is the same. Dump card state in both "Unexpected PKT busfree" paths. In ahd_reset(), set the chip to SCSI mode before reading SXFRCTL1. That register only exists in the SCSI mode. Also set the mode explicitly to the SCSI mode after chip reset due to paranoia. Re-arrange code so that SXFRCTL1 is restored as quickly after the chip reset as possible. S/G structurs must be 8byte aligned. Make this official by saying so in our DMA tag. Disable CIO bus stretch on MDFFSTAT if SHVALID is about to come true. This can cause a CIO bus lockup if a PCI or PCI-X error occurs while the stretch is occurring - the host cannot service the PCI-X error since the CIO bus is locked out and SHVALID will never resolve. The stretch was added in the Rev B to simplify the wait for SHVALID to resolve, but the code to do this in the open source sequencer is so simple it was never removed. Consistently use MAX_OFFSET for the user max syncrate set from non-volatile storage. This ensures that the offset does not conflict with AH?_OFFSET_UNKNOWN. Have ahd_pause_and_flushwork set the mode to ensure that it has access to the registers it checks. Also modify the checking of intstat so that the check against 0xFF can actually succeed if the INT_PEND mask is something other than 0xFF. Although there are no cardbus U320 controllers, this check may be needed to recover from a hot-plug PCI removal that occurs without informing the driver. Fix a typo. sg_prefetch_cnt -> sg_prefetch_align. This fixes an infinite loop at card initialization if the cacheline size is 0. aic79xx.h: Add AHD_EARLY_REQ_BUG bug flag. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. Change AH?_OFFSET_UNKNOWN to 0xFF. This is a value that the curr->offset can never be, unlike '0' which we previously used. This fixes code that only checks for a non-zero offset to determine if a sync negotiation is required since it will fire in the unknown case even if the goal is async. aic79xx.reg: Add comments for LQISTAT bits indicating their names in the 7902 data book. We use slightly different and more descriptive names in the firmware. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. aic79xx.seq: Update comments regarding rundown of the GSFIFO to reflect reality. Fix spelling errors. Since we use an 8byte address and 1 byte length, shorten the size of a block move for the legacy DMA'ed CDB case from 11 to 9 bytes. Remove code that, assuming the abort pending feature worked, would set MK_MESSAGE in the SCB's control byte on completion to catch invalid reselections. Since we don't see interrupts for completed selections, this status update could occur prior to us noticing the SELDO. The "select-out" queue logic will get confused by the MK_MESSAGE bit being set as this is used to catch packatized connections where we select-out with ATN. Since the abort pending feature doesn't work on any released controllers yet, this code was never executed. Add support for the AHD_EARLY_REQ_BUG. Don't ignore persistent REQ assertions just because they were asserted within the bus settle delay window. This allows us to tolerate devices like the GEM318 that violate the SCSI spec. Remove unintentional settnig of SG_CACHE_AVAIL. Writing this bit should have no effect, but who knows... On the Rev A, we must wait for HDMAENACK before loading additional segments to avoid clobbering the address of the first segment in the S/G FIFO. This resolves data-corruption issues with certain IBM (now Hitachi) and Fujitsu U320 drives. Rearrange calc_residual to avoid an extra jmp instruction. On RevA Silicon, if the target returns us to data-out after we have already trained for data-out, it is possible for us to transition the free running clock to data-valid before the required 100ns P1 setup time (8 P1 assertions in fast-160 mode). This will only happen if this L-Q is a continuation of a data transfer for which we have already prefetched data into our FIFO (LQ/Data followed by LQ/Data for the same write transaction). This can cause some target implementations to miss the first few data transfers on the bus. We detect this situation by noticing that this is the first data transfer after an LQ (LQIWORKONLQ true), that the data transfer is a continuation of a transfer already setup in our FIFO (SAVEPTRS interrupt), and that the transaction is a write (DIRECTION set in DFCNTRL). The delay is performed by disabling SCSIEN until we see the first REQ from the target. Only compile in snapshot savepointers handler for RevA silicon where it is enabled. Handle the cfg4icmd packetized interrupt. We just need to load the address and count, start the DMA, and CLRCHN once the transfer is complete. Fix an oversight in the overrun handler for packetized status operations. We need to wait for either CTXTDONE or an overrun when checking for an overrun. The previous code did not wait and thus could decide that no overrun had occurred even though an overrun will occur on the next data-valid req. Add some comment to this section for clarity. Use LAST_SEG_DONE instead of LASTSDONE for testing transfer completion in the packetized status case. LASTSDONE may come up more quickly since it only records completion on the SCSI side, but since LAST_SEG_DONE is used everywhere else (and needs to be), this is less confusing. Add a missing invalidation of the longjmp address in the non-pack handler. This code needs additional review. aic79xx_inline.h: Fix spelling error. aic79xx_osm.c: Set the cdb length for CDBs dma'ed from host memory. Add a comment indicating that, should CAM start supporting cdbs larger than 16bytes, the driver could store the CDB in the status buffer. aic79xx_pci.c: Add a table entry for the 39320A. Added a missing comma to an error string table. Fix spelling errors.
2003-02-27 23:23:16 +00:00
or DFCNTRL, PRELOADEN|HDMAEN|SCSIENWRDIS;
} else {
or DFCNTRL, PRELOADEN|HDMAEN;
}
/*
* Do we have another segment in the cache?
*/
add NONE, SG_PREFETCH_CNT_LIMIT, CCSGADDR;
jnc return;
and SG_STATE, ~SEGS_AVAIL ret;
/*
* Initialize the DMA address and counter from the SCB.
*/
load_first_seg:
bmov HADDR, SCB_DATAPTR, 11;
Upgrade to version 1.1 of the aic79xx U320 driver. aic79xx.c: o Remove redundant ahd_update_modes() call. o Correct panic in diagnostic should state corruption cause the SCB Id to be invalid during a selection timeout. o Add workaround for missing BUSFREEREV feature in Rev A silicon. o Corect formatting nits. o Use register pretty printing in more places. o Save and restore our SCB pointer when updating the waiting queue list for an "expected" LQ-out busfree. o In ahd_clear_intstat, deal with the missing autoclear in the CLRLQO* registers. o BE fixup in a diagnostic printf. o Make sure that we are in the proper mode before disabling selections in ahd_update_pending_scbs. o Add more diagnostics. o task_attribute_nonpkt_tag -> task_attribute: we don't need a nonpkt_tag field anymore for allowing all 512 SCBs to be used in non-packetized connections. o Negotiate HOLD_MCS to U320 devices. o Add a few additional mode assertions. o Restore the chip mode after clearing out the qinfifo so that code using ahd_abort_scbs sees a consistent mode. o Simplify the DMA engine shutdown routine prior to performing a bus reset. o Perform the sequencer restart after a chip reset prior to setting up our timer to poll for the reset to be complete. On some OSes, the timer could actually pre-empt us and order is important here. o Have our "reset poller" set the expected mode since there is no guarantee of what mode will be in force when we are called from the OS timer. o Save and restore the SCB pointer in ahd_dump_card_state(). This routine must not modify card state. o Ditto for ahd_dump_scbs(). aic79xx.h: o Add a few more chip bug definitions. o Align our tag on a 32bit boundary. aic79xx.reg: aic79xx.seq: o Start work on removing workarounds for Rev B. o Use a special location in scratch from for stroring our SCBPTR during legacy FIFO allocations. This corrects problems in mixed packetized/non-packetized configurations where calling into a FIFO task corrupted our SCBPTR. o Don't rely on DMA priority to guarantee that all data in our FIFOs will flush prior to a command completion notification going out of the command channel. We've never seen this assumption fail, but better safe than sorry. o Deal with missing BUSFREEREV feature in H2A. o Simplify disconnect list code now that the list will always have only a single entry. o Implement the AHD_REG_SLOW_SETTLE_BUG workaround. o Swith to using "REG_ISR" for local mode scratch during our ISR. o Add a missing jmp to the data_group_dma_loop after our data pointers have been re-initialized by the kernel. o Correct test in the bitbucket code so that we actually wait for the bitbucket to complete before signaling the kernel of the overrun condition. o Reposition pkt_saveptrs to avoid a jmp instruction. o Update a comment to reflect that the code now waits for a FIFO to drain prior to issuing a CLRCHN. aic79xx_inline.h: o Remove unused untagged queue handling code. o Don't attempt to htole64 what could be a 32bit value. aic79xx_pci.c: o Set additional bug flags for rev A chips.
2002-09-26 22:54:00 +00:00
and REG_ISR, ~SG_FULL_RESID, SCB_SGPTR[0];
test SCB_DATACNT[3], SG_LAST_SEG jz . + 2;
Upgrade to version 1.1 of the aic79xx U320 driver. aic79xx.c: o Remove redundant ahd_update_modes() call. o Correct panic in diagnostic should state corruption cause the SCB Id to be invalid during a selection timeout. o Add workaround for missing BUSFREEREV feature in Rev A silicon. o Corect formatting nits. o Use register pretty printing in more places. o Save and restore our SCB pointer when updating the waiting queue list for an "expected" LQ-out busfree. o In ahd_clear_intstat, deal with the missing autoclear in the CLRLQO* registers. o BE fixup in a diagnostic printf. o Make sure that we are in the proper mode before disabling selections in ahd_update_pending_scbs. o Add more diagnostics. o task_attribute_nonpkt_tag -> task_attribute: we don't need a nonpkt_tag field anymore for allowing all 512 SCBs to be used in non-packetized connections. o Negotiate HOLD_MCS to U320 devices. o Add a few additional mode assertions. o Restore the chip mode after clearing out the qinfifo so that code using ahd_abort_scbs sees a consistent mode. o Simplify the DMA engine shutdown routine prior to performing a bus reset. o Perform the sequencer restart after a chip reset prior to setting up our timer to poll for the reset to be complete. On some OSes, the timer could actually pre-empt us and order is important here. o Have our "reset poller" set the expected mode since there is no guarantee of what mode will be in force when we are called from the OS timer. o Save and restore the SCB pointer in ahd_dump_card_state(). This routine must not modify card state. o Ditto for ahd_dump_scbs(). aic79xx.h: o Add a few more chip bug definitions. o Align our tag on a 32bit boundary. aic79xx.reg: aic79xx.seq: o Start work on removing workarounds for Rev B. o Use a special location in scratch from for stroring our SCBPTR during legacy FIFO allocations. This corrects problems in mixed packetized/non-packetized configurations where calling into a FIFO task corrupted our SCBPTR. o Don't rely on DMA priority to guarantee that all data in our FIFOs will flush prior to a command completion notification going out of the command channel. We've never seen this assumption fail, but better safe than sorry. o Deal with missing BUSFREEREV feature in H2A. o Simplify disconnect list code now that the list will always have only a single entry. o Implement the AHD_REG_SLOW_SETTLE_BUG workaround. o Swith to using "REG_ISR" for local mode scratch during our ISR. o Add a missing jmp to the data_group_dma_loop after our data pointers have been re-initialized by the kernel. o Correct test in the bitbucket code so that we actually wait for the bitbucket to complete before signaling the kernel of the overrun condition. o Reposition pkt_saveptrs to avoid a jmp instruction. o Update a comment to reflect that the code now waits for a FIFO to drain prior to issuing a CLRCHN. aic79xx_inline.h: o Remove unused untagged queue handling code. o Don't attempt to htole64 what could be a 32bit value. aic79xx_pci.c: o Set additional bug flags for rev A chips.
2002-09-26 22:54:00 +00:00
or REG_ISR, LAST_SEG;
mov SG_CACHE_PRE, REG_ISR;
mvi DFCNTRL, (PRELOADEN|SCSIEN|HDMAEN);
/*
* Since we've are entering a data phase, we will
* rely on the SCB_RESID* fields. Initialize the
* residual and clear the full residual flag.
*/
and SCB_SGPTR[0], ~SG_FULL_RESID;
bmov SCB_RESIDUAL_DATACNT[3], SCB_DATACNT[3], 5;
/* If we need more S/G elements, tell the idle loop */
test SCB_RESIDUAL_DATACNT[3], SG_LAST_SEG jnz . + 2;
mvi SG_STATE, LOADING_NEEDED ret;
clr SG_STATE ret;
p_data_handle_xfer:
Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-05-04 00:20:07 +00:00
call setjmp;
test SG_STATE, LOADING_NEEDED jnz service_fifo;
p_data_clear_handler:
or LONGJMP_ADDR[1], INVALID_ADDR ret;
p_data:
test SEQ_FLAGS, NOT_IDENTIFIED|NO_CDB_SENT jz p_data_allowed;
SET_SEQINTCODE(PROTO_VIOLATION)
p_data_allowed:
test SEQ_FLAGS, DPHASE jz data_phase_initialize;
/*
* If we re-enter the data phase after going through another
* phase, our transfer location has almost certainly been
* corrupted by the interveining, non-data, transfers. Ask
* the host driver to fix us up based on the transfer residual
* unless we already know that we should be bitbucketing.
*/
test SCB_RESIDUAL_SGPTR[0], SG_LIST_NULL jnz p_data_bitbucket;
SET_SEQINTCODE(PDATA_REINIT)
jmp data_phase_inbounds;
p_data_bitbucket:
/*
* Turn on `Bit Bucket' mode, wait until the target takes
* us to another phase, and then notify the host.
*/
mov SAVED_MODE, MODE_PTR;
test MODE_PTR, ~(MK_MODE(M_DFF1, M_DFF1))
jnz bitbucket_not_m_dff;
/*
* Ensure that any FIFO contents are cleared out and the
* FIFO free'd prior to starting the BITBUCKET. BITBUCKET
* doesn't discard data already in the FIFO.
*/
mvi DFFSXFRCTL, RSTCHN|CLRSHCNT;
Upgrade to version 1.1 of the aic79xx U320 driver. aic79xx.c: o Remove redundant ahd_update_modes() call. o Correct panic in diagnostic should state corruption cause the SCB Id to be invalid during a selection timeout. o Add workaround for missing BUSFREEREV feature in Rev A silicon. o Corect formatting nits. o Use register pretty printing in more places. o Save and restore our SCB pointer when updating the waiting queue list for an "expected" LQ-out busfree. o In ahd_clear_intstat, deal with the missing autoclear in the CLRLQO* registers. o BE fixup in a diagnostic printf. o Make sure that we are in the proper mode before disabling selections in ahd_update_pending_scbs. o Add more diagnostics. o task_attribute_nonpkt_tag -> task_attribute: we don't need a nonpkt_tag field anymore for allowing all 512 SCBs to be used in non-packetized connections. o Negotiate HOLD_MCS to U320 devices. o Add a few additional mode assertions. o Restore the chip mode after clearing out the qinfifo so that code using ahd_abort_scbs sees a consistent mode. o Simplify the DMA engine shutdown routine prior to performing a bus reset. o Perform the sequencer restart after a chip reset prior to setting up our timer to poll for the reset to be complete. On some OSes, the timer could actually pre-empt us and order is important here. o Have our "reset poller" set the expected mode since there is no guarantee of what mode will be in force when we are called from the OS timer. o Save and restore the SCB pointer in ahd_dump_card_state(). This routine must not modify card state. o Ditto for ahd_dump_scbs(). aic79xx.h: o Add a few more chip bug definitions. o Align our tag on a 32bit boundary. aic79xx.reg: aic79xx.seq: o Start work on removing workarounds for Rev B. o Use a special location in scratch from for stroring our SCBPTR during legacy FIFO allocations. This corrects problems in mixed packetized/non-packetized configurations where calling into a FIFO task corrupted our SCBPTR. o Don't rely on DMA priority to guarantee that all data in our FIFOs will flush prior to a command completion notification going out of the command channel. We've never seen this assumption fail, but better safe than sorry. o Deal with missing BUSFREEREV feature in H2A. o Simplify disconnect list code now that the list will always have only a single entry. o Implement the AHD_REG_SLOW_SETTLE_BUG workaround. o Swith to using "REG_ISR" for local mode scratch during our ISR. o Add a missing jmp to the data_group_dma_loop after our data pointers have been re-initialized by the kernel. o Correct test in the bitbucket code so that we actually wait for the bitbucket to complete before signaling the kernel of the overrun condition. o Reposition pkt_saveptrs to avoid a jmp instruction. o Update a comment to reflect that the code now waits for a FIFO to drain prior to issuing a CLRCHN. aic79xx_inline.h: o Remove unused untagged queue handling code. o Don't attempt to htole64 what could be a 32bit value. aic79xx_pci.c: o Set additional bug flags for rev A chips.
2002-09-26 22:54:00 +00:00
SET_MODE(M_SCSI, M_SCSI)
bitbucket_not_m_dff:
or SXFRCTL1,BITBUCKET;
Upgrade to version 1.1 of the aic79xx U320 driver. aic79xx.c: o Remove redundant ahd_update_modes() call. o Correct panic in diagnostic should state corruption cause the SCB Id to be invalid during a selection timeout. o Add workaround for missing BUSFREEREV feature in Rev A silicon. o Corect formatting nits. o Use register pretty printing in more places. o Save and restore our SCB pointer when updating the waiting queue list for an "expected" LQ-out busfree. o In ahd_clear_intstat, deal with the missing autoclear in the CLRLQO* registers. o BE fixup in a diagnostic printf. o Make sure that we are in the proper mode before disabling selections in ahd_update_pending_scbs. o Add more diagnostics. o task_attribute_nonpkt_tag -> task_attribute: we don't need a nonpkt_tag field anymore for allowing all 512 SCBs to be used in non-packetized connections. o Negotiate HOLD_MCS to U320 devices. o Add a few additional mode assertions. o Restore the chip mode after clearing out the qinfifo so that code using ahd_abort_scbs sees a consistent mode. o Simplify the DMA engine shutdown routine prior to performing a bus reset. o Perform the sequencer restart after a chip reset prior to setting up our timer to poll for the reset to be complete. On some OSes, the timer could actually pre-empt us and order is important here. o Have our "reset poller" set the expected mode since there is no guarantee of what mode will be in force when we are called from the OS timer. o Save and restore the SCB pointer in ahd_dump_card_state(). This routine must not modify card state. o Ditto for ahd_dump_scbs(). aic79xx.h: o Add a few more chip bug definitions. o Align our tag on a 32bit boundary. aic79xx.reg: aic79xx.seq: o Start work on removing workarounds for Rev B. o Use a special location in scratch from for stroring our SCBPTR during legacy FIFO allocations. This corrects problems in mixed packetized/non-packetized configurations where calling into a FIFO task corrupted our SCBPTR. o Don't rely on DMA priority to guarantee that all data in our FIFOs will flush prior to a command completion notification going out of the command channel. We've never seen this assumption fail, but better safe than sorry. o Deal with missing BUSFREEREV feature in H2A. o Simplify disconnect list code now that the list will always have only a single entry. o Implement the AHD_REG_SLOW_SETTLE_BUG workaround. o Swith to using "REG_ISR" for local mode scratch during our ISR. o Add a missing jmp to the data_group_dma_loop after our data pointers have been re-initialized by the kernel. o Correct test in the bitbucket code so that we actually wait for the bitbucket to complete before signaling the kernel of the overrun condition. o Reposition pkt_saveptrs to avoid a jmp instruction. o Update a comment to reflect that the code now waits for a FIFO to drain prior to issuing a CLRCHN. aic79xx_inline.h: o Remove unused untagged queue handling code. o Don't attempt to htole64 what could be a 32bit value. aic79xx_pci.c: o Set additional bug flags for rev A chips.
2002-09-26 22:54:00 +00:00
/* Wait for non-data phase. */
test SCSIPHASE, ~DATA_PHASE_MASK jz .;
and SXFRCTL1, ~BITBUCKET;
RESTORE_MODE(SAVED_MODE)
SET_SRC_MODE M_DFF1;
SET_DST_MODE M_DFF1;
SET_SEQINTCODE(DATA_OVERRUN)
jmp ITloop;
data_phase_initialize:
test SCB_SGPTR[0], SG_LIST_NULL jnz p_data_bitbucket;
call load_first_seg;
data_phase_inbounds:
/* We have seen a data phase at least once. */
or SEQ_FLAGS, DPHASE;
mov SAVED_MODE, MODE_PTR;
test SG_STATE, LOADING_NEEDED jz data_group_dma_loop;
call p_data_handle_xfer;
data_group_dma_loop:
/*
* The transfer is complete if either the last segment
* completes or the target changes phase. Both conditions
* will clear SCSIEN.
*/
call idle_loop_service_fifos;
call idle_loop_cchan;
call idle_loop_gsfifo;
RESTORE_MODE(SAVED_MODE)
test DFCNTRL, SCSIEN jnz data_group_dma_loop;
data_group_dmafinish:
/*
* The transfer has terminated either due to a phase
* change, and/or the completion of the last segment.
* We have two goals here. Do as much other work
* as possible while the data fifo drains on a read
* and respond as quickly as possible to the standard
* messages (save data pointers/disconnect and command
* complete) that usually follow a data phase.
*/
call calc_residual;
/*
* Go ahead and shut down the DMA engine now.
*/
test DFCNTRL, DIRECTION jnz data_phase_finish;
data_group_fifoflush:
if ((ahd->bugs & AHD_AUTOFLUSH_BUG) != 0) {
or DFCNTRL, FIFOFLUSH;
}
/*
* We have enabled the auto-ack feature. This means
* that the controller may have already transferred
* some overrun bytes into the data FIFO and acked them
* on the bus. The only way to detect this situation is
* to wait for LAST_SEG_DONE to come true on a completed
* transfer and then test to see if the data FIFO is
* non-empty. We know there is more data yet to transfer
* if SG_LIST_NULL is not yet set, thus there cannot be
* an overrun.
*/
test SCB_RESIDUAL_SGPTR[0], SG_LIST_NULL jz data_phase_finish;
test SG_CACHE_SHADOW, LAST_SEG_DONE jz .;
test DFSTATUS, FIFOEMP jnz data_phase_finish;
/* Overrun */
jmp p_data;
data_phase_finish:
/*
* If the target has left us in data phase, loop through
* the dma code again. We will only loop if there is a
* data overrun.
*/
if ((ahd->flags & AHD_TARGETROLE) != 0) {
test SSTAT0, TARGET jnz data_phase_done;
}
if ((ahd->flags & AHD_INITIATORROLE) != 0) {
test SSTAT1, REQINIT jz .;
test SCSIPHASE, DATA_PHASE_MASK jnz p_data;
}
data_phase_done:
/* Kill off any pending prefetch */
call disable_ccsgen;
or LONGJMP_ADDR[1], INVALID_ADDR;
if ((ahd->flags & AHD_TARGETROLE) != 0) {
test SEQ_FLAGS, DPHASE_PENDING jz ITloop;
/*
and SEQ_FLAGS, ~DPHASE_PENDING;
* For data-in phases, wait for any pending acks from the
* initiator before changing phase. We only need to
* send Ignore Wide Residue messages for data-in phases.
test DFCNTRL, DIRECTION jz target_ITloop;
test SSTAT1, REQINIT jnz .;
test SCB_TASK_ATTRIBUTE, SCB_XFERLEN_ODD jz target_ITloop;
Upgrade to version 1.1 of the aic79xx U320 driver. aic79xx.c: o Remove redundant ahd_update_modes() call. o Correct panic in diagnostic should state corruption cause the SCB Id to be invalid during a selection timeout. o Add workaround for missing BUSFREEREV feature in Rev A silicon. o Corect formatting nits. o Use register pretty printing in more places. o Save and restore our SCB pointer when updating the waiting queue list for an "expected" LQ-out busfree. o In ahd_clear_intstat, deal with the missing autoclear in the CLRLQO* registers. o BE fixup in a diagnostic printf. o Make sure that we are in the proper mode before disabling selections in ahd_update_pending_scbs. o Add more diagnostics. o task_attribute_nonpkt_tag -> task_attribute: we don't need a nonpkt_tag field anymore for allowing all 512 SCBs to be used in non-packetized connections. o Negotiate HOLD_MCS to U320 devices. o Add a few additional mode assertions. o Restore the chip mode after clearing out the qinfifo so that code using ahd_abort_scbs sees a consistent mode. o Simplify the DMA engine shutdown routine prior to performing a bus reset. o Perform the sequencer restart after a chip reset prior to setting up our timer to poll for the reset to be complete. On some OSes, the timer could actually pre-empt us and order is important here. o Have our "reset poller" set the expected mode since there is no guarantee of what mode will be in force when we are called from the OS timer. o Save and restore the SCB pointer in ahd_dump_card_state(). This routine must not modify card state. o Ditto for ahd_dump_scbs(). aic79xx.h: o Add a few more chip bug definitions. o Align our tag on a 32bit boundary. aic79xx.reg: aic79xx.seq: o Start work on removing workarounds for Rev B. o Use a special location in scratch from for stroring our SCBPTR during legacy FIFO allocations. This corrects problems in mixed packetized/non-packetized configurations where calling into a FIFO task corrupted our SCBPTR. o Don't rely on DMA priority to guarantee that all data in our FIFOs will flush prior to a command completion notification going out of the command channel. We've never seen this assumption fail, but better safe than sorry. o Deal with missing BUSFREEREV feature in H2A. o Simplify disconnect list code now that the list will always have only a single entry. o Implement the AHD_REG_SLOW_SETTLE_BUG workaround. o Swith to using "REG_ISR" for local mode scratch during our ISR. o Add a missing jmp to the data_group_dma_loop after our data pointers have been re-initialized by the kernel. o Correct test in the bitbucket code so that we actually wait for the bitbucket to complete before signaling the kernel of the overrun condition. o Reposition pkt_saveptrs to avoid a jmp instruction. o Update a comment to reflect that the code now waits for a FIFO to drain prior to issuing a CLRCHN. aic79xx_inline.h: o Remove unused untagged queue handling code. o Don't attempt to htole64 what could be a 32bit value. aic79xx_pci.c: o Set additional bug flags for rev A chips.
2002-09-26 22:54:00 +00:00
SET_MODE(M_SCSI, M_SCSI)
test NEGCONOPTS, WIDEXFER jz target_ITloop;
*/
/*
* Issue an Ignore Wide Residue Message.
mvi P_MESGIN|BSYO call change_phase;
mvi MSG_IGN_WIDE_RESIDUE call target_outb;
mvi 1 call target_outb;
jmp target_ITloop;
*/
} else {
jmp ITloop;
}
/*
* We assume that, even though data may still be
* transferring to the host, that the SCSI side of
* the DMA engine is now in a static state. This
* allows us to update our notion of where we are
* in this transfer.
*
* If, by chance, we stopped before being able
* to fetch additional segments for this transfer,
* yet the last S/G was completely exhausted,
* call our idle loop until it is able to load
* another segment. This will allow us to immediately
* pickup on the next segment on the next data phase.
*
* If we happened to stop on the last segment, then
* our residual information is still correct from
* the idle loop and there is no need to perform
* any fixups.
*/
residual_before_last_seg:
test MDFFSTAT, SHVALID jnz sgptr_fixup;
/*
* Can never happen from an interrupt as the packetized
* hardware will only interrupt us once SHVALID or
* LAST_SEG_DONE.
*/
call idle_loop_service_fifos;
RESTORE_MODE(SAVED_MODE)
aic79xx.c: Clear the LQICRC_NLQ status should it pop up after we have already handled the SCSIPERR. During some streaming operations this status can be delayed until the stream ends. Without this change, the driver would complain about a "Missing case in ahd_handle_scsiint". In the LQOBUSFREE handler... Don't return the LQOMGR back to the idle state until after we have cleaned up ENSELO and any status related to this selection. The last thing we need is the LQO manager starting another select-out before we have updated the execution queue. It is not clear whether the LQOMGR would, or would not start a new selection early. Make sure ENSELO is off prior to clearing SELDO by flushing device writes. Move assignment of the next target SCB pointer inside of an if to make the code clearer. The effect is the same. Dump card state in both "Unexpected PKT busfree" paths. In ahd_reset(), set the chip to SCSI mode before reading SXFRCTL1. That register only exists in the SCSI mode. Also set the mode explicitly to the SCSI mode after chip reset due to paranoia. Re-arrange code so that SXFRCTL1 is restored as quickly after the chip reset as possible. S/G structurs must be 8byte aligned. Make this official by saying so in our DMA tag. Disable CIO bus stretch on MDFFSTAT if SHVALID is about to come true. This can cause a CIO bus lockup if a PCI or PCI-X error occurs while the stretch is occurring - the host cannot service the PCI-X error since the CIO bus is locked out and SHVALID will never resolve. The stretch was added in the Rev B to simplify the wait for SHVALID to resolve, but the code to do this in the open source sequencer is so simple it was never removed. Consistently use MAX_OFFSET for the user max syncrate set from non-volatile storage. This ensures that the offset does not conflict with AH?_OFFSET_UNKNOWN. Have ahd_pause_and_flushwork set the mode to ensure that it has access to the registers it checks. Also modify the checking of intstat so that the check against 0xFF can actually succeed if the INT_PEND mask is something other than 0xFF. Although there are no cardbus U320 controllers, this check may be needed to recover from a hot-plug PCI removal that occurs without informing the driver. Fix a typo. sg_prefetch_cnt -> sg_prefetch_align. This fixes an infinite loop at card initialization if the cacheline size is 0. aic79xx.h: Add AHD_EARLY_REQ_BUG bug flag. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. Change AH?_OFFSET_UNKNOWN to 0xFF. This is a value that the curr->offset can never be, unlike '0' which we previously used. This fixes code that only checks for a non-zero offset to determine if a sync negotiation is required since it will fire in the unknown case even if the goal is async. aic79xx.reg: Add comments for LQISTAT bits indicating their names in the 7902 data book. We use slightly different and more descriptive names in the firmware. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. aic79xx.seq: Update comments regarding rundown of the GSFIFO to reflect reality. Fix spelling errors. Since we use an 8byte address and 1 byte length, shorten the size of a block move for the legacy DMA'ed CDB case from 11 to 9 bytes. Remove code that, assuming the abort pending feature worked, would set MK_MESSAGE in the SCB's control byte on completion to catch invalid reselections. Since we don't see interrupts for completed selections, this status update could occur prior to us noticing the SELDO. The "select-out" queue logic will get confused by the MK_MESSAGE bit being set as this is used to catch packatized connections where we select-out with ATN. Since the abort pending feature doesn't work on any released controllers yet, this code was never executed. Add support for the AHD_EARLY_REQ_BUG. Don't ignore persistent REQ assertions just because they were asserted within the bus settle delay window. This allows us to tolerate devices like the GEM318 that violate the SCSI spec. Remove unintentional settnig of SG_CACHE_AVAIL. Writing this bit should have no effect, but who knows... On the Rev A, we must wait for HDMAENACK before loading additional segments to avoid clobbering the address of the first segment in the S/G FIFO. This resolves data-corruption issues with certain IBM (now Hitachi) and Fujitsu U320 drives. Rearrange calc_residual to avoid an extra jmp instruction. On RevA Silicon, if the target returns us to data-out after we have already trained for data-out, it is possible for us to transition the free running clock to data-valid before the required 100ns P1 setup time (8 P1 assertions in fast-160 mode). This will only happen if this L-Q is a continuation of a data transfer for which we have already prefetched data into our FIFO (LQ/Data followed by LQ/Data for the same write transaction). This can cause some target implementations to miss the first few data transfers on the bus. We detect this situation by noticing that this is the first data transfer after an LQ (LQIWORKONLQ true), that the data transfer is a continuation of a transfer already setup in our FIFO (SAVEPTRS interrupt), and that the transaction is a write (DIRECTION set in DFCNTRL). The delay is performed by disabling SCSIEN until we see the first REQ from the target. Only compile in snapshot savepointers handler for RevA silicon where it is enabled. Handle the cfg4icmd packetized interrupt. We just need to load the address and count, start the DMA, and CLRCHN once the transfer is complete. Fix an oversight in the overrun handler for packetized status operations. We need to wait for either CTXTDONE or an overrun when checking for an overrun. The previous code did not wait and thus could decide that no overrun had occurred even though an overrun will occur on the next data-valid req. Add some comment to this section for clarity. Use LAST_SEG_DONE instead of LASTSDONE for testing transfer completion in the packetized status case. LASTSDONE may come up more quickly since it only records completion on the SCSI side, but since LAST_SEG_DONE is used everywhere else (and needs to be), this is less confusing. Add a missing invalidation of the longjmp address in the non-pack handler. This code needs additional review. aic79xx_inline.h: Fix spelling error. aic79xx_osm.c: Set the cdb length for CDBs dma'ed from host memory. Add a comment indicating that, should CAM start supporting cdbs larger than 16bytes, the driver could store the CDB in the status buffer. aic79xx_pci.c: Add a table entry for the 39320A. Added a missing comma to an error string table. Fix spelling errors.
2003-02-27 23:23:16 +00:00
/* FALLTHROUGH */
calc_residual:
test SG_CACHE_SHADOW, LAST_SEG jz residual_before_last_seg;
/* Record if we've consumed all S/G entries */
test MDFFSTAT, SHVALID jz . + 2;
bmov SCB_RESIDUAL_DATACNT, SHCNT, 3 ret;
or SCB_RESIDUAL_SGPTR[0], SG_LIST_NULL ret;
sgptr_fixup:
/*
* Fixup the residual next S/G pointer. The S/G preload
* feature of the chip allows us to load two elements
* in addition to the currently active element. We
* store the bottom byte of the next S/G pointer in
* the SG_CACHE_PTR register so we can restore the
* correct value when the DMA completes. If the next
* sg ptr value has advanced to the point where higher
* bytes in the address have been affected, fix them
* too.
*/
test SG_CACHE_SHADOW, 0x80 jz sgptr_fixup_done;
test SCB_RESIDUAL_SGPTR[0], 0x80 jnz sgptr_fixup_done;
add SCB_RESIDUAL_SGPTR[1], -1;
adc SCB_RESIDUAL_SGPTR[2], -1;
adc SCB_RESIDUAL_SGPTR[3], -1;
sgptr_fixup_done:
and SCB_RESIDUAL_SGPTR[0], SG_ADDR_MASK, SG_CACHE_SHADOW;
clr SCB_RESIDUAL_DATACNT[3]; /* We are not the last seg */
bmov SCB_RESIDUAL_DATACNT, SHCNT, 3 ret;
aic79xx.c: Implement the SCB_SILENT flag. This is useful for hushing up the driver during DV or other operations that we expect to cause transmission errors. The messages will still print if the SHOW_MASKED_ERRORS debug option is enabled. Save and restore the NEGOADDR address when setting new transfer settings. The sequencer performs lookups in the negotiation table too and it expects NEGOADDR to remain consistent across pause/unpause sessions. Consistently use "offset" instead of "period" to determine if we are running sync or not. Add a SHOW_MESSAGES diagnostic for when we assert ATN during message processing. Print out IU, QAS, and RTI features when showing transfer options. Limit the syncrate after all option conformance changes have taken place in ahd_devlimited_syncrate. Changes in options may change the final syncrate we accept. Keep a copy of the hs_mailbox in our softc so that we can perform read/modify/write operations on the hs_mailbox without having to pause the sequencer to read the last written value. Use the ENINT_COALESS flag in the hs_mailbox to toggle interrupt coalessing. Add entrypoints for enabling interrupt coalessing and setting both a timeout (how long to wait for commands to be coalessed) and a maximum commands to coaless value. Add a statistics timer that decides when to enable or disable interrupt coalessing based on load. Add a routine, ahd_reset_cmds_pending() which is used to update the CMDS_PENDING sequencer variable whenever error recovery compeltes SCBs without notifying the sequencer. Since ahd_reset_cmds_pending is called during ahd_unpause() only if we've aborted SCBs, its call to ahd_flush_qoutfifo should not cause recursion through ahd_run_qoutfifo(). A panic has been added to ensure that this recursion does not occur. In ahd_search_qinfifo, update the CMDS_PENDING sequencer variable directly. ahd_search_qinififo can be called in situations where using ahd_reset_cmds_pending() might cause recursion. Since we can safely determine the exact number to reduce CMDS_PENDING by in this scenario without running the qoutfifo, the manual update is sufficient. Clean up diagnostics. Add ahd_flush_qoutfifo() which will run the qoutfifo as well as complete any commands sitting on the sequencer's COMPLETE_SCB lists or the good status FIFO. Use this routine in several places that did similar things in an add-hoc, but incomplete, fashion. A call to this routine was also added to ahd_abort_scbs() to close a race. In ahd_pause_and_flushwork() only return once selections are safely disabled. Flush all completed commands via ahd_flush_qoutfifo(). Remove "Now packetized" diagnostic now that this information is incorperated into the actual negotiation messages that are displayed. When forcing renegotiation, don't clober the current ppr_options. Much of the driver uses this information to determine if we are currently packetized or not. Remove some stray spaces at column 1 in ahd_set_tags. When complaining about getting a host message loop request with no pending messages, print out the SCB_CONTROL register down on the card. Modify the ahd_sent_msg() routine to handle a search for an outgoing identify message. Use this to detect a msg reject on an identify message which typically indicates that the target thought we were packetized. Force a renegotiation in this case. In ahd_search_qinfifo(), wait more effectively for SCB DMA activities to cease. We also disable SCB fetch operations since we are about to change the qinfifo and any fetch in progress will likely be invalidated. In ahd_qinfifo_count(), fix the qinfifo empty case. In ahd_dump_card_state(), print out CCSCBCTL in the correct mode. If we are a narrow controller, don't set the current width to unknown when forcing a future negotiation. This just confuses the code into attempting a wide negotiation on a narrow bus. Add support for task management function completions. Modify ahd_handle_devreset so that it can handle lun resets in addition to target resets. Use ahd_handle_devreset for lun and target reset task management functions. Handle the abort task TMF race case better. We now wait until any current selections are over and then set the TMF back to zero. This should cause the sequencer to ignore the abort TMF completion should it occur. Correct a bug in the illegal phase handler that caused us to drop down to narrow when handling the unexpected command phase case after 3rd party reset of a packetized device. Indicate the features, bugs, and flags set in the softc that are used to control firmware patch download when booting verbose. aic79xx.h: Add coalessing and HS_MAILBOX fields. Add per-softc variables for the stats "daemon". Add a debug option for interrupt coalessing activities. Add two new softc flags: o AHD_UPDATE_PEND_CMDS Run ahd_reset_cmds_pending() on the next unpause. o AHD_RUNNING_QOUTFIFO Used to catch recursion through ahd_run_qoutfifo(). aic79xx.reg: Correct register addresses related to the software timer and the DFDBCTL register. Add constants paramaterizing the software timer. Add scratch ram locations for storing interrupt coalessing tunables. Break INTMASK in SEQITNCTL out into INTMASK1 and INTMASK2. In at least the REV A, these are writable bits. We make use of that for a swtimer workaround in the sequencer. Since HS_MAILBOX autoclears, provide a sequencer variable to store its contents. Add SEQINT codes for handling task management completions. aic79xx.seq: Correct ignore wide residue processing check for a wide negotiation being in effect. We must be in the SCSI register window in order to access the negotiation table. Use the software timer and a commands completed count to implement interrupt coalessing. The command complete is deferred until either the maximum command threshold or a the expiration of a command deferral timer. If we have more SCBs to complete to the host (sitting in COMPLETE_SCB lists), always try to coaless them up to our coalessing limit. If coalessing is enabled, but we have fewer commands oustanting than the host's min coalessing limit, complete the command immediately. Add code to track the number of commands outstanding. Commands are outstanding from the time they are placed into the execution queue until the DMA to post completion is setup. Add a workaround for intvec_2 interrupts on the H2A4. In H2A4, the mode pointer is not saved for intvec2, but is restored on iret. This can lead to the restoration of a bogus mode ptr. Manually clear the intmask bits and do a normal return to compensate. We use intvec_2 to track interrupt coalessing timeouts. Since we cannot disable the swtimer's countdown, simply mask its interrupt once we no longer care about it firing. In idle_loop_cchan, update LOCAL_HS_MAILBOX everytime we are notified of an HS_MAILBOX update via the HS_MAILBOX_ACT bit in QOFF_CTLSTA. We have to use a local copy of persistant portions of the HS_MAILBOX as the mailbox auto-clears on any read. Move the test for the cfg4istat interrupt up an instruction to hopefully close a race between the next outgoing selection and our disabling of selections. Add a missing ret to the last instruction in load_overrun_buf. Add notifications to the host of task management completions as well as the completions for commands that completed successfully before their corresponding TMF could be sent. Hold a critical section during select-out processing until we have a fully identified connection. This removes a race condition with the legacy abort handler. Correct a few spelling errors in some comments. aic79xx_inline.h: Call ahd_reset_cmds_pending() in ahd_unpause if required. Update cmdcmplt interrupt statistics in our interrupt handler. Allow callers to ahd_send_scb() to set the task management function. aic79xx_pci.c: Disable SERR and pause the controller prior to performing our mmapped I/O test. The U320 controllers do not support "auto-access-pause". aic79xx_osm.c: Set the task management function now that ahd_send_scb() doesn't do it for us. We also perform a lun reset in response to BDR requests to packetized devices.
2003-01-20 20:17:35 +00:00
export timer_isr:
call issue_cmdcmplt;
mvi CLRSEQINTSTAT, CLRSEQ_SWTMRTO;
if ((ahd->bugs & AHD_SET_MODE_BUG) != 0) {
/*
* In H2A4, the mode pointer is not saved
* for intvec2, but is restored on iret.
* This can lead to the restoration of a
* bogus mode ptr. Manually clear the
* intmask bits and do a normal return
* to compensate.
*/
and SEQINTCTL, ~(INTMASK2|INTMASK1) ret;
} else {
or SEQINTCTL, IRET ret;
}
export seq_isr:
aic79xx.c: Clear the LQICRC_NLQ status should it pop up after we have already handled the SCSIPERR. During some streaming operations this status can be delayed until the stream ends. Without this change, the driver would complain about a "Missing case in ahd_handle_scsiint". In the LQOBUSFREE handler... Don't return the LQOMGR back to the idle state until after we have cleaned up ENSELO and any status related to this selection. The last thing we need is the LQO manager starting another select-out before we have updated the execution queue. It is not clear whether the LQOMGR would, or would not start a new selection early. Make sure ENSELO is off prior to clearing SELDO by flushing device writes. Move assignment of the next target SCB pointer inside of an if to make the code clearer. The effect is the same. Dump card state in both "Unexpected PKT busfree" paths. In ahd_reset(), set the chip to SCSI mode before reading SXFRCTL1. That register only exists in the SCSI mode. Also set the mode explicitly to the SCSI mode after chip reset due to paranoia. Re-arrange code so that SXFRCTL1 is restored as quickly after the chip reset as possible. S/G structurs must be 8byte aligned. Make this official by saying so in our DMA tag. Disable CIO bus stretch on MDFFSTAT if SHVALID is about to come true. This can cause a CIO bus lockup if a PCI or PCI-X error occurs while the stretch is occurring - the host cannot service the PCI-X error since the CIO bus is locked out and SHVALID will never resolve. The stretch was added in the Rev B to simplify the wait for SHVALID to resolve, but the code to do this in the open source sequencer is so simple it was never removed. Consistently use MAX_OFFSET for the user max syncrate set from non-volatile storage. This ensures that the offset does not conflict with AH?_OFFSET_UNKNOWN. Have ahd_pause_and_flushwork set the mode to ensure that it has access to the registers it checks. Also modify the checking of intstat so that the check against 0xFF can actually succeed if the INT_PEND mask is something other than 0xFF. Although there are no cardbus U320 controllers, this check may be needed to recover from a hot-plug PCI removal that occurs without informing the driver. Fix a typo. sg_prefetch_cnt -> sg_prefetch_align. This fixes an infinite loop at card initialization if the cacheline size is 0. aic79xx.h: Add AHD_EARLY_REQ_BUG bug flag. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. Change AH?_OFFSET_UNKNOWN to 0xFF. This is a value that the curr->offset can never be, unlike '0' which we previously used. This fixes code that only checks for a non-zero offset to determine if a sync negotiation is required since it will fire in the unknown case even if the goal is async. aic79xx.reg: Add comments for LQISTAT bits indicating their names in the 7902 data book. We use slightly different and more descriptive names in the firmware. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. aic79xx.seq: Update comments regarding rundown of the GSFIFO to reflect reality. Fix spelling errors. Since we use an 8byte address and 1 byte length, shorten the size of a block move for the legacy DMA'ed CDB case from 11 to 9 bytes. Remove code that, assuming the abort pending feature worked, would set MK_MESSAGE in the SCB's control byte on completion to catch invalid reselections. Since we don't see interrupts for completed selections, this status update could occur prior to us noticing the SELDO. The "select-out" queue logic will get confused by the MK_MESSAGE bit being set as this is used to catch packatized connections where we select-out with ATN. Since the abort pending feature doesn't work on any released controllers yet, this code was never executed. Add support for the AHD_EARLY_REQ_BUG. Don't ignore persistent REQ assertions just because they were asserted within the bus settle delay window. This allows us to tolerate devices like the GEM318 that violate the SCSI spec. Remove unintentional settnig of SG_CACHE_AVAIL. Writing this bit should have no effect, but who knows... On the Rev A, we must wait for HDMAENACK before loading additional segments to avoid clobbering the address of the first segment in the S/G FIFO. This resolves data-corruption issues with certain IBM (now Hitachi) and Fujitsu U320 drives. Rearrange calc_residual to avoid an extra jmp instruction. On RevA Silicon, if the target returns us to data-out after we have already trained for data-out, it is possible for us to transition the free running clock to data-valid before the required 100ns P1 setup time (8 P1 assertions in fast-160 mode). This will only happen if this L-Q is a continuation of a data transfer for which we have already prefetched data into our FIFO (LQ/Data followed by LQ/Data for the same write transaction). This can cause some target implementations to miss the first few data transfers on the bus. We detect this situation by noticing that this is the first data transfer after an LQ (LQIWORKONLQ true), that the data transfer is a continuation of a transfer already setup in our FIFO (SAVEPTRS interrupt), and that the transaction is a write (DIRECTION set in DFCNTRL). The delay is performed by disabling SCSIEN until we see the first REQ from the target. Only compile in snapshot savepointers handler for RevA silicon where it is enabled. Handle the cfg4icmd packetized interrupt. We just need to load the address and count, start the DMA, and CLRCHN once the transfer is complete. Fix an oversight in the overrun handler for packetized status operations. We need to wait for either CTXTDONE or an overrun when checking for an overrun. The previous code did not wait and thus could decide that no overrun had occurred even though an overrun will occur on the next data-valid req. Add some comment to this section for clarity. Use LAST_SEG_DONE instead of LASTSDONE for testing transfer completion in the packetized status case. LASTSDONE may come up more quickly since it only records completion on the SCSI side, but since LAST_SEG_DONE is used everywhere else (and needs to be), this is less confusing. Add a missing invalidation of the longjmp address in the non-pack handler. This code needs additional review. aic79xx_inline.h: Fix spelling error. aic79xx_osm.c: Set the cdb length for CDBs dma'ed from host memory. Add a comment indicating that, should CAM start supporting cdbs larger than 16bytes, the driver could store the CDB in the status buffer. aic79xx_pci.c: Add a table entry for the 39320A. Added a missing comma to an error string table. Fix spelling errors.
2003-02-27 23:23:16 +00:00
if ((ahd->features & AHD_RTI) == 0) {
/*
* On RevA Silicon, if the target returns us to data-out
* after we have already trained for data-out, it is
* possible for us to transition the free running clock to
* data-valid before the required 100ns P1 setup time (8 P1
* assertions in fast-160 mode). This will only happen if
* this L-Q is a continuation of a data transfer for which
* we have already prefetched data into our FIFO (LQ/Data
* followed by LQ/Data for the same write transaction).
* This can cause some target implementations to miss the
* first few data transfers on the bus. We detect this
* situation by noticing that this is the first data transfer
* after an LQ (LQIWORKONLQ true), that the data transfer is
* a continuation of a transfer already setup in our FIFO
* (SAVEPTRS interrupt), and that the transaction is a write
* (DIRECTION set in DFCNTRL). The delay is performed by
* disabling SCSIEN until we see the first REQ from the
* target.
*
* First instruction in an ISR cannot be a branch on
* Rev A. Snapshot LQISTAT2 so the status is not missed
* and deffer the test by one instruction.
*/
mov REG_ISR, LQISTAT2;
Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-05-04 00:20:07 +00:00
test REG_ISR, LQIWORKONLQ jz main_isr;
test SEQINTSRC, SAVEPTRS jz main_isr;
aic79xx.c: Clear the LQICRC_NLQ status should it pop up after we have already handled the SCSIPERR. During some streaming operations this status can be delayed until the stream ends. Without this change, the driver would complain about a "Missing case in ahd_handle_scsiint". In the LQOBUSFREE handler... Don't return the LQOMGR back to the idle state until after we have cleaned up ENSELO and any status related to this selection. The last thing we need is the LQO manager starting another select-out before we have updated the execution queue. It is not clear whether the LQOMGR would, or would not start a new selection early. Make sure ENSELO is off prior to clearing SELDO by flushing device writes. Move assignment of the next target SCB pointer inside of an if to make the code clearer. The effect is the same. Dump card state in both "Unexpected PKT busfree" paths. In ahd_reset(), set the chip to SCSI mode before reading SXFRCTL1. That register only exists in the SCSI mode. Also set the mode explicitly to the SCSI mode after chip reset due to paranoia. Re-arrange code so that SXFRCTL1 is restored as quickly after the chip reset as possible. S/G structurs must be 8byte aligned. Make this official by saying so in our DMA tag. Disable CIO bus stretch on MDFFSTAT if SHVALID is about to come true. This can cause a CIO bus lockup if a PCI or PCI-X error occurs while the stretch is occurring - the host cannot service the PCI-X error since the CIO bus is locked out and SHVALID will never resolve. The stretch was added in the Rev B to simplify the wait for SHVALID to resolve, but the code to do this in the open source sequencer is so simple it was never removed. Consistently use MAX_OFFSET for the user max syncrate set from non-volatile storage. This ensures that the offset does not conflict with AH?_OFFSET_UNKNOWN. Have ahd_pause_and_flushwork set the mode to ensure that it has access to the registers it checks. Also modify the checking of intstat so that the check against 0xFF can actually succeed if the INT_PEND mask is something other than 0xFF. Although there are no cardbus U320 controllers, this check may be needed to recover from a hot-plug PCI removal that occurs without informing the driver. Fix a typo. sg_prefetch_cnt -> sg_prefetch_align. This fixes an infinite loop at card initialization if the cacheline size is 0. aic79xx.h: Add AHD_EARLY_REQ_BUG bug flag. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. Change AH?_OFFSET_UNKNOWN to 0xFF. This is a value that the curr->offset can never be, unlike '0' which we previously used. This fixes code that only checks for a non-zero offset to determine if a sync negotiation is required since it will fire in the unknown case even if the goal is async. aic79xx.reg: Add comments for LQISTAT bits indicating their names in the 7902 data book. We use slightly different and more descriptive names in the firmware. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. aic79xx.seq: Update comments regarding rundown of the GSFIFO to reflect reality. Fix spelling errors. Since we use an 8byte address and 1 byte length, shorten the size of a block move for the legacy DMA'ed CDB case from 11 to 9 bytes. Remove code that, assuming the abort pending feature worked, would set MK_MESSAGE in the SCB's control byte on completion to catch invalid reselections. Since we don't see interrupts for completed selections, this status update could occur prior to us noticing the SELDO. The "select-out" queue logic will get confused by the MK_MESSAGE bit being set as this is used to catch packatized connections where we select-out with ATN. Since the abort pending feature doesn't work on any released controllers yet, this code was never executed. Add support for the AHD_EARLY_REQ_BUG. Don't ignore persistent REQ assertions just because they were asserted within the bus settle delay window. This allows us to tolerate devices like the GEM318 that violate the SCSI spec. Remove unintentional settnig of SG_CACHE_AVAIL. Writing this bit should have no effect, but who knows... On the Rev A, we must wait for HDMAENACK before loading additional segments to avoid clobbering the address of the first segment in the S/G FIFO. This resolves data-corruption issues with certain IBM (now Hitachi) and Fujitsu U320 drives. Rearrange calc_residual to avoid an extra jmp instruction. On RevA Silicon, if the target returns us to data-out after we have already trained for data-out, it is possible for us to transition the free running clock to data-valid before the required 100ns P1 setup time (8 P1 assertions in fast-160 mode). This will only happen if this L-Q is a continuation of a data transfer for which we have already prefetched data into our FIFO (LQ/Data followed by LQ/Data for the same write transaction). This can cause some target implementations to miss the first few data transfers on the bus. We detect this situation by noticing that this is the first data transfer after an LQ (LQIWORKONLQ true), that the data transfer is a continuation of a transfer already setup in our FIFO (SAVEPTRS interrupt), and that the transaction is a write (DIRECTION set in DFCNTRL). The delay is performed by disabling SCSIEN until we see the first REQ from the target. Only compile in snapshot savepointers handler for RevA silicon where it is enabled. Handle the cfg4icmd packetized interrupt. We just need to load the address and count, start the DMA, and CLRCHN once the transfer is complete. Fix an oversight in the overrun handler for packetized status operations. We need to wait for either CTXTDONE or an overrun when checking for an overrun. The previous code did not wait and thus could decide that no overrun had occurred even though an overrun will occur on the next data-valid req. Add some comment to this section for clarity. Use LAST_SEG_DONE instead of LASTSDONE for testing transfer completion in the packetized status case. LASTSDONE may come up more quickly since it only records completion on the SCSI side, but since LAST_SEG_DONE is used everywhere else (and needs to be), this is less confusing. Add a missing invalidation of the longjmp address in the non-pack handler. This code needs additional review. aic79xx_inline.h: Fix spelling error. aic79xx_osm.c: Set the cdb length for CDBs dma'ed from host memory. Add a comment indicating that, should CAM start supporting cdbs larger than 16bytes, the driver could store the CDB in the status buffer. aic79xx_pci.c: Add a table entry for the 39320A. Added a missing comma to an error string table. Fix spelling errors.
2003-02-27 23:23:16 +00:00
test LONGJMP_ADDR[1], INVALID_ADDR jz saveptr_active_fifo;
/*
Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-05-04 00:20:07 +00:00
* Switch to the active FIFO after clearing the snapshot
* savepointer in the current FIFO. We do this so that
* a pending CTXTDONE or SAVEPTR is visible in the active
* FIFO. This status is the only way we can detect if we
ahc_eisa.c: ahc_pci.c: ahd_pci.c: aic7xxx.c: aic79xx.c: aic_osm_lib.c: aic_osm_lib.h: Use common OSM routines from aic_osm_lib for bus dma operations, delay routines, accessing CCBs, byte swapping, etc. aic7xxx_pci.c: Provide a better description for the 2915/30LP on attach. aic7xxx.c: aic79xx.c: aic7770.c: aic79xx_pci.c: aic7xxx_pci.c: aic7xxx_93cx6.c: Move FBSDID behind an ifdef so that these core files will still compile under other OSes. aic79xx.h: aic79xx_pci.c: aic79xx.seq: To speed up non-packetized CDB delivery in Rev B, all CDB acks are "released" to the output sync as soon as the command phase starts. There is only one problem with this approach. If the target changes phase before all data are sent, we have left over acks that can go out on the bus in a data phase. Due to other chip contraints, this only happens if the target goes to data-in, but if the acks go out before we can test SDONE, we'll think that the transfer has completed successfully. Work around this by taking advantage of the 400ns or 800ns dead time between command phase and the REQ of the new phase. If the transfer has completed successfully, SCSIEN should fall *long* before we see a phase change. We thus treat any phasemiss that occurs before SCSIEN falls as an incomplete transfer. aic79xx.h: Add the AHD_FAST_CDB_DELIVERY feature. aic79xx_pci.c: Set AHD_FAST_CDB_DELIVERY for all Rev. B parts. aic79xx.seq: Test for PHASEMIS in the command phase for all AHD_FAST_CDB_DELIVERY controlelrs. ahd_pci.c: ahc_pci.c: aic7xxx.h: aic79xx.h: Move definition of controller BAR offsets to core header files. aic7xxx.c: aic79xx.c: In the softc free routine, leave removal of a softc from the global list of softcs to the OSM (the caller of this routine). This allows us to avoid holding the softc list_lock during device destruction where we may have to sleep waiting for our recovery thread to halt. ahc_pci.c: Use ahc_pci_test_register access to validate I/O mapped in addition to the tests already performed for memory mapped access. Remove unused ahc_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. ahd_pci.c: Remove reduntant definition of controller BAR offsets. These are also defined in aic79xx.h. Remove unused ahd_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. aic7xxx.c: aic79xx.c: aic79xx.h: aic7xxx.h: aic7xxx_osm.c: aic79xx_osm.c: Move timeout handling to the driver cores. In the case of the aic79xx driver, the algorithm has been enhanced to try target resets before performing a bus reset. For the aic7xxx driver, the algorithm is unchanged. Although the drivers do not currently sleep during recovery (recovery is timeout driven), the cores do expect all processing to be performed via a recovery thread. Our timeout handlers are now little stubs that wakeup the recovery thread. aic79xx.c: aic79xx.h: aic79xx_inline.h: Change shared_data allocation to use a map_node so that the sentinel hscb can use this map node in ahd_swap_with_next_hscb. This routine now swaps the hscb_map pointer in additon to the hscb contents so that any sync operations occur on the correct map. physaddr -> busaddr Pointed out by: Jason Thorpe <thorpej@wasabisystems.com> aic79xx.c: Make more use of the in/out/w/l/q macros for accessing byte registers in the chip. Correct some issues in the ahd_flush_qoutfifo() routine. o Run the qoutfifo only once the command channel DMA engine has been halted. This closes a window where we might have missed some entries. o Change ahd_run_data_fifo() to not loop to completion. If we happen to start on the wrong FIFO and the other FIFO has a snapshot savepointers, we might deadlock. This required our delay between FIFO tests to be moved to the ahd_flush_qoutfifo() routine. o Update/add comments. o Remove spurious test for COMPLETE_DMA list being empty when completing transactions from the GSFIFO with residuals. The SCB must be put on the COMPLETE_DMA scb list unconditionally. o When halting command channel DMA activity, we must disable the DMA channel in all cases but an update of the QOUTFIFO. The latter case is required so that the sequencer will update its position in the QOUTFIFO. Previously, we left the channel enabled for all "push" DMAs. This left us vulnerable to the sequencer handling an SCB push long after that SCB was already processed manually by this routine. o Correct the polarity of tests involving ahd_scb_active_in_fifo(). This routine returns non-zero for true. Return to processing bad status completions through the qoutfifo. This reduces the time that the sequencer is kept paused when handling transactions with bad status or underruns. When waiting for the controller to quiece selections, add a delay to our loop. Otherwise we may fail to wait long enough for the sequencer to comply. On H2A4 hardware, use the slow slewrate for non-paced transfers. This mirrors what the Adaptec Windows drivers do. On the Rev B. only slow down the CRC timing for older U160 devices that might need the slower timing. We define "older" as devices that do not support packetized protocol. Wait up to 5000 * 5us for the SEEPROM to become unbusy. Write ops seem to take much longer than read ops. aic79xx.seq: For controllers with the FAINT_LED bug, turn the diagnostic led feature on during selection and reselection. This covers the non-packetized case. The LED will be disabled for non-packetized transfers once we return to the top level idle loop. Add more comments about the busy LED workaround. Extend a critical section around the entire command channel idle loop process. Previously the portion of this handler that directly manipulated the linked list of completed SCBs was not protected. This is the likely cause of the recent reports of commands being completed twice by the driver. Extend critical sections across the test for, and the longjump to, longjump routines. This prevents the firmware from trying to jump to a longjmp handler that was just cleared by the host. Improve the locations of several critical section begin and end points. Typically these changes remove instructions that did not need to be inside a critical section. Close the "busfree after selection, but before busfree interrupts can be enabled" race to just a single sequencer instruction. We now test the BSY line explicitly before clearing the busfree status and enabling the busfree interrupt. Close a race condition in the processing of HS_MAILBOX updates. We now clear the "updated" status before the copy. This ensures that we don't accidentally clear the status incorrectly when the host sneaks in an update just after our last copy, but before we clear the status. This race has never been observed. Don't re-enable SCSIEN if we lose the race to disable SCSIEN in our interrupt handler's workaround for the RevA data-valid too early issue. aic79xx_inline.h: Add comments indicating that the order in which bytes are read or written in ahd_inw and ahd_outw is important. This allows us to use these inlines when accessing registers with side-effects. aic79xx_pci.c: The 29320 and the 29320B are 7902 not 7901 based products. Correct the driver banner. aic7xxx.h: Enable the use of the auto-access pause feature on the aic7870 and aic7880. It was disabled due to an oversight. aic7xxx.reg: Move TARG_IMMEDIATE_SCB to alias LAST_MSG to avoid leaving garbage in MWI_RESIDUAL. This prevents spurious overflows whn operating target mode on controllers that require the MWI_RESIDUAL work-around. aic7xxx.seq: AHC_TMODE_WIDEODD_BUG is a bug, not a softc flag. Reference the correct softc field when testing for its presence. Set the NOT_IDENTIFIED and NO_CDB_SENT bits in SEQ_FLAGS to indicate that the nexus is invalid in await busfree. aic7xxx_93cx6.c: Add support for the C56/C66 versions of the EWEN and EWDS commands. aic7xxx.c: aic7xxx_pci.c: Move test for the validity of left over BIOS data to ahc_test_register_access(). This guarantees that any left over CHIPRST value is not clobbered by our register access test and lost to the test that was in ahc_reset.
2003-12-17 00:02:10 +00:00
* have lost the race (e.g. host paused us) and our attempts
Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-05-04 00:20:07 +00:00
* to disable the channel occurred after all REQs were
* already seen and acked (REQINIT never comes true).
aic79xx.c: Clear the LQICRC_NLQ status should it pop up after we have already handled the SCSIPERR. During some streaming operations this status can be delayed until the stream ends. Without this change, the driver would complain about a "Missing case in ahd_handle_scsiint". In the LQOBUSFREE handler... Don't return the LQOMGR back to the idle state until after we have cleaned up ENSELO and any status related to this selection. The last thing we need is the LQO manager starting another select-out before we have updated the execution queue. It is not clear whether the LQOMGR would, or would not start a new selection early. Make sure ENSELO is off prior to clearing SELDO by flushing device writes. Move assignment of the next target SCB pointer inside of an if to make the code clearer. The effect is the same. Dump card state in both "Unexpected PKT busfree" paths. In ahd_reset(), set the chip to SCSI mode before reading SXFRCTL1. That register only exists in the SCSI mode. Also set the mode explicitly to the SCSI mode after chip reset due to paranoia. Re-arrange code so that SXFRCTL1 is restored as quickly after the chip reset as possible. S/G structurs must be 8byte aligned. Make this official by saying so in our DMA tag. Disable CIO bus stretch on MDFFSTAT if SHVALID is about to come true. This can cause a CIO bus lockup if a PCI or PCI-X error occurs while the stretch is occurring - the host cannot service the PCI-X error since the CIO bus is locked out and SHVALID will never resolve. The stretch was added in the Rev B to simplify the wait for SHVALID to resolve, but the code to do this in the open source sequencer is so simple it was never removed. Consistently use MAX_OFFSET for the user max syncrate set from non-volatile storage. This ensures that the offset does not conflict with AH?_OFFSET_UNKNOWN. Have ahd_pause_and_flushwork set the mode to ensure that it has access to the registers it checks. Also modify the checking of intstat so that the check against 0xFF can actually succeed if the INT_PEND mask is something other than 0xFF. Although there are no cardbus U320 controllers, this check may be needed to recover from a hot-plug PCI removal that occurs without informing the driver. Fix a typo. sg_prefetch_cnt -> sg_prefetch_align. This fixes an infinite loop at card initialization if the cacheline size is 0. aic79xx.h: Add AHD_EARLY_REQ_BUG bug flag. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. Change AH?_OFFSET_UNKNOWN to 0xFF. This is a value that the curr->offset can never be, unlike '0' which we previously used. This fixes code that only checks for a non-zero offset to determine if a sync negotiation is required since it will fire in the unknown case even if the goal is async. aic79xx.reg: Add comments for LQISTAT bits indicating their names in the 7902 data book. We use slightly different and more descriptive names in the firmware. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. aic79xx.seq: Update comments regarding rundown of the GSFIFO to reflect reality. Fix spelling errors. Since we use an 8byte address and 1 byte length, shorten the size of a block move for the legacy DMA'ed CDB case from 11 to 9 bytes. Remove code that, assuming the abort pending feature worked, would set MK_MESSAGE in the SCB's control byte on completion to catch invalid reselections. Since we don't see interrupts for completed selections, this status update could occur prior to us noticing the SELDO. The "select-out" queue logic will get confused by the MK_MESSAGE bit being set as this is used to catch packatized connections where we select-out with ATN. Since the abort pending feature doesn't work on any released controllers yet, this code was never executed. Add support for the AHD_EARLY_REQ_BUG. Don't ignore persistent REQ assertions just because they were asserted within the bus settle delay window. This allows us to tolerate devices like the GEM318 that violate the SCSI spec. Remove unintentional settnig of SG_CACHE_AVAIL. Writing this bit should have no effect, but who knows... On the Rev A, we must wait for HDMAENACK before loading additional segments to avoid clobbering the address of the first segment in the S/G FIFO. This resolves data-corruption issues with certain IBM (now Hitachi) and Fujitsu U320 drives. Rearrange calc_residual to avoid an extra jmp instruction. On RevA Silicon, if the target returns us to data-out after we have already trained for data-out, it is possible for us to transition the free running clock to data-valid before the required 100ns P1 setup time (8 P1 assertions in fast-160 mode). This will only happen if this L-Q is a continuation of a data transfer for which we have already prefetched data into our FIFO (LQ/Data followed by LQ/Data for the same write transaction). This can cause some target implementations to miss the first few data transfers on the bus. We detect this situation by noticing that this is the first data transfer after an LQ (LQIWORKONLQ true), that the data transfer is a continuation of a transfer already setup in our FIFO (SAVEPTRS interrupt), and that the transaction is a write (DIRECTION set in DFCNTRL). The delay is performed by disabling SCSIEN until we see the first REQ from the target. Only compile in snapshot savepointers handler for RevA silicon where it is enabled. Handle the cfg4icmd packetized interrupt. We just need to load the address and count, start the DMA, and CLRCHN once the transfer is complete. Fix an oversight in the overrun handler for packetized status operations. We need to wait for either CTXTDONE or an overrun when checking for an overrun. The previous code did not wait and thus could decide that no overrun had occurred even though an overrun will occur on the next data-valid req. Add some comment to this section for clarity. Use LAST_SEG_DONE instead of LASTSDONE for testing transfer completion in the packetized status case. LASTSDONE may come up more quickly since it only records completion on the SCSI side, but since LAST_SEG_DONE is used everywhere else (and needs to be), this is less confusing. Add a missing invalidation of the longjmp address in the non-pack handler. This code needs additional review. aic79xx_inline.h: Fix spelling error. aic79xx_osm.c: Set the cdb length for CDBs dma'ed from host memory. Add a comment indicating that, should CAM start supporting cdbs larger than 16bytes, the driver could store the CDB in the status buffer. aic79xx_pci.c: Add a table entry for the 39320A. Added a missing comma to an error string table. Fix spelling errors.
2003-02-27 23:23:16 +00:00
*/
Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-05-04 00:20:07 +00:00
mvi DFFSXFRCTL, CLRCHN;
aic79xx.c: Clear the LQICRC_NLQ status should it pop up after we have already handled the SCSIPERR. During some streaming operations this status can be delayed until the stream ends. Without this change, the driver would complain about a "Missing case in ahd_handle_scsiint". In the LQOBUSFREE handler... Don't return the LQOMGR back to the idle state until after we have cleaned up ENSELO and any status related to this selection. The last thing we need is the LQO manager starting another select-out before we have updated the execution queue. It is not clear whether the LQOMGR would, or would not start a new selection early. Make sure ENSELO is off prior to clearing SELDO by flushing device writes. Move assignment of the next target SCB pointer inside of an if to make the code clearer. The effect is the same. Dump card state in both "Unexpected PKT busfree" paths. In ahd_reset(), set the chip to SCSI mode before reading SXFRCTL1. That register only exists in the SCSI mode. Also set the mode explicitly to the SCSI mode after chip reset due to paranoia. Re-arrange code so that SXFRCTL1 is restored as quickly after the chip reset as possible. S/G structurs must be 8byte aligned. Make this official by saying so in our DMA tag. Disable CIO bus stretch on MDFFSTAT if SHVALID is about to come true. This can cause a CIO bus lockup if a PCI or PCI-X error occurs while the stretch is occurring - the host cannot service the PCI-X error since the CIO bus is locked out and SHVALID will never resolve. The stretch was added in the Rev B to simplify the wait for SHVALID to resolve, but the code to do this in the open source sequencer is so simple it was never removed. Consistently use MAX_OFFSET for the user max syncrate set from non-volatile storage. This ensures that the offset does not conflict with AH?_OFFSET_UNKNOWN. Have ahd_pause_and_flushwork set the mode to ensure that it has access to the registers it checks. Also modify the checking of intstat so that the check against 0xFF can actually succeed if the INT_PEND mask is something other than 0xFF. Although there are no cardbus U320 controllers, this check may be needed to recover from a hot-plug PCI removal that occurs without informing the driver. Fix a typo. sg_prefetch_cnt -> sg_prefetch_align. This fixes an infinite loop at card initialization if the cacheline size is 0. aic79xx.h: Add AHD_EARLY_REQ_BUG bug flag. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. Change AH?_OFFSET_UNKNOWN to 0xFF. This is a value that the curr->offset can never be, unlike '0' which we previously used. This fixes code that only checks for a non-zero offset to determine if a sync negotiation is required since it will fire in the unknown case even if the goal is async. aic79xx.reg: Add comments for LQISTAT bits indicating their names in the 7902 data book. We use slightly different and more descriptive names in the firmware. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. aic79xx.seq: Update comments regarding rundown of the GSFIFO to reflect reality. Fix spelling errors. Since we use an 8byte address and 1 byte length, shorten the size of a block move for the legacy DMA'ed CDB case from 11 to 9 bytes. Remove code that, assuming the abort pending feature worked, would set MK_MESSAGE in the SCB's control byte on completion to catch invalid reselections. Since we don't see interrupts for completed selections, this status update could occur prior to us noticing the SELDO. The "select-out" queue logic will get confused by the MK_MESSAGE bit being set as this is used to catch packatized connections where we select-out with ATN. Since the abort pending feature doesn't work on any released controllers yet, this code was never executed. Add support for the AHD_EARLY_REQ_BUG. Don't ignore persistent REQ assertions just because they were asserted within the bus settle delay window. This allows us to tolerate devices like the GEM318 that violate the SCSI spec. Remove unintentional settnig of SG_CACHE_AVAIL. Writing this bit should have no effect, but who knows... On the Rev A, we must wait for HDMAENACK before loading additional segments to avoid clobbering the address of the first segment in the S/G FIFO. This resolves data-corruption issues with certain IBM (now Hitachi) and Fujitsu U320 drives. Rearrange calc_residual to avoid an extra jmp instruction. On RevA Silicon, if the target returns us to data-out after we have already trained for data-out, it is possible for us to transition the free running clock to data-valid before the required 100ns P1 setup time (8 P1 assertions in fast-160 mode). This will only happen if this L-Q is a continuation of a data transfer for which we have already prefetched data into our FIFO (LQ/Data followed by LQ/Data for the same write transaction). This can cause some target implementations to miss the first few data transfers on the bus. We detect this situation by noticing that this is the first data transfer after an LQ (LQIWORKONLQ true), that the data transfer is a continuation of a transfer already setup in our FIFO (SAVEPTRS interrupt), and that the transaction is a write (DIRECTION set in DFCNTRL). The delay is performed by disabling SCSIEN until we see the first REQ from the target. Only compile in snapshot savepointers handler for RevA silicon where it is enabled. Handle the cfg4icmd packetized interrupt. We just need to load the address and count, start the DMA, and CLRCHN once the transfer is complete. Fix an oversight in the overrun handler for packetized status operations. We need to wait for either CTXTDONE or an overrun when checking for an overrun. The previous code did not wait and thus could decide that no overrun had occurred even though an overrun will occur on the next data-valid req. Add some comment to this section for clarity. Use LAST_SEG_DONE instead of LASTSDONE for testing transfer completion in the packetized status case. LASTSDONE may come up more quickly since it only records completion on the SCSI side, but since LAST_SEG_DONE is used everywhere else (and needs to be), this is less confusing. Add a missing invalidation of the longjmp address in the non-pack handler. This code needs additional review. aic79xx_inline.h: Fix spelling error. aic79xx_osm.c: Set the cdb length for CDBs dma'ed from host memory. Add a comment indicating that, should CAM start supporting cdbs larger than 16bytes, the driver could store the CDB in the status buffer. aic79xx_pci.c: Add a table entry for the 39320A. Added a missing comma to an error string table. Fix spelling errors.
2003-02-27 23:23:16 +00:00
xor MODE_PTR, MK_MODE(M_DFF1, M_DFF1);
Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-05-04 00:20:07 +00:00
test DFCNTRL, DIRECTION jz interrupt_return;
aic79xx.c: Clear the LQICRC_NLQ status should it pop up after we have already handled the SCSIPERR. During some streaming operations this status can be delayed until the stream ends. Without this change, the driver would complain about a "Missing case in ahd_handle_scsiint". In the LQOBUSFREE handler... Don't return the LQOMGR back to the idle state until after we have cleaned up ENSELO and any status related to this selection. The last thing we need is the LQO manager starting another select-out before we have updated the execution queue. It is not clear whether the LQOMGR would, or would not start a new selection early. Make sure ENSELO is off prior to clearing SELDO by flushing device writes. Move assignment of the next target SCB pointer inside of an if to make the code clearer. The effect is the same. Dump card state in both "Unexpected PKT busfree" paths. In ahd_reset(), set the chip to SCSI mode before reading SXFRCTL1. That register only exists in the SCSI mode. Also set the mode explicitly to the SCSI mode after chip reset due to paranoia. Re-arrange code so that SXFRCTL1 is restored as quickly after the chip reset as possible. S/G structurs must be 8byte aligned. Make this official by saying so in our DMA tag. Disable CIO bus stretch on MDFFSTAT if SHVALID is about to come true. This can cause a CIO bus lockup if a PCI or PCI-X error occurs while the stretch is occurring - the host cannot service the PCI-X error since the CIO bus is locked out and SHVALID will never resolve. The stretch was added in the Rev B to simplify the wait for SHVALID to resolve, but the code to do this in the open source sequencer is so simple it was never removed. Consistently use MAX_OFFSET for the user max syncrate set from non-volatile storage. This ensures that the offset does not conflict with AH?_OFFSET_UNKNOWN. Have ahd_pause_and_flushwork set the mode to ensure that it has access to the registers it checks. Also modify the checking of intstat so that the check against 0xFF can actually succeed if the INT_PEND mask is something other than 0xFF. Although there are no cardbus U320 controllers, this check may be needed to recover from a hot-plug PCI removal that occurs without informing the driver. Fix a typo. sg_prefetch_cnt -> sg_prefetch_align. This fixes an infinite loop at card initialization if the cacheline size is 0. aic79xx.h: Add AHD_EARLY_REQ_BUG bug flag. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. Change AH?_OFFSET_UNKNOWN to 0xFF. This is a value that the curr->offset can never be, unlike '0' which we previously used. This fixes code that only checks for a non-zero offset to determine if a sync negotiation is required since it will fire in the unknown case even if the goal is async. aic79xx.reg: Add comments for LQISTAT bits indicating their names in the 7902 data book. We use slightly different and more descriptive names in the firmware. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. aic79xx.seq: Update comments regarding rundown of the GSFIFO to reflect reality. Fix spelling errors. Since we use an 8byte address and 1 byte length, shorten the size of a block move for the legacy DMA'ed CDB case from 11 to 9 bytes. Remove code that, assuming the abort pending feature worked, would set MK_MESSAGE in the SCB's control byte on completion to catch invalid reselections. Since we don't see interrupts for completed selections, this status update could occur prior to us noticing the SELDO. The "select-out" queue logic will get confused by the MK_MESSAGE bit being set as this is used to catch packatized connections where we select-out with ATN. Since the abort pending feature doesn't work on any released controllers yet, this code was never executed. Add support for the AHD_EARLY_REQ_BUG. Don't ignore persistent REQ assertions just because they were asserted within the bus settle delay window. This allows us to tolerate devices like the GEM318 that violate the SCSI spec. Remove unintentional settnig of SG_CACHE_AVAIL. Writing this bit should have no effect, but who knows... On the Rev A, we must wait for HDMAENACK before loading additional segments to avoid clobbering the address of the first segment in the S/G FIFO. This resolves data-corruption issues with certain IBM (now Hitachi) and Fujitsu U320 drives. Rearrange calc_residual to avoid an extra jmp instruction. On RevA Silicon, if the target returns us to data-out after we have already trained for data-out, it is possible for us to transition the free running clock to data-valid before the required 100ns P1 setup time (8 P1 assertions in fast-160 mode). This will only happen if this L-Q is a continuation of a data transfer for which we have already prefetched data into our FIFO (LQ/Data followed by LQ/Data for the same write transaction). This can cause some target implementations to miss the first few data transfers on the bus. We detect this situation by noticing that this is the first data transfer after an LQ (LQIWORKONLQ true), that the data transfer is a continuation of a transfer already setup in our FIFO (SAVEPTRS interrupt), and that the transaction is a write (DIRECTION set in DFCNTRL). The delay is performed by disabling SCSIEN until we see the first REQ from the target. Only compile in snapshot savepointers handler for RevA silicon where it is enabled. Handle the cfg4icmd packetized interrupt. We just need to load the address and count, start the DMA, and CLRCHN once the transfer is complete. Fix an oversight in the overrun handler for packetized status operations. We need to wait for either CTXTDONE or an overrun when checking for an overrun. The previous code did not wait and thus could decide that no overrun had occurred even though an overrun will occur on the next data-valid req. Add some comment to this section for clarity. Use LAST_SEG_DONE instead of LASTSDONE for testing transfer completion in the packetized status case. LASTSDONE may come up more quickly since it only records completion on the SCSI side, but since LAST_SEG_DONE is used everywhere else (and needs to be), this is less confusing. Add a missing invalidation of the longjmp address in the non-pack handler. This code needs additional review. aic79xx_inline.h: Fix spelling error. aic79xx_osm.c: Set the cdb length for CDBs dma'ed from host memory. Add a comment indicating that, should CAM start supporting cdbs larger than 16bytes, the driver could store the CDB in the status buffer. aic79xx_pci.c: Add a table entry for the 39320A. Added a missing comma to an error string table. Fix spelling errors.
2003-02-27 23:23:16 +00:00
and DFCNTRL, ~SCSIEN;
Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-05-04 00:20:07 +00:00
snapshot_wait_data_valid:
ahc_eisa.c: ahc_pci.c: ahd_pci.c: aic7xxx.c: aic79xx.c: aic_osm_lib.c: aic_osm_lib.h: Use common OSM routines from aic_osm_lib for bus dma operations, delay routines, accessing CCBs, byte swapping, etc. aic7xxx_pci.c: Provide a better description for the 2915/30LP on attach. aic7xxx.c: aic79xx.c: aic7770.c: aic79xx_pci.c: aic7xxx_pci.c: aic7xxx_93cx6.c: Move FBSDID behind an ifdef so that these core files will still compile under other OSes. aic79xx.h: aic79xx_pci.c: aic79xx.seq: To speed up non-packetized CDB delivery in Rev B, all CDB acks are "released" to the output sync as soon as the command phase starts. There is only one problem with this approach. If the target changes phase before all data are sent, we have left over acks that can go out on the bus in a data phase. Due to other chip contraints, this only happens if the target goes to data-in, but if the acks go out before we can test SDONE, we'll think that the transfer has completed successfully. Work around this by taking advantage of the 400ns or 800ns dead time between command phase and the REQ of the new phase. If the transfer has completed successfully, SCSIEN should fall *long* before we see a phase change. We thus treat any phasemiss that occurs before SCSIEN falls as an incomplete transfer. aic79xx.h: Add the AHD_FAST_CDB_DELIVERY feature. aic79xx_pci.c: Set AHD_FAST_CDB_DELIVERY for all Rev. B parts. aic79xx.seq: Test for PHASEMIS in the command phase for all AHD_FAST_CDB_DELIVERY controlelrs. ahd_pci.c: ahc_pci.c: aic7xxx.h: aic79xx.h: Move definition of controller BAR offsets to core header files. aic7xxx.c: aic79xx.c: In the softc free routine, leave removal of a softc from the global list of softcs to the OSM (the caller of this routine). This allows us to avoid holding the softc list_lock during device destruction where we may have to sleep waiting for our recovery thread to halt. ahc_pci.c: Use ahc_pci_test_register access to validate I/O mapped in addition to the tests already performed for memory mapped access. Remove unused ahc_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. ahd_pci.c: Remove reduntant definition of controller BAR offsets. These are also defined in aic79xx.h. Remove unused ahd_power_state_change() function. The PCI layer in both 4.X and 5.X now offer this functionality. aic7xxx.c: aic79xx.c: aic79xx.h: aic7xxx.h: aic7xxx_osm.c: aic79xx_osm.c: Move timeout handling to the driver cores. In the case of the aic79xx driver, the algorithm has been enhanced to try target resets before performing a bus reset. For the aic7xxx driver, the algorithm is unchanged. Although the drivers do not currently sleep during recovery (recovery is timeout driven), the cores do expect all processing to be performed via a recovery thread. Our timeout handlers are now little stubs that wakeup the recovery thread. aic79xx.c: aic79xx.h: aic79xx_inline.h: Change shared_data allocation to use a map_node so that the sentinel hscb can use this map node in ahd_swap_with_next_hscb. This routine now swaps the hscb_map pointer in additon to the hscb contents so that any sync operations occur on the correct map. physaddr -> busaddr Pointed out by: Jason Thorpe <thorpej@wasabisystems.com> aic79xx.c: Make more use of the in/out/w/l/q macros for accessing byte registers in the chip. Correct some issues in the ahd_flush_qoutfifo() routine. o Run the qoutfifo only once the command channel DMA engine has been halted. This closes a window where we might have missed some entries. o Change ahd_run_data_fifo() to not loop to completion. If we happen to start on the wrong FIFO and the other FIFO has a snapshot savepointers, we might deadlock. This required our delay between FIFO tests to be moved to the ahd_flush_qoutfifo() routine. o Update/add comments. o Remove spurious test for COMPLETE_DMA list being empty when completing transactions from the GSFIFO with residuals. The SCB must be put on the COMPLETE_DMA scb list unconditionally. o When halting command channel DMA activity, we must disable the DMA channel in all cases but an update of the QOUTFIFO. The latter case is required so that the sequencer will update its position in the QOUTFIFO. Previously, we left the channel enabled for all "push" DMAs. This left us vulnerable to the sequencer handling an SCB push long after that SCB was already processed manually by this routine. o Correct the polarity of tests involving ahd_scb_active_in_fifo(). This routine returns non-zero for true. Return to processing bad status completions through the qoutfifo. This reduces the time that the sequencer is kept paused when handling transactions with bad status or underruns. When waiting for the controller to quiece selections, add a delay to our loop. Otherwise we may fail to wait long enough for the sequencer to comply. On H2A4 hardware, use the slow slewrate for non-paced transfers. This mirrors what the Adaptec Windows drivers do. On the Rev B. only slow down the CRC timing for older U160 devices that might need the slower timing. We define "older" as devices that do not support packetized protocol. Wait up to 5000 * 5us for the SEEPROM to become unbusy. Write ops seem to take much longer than read ops. aic79xx.seq: For controllers with the FAINT_LED bug, turn the diagnostic led feature on during selection and reselection. This covers the non-packetized case. The LED will be disabled for non-packetized transfers once we return to the top level idle loop. Add more comments about the busy LED workaround. Extend a critical section around the entire command channel idle loop process. Previously the portion of this handler that directly manipulated the linked list of completed SCBs was not protected. This is the likely cause of the recent reports of commands being completed twice by the driver. Extend critical sections across the test for, and the longjump to, longjump routines. This prevents the firmware from trying to jump to a longjmp handler that was just cleared by the host. Improve the locations of several critical section begin and end points. Typically these changes remove instructions that did not need to be inside a critical section. Close the "busfree after selection, but before busfree interrupts can be enabled" race to just a single sequencer instruction. We now test the BSY line explicitly before clearing the busfree status and enabling the busfree interrupt. Close a race condition in the processing of HS_MAILBOX updates. We now clear the "updated" status before the copy. This ensures that we don't accidentally clear the status incorrectly when the host sneaks in an update just after our last copy, but before we clear the status. This race has never been observed. Don't re-enable SCSIEN if we lose the race to disable SCSIEN in our interrupt handler's workaround for the RevA data-valid too early issue. aic79xx_inline.h: Add comments indicating that the order in which bytes are read or written in ahd_inw and ahd_outw is important. This allows us to use these inlines when accessing registers with side-effects. aic79xx_pci.c: The 29320 and the 29320B are 7902 not 7901 based products. Correct the driver banner. aic7xxx.h: Enable the use of the auto-access pause feature on the aic7870 and aic7880. It was disabled due to an oversight. aic7xxx.reg: Move TARG_IMMEDIATE_SCB to alias LAST_MSG to avoid leaving garbage in MWI_RESIDUAL. This prevents spurious overflows whn operating target mode on controllers that require the MWI_RESIDUAL work-around. aic7xxx.seq: AHC_TMODE_WIDEODD_BUG is a bug, not a softc flag. Reference the correct softc field when testing for its presence. Set the NOT_IDENTIFIED and NO_CDB_SENT bits in SEQ_FLAGS to indicate that the nexus is invalid in await busfree. aic7xxx_93cx6.c: Add support for the C56/C66 versions of the EWEN and EWDS commands. aic7xxx.c: aic7xxx_pci.c: Move test for the validity of left over BIOS data to ahc_test_register_access(). This guarantees that any left over CHIPRST value is not clobbered by our register access test and lost to the test that was in ahc_reset.
2003-12-17 00:02:10 +00:00
test SEQINTSRC, (CTXTDONE|SAVEPTRS) jnz interrupt_return;
Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-05-04 00:20:07 +00:00
test SSTAT1, REQINIT jz snapshot_wait_data_valid;
snapshot_data_valid:
aic79xx.c: Clear the LQICRC_NLQ status should it pop up after we have already handled the SCSIPERR. During some streaming operations this status can be delayed until the stream ends. Without this change, the driver would complain about a "Missing case in ahd_handle_scsiint". In the LQOBUSFREE handler... Don't return the LQOMGR back to the idle state until after we have cleaned up ENSELO and any status related to this selection. The last thing we need is the LQO manager starting another select-out before we have updated the execution queue. It is not clear whether the LQOMGR would, or would not start a new selection early. Make sure ENSELO is off prior to clearing SELDO by flushing device writes. Move assignment of the next target SCB pointer inside of an if to make the code clearer. The effect is the same. Dump card state in both "Unexpected PKT busfree" paths. In ahd_reset(), set the chip to SCSI mode before reading SXFRCTL1. That register only exists in the SCSI mode. Also set the mode explicitly to the SCSI mode after chip reset due to paranoia. Re-arrange code so that SXFRCTL1 is restored as quickly after the chip reset as possible. S/G structurs must be 8byte aligned. Make this official by saying so in our DMA tag. Disable CIO bus stretch on MDFFSTAT if SHVALID is about to come true. This can cause a CIO bus lockup if a PCI or PCI-X error occurs while the stretch is occurring - the host cannot service the PCI-X error since the CIO bus is locked out and SHVALID will never resolve. The stretch was added in the Rev B to simplify the wait for SHVALID to resolve, but the code to do this in the open source sequencer is so simple it was never removed. Consistently use MAX_OFFSET for the user max syncrate set from non-volatile storage. This ensures that the offset does not conflict with AH?_OFFSET_UNKNOWN. Have ahd_pause_and_flushwork set the mode to ensure that it has access to the registers it checks. Also modify the checking of intstat so that the check against 0xFF can actually succeed if the INT_PEND mask is something other than 0xFF. Although there are no cardbus U320 controllers, this check may be needed to recover from a hot-plug PCI removal that occurs without informing the driver. Fix a typo. sg_prefetch_cnt -> sg_prefetch_align. This fixes an infinite loop at card initialization if the cacheline size is 0. aic79xx.h: Add AHD_EARLY_REQ_BUG bug flag. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. Change AH?_OFFSET_UNKNOWN to 0xFF. This is a value that the curr->offset can never be, unlike '0' which we previously used. This fixes code that only checks for a non-zero offset to determine if a sync negotiation is required since it will fire in the unknown case even if the goal is async. aic79xx.reg: Add comments for LQISTAT bits indicating their names in the 7902 data book. We use slightly different and more descriptive names in the firmware. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. aic79xx.seq: Update comments regarding rundown of the GSFIFO to reflect reality. Fix spelling errors. Since we use an 8byte address and 1 byte length, shorten the size of a block move for the legacy DMA'ed CDB case from 11 to 9 bytes. Remove code that, assuming the abort pending feature worked, would set MK_MESSAGE in the SCB's control byte on completion to catch invalid reselections. Since we don't see interrupts for completed selections, this status update could occur prior to us noticing the SELDO. The "select-out" queue logic will get confused by the MK_MESSAGE bit being set as this is used to catch packatized connections where we select-out with ATN. Since the abort pending feature doesn't work on any released controllers yet, this code was never executed. Add support for the AHD_EARLY_REQ_BUG. Don't ignore persistent REQ assertions just because they were asserted within the bus settle delay window. This allows us to tolerate devices like the GEM318 that violate the SCSI spec. Remove unintentional settnig of SG_CACHE_AVAIL. Writing this bit should have no effect, but who knows... On the Rev A, we must wait for HDMAENACK before loading additional segments to avoid clobbering the address of the first segment in the S/G FIFO. This resolves data-corruption issues with certain IBM (now Hitachi) and Fujitsu U320 drives. Rearrange calc_residual to avoid an extra jmp instruction. On RevA Silicon, if the target returns us to data-out after we have already trained for data-out, it is possible for us to transition the free running clock to data-valid before the required 100ns P1 setup time (8 P1 assertions in fast-160 mode). This will only happen if this L-Q is a continuation of a data transfer for which we have already prefetched data into our FIFO (LQ/Data followed by LQ/Data for the same write transaction). This can cause some target implementations to miss the first few data transfers on the bus. We detect this situation by noticing that this is the first data transfer after an LQ (LQIWORKONLQ true), that the data transfer is a continuation of a transfer already setup in our FIFO (SAVEPTRS interrupt), and that the transaction is a write (DIRECTION set in DFCNTRL). The delay is performed by disabling SCSIEN until we see the first REQ from the target. Only compile in snapshot savepointers handler for RevA silicon where it is enabled. Handle the cfg4icmd packetized interrupt. We just need to load the address and count, start the DMA, and CLRCHN once the transfer is complete. Fix an oversight in the overrun handler for packetized status operations. We need to wait for either CTXTDONE or an overrun when checking for an overrun. The previous code did not wait and thus could decide that no overrun had occurred even though an overrun will occur on the next data-valid req. Add some comment to this section for clarity. Use LAST_SEG_DONE instead of LASTSDONE for testing transfer completion in the packetized status case. LASTSDONE may come up more quickly since it only records completion on the SCSI side, but since LAST_SEG_DONE is used everywhere else (and needs to be), this is less confusing. Add a missing invalidation of the longjmp address in the non-pack handler. This code needs additional review. aic79xx_inline.h: Fix spelling error. aic79xx_osm.c: Set the cdb length for CDBs dma'ed from host memory. Add a comment indicating that, should CAM start supporting cdbs larger than 16bytes, the driver could store the CDB in the status buffer. aic79xx_pci.c: Add a table entry for the 39320A. Added a missing comma to an error string table. Fix spelling errors.
2003-02-27 23:23:16 +00:00
or DFCNTRL, SCSIEN;
Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-05-04 00:20:07 +00:00
or SEQINTCTL, IRET ret;
aic79xx.c: Clear the LQICRC_NLQ status should it pop up after we have already handled the SCSIPERR. During some streaming operations this status can be delayed until the stream ends. Without this change, the driver would complain about a "Missing case in ahd_handle_scsiint". In the LQOBUSFREE handler... Don't return the LQOMGR back to the idle state until after we have cleaned up ENSELO and any status related to this selection. The last thing we need is the LQO manager starting another select-out before we have updated the execution queue. It is not clear whether the LQOMGR would, or would not start a new selection early. Make sure ENSELO is off prior to clearing SELDO by flushing device writes. Move assignment of the next target SCB pointer inside of an if to make the code clearer. The effect is the same. Dump card state in both "Unexpected PKT busfree" paths. In ahd_reset(), set the chip to SCSI mode before reading SXFRCTL1. That register only exists in the SCSI mode. Also set the mode explicitly to the SCSI mode after chip reset due to paranoia. Re-arrange code so that SXFRCTL1 is restored as quickly after the chip reset as possible. S/G structurs must be 8byte aligned. Make this official by saying so in our DMA tag. Disable CIO bus stretch on MDFFSTAT if SHVALID is about to come true. This can cause a CIO bus lockup if a PCI or PCI-X error occurs while the stretch is occurring - the host cannot service the PCI-X error since the CIO bus is locked out and SHVALID will never resolve. The stretch was added in the Rev B to simplify the wait for SHVALID to resolve, but the code to do this in the open source sequencer is so simple it was never removed. Consistently use MAX_OFFSET for the user max syncrate set from non-volatile storage. This ensures that the offset does not conflict with AH?_OFFSET_UNKNOWN. Have ahd_pause_and_flushwork set the mode to ensure that it has access to the registers it checks. Also modify the checking of intstat so that the check against 0xFF can actually succeed if the INT_PEND mask is something other than 0xFF. Although there are no cardbus U320 controllers, this check may be needed to recover from a hot-plug PCI removal that occurs without informing the driver. Fix a typo. sg_prefetch_cnt -> sg_prefetch_align. This fixes an infinite loop at card initialization if the cacheline size is 0. aic79xx.h: Add AHD_EARLY_REQ_BUG bug flag. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. Change AH?_OFFSET_UNKNOWN to 0xFF. This is a value that the curr->offset can never be, unlike '0' which we previously used. This fixes code that only checks for a non-zero offset to determine if a sync negotiation is required since it will fire in the unknown case even if the goal is async. aic79xx.reg: Add comments for LQISTAT bits indicating their names in the 7902 data book. We use slightly different and more descriptive names in the firmware. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. aic79xx.seq: Update comments regarding rundown of the GSFIFO to reflect reality. Fix spelling errors. Since we use an 8byte address and 1 byte length, shorten the size of a block move for the legacy DMA'ed CDB case from 11 to 9 bytes. Remove code that, assuming the abort pending feature worked, would set MK_MESSAGE in the SCB's control byte on completion to catch invalid reselections. Since we don't see interrupts for completed selections, this status update could occur prior to us noticing the SELDO. The "select-out" queue logic will get confused by the MK_MESSAGE bit being set as this is used to catch packatized connections where we select-out with ATN. Since the abort pending feature doesn't work on any released controllers yet, this code was never executed. Add support for the AHD_EARLY_REQ_BUG. Don't ignore persistent REQ assertions just because they were asserted within the bus settle delay window. This allows us to tolerate devices like the GEM318 that violate the SCSI spec. Remove unintentional settnig of SG_CACHE_AVAIL. Writing this bit should have no effect, but who knows... On the Rev A, we must wait for HDMAENACK before loading additional segments to avoid clobbering the address of the first segment in the S/G FIFO. This resolves data-corruption issues with certain IBM (now Hitachi) and Fujitsu U320 drives. Rearrange calc_residual to avoid an extra jmp instruction. On RevA Silicon, if the target returns us to data-out after we have already trained for data-out, it is possible for us to transition the free running clock to data-valid before the required 100ns P1 setup time (8 P1 assertions in fast-160 mode). This will only happen if this L-Q is a continuation of a data transfer for which we have already prefetched data into our FIFO (LQ/Data followed by LQ/Data for the same write transaction). This can cause some target implementations to miss the first few data transfers on the bus. We detect this situation by noticing that this is the first data transfer after an LQ (LQIWORKONLQ true), that the data transfer is a continuation of a transfer already setup in our FIFO (SAVEPTRS interrupt), and that the transaction is a write (DIRECTION set in DFCNTRL). The delay is performed by disabling SCSIEN until we see the first REQ from the target. Only compile in snapshot savepointers handler for RevA silicon where it is enabled. Handle the cfg4icmd packetized interrupt. We just need to load the address and count, start the DMA, and CLRCHN once the transfer is complete. Fix an oversight in the overrun handler for packetized status operations. We need to wait for either CTXTDONE or an overrun when checking for an overrun. The previous code did not wait and thus could decide that no overrun had occurred even though an overrun will occur on the next data-valid req. Add some comment to this section for clarity. Use LAST_SEG_DONE instead of LASTSDONE for testing transfer completion in the packetized status case. LASTSDONE may come up more quickly since it only records completion on the SCSI side, but since LAST_SEG_DONE is used everywhere else (and needs to be), this is less confusing. Add a missing invalidation of the longjmp address in the non-pack handler. This code needs additional review. aic79xx_inline.h: Fix spelling error. aic79xx_osm.c: Set the cdb length for CDBs dma'ed from host memory. Add a comment indicating that, should CAM start supporting cdbs larger than 16bytes, the driver could store the CDB in the status buffer. aic79xx_pci.c: Add a table entry for the 39320A. Added a missing comma to an error string table. Fix spelling errors.
2003-02-27 23:23:16 +00:00
snapshot_saveptr:
mvi DFFSXFRCTL, CLRCHN;
or SEQINTCTL, IRET ret;
Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-05-04 00:20:07 +00:00
main_isr:
aic79xx.c: Clear the LQICRC_NLQ status should it pop up after we have already handled the SCSIPERR. During some streaming operations this status can be delayed until the stream ends. Without this change, the driver would complain about a "Missing case in ahd_handle_scsiint". In the LQOBUSFREE handler... Don't return the LQOMGR back to the idle state until after we have cleaned up ENSELO and any status related to this selection. The last thing we need is the LQO manager starting another select-out before we have updated the execution queue. It is not clear whether the LQOMGR would, or would not start a new selection early. Make sure ENSELO is off prior to clearing SELDO by flushing device writes. Move assignment of the next target SCB pointer inside of an if to make the code clearer. The effect is the same. Dump card state in both "Unexpected PKT busfree" paths. In ahd_reset(), set the chip to SCSI mode before reading SXFRCTL1. That register only exists in the SCSI mode. Also set the mode explicitly to the SCSI mode after chip reset due to paranoia. Re-arrange code so that SXFRCTL1 is restored as quickly after the chip reset as possible. S/G structurs must be 8byte aligned. Make this official by saying so in our DMA tag. Disable CIO bus stretch on MDFFSTAT if SHVALID is about to come true. This can cause a CIO bus lockup if a PCI or PCI-X error occurs while the stretch is occurring - the host cannot service the PCI-X error since the CIO bus is locked out and SHVALID will never resolve. The stretch was added in the Rev B to simplify the wait for SHVALID to resolve, but the code to do this in the open source sequencer is so simple it was never removed. Consistently use MAX_OFFSET for the user max syncrate set from non-volatile storage. This ensures that the offset does not conflict with AH?_OFFSET_UNKNOWN. Have ahd_pause_and_flushwork set the mode to ensure that it has access to the registers it checks. Also modify the checking of intstat so that the check against 0xFF can actually succeed if the INT_PEND mask is something other than 0xFF. Although there are no cardbus U320 controllers, this check may be needed to recover from a hot-plug PCI removal that occurs without informing the driver. Fix a typo. sg_prefetch_cnt -> sg_prefetch_align. This fixes an infinite loop at card initialization if the cacheline size is 0. aic79xx.h: Add AHD_EARLY_REQ_BUG bug flag. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. Change AH?_OFFSET_UNKNOWN to 0xFF. This is a value that the curr->offset can never be, unlike '0' which we previously used. This fixes code that only checks for a non-zero offset to determine if a sync negotiation is required since it will fire in the unknown case even if the goal is async. aic79xx.reg: Add comments for LQISTAT bits indicating their names in the 7902 data book. We use slightly different and more descriptive names in the firmware. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. aic79xx.seq: Update comments regarding rundown of the GSFIFO to reflect reality. Fix spelling errors. Since we use an 8byte address and 1 byte length, shorten the size of a block move for the legacy DMA'ed CDB case from 11 to 9 bytes. Remove code that, assuming the abort pending feature worked, would set MK_MESSAGE in the SCB's control byte on completion to catch invalid reselections. Since we don't see interrupts for completed selections, this status update could occur prior to us noticing the SELDO. The "select-out" queue logic will get confused by the MK_MESSAGE bit being set as this is used to catch packatized connections where we select-out with ATN. Since the abort pending feature doesn't work on any released controllers yet, this code was never executed. Add support for the AHD_EARLY_REQ_BUG. Don't ignore persistent REQ assertions just because they were asserted within the bus settle delay window. This allows us to tolerate devices like the GEM318 that violate the SCSI spec. Remove unintentional settnig of SG_CACHE_AVAIL. Writing this bit should have no effect, but who knows... On the Rev A, we must wait for HDMAENACK before loading additional segments to avoid clobbering the address of the first segment in the S/G FIFO. This resolves data-corruption issues with certain IBM (now Hitachi) and Fujitsu U320 drives. Rearrange calc_residual to avoid an extra jmp instruction. On RevA Silicon, if the target returns us to data-out after we have already trained for data-out, it is possible for us to transition the free running clock to data-valid before the required 100ns P1 setup time (8 P1 assertions in fast-160 mode). This will only happen if this L-Q is a continuation of a data transfer for which we have already prefetched data into our FIFO (LQ/Data followed by LQ/Data for the same write transaction). This can cause some target implementations to miss the first few data transfers on the bus. We detect this situation by noticing that this is the first data transfer after an LQ (LQIWORKONLQ true), that the data transfer is a continuation of a transfer already setup in our FIFO (SAVEPTRS interrupt), and that the transaction is a write (DIRECTION set in DFCNTRL). The delay is performed by disabling SCSIEN until we see the first REQ from the target. Only compile in snapshot savepointers handler for RevA silicon where it is enabled. Handle the cfg4icmd packetized interrupt. We just need to load the address and count, start the DMA, and CLRCHN once the transfer is complete. Fix an oversight in the overrun handler for packetized status operations. We need to wait for either CTXTDONE or an overrun when checking for an overrun. The previous code did not wait and thus could decide that no overrun had occurred even though an overrun will occur on the next data-valid req. Add some comment to this section for clarity. Use LAST_SEG_DONE instead of LASTSDONE for testing transfer completion in the packetized status case. LASTSDONE may come up more quickly since it only records completion on the SCSI side, but since LAST_SEG_DONE is used everywhere else (and needs to be), this is less confusing. Add a missing invalidation of the longjmp address in the non-pack handler. This code needs additional review. aic79xx_inline.h: Fix spelling error. aic79xx_osm.c: Set the cdb length for CDBs dma'ed from host memory. Add a comment indicating that, should CAM start supporting cdbs larger than 16bytes, the driver could store the CDB in the status buffer. aic79xx_pci.c: Add a table entry for the 39320A. Added a missing comma to an error string table. Fix spelling errors.
2003-02-27 23:23:16 +00:00
}
test SEQINTSRC, CFG4DATA jnz cfg4data_intr;
test SEQINTSRC, CFG4ISTAT jnz cfg4istat_intr;
aic79xx.c: Implement the SCB_SILENT flag. This is useful for hushing up the driver during DV or other operations that we expect to cause transmission errors. The messages will still print if the SHOW_MASKED_ERRORS debug option is enabled. Save and restore the NEGOADDR address when setting new transfer settings. The sequencer performs lookups in the negotiation table too and it expects NEGOADDR to remain consistent across pause/unpause sessions. Consistently use "offset" instead of "period" to determine if we are running sync or not. Add a SHOW_MESSAGES diagnostic for when we assert ATN during message processing. Print out IU, QAS, and RTI features when showing transfer options. Limit the syncrate after all option conformance changes have taken place in ahd_devlimited_syncrate. Changes in options may change the final syncrate we accept. Keep a copy of the hs_mailbox in our softc so that we can perform read/modify/write operations on the hs_mailbox without having to pause the sequencer to read the last written value. Use the ENINT_COALESS flag in the hs_mailbox to toggle interrupt coalessing. Add entrypoints for enabling interrupt coalessing and setting both a timeout (how long to wait for commands to be coalessed) and a maximum commands to coaless value. Add a statistics timer that decides when to enable or disable interrupt coalessing based on load. Add a routine, ahd_reset_cmds_pending() which is used to update the CMDS_PENDING sequencer variable whenever error recovery compeltes SCBs without notifying the sequencer. Since ahd_reset_cmds_pending is called during ahd_unpause() only if we've aborted SCBs, its call to ahd_flush_qoutfifo should not cause recursion through ahd_run_qoutfifo(). A panic has been added to ensure that this recursion does not occur. In ahd_search_qinfifo, update the CMDS_PENDING sequencer variable directly. ahd_search_qinififo can be called in situations where using ahd_reset_cmds_pending() might cause recursion. Since we can safely determine the exact number to reduce CMDS_PENDING by in this scenario without running the qoutfifo, the manual update is sufficient. Clean up diagnostics. Add ahd_flush_qoutfifo() which will run the qoutfifo as well as complete any commands sitting on the sequencer's COMPLETE_SCB lists or the good status FIFO. Use this routine in several places that did similar things in an add-hoc, but incomplete, fashion. A call to this routine was also added to ahd_abort_scbs() to close a race. In ahd_pause_and_flushwork() only return once selections are safely disabled. Flush all completed commands via ahd_flush_qoutfifo(). Remove "Now packetized" diagnostic now that this information is incorperated into the actual negotiation messages that are displayed. When forcing renegotiation, don't clober the current ppr_options. Much of the driver uses this information to determine if we are currently packetized or not. Remove some stray spaces at column 1 in ahd_set_tags. When complaining about getting a host message loop request with no pending messages, print out the SCB_CONTROL register down on the card. Modify the ahd_sent_msg() routine to handle a search for an outgoing identify message. Use this to detect a msg reject on an identify message which typically indicates that the target thought we were packetized. Force a renegotiation in this case. In ahd_search_qinfifo(), wait more effectively for SCB DMA activities to cease. We also disable SCB fetch operations since we are about to change the qinfifo and any fetch in progress will likely be invalidated. In ahd_qinfifo_count(), fix the qinfifo empty case. In ahd_dump_card_state(), print out CCSCBCTL in the correct mode. If we are a narrow controller, don't set the current width to unknown when forcing a future negotiation. This just confuses the code into attempting a wide negotiation on a narrow bus. Add support for task management function completions. Modify ahd_handle_devreset so that it can handle lun resets in addition to target resets. Use ahd_handle_devreset for lun and target reset task management functions. Handle the abort task TMF race case better. We now wait until any current selections are over and then set the TMF back to zero. This should cause the sequencer to ignore the abort TMF completion should it occur. Correct a bug in the illegal phase handler that caused us to drop down to narrow when handling the unexpected command phase case after 3rd party reset of a packetized device. Indicate the features, bugs, and flags set in the softc that are used to control firmware patch download when booting verbose. aic79xx.h: Add coalessing and HS_MAILBOX fields. Add per-softc variables for the stats "daemon". Add a debug option for interrupt coalessing activities. Add two new softc flags: o AHD_UPDATE_PEND_CMDS Run ahd_reset_cmds_pending() on the next unpause. o AHD_RUNNING_QOUTFIFO Used to catch recursion through ahd_run_qoutfifo(). aic79xx.reg: Correct register addresses related to the software timer and the DFDBCTL register. Add constants paramaterizing the software timer. Add scratch ram locations for storing interrupt coalessing tunables. Break INTMASK in SEQITNCTL out into INTMASK1 and INTMASK2. In at least the REV A, these are writable bits. We make use of that for a swtimer workaround in the sequencer. Since HS_MAILBOX autoclears, provide a sequencer variable to store its contents. Add SEQINT codes for handling task management completions. aic79xx.seq: Correct ignore wide residue processing check for a wide negotiation being in effect. We must be in the SCSI register window in order to access the negotiation table. Use the software timer and a commands completed count to implement interrupt coalessing. The command complete is deferred until either the maximum command threshold or a the expiration of a command deferral timer. If we have more SCBs to complete to the host (sitting in COMPLETE_SCB lists), always try to coaless them up to our coalessing limit. If coalessing is enabled, but we have fewer commands oustanting than the host's min coalessing limit, complete the command immediately. Add code to track the number of commands outstanding. Commands are outstanding from the time they are placed into the execution queue until the DMA to post completion is setup. Add a workaround for intvec_2 interrupts on the H2A4. In H2A4, the mode pointer is not saved for intvec2, but is restored on iret. This can lead to the restoration of a bogus mode ptr. Manually clear the intmask bits and do a normal return to compensate. We use intvec_2 to track interrupt coalessing timeouts. Since we cannot disable the swtimer's countdown, simply mask its interrupt once we no longer care about it firing. In idle_loop_cchan, update LOCAL_HS_MAILBOX everytime we are notified of an HS_MAILBOX update via the HS_MAILBOX_ACT bit in QOFF_CTLSTA. We have to use a local copy of persistant portions of the HS_MAILBOX as the mailbox auto-clears on any read. Move the test for the cfg4istat interrupt up an instruction to hopefully close a race between the next outgoing selection and our disabling of selections. Add a missing ret to the last instruction in load_overrun_buf. Add notifications to the host of task management completions as well as the completions for commands that completed successfully before their corresponding TMF could be sent. Hold a critical section during select-out processing until we have a fully identified connection. This removes a race condition with the legacy abort handler. Correct a few spelling errors in some comments. aic79xx_inline.h: Call ahd_reset_cmds_pending() in ahd_unpause if required. Update cmdcmplt interrupt statistics in our interrupt handler. Allow callers to ahd_send_scb() to set the task management function. aic79xx_pci.c: Disable SERR and pause the controller prior to performing our mmapped I/O test. The U320 controllers do not support "auto-access-pause". aic79xx_osm.c: Set the task management function now that ahd_send_scb() doesn't do it for us. We also perform a lun reset in response to BDR requests to packetized devices.
2003-01-20 20:17:35 +00:00
test SEQINTSRC, SAVEPTRS jnz saveptr_intr;
test SEQINTSRC, CFG4ICMD jnz cfg4icmd_intr;
SET_SEQINTCODE(INVALID_SEQINT)
/*
* There are two types of save pointers interrupts:
* The first is a snapshot save pointers where the current FIFO is not
* active and contains a snapshot of the current poniter information.
* This happens between packets in a stream for a single L_Q. Since we
* are not performing a pointer save, we can safely clear the channel
aic79xx.c: Clear the LQICRC_NLQ status should it pop up after we have already handled the SCSIPERR. During some streaming operations this status can be delayed until the stream ends. Without this change, the driver would complain about a "Missing case in ahd_handle_scsiint". In the LQOBUSFREE handler... Don't return the LQOMGR back to the idle state until after we have cleaned up ENSELO and any status related to this selection. The last thing we need is the LQO manager starting another select-out before we have updated the execution queue. It is not clear whether the LQOMGR would, or would not start a new selection early. Make sure ENSELO is off prior to clearing SELDO by flushing device writes. Move assignment of the next target SCB pointer inside of an if to make the code clearer. The effect is the same. Dump card state in both "Unexpected PKT busfree" paths. In ahd_reset(), set the chip to SCSI mode before reading SXFRCTL1. That register only exists in the SCSI mode. Also set the mode explicitly to the SCSI mode after chip reset due to paranoia. Re-arrange code so that SXFRCTL1 is restored as quickly after the chip reset as possible. S/G structurs must be 8byte aligned. Make this official by saying so in our DMA tag. Disable CIO bus stretch on MDFFSTAT if SHVALID is about to come true. This can cause a CIO bus lockup if a PCI or PCI-X error occurs while the stretch is occurring - the host cannot service the PCI-X error since the CIO bus is locked out and SHVALID will never resolve. The stretch was added in the Rev B to simplify the wait for SHVALID to resolve, but the code to do this in the open source sequencer is so simple it was never removed. Consistently use MAX_OFFSET for the user max syncrate set from non-volatile storage. This ensures that the offset does not conflict with AH?_OFFSET_UNKNOWN. Have ahd_pause_and_flushwork set the mode to ensure that it has access to the registers it checks. Also modify the checking of intstat so that the check against 0xFF can actually succeed if the INT_PEND mask is something other than 0xFF. Although there are no cardbus U320 controllers, this check may be needed to recover from a hot-plug PCI removal that occurs without informing the driver. Fix a typo. sg_prefetch_cnt -> sg_prefetch_align. This fixes an infinite loop at card initialization if the cacheline size is 0. aic79xx.h: Add AHD_EARLY_REQ_BUG bug flag. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. Change AH?_OFFSET_UNKNOWN to 0xFF. This is a value that the curr->offset can never be, unlike '0' which we previously used. This fixes code that only checks for a non-zero offset to determine if a sync negotiation is required since it will fire in the unknown case even if the goal is async. aic79xx.reg: Add comments for LQISTAT bits indicating their names in the 7902 data book. We use slightly different and more descriptive names in the firmware. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. aic79xx.seq: Update comments regarding rundown of the GSFIFO to reflect reality. Fix spelling errors. Since we use an 8byte address and 1 byte length, shorten the size of a block move for the legacy DMA'ed CDB case from 11 to 9 bytes. Remove code that, assuming the abort pending feature worked, would set MK_MESSAGE in the SCB's control byte on completion to catch invalid reselections. Since we don't see interrupts for completed selections, this status update could occur prior to us noticing the SELDO. The "select-out" queue logic will get confused by the MK_MESSAGE bit being set as this is used to catch packatized connections where we select-out with ATN. Since the abort pending feature doesn't work on any released controllers yet, this code was never executed. Add support for the AHD_EARLY_REQ_BUG. Don't ignore persistent REQ assertions just because they were asserted within the bus settle delay window. This allows us to tolerate devices like the GEM318 that violate the SCSI spec. Remove unintentional settnig of SG_CACHE_AVAIL. Writing this bit should have no effect, but who knows... On the Rev A, we must wait for HDMAENACK before loading additional segments to avoid clobbering the address of the first segment in the S/G FIFO. This resolves data-corruption issues with certain IBM (now Hitachi) and Fujitsu U320 drives. Rearrange calc_residual to avoid an extra jmp instruction. On RevA Silicon, if the target returns us to data-out after we have already trained for data-out, it is possible for us to transition the free running clock to data-valid before the required 100ns P1 setup time (8 P1 assertions in fast-160 mode). This will only happen if this L-Q is a continuation of a data transfer for which we have already prefetched data into our FIFO (LQ/Data followed by LQ/Data for the same write transaction). This can cause some target implementations to miss the first few data transfers on the bus. We detect this situation by noticing that this is the first data transfer after an LQ (LQIWORKONLQ true), that the data transfer is a continuation of a transfer already setup in our FIFO (SAVEPTRS interrupt), and that the transaction is a write (DIRECTION set in DFCNTRL). The delay is performed by disabling SCSIEN until we see the first REQ from the target. Only compile in snapshot savepointers handler for RevA silicon where it is enabled. Handle the cfg4icmd packetized interrupt. We just need to load the address and count, start the DMA, and CLRCHN once the transfer is complete. Fix an oversight in the overrun handler for packetized status operations. We need to wait for either CTXTDONE or an overrun when checking for an overrun. The previous code did not wait and thus could decide that no overrun had occurred even though an overrun will occur on the next data-valid req. Add some comment to this section for clarity. Use LAST_SEG_DONE instead of LASTSDONE for testing transfer completion in the packetized status case. LASTSDONE may come up more quickly since it only records completion on the SCSI side, but since LAST_SEG_DONE is used everywhere else (and needs to be), this is less confusing. Add a missing invalidation of the longjmp address in the non-pack handler. This code needs additional review. aic79xx_inline.h: Fix spelling error. aic79xx_osm.c: Set the cdb length for CDBs dma'ed from host memory. Add a comment indicating that, should CAM start supporting cdbs larger than 16bytes, the driver could store the CDB in the status buffer. aic79xx_pci.c: Add a table entry for the 39320A. Added a missing comma to an error string table. Fix spelling errors.
2003-02-27 23:23:16 +00:00
* so it can be used for other transactions. On RTI capable controllers,
* where snapshots can, and are, disabled, the code to handle this type
* of snapshot is not active.
*
* The second case is a save pointers on an active FIFO which occurs
aic79xx.c: Clear the LQICRC_NLQ status should it pop up after we have already handled the SCSIPERR. During some streaming operations this status can be delayed until the stream ends. Without this change, the driver would complain about a "Missing case in ahd_handle_scsiint". In the LQOBUSFREE handler... Don't return the LQOMGR back to the idle state until after we have cleaned up ENSELO and any status related to this selection. The last thing we need is the LQO manager starting another select-out before we have updated the execution queue. It is not clear whether the LQOMGR would, or would not start a new selection early. Make sure ENSELO is off prior to clearing SELDO by flushing device writes. Move assignment of the next target SCB pointer inside of an if to make the code clearer. The effect is the same. Dump card state in both "Unexpected PKT busfree" paths. In ahd_reset(), set the chip to SCSI mode before reading SXFRCTL1. That register only exists in the SCSI mode. Also set the mode explicitly to the SCSI mode after chip reset due to paranoia. Re-arrange code so that SXFRCTL1 is restored as quickly after the chip reset as possible. S/G structurs must be 8byte aligned. Make this official by saying so in our DMA tag. Disable CIO bus stretch on MDFFSTAT if SHVALID is about to come true. This can cause a CIO bus lockup if a PCI or PCI-X error occurs while the stretch is occurring - the host cannot service the PCI-X error since the CIO bus is locked out and SHVALID will never resolve. The stretch was added in the Rev B to simplify the wait for SHVALID to resolve, but the code to do this in the open source sequencer is so simple it was never removed. Consistently use MAX_OFFSET for the user max syncrate set from non-volatile storage. This ensures that the offset does not conflict with AH?_OFFSET_UNKNOWN. Have ahd_pause_and_flushwork set the mode to ensure that it has access to the registers it checks. Also modify the checking of intstat so that the check against 0xFF can actually succeed if the INT_PEND mask is something other than 0xFF. Although there are no cardbus U320 controllers, this check may be needed to recover from a hot-plug PCI removal that occurs without informing the driver. Fix a typo. sg_prefetch_cnt -> sg_prefetch_align. This fixes an infinite loop at card initialization if the cacheline size is 0. aic79xx.h: Add AHD_EARLY_REQ_BUG bug flag. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. Change AH?_OFFSET_UNKNOWN to 0xFF. This is a value that the curr->offset can never be, unlike '0' which we previously used. This fixes code that only checks for a non-zero offset to determine if a sync negotiation is required since it will fire in the unknown case even if the goal is async. aic79xx.reg: Add comments for LQISTAT bits indicating their names in the 7902 data book. We use slightly different and more descriptive names in the firmware. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. aic79xx.seq: Update comments regarding rundown of the GSFIFO to reflect reality. Fix spelling errors. Since we use an 8byte address and 1 byte length, shorten the size of a block move for the legacy DMA'ed CDB case from 11 to 9 bytes. Remove code that, assuming the abort pending feature worked, would set MK_MESSAGE in the SCB's control byte on completion to catch invalid reselections. Since we don't see interrupts for completed selections, this status update could occur prior to us noticing the SELDO. The "select-out" queue logic will get confused by the MK_MESSAGE bit being set as this is used to catch packatized connections where we select-out with ATN. Since the abort pending feature doesn't work on any released controllers yet, this code was never executed. Add support for the AHD_EARLY_REQ_BUG. Don't ignore persistent REQ assertions just because they were asserted within the bus settle delay window. This allows us to tolerate devices like the GEM318 that violate the SCSI spec. Remove unintentional settnig of SG_CACHE_AVAIL. Writing this bit should have no effect, but who knows... On the Rev A, we must wait for HDMAENACK before loading additional segments to avoid clobbering the address of the first segment in the S/G FIFO. This resolves data-corruption issues with certain IBM (now Hitachi) and Fujitsu U320 drives. Rearrange calc_residual to avoid an extra jmp instruction. On RevA Silicon, if the target returns us to data-out after we have already trained for data-out, it is possible for us to transition the free running clock to data-valid before the required 100ns P1 setup time (8 P1 assertions in fast-160 mode). This will only happen if this L-Q is a continuation of a data transfer for which we have already prefetched data into our FIFO (LQ/Data followed by LQ/Data for the same write transaction). This can cause some target implementations to miss the first few data transfers on the bus. We detect this situation by noticing that this is the first data transfer after an LQ (LQIWORKONLQ true), that the data transfer is a continuation of a transfer already setup in our FIFO (SAVEPTRS interrupt), and that the transaction is a write (DIRECTION set in DFCNTRL). The delay is performed by disabling SCSIEN until we see the first REQ from the target. Only compile in snapshot savepointers handler for RevA silicon where it is enabled. Handle the cfg4icmd packetized interrupt. We just need to load the address and count, start the DMA, and CLRCHN once the transfer is complete. Fix an oversight in the overrun handler for packetized status operations. We need to wait for either CTXTDONE or an overrun when checking for an overrun. The previous code did not wait and thus could decide that no overrun had occurred even though an overrun will occur on the next data-valid req. Add some comment to this section for clarity. Use LAST_SEG_DONE instead of LASTSDONE for testing transfer completion in the packetized status case. LASTSDONE may come up more quickly since it only records completion on the SCSI side, but since LAST_SEG_DONE is used everywhere else (and needs to be), this is less confusing. Add a missing invalidation of the longjmp address in the non-pack handler. This code needs additional review. aic79xx_inline.h: Fix spelling error. aic79xx_osm.c: Set the cdb length for CDBs dma'ed from host memory. Add a comment indicating that, should CAM start supporting cdbs larger than 16bytes, the driver could store the CDB in the status buffer. aic79xx_pci.c: Add a table entry for the 39320A. Added a missing comma to an error string table. Fix spelling errors.
2003-02-27 23:23:16 +00:00
* if the target changes to a new L_Q or busfrees/QASes and the transfer
* has a residual. This should occur coincident with a ctxtdone. We
* disable the interrupt and allow our active routine to handle the
* save.
*/
saveptr_intr:
aic79xx.c: Clear the LQICRC_NLQ status should it pop up after we have already handled the SCSIPERR. During some streaming operations this status can be delayed until the stream ends. Without this change, the driver would complain about a "Missing case in ahd_handle_scsiint". In the LQOBUSFREE handler... Don't return the LQOMGR back to the idle state until after we have cleaned up ENSELO and any status related to this selection. The last thing we need is the LQO manager starting another select-out before we have updated the execution queue. It is not clear whether the LQOMGR would, or would not start a new selection early. Make sure ENSELO is off prior to clearing SELDO by flushing device writes. Move assignment of the next target SCB pointer inside of an if to make the code clearer. The effect is the same. Dump card state in both "Unexpected PKT busfree" paths. In ahd_reset(), set the chip to SCSI mode before reading SXFRCTL1. That register only exists in the SCSI mode. Also set the mode explicitly to the SCSI mode after chip reset due to paranoia. Re-arrange code so that SXFRCTL1 is restored as quickly after the chip reset as possible. S/G structurs must be 8byte aligned. Make this official by saying so in our DMA tag. Disable CIO bus stretch on MDFFSTAT if SHVALID is about to come true. This can cause a CIO bus lockup if a PCI or PCI-X error occurs while the stretch is occurring - the host cannot service the PCI-X error since the CIO bus is locked out and SHVALID will never resolve. The stretch was added in the Rev B to simplify the wait for SHVALID to resolve, but the code to do this in the open source sequencer is so simple it was never removed. Consistently use MAX_OFFSET for the user max syncrate set from non-volatile storage. This ensures that the offset does not conflict with AH?_OFFSET_UNKNOWN. Have ahd_pause_and_flushwork set the mode to ensure that it has access to the registers it checks. Also modify the checking of intstat so that the check against 0xFF can actually succeed if the INT_PEND mask is something other than 0xFF. Although there are no cardbus U320 controllers, this check may be needed to recover from a hot-plug PCI removal that occurs without informing the driver. Fix a typo. sg_prefetch_cnt -> sg_prefetch_align. This fixes an infinite loop at card initialization if the cacheline size is 0. aic79xx.h: Add AHD_EARLY_REQ_BUG bug flag. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. Change AH?_OFFSET_UNKNOWN to 0xFF. This is a value that the curr->offset can never be, unlike '0' which we previously used. This fixes code that only checks for a non-zero offset to determine if a sync negotiation is required since it will fire in the unknown case even if the goal is async. aic79xx.reg: Add comments for LQISTAT bits indicating their names in the 7902 data book. We use slightly different and more descriptive names in the firmware. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. aic79xx.seq: Update comments regarding rundown of the GSFIFO to reflect reality. Fix spelling errors. Since we use an 8byte address and 1 byte length, shorten the size of a block move for the legacy DMA'ed CDB case from 11 to 9 bytes. Remove code that, assuming the abort pending feature worked, would set MK_MESSAGE in the SCB's control byte on completion to catch invalid reselections. Since we don't see interrupts for completed selections, this status update could occur prior to us noticing the SELDO. The "select-out" queue logic will get confused by the MK_MESSAGE bit being set as this is used to catch packatized connections where we select-out with ATN. Since the abort pending feature doesn't work on any released controllers yet, this code was never executed. Add support for the AHD_EARLY_REQ_BUG. Don't ignore persistent REQ assertions just because they were asserted within the bus settle delay window. This allows us to tolerate devices like the GEM318 that violate the SCSI spec. Remove unintentional settnig of SG_CACHE_AVAIL. Writing this bit should have no effect, but who knows... On the Rev A, we must wait for HDMAENACK before loading additional segments to avoid clobbering the address of the first segment in the S/G FIFO. This resolves data-corruption issues with certain IBM (now Hitachi) and Fujitsu U320 drives. Rearrange calc_residual to avoid an extra jmp instruction. On RevA Silicon, if the target returns us to data-out after we have already trained for data-out, it is possible for us to transition the free running clock to data-valid before the required 100ns P1 setup time (8 P1 assertions in fast-160 mode). This will only happen if this L-Q is a continuation of a data transfer for which we have already prefetched data into our FIFO (LQ/Data followed by LQ/Data for the same write transaction). This can cause some target implementations to miss the first few data transfers on the bus. We detect this situation by noticing that this is the first data transfer after an LQ (LQIWORKONLQ true), that the data transfer is a continuation of a transfer already setup in our FIFO (SAVEPTRS interrupt), and that the transaction is a write (DIRECTION set in DFCNTRL). The delay is performed by disabling SCSIEN until we see the first REQ from the target. Only compile in snapshot savepointers handler for RevA silicon where it is enabled. Handle the cfg4icmd packetized interrupt. We just need to load the address and count, start the DMA, and CLRCHN once the transfer is complete. Fix an oversight in the overrun handler for packetized status operations. We need to wait for either CTXTDONE or an overrun when checking for an overrun. The previous code did not wait and thus could decide that no overrun had occurred even though an overrun will occur on the next data-valid req. Add some comment to this section for clarity. Use LAST_SEG_DONE instead of LASTSDONE for testing transfer completion in the packetized status case. LASTSDONE may come up more quickly since it only records completion on the SCSI side, but since LAST_SEG_DONE is used everywhere else (and needs to be), this is less confusing. Add a missing invalidation of the longjmp address in the non-pack handler. This code needs additional review. aic79xx_inline.h: Fix spelling error. aic79xx_osm.c: Set the cdb length for CDBs dma'ed from host memory. Add a comment indicating that, should CAM start supporting cdbs larger than 16bytes, the driver could store the CDB in the status buffer. aic79xx_pci.c: Add a table entry for the 39320A. Added a missing comma to an error string table. Fix spelling errors.
2003-02-27 23:23:16 +00:00
if ((ahd->features & AHD_RTI) == 0) {
test LONGJMP_ADDR[1], INVALID_ADDR jnz snapshot_saveptr;
}
saveptr_active_fifo:
and SEQIMODE, ~ENSAVEPTRS;
or SEQINTCTL, IRET ret;
cfg4data_intr:
Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-05-04 00:20:07 +00:00
test SCB_SGPTR[0], SG_LIST_NULL jnz pkt_handle_overrun_inc_use_count;
call load_first_seg;
call pkt_handle_xfer;
Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-05-04 00:20:07 +00:00
inc SCB_FIFO_USE_COUNT;
interrupt_return:
or SEQINTCTL, IRET ret;
cfg4istat_intr:
call freeze_queue;
add NONE, -13, SCB_CDB_LEN;
jnc cfg4istat_have_sense_addr;
test SCB_CDB_LEN, SCB_CDB_LEN_PTR jnz cfg4istat_have_sense_addr;
/*
* Host sets up address/count and enables transfer.
*/
SET_SEQINTCODE(CFG4ISTAT_INTR)
jmp cfg4istat_setup_handler;
cfg4istat_have_sense_addr:
bmov HADDR, SCB_SENSE_BUSADDR, 4;
mvi HCNT[1], (AHD_SENSE_BUFSIZE >> 8);
mvi SG_CACHE_PRE, LAST_SEG;
mvi DFCNTRL, PRELOADEN|SCSIEN|HDMAEN;
cfg4istat_setup_handler:
/*
* Status pkt is transferring to host.
* Wait in idle loop for transfer to complete.
aic79xx.c: Implement the SCB_SILENT flag. This is useful for hushing up the driver during DV or other operations that we expect to cause transmission errors. The messages will still print if the SHOW_MASKED_ERRORS debug option is enabled. Save and restore the NEGOADDR address when setting new transfer settings. The sequencer performs lookups in the negotiation table too and it expects NEGOADDR to remain consistent across pause/unpause sessions. Consistently use "offset" instead of "period" to determine if we are running sync or not. Add a SHOW_MESSAGES diagnostic for when we assert ATN during message processing. Print out IU, QAS, and RTI features when showing transfer options. Limit the syncrate after all option conformance changes have taken place in ahd_devlimited_syncrate. Changes in options may change the final syncrate we accept. Keep a copy of the hs_mailbox in our softc so that we can perform read/modify/write operations on the hs_mailbox without having to pause the sequencer to read the last written value. Use the ENINT_COALESS flag in the hs_mailbox to toggle interrupt coalessing. Add entrypoints for enabling interrupt coalessing and setting both a timeout (how long to wait for commands to be coalessed) and a maximum commands to coaless value. Add a statistics timer that decides when to enable or disable interrupt coalessing based on load. Add a routine, ahd_reset_cmds_pending() which is used to update the CMDS_PENDING sequencer variable whenever error recovery compeltes SCBs without notifying the sequencer. Since ahd_reset_cmds_pending is called during ahd_unpause() only if we've aborted SCBs, its call to ahd_flush_qoutfifo should not cause recursion through ahd_run_qoutfifo(). A panic has been added to ensure that this recursion does not occur. In ahd_search_qinfifo, update the CMDS_PENDING sequencer variable directly. ahd_search_qinififo can be called in situations where using ahd_reset_cmds_pending() might cause recursion. Since we can safely determine the exact number to reduce CMDS_PENDING by in this scenario without running the qoutfifo, the manual update is sufficient. Clean up diagnostics. Add ahd_flush_qoutfifo() which will run the qoutfifo as well as complete any commands sitting on the sequencer's COMPLETE_SCB lists or the good status FIFO. Use this routine in several places that did similar things in an add-hoc, but incomplete, fashion. A call to this routine was also added to ahd_abort_scbs() to close a race. In ahd_pause_and_flushwork() only return once selections are safely disabled. Flush all completed commands via ahd_flush_qoutfifo(). Remove "Now packetized" diagnostic now that this information is incorperated into the actual negotiation messages that are displayed. When forcing renegotiation, don't clober the current ppr_options. Much of the driver uses this information to determine if we are currently packetized or not. Remove some stray spaces at column 1 in ahd_set_tags. When complaining about getting a host message loop request with no pending messages, print out the SCB_CONTROL register down on the card. Modify the ahd_sent_msg() routine to handle a search for an outgoing identify message. Use this to detect a msg reject on an identify message which typically indicates that the target thought we were packetized. Force a renegotiation in this case. In ahd_search_qinfifo(), wait more effectively for SCB DMA activities to cease. We also disable SCB fetch operations since we are about to change the qinfifo and any fetch in progress will likely be invalidated. In ahd_qinfifo_count(), fix the qinfifo empty case. In ahd_dump_card_state(), print out CCSCBCTL in the correct mode. If we are a narrow controller, don't set the current width to unknown when forcing a future negotiation. This just confuses the code into attempting a wide negotiation on a narrow bus. Add support for task management function completions. Modify ahd_handle_devreset so that it can handle lun resets in addition to target resets. Use ahd_handle_devreset for lun and target reset task management functions. Handle the abort task TMF race case better. We now wait until any current selections are over and then set the TMF back to zero. This should cause the sequencer to ignore the abort TMF completion should it occur. Correct a bug in the illegal phase handler that caused us to drop down to narrow when handling the unexpected command phase case after 3rd party reset of a packetized device. Indicate the features, bugs, and flags set in the softc that are used to control firmware patch download when booting verbose. aic79xx.h: Add coalessing and HS_MAILBOX fields. Add per-softc variables for the stats "daemon". Add a debug option for interrupt coalessing activities. Add two new softc flags: o AHD_UPDATE_PEND_CMDS Run ahd_reset_cmds_pending() on the next unpause. o AHD_RUNNING_QOUTFIFO Used to catch recursion through ahd_run_qoutfifo(). aic79xx.reg: Correct register addresses related to the software timer and the DFDBCTL register. Add constants paramaterizing the software timer. Add scratch ram locations for storing interrupt coalessing tunables. Break INTMASK in SEQITNCTL out into INTMASK1 and INTMASK2. In at least the REV A, these are writable bits. We make use of that for a swtimer workaround in the sequencer. Since HS_MAILBOX autoclears, provide a sequencer variable to store its contents. Add SEQINT codes for handling task management completions. aic79xx.seq: Correct ignore wide residue processing check for a wide negotiation being in effect. We must be in the SCSI register window in order to access the negotiation table. Use the software timer and a commands completed count to implement interrupt coalessing. The command complete is deferred until either the maximum command threshold or a the expiration of a command deferral timer. If we have more SCBs to complete to the host (sitting in COMPLETE_SCB lists), always try to coaless them up to our coalessing limit. If coalessing is enabled, but we have fewer commands oustanting than the host's min coalessing limit, complete the command immediately. Add code to track the number of commands outstanding. Commands are outstanding from the time they are placed into the execution queue until the DMA to post completion is setup. Add a workaround for intvec_2 interrupts on the H2A4. In H2A4, the mode pointer is not saved for intvec2, but is restored on iret. This can lead to the restoration of a bogus mode ptr. Manually clear the intmask bits and do a normal return to compensate. We use intvec_2 to track interrupt coalessing timeouts. Since we cannot disable the swtimer's countdown, simply mask its interrupt once we no longer care about it firing. In idle_loop_cchan, update LOCAL_HS_MAILBOX everytime we are notified of an HS_MAILBOX update via the HS_MAILBOX_ACT bit in QOFF_CTLSTA. We have to use a local copy of persistant portions of the HS_MAILBOX as the mailbox auto-clears on any read. Move the test for the cfg4istat interrupt up an instruction to hopefully close a race between the next outgoing selection and our disabling of selections. Add a missing ret to the last instruction in load_overrun_buf. Add notifications to the host of task management completions as well as the completions for commands that completed successfully before their corresponding TMF could be sent. Hold a critical section during select-out processing until we have a fully identified connection. This removes a race condition with the legacy abort handler. Correct a few spelling errors in some comments. aic79xx_inline.h: Call ahd_reset_cmds_pending() in ahd_unpause if required. Update cmdcmplt interrupt statistics in our interrupt handler. Allow callers to ahd_send_scb() to set the task management function. aic79xx_pci.c: Disable SERR and pause the controller prior to performing our mmapped I/O test. The U320 controllers do not support "auto-access-pause". aic79xx_osm.c: Set the task management function now that ahd_send_scb() doesn't do it for us. We also perform a lun reset in response to BDR requests to packetized devices.
2003-01-20 20:17:35 +00:00
* If a command completed before an attempted
* task management function completed, notify the host.
*/
aic79xx.c: Implement the SCB_SILENT flag. This is useful for hushing up the driver during DV or other operations that we expect to cause transmission errors. The messages will still print if the SHOW_MASKED_ERRORS debug option is enabled. Save and restore the NEGOADDR address when setting new transfer settings. The sequencer performs lookups in the negotiation table too and it expects NEGOADDR to remain consistent across pause/unpause sessions. Consistently use "offset" instead of "period" to determine if we are running sync or not. Add a SHOW_MESSAGES diagnostic for when we assert ATN during message processing. Print out IU, QAS, and RTI features when showing transfer options. Limit the syncrate after all option conformance changes have taken place in ahd_devlimited_syncrate. Changes in options may change the final syncrate we accept. Keep a copy of the hs_mailbox in our softc so that we can perform read/modify/write operations on the hs_mailbox without having to pause the sequencer to read the last written value. Use the ENINT_COALESS flag in the hs_mailbox to toggle interrupt coalessing. Add entrypoints for enabling interrupt coalessing and setting both a timeout (how long to wait for commands to be coalessed) and a maximum commands to coaless value. Add a statistics timer that decides when to enable or disable interrupt coalessing based on load. Add a routine, ahd_reset_cmds_pending() which is used to update the CMDS_PENDING sequencer variable whenever error recovery compeltes SCBs without notifying the sequencer. Since ahd_reset_cmds_pending is called during ahd_unpause() only if we've aborted SCBs, its call to ahd_flush_qoutfifo should not cause recursion through ahd_run_qoutfifo(). A panic has been added to ensure that this recursion does not occur. In ahd_search_qinfifo, update the CMDS_PENDING sequencer variable directly. ahd_search_qinififo can be called in situations where using ahd_reset_cmds_pending() might cause recursion. Since we can safely determine the exact number to reduce CMDS_PENDING by in this scenario without running the qoutfifo, the manual update is sufficient. Clean up diagnostics. Add ahd_flush_qoutfifo() which will run the qoutfifo as well as complete any commands sitting on the sequencer's COMPLETE_SCB lists or the good status FIFO. Use this routine in several places that did similar things in an add-hoc, but incomplete, fashion. A call to this routine was also added to ahd_abort_scbs() to close a race. In ahd_pause_and_flushwork() only return once selections are safely disabled. Flush all completed commands via ahd_flush_qoutfifo(). Remove "Now packetized" diagnostic now that this information is incorperated into the actual negotiation messages that are displayed. When forcing renegotiation, don't clober the current ppr_options. Much of the driver uses this information to determine if we are currently packetized or not. Remove some stray spaces at column 1 in ahd_set_tags. When complaining about getting a host message loop request with no pending messages, print out the SCB_CONTROL register down on the card. Modify the ahd_sent_msg() routine to handle a search for an outgoing identify message. Use this to detect a msg reject on an identify message which typically indicates that the target thought we were packetized. Force a renegotiation in this case. In ahd_search_qinfifo(), wait more effectively for SCB DMA activities to cease. We also disable SCB fetch operations since we are about to change the qinfifo and any fetch in progress will likely be invalidated. In ahd_qinfifo_count(), fix the qinfifo empty case. In ahd_dump_card_state(), print out CCSCBCTL in the correct mode. If we are a narrow controller, don't set the current width to unknown when forcing a future negotiation. This just confuses the code into attempting a wide negotiation on a narrow bus. Add support for task management function completions. Modify ahd_handle_devreset so that it can handle lun resets in addition to target resets. Use ahd_handle_devreset for lun and target reset task management functions. Handle the abort task TMF race case better. We now wait until any current selections are over and then set the TMF back to zero. This should cause the sequencer to ignore the abort TMF completion should it occur. Correct a bug in the illegal phase handler that caused us to drop down to narrow when handling the unexpected command phase case after 3rd party reset of a packetized device. Indicate the features, bugs, and flags set in the softc that are used to control firmware patch download when booting verbose. aic79xx.h: Add coalessing and HS_MAILBOX fields. Add per-softc variables for the stats "daemon". Add a debug option for interrupt coalessing activities. Add two new softc flags: o AHD_UPDATE_PEND_CMDS Run ahd_reset_cmds_pending() on the next unpause. o AHD_RUNNING_QOUTFIFO Used to catch recursion through ahd_run_qoutfifo(). aic79xx.reg: Correct register addresses related to the software timer and the DFDBCTL register. Add constants paramaterizing the software timer. Add scratch ram locations for storing interrupt coalessing tunables. Break INTMASK in SEQITNCTL out into INTMASK1 and INTMASK2. In at least the REV A, these are writable bits. We make use of that for a swtimer workaround in the sequencer. Since HS_MAILBOX autoclears, provide a sequencer variable to store its contents. Add SEQINT codes for handling task management completions. aic79xx.seq: Correct ignore wide residue processing check for a wide negotiation being in effect. We must be in the SCSI register window in order to access the negotiation table. Use the software timer and a commands completed count to implement interrupt coalessing. The command complete is deferred until either the maximum command threshold or a the expiration of a command deferral timer. If we have more SCBs to complete to the host (sitting in COMPLETE_SCB lists), always try to coaless them up to our coalessing limit. If coalessing is enabled, but we have fewer commands oustanting than the host's min coalessing limit, complete the command immediately. Add code to track the number of commands outstanding. Commands are outstanding from the time they are placed into the execution queue until the DMA to post completion is setup. Add a workaround for intvec_2 interrupts on the H2A4. In H2A4, the mode pointer is not saved for intvec2, but is restored on iret. This can lead to the restoration of a bogus mode ptr. Manually clear the intmask bits and do a normal return to compensate. We use intvec_2 to track interrupt coalessing timeouts. Since we cannot disable the swtimer's countdown, simply mask its interrupt once we no longer care about it firing. In idle_loop_cchan, update LOCAL_HS_MAILBOX everytime we are notified of an HS_MAILBOX update via the HS_MAILBOX_ACT bit in QOFF_CTLSTA. We have to use a local copy of persistant portions of the HS_MAILBOX as the mailbox auto-clears on any read. Move the test for the cfg4istat interrupt up an instruction to hopefully close a race between the next outgoing selection and our disabling of selections. Add a missing ret to the last instruction in load_overrun_buf. Add notifications to the host of task management completions as well as the completions for commands that completed successfully before their corresponding TMF could be sent. Hold a critical section during select-out processing until we have a fully identified connection. This removes a race condition with the legacy abort handler. Correct a few spelling errors in some comments. aic79xx_inline.h: Call ahd_reset_cmds_pending() in ahd_unpause if required. Update cmdcmplt interrupt statistics in our interrupt handler. Allow callers to ahd_send_scb() to set the task management function. aic79xx_pci.c: Disable SERR and pause the controller prior to performing our mmapped I/O test. The U320 controllers do not support "auto-access-pause". aic79xx_osm.c: Set the task management function now that ahd_send_scb() doesn't do it for us. We also perform a lun reset in response to BDR requests to packetized devices.
2003-01-20 20:17:35 +00:00
test SCB_TASK_MANAGEMENT, 0xFF jz cfg4istat_no_taskmgmt_func;
SET_SEQINTCODE(TASKMGMT_CMD_CMPLT_OKAY)
cfg4istat_no_taskmgmt_func:
call pkt_handle_status;
or SEQINTCTL, IRET ret;
aic79xx.c: Clear the LQICRC_NLQ status should it pop up after we have already handled the SCSIPERR. During some streaming operations this status can be delayed until the stream ends. Without this change, the driver would complain about a "Missing case in ahd_handle_scsiint". In the LQOBUSFREE handler... Don't return the LQOMGR back to the idle state until after we have cleaned up ENSELO and any status related to this selection. The last thing we need is the LQO manager starting another select-out before we have updated the execution queue. It is not clear whether the LQOMGR would, or would not start a new selection early. Make sure ENSELO is off prior to clearing SELDO by flushing device writes. Move assignment of the next target SCB pointer inside of an if to make the code clearer. The effect is the same. Dump card state in both "Unexpected PKT busfree" paths. In ahd_reset(), set the chip to SCSI mode before reading SXFRCTL1. That register only exists in the SCSI mode. Also set the mode explicitly to the SCSI mode after chip reset due to paranoia. Re-arrange code so that SXFRCTL1 is restored as quickly after the chip reset as possible. S/G structurs must be 8byte aligned. Make this official by saying so in our DMA tag. Disable CIO bus stretch on MDFFSTAT if SHVALID is about to come true. This can cause a CIO bus lockup if a PCI or PCI-X error occurs while the stretch is occurring - the host cannot service the PCI-X error since the CIO bus is locked out and SHVALID will never resolve. The stretch was added in the Rev B to simplify the wait for SHVALID to resolve, but the code to do this in the open source sequencer is so simple it was never removed. Consistently use MAX_OFFSET for the user max syncrate set from non-volatile storage. This ensures that the offset does not conflict with AH?_OFFSET_UNKNOWN. Have ahd_pause_and_flushwork set the mode to ensure that it has access to the registers it checks. Also modify the checking of intstat so that the check against 0xFF can actually succeed if the INT_PEND mask is something other than 0xFF. Although there are no cardbus U320 controllers, this check may be needed to recover from a hot-plug PCI removal that occurs without informing the driver. Fix a typo. sg_prefetch_cnt -> sg_prefetch_align. This fixes an infinite loop at card initialization if the cacheline size is 0. aic79xx.h: Add AHD_EARLY_REQ_BUG bug flag. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. Change AH?_OFFSET_UNKNOWN to 0xFF. This is a value that the curr->offset can never be, unlike '0' which we previously used. This fixes code that only checks for a non-zero offset to determine if a sync negotiation is required since it will fire in the unknown case even if the goal is async. aic79xx.reg: Add comments for LQISTAT bits indicating their names in the 7902 data book. We use slightly different and more descriptive names in the firmware. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. aic79xx.seq: Update comments regarding rundown of the GSFIFO to reflect reality. Fix spelling errors. Since we use an 8byte address and 1 byte length, shorten the size of a block move for the legacy DMA'ed CDB case from 11 to 9 bytes. Remove code that, assuming the abort pending feature worked, would set MK_MESSAGE in the SCB's control byte on completion to catch invalid reselections. Since we don't see interrupts for completed selections, this status update could occur prior to us noticing the SELDO. The "select-out" queue logic will get confused by the MK_MESSAGE bit being set as this is used to catch packatized connections where we select-out with ATN. Since the abort pending feature doesn't work on any released controllers yet, this code was never executed. Add support for the AHD_EARLY_REQ_BUG. Don't ignore persistent REQ assertions just because they were asserted within the bus settle delay window. This allows us to tolerate devices like the GEM318 that violate the SCSI spec. Remove unintentional settnig of SG_CACHE_AVAIL. Writing this bit should have no effect, but who knows... On the Rev A, we must wait for HDMAENACK before loading additional segments to avoid clobbering the address of the first segment in the S/G FIFO. This resolves data-corruption issues with certain IBM (now Hitachi) and Fujitsu U320 drives. Rearrange calc_residual to avoid an extra jmp instruction. On RevA Silicon, if the target returns us to data-out after we have already trained for data-out, it is possible for us to transition the free running clock to data-valid before the required 100ns P1 setup time (8 P1 assertions in fast-160 mode). This will only happen if this L-Q is a continuation of a data transfer for which we have already prefetched data into our FIFO (LQ/Data followed by LQ/Data for the same write transaction). This can cause some target implementations to miss the first few data transfers on the bus. We detect this situation by noticing that this is the first data transfer after an LQ (LQIWORKONLQ true), that the data transfer is a continuation of a transfer already setup in our FIFO (SAVEPTRS interrupt), and that the transaction is a write (DIRECTION set in DFCNTRL). The delay is performed by disabling SCSIEN until we see the first REQ from the target. Only compile in snapshot savepointers handler for RevA silicon where it is enabled. Handle the cfg4icmd packetized interrupt. We just need to load the address and count, start the DMA, and CLRCHN once the transfer is complete. Fix an oversight in the overrun handler for packetized status operations. We need to wait for either CTXTDONE or an overrun when checking for an overrun. The previous code did not wait and thus could decide that no overrun had occurred even though an overrun will occur on the next data-valid req. Add some comment to this section for clarity. Use LAST_SEG_DONE instead of LASTSDONE for testing transfer completion in the packetized status case. LASTSDONE may come up more quickly since it only records completion on the SCSI side, but since LAST_SEG_DONE is used everywhere else (and needs to be), this is less confusing. Add a missing invalidation of the longjmp address in the non-pack handler. This code needs additional review. aic79xx_inline.h: Fix spelling error. aic79xx_osm.c: Set the cdb length for CDBs dma'ed from host memory. Add a comment indicating that, should CAM start supporting cdbs larger than 16bytes, the driver could store the CDB in the status buffer. aic79xx_pci.c: Add a table entry for the 39320A. Added a missing comma to an error string table. Fix spelling errors.
2003-02-27 23:23:16 +00:00
cfg4icmd_intr:
/*
* In the case of DMAing a CDB from the host, the normal
* CDB buffer is formatted with an 8 byte address followed
* by a 1 byte count.
*/
bmov HADDR[0], SCB_HOST_CDB_PTR, 9;
mvi SG_CACHE_PRE, LAST_SEG;
mvi DFCNTRL, (PRELOADEN|SCSIEN|HDMAEN);
call pkt_handle_cdb;
or SEQINTCTL, IRET ret;
/*
* See if the target has gone on in this context creating an
* overrun condition. For the write case, the hardware cannot
Upgrade to version 1.1 of the aic79xx U320 driver. aic79xx.c: o Remove redundant ahd_update_modes() call. o Correct panic in diagnostic should state corruption cause the SCB Id to be invalid during a selection timeout. o Add workaround for missing BUSFREEREV feature in Rev A silicon. o Corect formatting nits. o Use register pretty printing in more places. o Save and restore our SCB pointer when updating the waiting queue list for an "expected" LQ-out busfree. o In ahd_clear_intstat, deal with the missing autoclear in the CLRLQO* registers. o BE fixup in a diagnostic printf. o Make sure that we are in the proper mode before disabling selections in ahd_update_pending_scbs. o Add more diagnostics. o task_attribute_nonpkt_tag -> task_attribute: we don't need a nonpkt_tag field anymore for allowing all 512 SCBs to be used in non-packetized connections. o Negotiate HOLD_MCS to U320 devices. o Add a few additional mode assertions. o Restore the chip mode after clearing out the qinfifo so that code using ahd_abort_scbs sees a consistent mode. o Simplify the DMA engine shutdown routine prior to performing a bus reset. o Perform the sequencer restart after a chip reset prior to setting up our timer to poll for the reset to be complete. On some OSes, the timer could actually pre-empt us and order is important here. o Have our "reset poller" set the expected mode since there is no guarantee of what mode will be in force when we are called from the OS timer. o Save and restore the SCB pointer in ahd_dump_card_state(). This routine must not modify card state. o Ditto for ahd_dump_scbs(). aic79xx.h: o Add a few more chip bug definitions. o Align our tag on a 32bit boundary. aic79xx.reg: aic79xx.seq: o Start work on removing workarounds for Rev B. o Use a special location in scratch from for stroring our SCBPTR during legacy FIFO allocations. This corrects problems in mixed packetized/non-packetized configurations where calling into a FIFO task corrupted our SCBPTR. o Don't rely on DMA priority to guarantee that all data in our FIFOs will flush prior to a command completion notification going out of the command channel. We've never seen this assumption fail, but better safe than sorry. o Deal with missing BUSFREEREV feature in H2A. o Simplify disconnect list code now that the list will always have only a single entry. o Implement the AHD_REG_SLOW_SETTLE_BUG workaround. o Swith to using "REG_ISR" for local mode scratch during our ISR. o Add a missing jmp to the data_group_dma_loop after our data pointers have been re-initialized by the kernel. o Correct test in the bitbucket code so that we actually wait for the bitbucket to complete before signaling the kernel of the overrun condition. o Reposition pkt_saveptrs to avoid a jmp instruction. o Update a comment to reflect that the code now waits for a FIFO to drain prior to issuing a CLRCHN. aic79xx_inline.h: o Remove unused untagged queue handling code. o Don't attempt to htole64 what could be a 32bit value. aic79xx_pci.c: o Set additional bug flags for rev A chips.
2002-09-26 22:54:00 +00:00
* ack bytes until data are provided. So, if the target begins
* another packet without changing contexts, implying we are
* not sitting on a packet boundary, we are in an overrun
* situation. For the read case, the hardware will continue to
* ack bytes into the FIFO, and may even ack the last overrun packet
* into the FIFO. If the FIFO should become non-empty, we are in
* a read overrun case.
*/
#define check_overrun \
/* Not on a packet boundary. */ \
test MDFFSTAT, DLZERO jz pkt_handle_overrun; \
test DFSTATUS, FIFOEMP jz pkt_handle_overrun
pkt_handle_xfer:
test SG_STATE, LOADING_NEEDED jz pkt_last_seg;
call setjmp;
test SEQINTSRC, SAVEPTRS jnz pkt_saveptrs;
test SCSIPHASE, ~DATA_PHASE_MASK jz . + 2;
test SCSISIGO, ATNO jnz . + 2;
test SSTAT2, NONPACKREQ jz pkt_service_fifo;
/*
* Defer handling of this NONPACKREQ until we
* can be sure it pertains to this FIFO. SAVEPTRS
* will not be asserted if the NONPACKREQ is for us,
* so we must simulate it if shaddow is valid. If
* shaddow is not valid, keep running this FIFO until we
* have satisfied the transfer by loading segments and
* waiting for either shaddow valid or last_seg_done.
*/
test MDFFSTAT, SHVALID jnz pkt_saveptrs;
pkt_service_fifo:
test SG_STATE, LOADING_NEEDED jnz service_fifo;
pkt_last_seg:
call setjmp;
test SEQINTSRC, SAVEPTRS jnz pkt_saveptrs;
Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-05-04 00:20:07 +00:00
test SG_CACHE_SHADOW, LAST_SEG_DONE jnz pkt_last_seg_done;
test SCSIPHASE, ~DATA_PHASE_MASK jz . + 2;
test SCSISIGO, ATNO jnz . + 2;
test SSTAT2, NONPACKREQ jz return;
Upgrade to version 1.1 of the aic79xx U320 driver. aic79xx.c: o Remove redundant ahd_update_modes() call. o Correct panic in diagnostic should state corruption cause the SCB Id to be invalid during a selection timeout. o Add workaround for missing BUSFREEREV feature in Rev A silicon. o Corect formatting nits. o Use register pretty printing in more places. o Save and restore our SCB pointer when updating the waiting queue list for an "expected" LQ-out busfree. o In ahd_clear_intstat, deal with the missing autoclear in the CLRLQO* registers. o BE fixup in a diagnostic printf. o Make sure that we are in the proper mode before disabling selections in ahd_update_pending_scbs. o Add more diagnostics. o task_attribute_nonpkt_tag -> task_attribute: we don't need a nonpkt_tag field anymore for allowing all 512 SCBs to be used in non-packetized connections. o Negotiate HOLD_MCS to U320 devices. o Add a few additional mode assertions. o Restore the chip mode after clearing out the qinfifo so that code using ahd_abort_scbs sees a consistent mode. o Simplify the DMA engine shutdown routine prior to performing a bus reset. o Perform the sequencer restart after a chip reset prior to setting up our timer to poll for the reset to be complete. On some OSes, the timer could actually pre-empt us and order is important here. o Have our "reset poller" set the expected mode since there is no guarantee of what mode will be in force when we are called from the OS timer. o Save and restore the SCB pointer in ahd_dump_card_state(). This routine must not modify card state. o Ditto for ahd_dump_scbs(). aic79xx.h: o Add a few more chip bug definitions. o Align our tag on a 32bit boundary. aic79xx.reg: aic79xx.seq: o Start work on removing workarounds for Rev B. o Use a special location in scratch from for stroring our SCBPTR during legacy FIFO allocations. This corrects problems in mixed packetized/non-packetized configurations where calling into a FIFO task corrupted our SCBPTR. o Don't rely on DMA priority to guarantee that all data in our FIFOs will flush prior to a command completion notification going out of the command channel. We've never seen this assumption fail, but better safe than sorry. o Deal with missing BUSFREEREV feature in H2A. o Simplify disconnect list code now that the list will always have only a single entry. o Implement the AHD_REG_SLOW_SETTLE_BUG workaround. o Swith to using "REG_ISR" for local mode scratch during our ISR. o Add a missing jmp to the data_group_dma_loop after our data pointers have been re-initialized by the kernel. o Correct test in the bitbucket code so that we actually wait for the bitbucket to complete before signaling the kernel of the overrun condition. o Reposition pkt_saveptrs to avoid a jmp instruction. o Update a comment to reflect that the code now waits for a FIFO to drain prior to issuing a CLRCHN. aic79xx_inline.h: o Remove unused untagged queue handling code. o Don't attempt to htole64 what could be a 32bit value. aic79xx_pci.c: o Set additional bug flags for rev A chips.
2002-09-26 22:54:00 +00:00
test MDFFSTAT, SHVALID jz return;
/* FALLTHROUGH */
/*
* Either a SAVEPTRS interrupt condition is pending for this FIFO
Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-05-04 00:20:07 +00:00
* or we have a pending NONPACKREQ for this FIFO. We differentiate
Upgrade to version 1.1 of the aic79xx U320 driver. aic79xx.c: o Remove redundant ahd_update_modes() call. o Correct panic in diagnostic should state corruption cause the SCB Id to be invalid during a selection timeout. o Add workaround for missing BUSFREEREV feature in Rev A silicon. o Corect formatting nits. o Use register pretty printing in more places. o Save and restore our SCB pointer when updating the waiting queue list for an "expected" LQ-out busfree. o In ahd_clear_intstat, deal with the missing autoclear in the CLRLQO* registers. o BE fixup in a diagnostic printf. o Make sure that we are in the proper mode before disabling selections in ahd_update_pending_scbs. o Add more diagnostics. o task_attribute_nonpkt_tag -> task_attribute: we don't need a nonpkt_tag field anymore for allowing all 512 SCBs to be used in non-packetized connections. o Negotiate HOLD_MCS to U320 devices. o Add a few additional mode assertions. o Restore the chip mode after clearing out the qinfifo so that code using ahd_abort_scbs sees a consistent mode. o Simplify the DMA engine shutdown routine prior to performing a bus reset. o Perform the sequencer restart after a chip reset prior to setting up our timer to poll for the reset to be complete. On some OSes, the timer could actually pre-empt us and order is important here. o Have our "reset poller" set the expected mode since there is no guarantee of what mode will be in force when we are called from the OS timer. o Save and restore the SCB pointer in ahd_dump_card_state(). This routine must not modify card state. o Ditto for ahd_dump_scbs(). aic79xx.h: o Add a few more chip bug definitions. o Align our tag on a 32bit boundary. aic79xx.reg: aic79xx.seq: o Start work on removing workarounds for Rev B. o Use a special location in scratch from for stroring our SCBPTR during legacy FIFO allocations. This corrects problems in mixed packetized/non-packetized configurations where calling into a FIFO task corrupted our SCBPTR. o Don't rely on DMA priority to guarantee that all data in our FIFOs will flush prior to a command completion notification going out of the command channel. We've never seen this assumption fail, but better safe than sorry. o Deal with missing BUSFREEREV feature in H2A. o Simplify disconnect list code now that the list will always have only a single entry. o Implement the AHD_REG_SLOW_SETTLE_BUG workaround. o Swith to using "REG_ISR" for local mode scratch during our ISR. o Add a missing jmp to the data_group_dma_loop after our data pointers have been re-initialized by the kernel. o Correct test in the bitbucket code so that we actually wait for the bitbucket to complete before signaling the kernel of the overrun condition. o Reposition pkt_saveptrs to avoid a jmp instruction. o Update a comment to reflect that the code now waits for a FIFO to drain prior to issuing a CLRCHN. aic79xx_inline.h: o Remove unused untagged queue handling code. o Don't attempt to htole64 what could be a 32bit value. aic79xx_pci.c: o Set additional bug flags for rev A chips.
2002-09-26 22:54:00 +00:00
* between the two by capturing the state of the SAVEPTRS interrupt
* prior to clearing this status and executing the common code for
* these two cases.
*/
pkt_saveptrs:
BEGIN_CRITICAL;
if ((ahd->bugs & AHD_AUTOFLUSH_BUG) != 0) {
or DFCNTRL, FIFOFLUSH;
}
mov REG0, SEQINTSRC;
call calc_residual;
call save_pointers;
mvi CLRSEQINTSRC, CLRSAVEPTRS;
call disable_ccsgen;
or SEQIMODE, ENSAVEPTRS;
test DFCNTRL, DIRECTION jnz pkt_saveptrs_check_status;
test DFSTATUS, FIFOEMP jnz pkt_saveptrs_check_status;
/*
* Keep a handler around for this FIFO until it drains
* to the host to guarantee that we don't complete the
* command to the host before the data arrives.
*/
pkt_saveptrs_wait_fifoemp:
call setjmp;
test DFSTATUS, FIFOEMP jz return;
pkt_saveptrs_check_status:
or LONGJMP_ADDR[1], INVALID_ADDR;
test REG0, SAVEPTRS jz unexpected_nonpkt_phase;
Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-05-04 00:20:07 +00:00
dec SCB_FIFO_USE_COUNT;
test SCB_CONTROL, STATUS_RCVD jnz pkt_complete_scb_if_fifos_idle;
Upgrade to version 1.1 of the aic79xx U320 driver. aic79xx.c: o Remove redundant ahd_update_modes() call. o Correct panic in diagnostic should state corruption cause the SCB Id to be invalid during a selection timeout. o Add workaround for missing BUSFREEREV feature in Rev A silicon. o Corect formatting nits. o Use register pretty printing in more places. o Save and restore our SCB pointer when updating the waiting queue list for an "expected" LQ-out busfree. o In ahd_clear_intstat, deal with the missing autoclear in the CLRLQO* registers. o BE fixup in a diagnostic printf. o Make sure that we are in the proper mode before disabling selections in ahd_update_pending_scbs. o Add more diagnostics. o task_attribute_nonpkt_tag -> task_attribute: we don't need a nonpkt_tag field anymore for allowing all 512 SCBs to be used in non-packetized connections. o Negotiate HOLD_MCS to U320 devices. o Add a few additional mode assertions. o Restore the chip mode after clearing out the qinfifo so that code using ahd_abort_scbs sees a consistent mode. o Simplify the DMA engine shutdown routine prior to performing a bus reset. o Perform the sequencer restart after a chip reset prior to setting up our timer to poll for the reset to be complete. On some OSes, the timer could actually pre-empt us and order is important here. o Have our "reset poller" set the expected mode since there is no guarantee of what mode will be in force when we are called from the OS timer. o Save and restore the SCB pointer in ahd_dump_card_state(). This routine must not modify card state. o Ditto for ahd_dump_scbs(). aic79xx.h: o Add a few more chip bug definitions. o Align our tag on a 32bit boundary. aic79xx.reg: aic79xx.seq: o Start work on removing workarounds for Rev B. o Use a special location in scratch from for stroring our SCBPTR during legacy FIFO allocations. This corrects problems in mixed packetized/non-packetized configurations where calling into a FIFO task corrupted our SCBPTR. o Don't rely on DMA priority to guarantee that all data in our FIFOs will flush prior to a command completion notification going out of the command channel. We've never seen this assumption fail, but better safe than sorry. o Deal with missing BUSFREEREV feature in H2A. o Simplify disconnect list code now that the list will always have only a single entry. o Implement the AHD_REG_SLOW_SETTLE_BUG workaround. o Swith to using "REG_ISR" for local mode scratch during our ISR. o Add a missing jmp to the data_group_dma_loop after our data pointers have been re-initialized by the kernel. o Correct test in the bitbucket code so that we actually wait for the bitbucket to complete before signaling the kernel of the overrun condition. o Reposition pkt_saveptrs to avoid a jmp instruction. o Update a comment to reflect that the code now waits for a FIFO to drain prior to issuing a CLRCHN. aic79xx_inline.h: o Remove unused untagged queue handling code. o Don't attempt to htole64 what could be a 32bit value. aic79xx_pci.c: o Set additional bug flags for rev A chips.
2002-09-26 22:54:00 +00:00
mvi DFFSXFRCTL, CLRCHN ret;
Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-05-04 00:20:07 +00:00
/*
* LAST_SEG_DONE status has been seen in the current FIFO.
* This indicates that all of the allowed data for this
* command has transferred across the SCSI and host buses.
* Check for overrun and see if we can complete this command.
*/
pkt_last_seg_done:
/*
Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-05-04 00:20:07 +00:00
* Mark transfer as completed.
*/
Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-05-04 00:20:07 +00:00
or SCB_SGPTR, SG_LIST_NULL;
Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-05-04 00:20:07 +00:00
/*
* Wait for the current context to finish to verify that
* no overrun condition has occurred.
*/
test SEQINTSRC, CTXTDONE jnz pkt_ctxt_done;
call setjmp;
Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-05-04 00:20:07 +00:00
pkt_wait_ctxt_done_loop:
test SEQINTSRC, CTXTDONE jnz pkt_ctxt_done;
/*
* A sufficiently large overrun or a NONPACKREQ may
* prevent CTXTDONE from ever asserting, so we must
* poll for these statuses too.
*/
check_overrun;
test SSTAT2, NONPACKREQ jz return;
test SEQINTSRC, CTXTDONE jz unexpected_nonpkt_phase;
Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-05-04 00:20:07 +00:00
/* FALLTHROUGH */
pkt_ctxt_done:
check_overrun;
or LONGJMP_ADDR[1], INVALID_ADDR;
Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-05-04 00:20:07 +00:00
/*
* If status has been received, it is safe to skip
* the check to see if another FIFO is active because
* LAST_SEG_DONE has been observed. However, we check
* the FIFO anyway since it costs us only one extra
* instruction to leverage common code to perform the
* SCB completion.
*/
dec SCB_FIFO_USE_COUNT;
test SCB_CONTROL, STATUS_RCVD jnz pkt_complete_scb_if_fifos_idle;
mvi DFFSXFRCTL, CLRCHN ret;
END_CRITICAL;
Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-05-04 00:20:07 +00:00
/*
* Must wait until CDB xfer is over before issuing the
* clear channel.
*/
pkt_handle_cdb:
call setjmp;
test SG_CACHE_SHADOW, LAST_SEG_DONE jz return;
or LONGJMP_ADDR[1], INVALID_ADDR;
mvi DFFSXFRCTL, CLRCHN ret;
aic79xx.c: Clear the LQICRC_NLQ status should it pop up after we have already handled the SCSIPERR. During some streaming operations this status can be delayed until the stream ends. Without this change, the driver would complain about a "Missing case in ahd_handle_scsiint". In the LQOBUSFREE handler... Don't return the LQOMGR back to the idle state until after we have cleaned up ENSELO and any status related to this selection. The last thing we need is the LQO manager starting another select-out before we have updated the execution queue. It is not clear whether the LQOMGR would, or would not start a new selection early. Make sure ENSELO is off prior to clearing SELDO by flushing device writes. Move assignment of the next target SCB pointer inside of an if to make the code clearer. The effect is the same. Dump card state in both "Unexpected PKT busfree" paths. In ahd_reset(), set the chip to SCSI mode before reading SXFRCTL1. That register only exists in the SCSI mode. Also set the mode explicitly to the SCSI mode after chip reset due to paranoia. Re-arrange code so that SXFRCTL1 is restored as quickly after the chip reset as possible. S/G structurs must be 8byte aligned. Make this official by saying so in our DMA tag. Disable CIO bus stretch on MDFFSTAT if SHVALID is about to come true. This can cause a CIO bus lockup if a PCI or PCI-X error occurs while the stretch is occurring - the host cannot service the PCI-X error since the CIO bus is locked out and SHVALID will never resolve. The stretch was added in the Rev B to simplify the wait for SHVALID to resolve, but the code to do this in the open source sequencer is so simple it was never removed. Consistently use MAX_OFFSET for the user max syncrate set from non-volatile storage. This ensures that the offset does not conflict with AH?_OFFSET_UNKNOWN. Have ahd_pause_and_flushwork set the mode to ensure that it has access to the registers it checks. Also modify the checking of intstat so that the check against 0xFF can actually succeed if the INT_PEND mask is something other than 0xFF. Although there are no cardbus U320 controllers, this check may be needed to recover from a hot-plug PCI removal that occurs without informing the driver. Fix a typo. sg_prefetch_cnt -> sg_prefetch_align. This fixes an infinite loop at card initialization if the cacheline size is 0. aic79xx.h: Add AHD_EARLY_REQ_BUG bug flag. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. Change AH?_OFFSET_UNKNOWN to 0xFF. This is a value that the curr->offset can never be, unlike '0' which we previously used. This fixes code that only checks for a non-zero offset to determine if a sync negotiation is required since it will fire in the unknown case even if the goal is async. aic79xx.reg: Add comments for LQISTAT bits indicating their names in the 7902 data book. We use slightly different and more descriptive names in the firmware. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. aic79xx.seq: Update comments regarding rundown of the GSFIFO to reflect reality. Fix spelling errors. Since we use an 8byte address and 1 byte length, shorten the size of a block move for the legacy DMA'ed CDB case from 11 to 9 bytes. Remove code that, assuming the abort pending feature worked, would set MK_MESSAGE in the SCB's control byte on completion to catch invalid reselections. Since we don't see interrupts for completed selections, this status update could occur prior to us noticing the SELDO. The "select-out" queue logic will get confused by the MK_MESSAGE bit being set as this is used to catch packatized connections where we select-out with ATN. Since the abort pending feature doesn't work on any released controllers yet, this code was never executed. Add support for the AHD_EARLY_REQ_BUG. Don't ignore persistent REQ assertions just because they were asserted within the bus settle delay window. This allows us to tolerate devices like the GEM318 that violate the SCSI spec. Remove unintentional settnig of SG_CACHE_AVAIL. Writing this bit should have no effect, but who knows... On the Rev A, we must wait for HDMAENACK before loading additional segments to avoid clobbering the address of the first segment in the S/G FIFO. This resolves data-corruption issues with certain IBM (now Hitachi) and Fujitsu U320 drives. Rearrange calc_residual to avoid an extra jmp instruction. On RevA Silicon, if the target returns us to data-out after we have already trained for data-out, it is possible for us to transition the free running clock to data-valid before the required 100ns P1 setup time (8 P1 assertions in fast-160 mode). This will only happen if this L-Q is a continuation of a data transfer for which we have already prefetched data into our FIFO (LQ/Data followed by LQ/Data for the same write transaction). This can cause some target implementations to miss the first few data transfers on the bus. We detect this situation by noticing that this is the first data transfer after an LQ (LQIWORKONLQ true), that the data transfer is a continuation of a transfer already setup in our FIFO (SAVEPTRS interrupt), and that the transaction is a write (DIRECTION set in DFCNTRL). The delay is performed by disabling SCSIEN until we see the first REQ from the target. Only compile in snapshot savepointers handler for RevA silicon where it is enabled. Handle the cfg4icmd packetized interrupt. We just need to load the address and count, start the DMA, and CLRCHN once the transfer is complete. Fix an oversight in the overrun handler for packetized status operations. We need to wait for either CTXTDONE or an overrun when checking for an overrun. The previous code did not wait and thus could decide that no overrun had occurred even though an overrun will occur on the next data-valid req. Add some comment to this section for clarity. Use LAST_SEG_DONE instead of LASTSDONE for testing transfer completion in the packetized status case. LASTSDONE may come up more quickly since it only records completion on the SCSI side, but since LAST_SEG_DONE is used everywhere else (and needs to be), this is less confusing. Add a missing invalidation of the longjmp address in the non-pack handler. This code needs additional review. aic79xx_inline.h: Fix spelling error. aic79xx_osm.c: Set the cdb length for CDBs dma'ed from host memory. Add a comment indicating that, should CAM start supporting cdbs larger than 16bytes, the driver could store the CDB in the status buffer. aic79xx_pci.c: Add a table entry for the 39320A. Added a missing comma to an error string table. Fix spelling errors.
2003-02-27 23:23:16 +00:00
/*
* Watch over the status transfer. Our host sense buffer is
* large enough to take the maximum allowed status packet.
* None-the-less, we must still catch and report overruns to
Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-05-04 00:20:07 +00:00
* the host. Additionally, properly catch unexpected non-packet
* phases that are typically caused by CRC errors in status packet
* transmission.
aic79xx.c: Clear the LQICRC_NLQ status should it pop up after we have already handled the SCSIPERR. During some streaming operations this status can be delayed until the stream ends. Without this change, the driver would complain about a "Missing case in ahd_handle_scsiint". In the LQOBUSFREE handler... Don't return the LQOMGR back to the idle state until after we have cleaned up ENSELO and any status related to this selection. The last thing we need is the LQO manager starting another select-out before we have updated the execution queue. It is not clear whether the LQOMGR would, or would not start a new selection early. Make sure ENSELO is off prior to clearing SELDO by flushing device writes. Move assignment of the next target SCB pointer inside of an if to make the code clearer. The effect is the same. Dump card state in both "Unexpected PKT busfree" paths. In ahd_reset(), set the chip to SCSI mode before reading SXFRCTL1. That register only exists in the SCSI mode. Also set the mode explicitly to the SCSI mode after chip reset due to paranoia. Re-arrange code so that SXFRCTL1 is restored as quickly after the chip reset as possible. S/G structurs must be 8byte aligned. Make this official by saying so in our DMA tag. Disable CIO bus stretch on MDFFSTAT if SHVALID is about to come true. This can cause a CIO bus lockup if a PCI or PCI-X error occurs while the stretch is occurring - the host cannot service the PCI-X error since the CIO bus is locked out and SHVALID will never resolve. The stretch was added in the Rev B to simplify the wait for SHVALID to resolve, but the code to do this in the open source sequencer is so simple it was never removed. Consistently use MAX_OFFSET for the user max syncrate set from non-volatile storage. This ensures that the offset does not conflict with AH?_OFFSET_UNKNOWN. Have ahd_pause_and_flushwork set the mode to ensure that it has access to the registers it checks. Also modify the checking of intstat so that the check against 0xFF can actually succeed if the INT_PEND mask is something other than 0xFF. Although there are no cardbus U320 controllers, this check may be needed to recover from a hot-plug PCI removal that occurs without informing the driver. Fix a typo. sg_prefetch_cnt -> sg_prefetch_align. This fixes an infinite loop at card initialization if the cacheline size is 0. aic79xx.h: Add AHD_EARLY_REQ_BUG bug flag. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. Change AH?_OFFSET_UNKNOWN to 0xFF. This is a value that the curr->offset can never be, unlike '0' which we previously used. This fixes code that only checks for a non-zero offset to determine if a sync negotiation is required since it will fire in the unknown case even if the goal is async. aic79xx.reg: Add comments for LQISTAT bits indicating their names in the 7902 data book. We use slightly different and more descriptive names in the firmware. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. aic79xx.seq: Update comments regarding rundown of the GSFIFO to reflect reality. Fix spelling errors. Since we use an 8byte address and 1 byte length, shorten the size of a block move for the legacy DMA'ed CDB case from 11 to 9 bytes. Remove code that, assuming the abort pending feature worked, would set MK_MESSAGE in the SCB's control byte on completion to catch invalid reselections. Since we don't see interrupts for completed selections, this status update could occur prior to us noticing the SELDO. The "select-out" queue logic will get confused by the MK_MESSAGE bit being set as this is used to catch packatized connections where we select-out with ATN. Since the abort pending feature doesn't work on any released controllers yet, this code was never executed. Add support for the AHD_EARLY_REQ_BUG. Don't ignore persistent REQ assertions just because they were asserted within the bus settle delay window. This allows us to tolerate devices like the GEM318 that violate the SCSI spec. Remove unintentional settnig of SG_CACHE_AVAIL. Writing this bit should have no effect, but who knows... On the Rev A, we must wait for HDMAENACK before loading additional segments to avoid clobbering the address of the first segment in the S/G FIFO. This resolves data-corruption issues with certain IBM (now Hitachi) and Fujitsu U320 drives. Rearrange calc_residual to avoid an extra jmp instruction. On RevA Silicon, if the target returns us to data-out after we have already trained for data-out, it is possible for us to transition the free running clock to data-valid before the required 100ns P1 setup time (8 P1 assertions in fast-160 mode). This will only happen if this L-Q is a continuation of a data transfer for which we have already prefetched data into our FIFO (LQ/Data followed by LQ/Data for the same write transaction). This can cause some target implementations to miss the first few data transfers on the bus. We detect this situation by noticing that this is the first data transfer after an LQ (LQIWORKONLQ true), that the data transfer is a continuation of a transfer already setup in our FIFO (SAVEPTRS interrupt), and that the transaction is a write (DIRECTION set in DFCNTRL). The delay is performed by disabling SCSIEN until we see the first REQ from the target. Only compile in snapshot savepointers handler for RevA silicon where it is enabled. Handle the cfg4icmd packetized interrupt. We just need to load the address and count, start the DMA, and CLRCHN once the transfer is complete. Fix an oversight in the overrun handler for packetized status operations. We need to wait for either CTXTDONE or an overrun when checking for an overrun. The previous code did not wait and thus could decide that no overrun had occurred even though an overrun will occur on the next data-valid req. Add some comment to this section for clarity. Use LAST_SEG_DONE instead of LASTSDONE for testing transfer completion in the packetized status case. LASTSDONE may come up more quickly since it only records completion on the SCSI side, but since LAST_SEG_DONE is used everywhere else (and needs to be), this is less confusing. Add a missing invalidation of the longjmp address in the non-pack handler. This code needs additional review. aic79xx_inline.h: Fix spelling error. aic79xx_osm.c: Set the cdb length for CDBs dma'ed from host memory. Add a comment indicating that, should CAM start supporting cdbs larger than 16bytes, the driver could store the CDB in the status buffer. aic79xx_pci.c: Add a table entry for the 39320A. Added a missing comma to an error string table. Fix spelling errors.
2003-02-27 23:23:16 +00:00
*/
pkt_handle_status:
Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-05-04 00:20:07 +00:00
call setjmp;
test SG_CACHE_SHADOW, LAST_SEG_DONE jnz pkt_status_check_overrun;
test SEQINTSRC, CTXTDONE jz pkt_status_check_nonpackreq;
test SG_CACHE_SHADOW, LAST_SEG_DONE jnz pkt_status_check_overrun;
pkt_status_IU_done:
if ((ahd->bugs & AHD_AUTOFLUSH_BUG) != 0) {
or DFCNTRL, FIFOFLUSH;
}
Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-05-04 00:20:07 +00:00
test DFSTATUS, FIFOEMP jz return;
BEGIN_CRITICAL;
or LONGJMP_ADDR[1], INVALID_ADDR;
mvi SCB_SCSI_STATUS, STATUS_PKT_SENSE;
or SCB_CONTROL, STATUS_RCVD;
Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-05-04 00:20:07 +00:00
jmp pkt_complete_scb_if_fifos_idle;
END_CRITICAL;
Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-05-04 00:20:07 +00:00
pkt_status_check_overrun:
/*
* Status PKT overruns are unceremoniously recovered with a
Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-05-04 00:20:07 +00:00
* bus reset. If we've overrun, let the host know so that
* recovery can be performed.
*
* LAST_SEG_DONE has been observed. If either CTXTDONE or
* a NONPACKREQ phase change have occurred and the FIFO is
* empty, there is no overrun.
*/
test DFSTATUS, FIFOEMP jz pkt_status_report_overrun;
test SEQINTSRC, CTXTDONE jz . + 2;
test DFSTATUS, FIFOEMP jnz pkt_status_IU_done;
test SCSIPHASE, ~DATA_PHASE_MASK jz return;
test DFSTATUS, FIFOEMP jnz pkt_status_check_nonpackreq;
pkt_status_report_overrun:
aic79xx.c: Clear the LQICRC_NLQ status should it pop up after we have already handled the SCSIPERR. During some streaming operations this status can be delayed until the stream ends. Without this change, the driver would complain about a "Missing case in ahd_handle_scsiint". In the LQOBUSFREE handler... Don't return the LQOMGR back to the idle state until after we have cleaned up ENSELO and any status related to this selection. The last thing we need is the LQO manager starting another select-out before we have updated the execution queue. It is not clear whether the LQOMGR would, or would not start a new selection early. Make sure ENSELO is off prior to clearing SELDO by flushing device writes. Move assignment of the next target SCB pointer inside of an if to make the code clearer. The effect is the same. Dump card state in both "Unexpected PKT busfree" paths. In ahd_reset(), set the chip to SCSI mode before reading SXFRCTL1. That register only exists in the SCSI mode. Also set the mode explicitly to the SCSI mode after chip reset due to paranoia. Re-arrange code so that SXFRCTL1 is restored as quickly after the chip reset as possible. S/G structurs must be 8byte aligned. Make this official by saying so in our DMA tag. Disable CIO bus stretch on MDFFSTAT if SHVALID is about to come true. This can cause a CIO bus lockup if a PCI or PCI-X error occurs while the stretch is occurring - the host cannot service the PCI-X error since the CIO bus is locked out and SHVALID will never resolve. The stretch was added in the Rev B to simplify the wait for SHVALID to resolve, but the code to do this in the open source sequencer is so simple it was never removed. Consistently use MAX_OFFSET for the user max syncrate set from non-volatile storage. This ensures that the offset does not conflict with AH?_OFFSET_UNKNOWN. Have ahd_pause_and_flushwork set the mode to ensure that it has access to the registers it checks. Also modify the checking of intstat so that the check against 0xFF can actually succeed if the INT_PEND mask is something other than 0xFF. Although there are no cardbus U320 controllers, this check may be needed to recover from a hot-plug PCI removal that occurs without informing the driver. Fix a typo. sg_prefetch_cnt -> sg_prefetch_align. This fixes an infinite loop at card initialization if the cacheline size is 0. aic79xx.h: Add AHD_EARLY_REQ_BUG bug flag. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. Change AH?_OFFSET_UNKNOWN to 0xFF. This is a value that the curr->offset can never be, unlike '0' which we previously used. This fixes code that only checks for a non-zero offset to determine if a sync negotiation is required since it will fire in the unknown case even if the goal is async. aic79xx.reg: Add comments for LQISTAT bits indicating their names in the 7902 data book. We use slightly different and more descriptive names in the firmware. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. aic79xx.seq: Update comments regarding rundown of the GSFIFO to reflect reality. Fix spelling errors. Since we use an 8byte address and 1 byte length, shorten the size of a block move for the legacy DMA'ed CDB case from 11 to 9 bytes. Remove code that, assuming the abort pending feature worked, would set MK_MESSAGE in the SCB's control byte on completion to catch invalid reselections. Since we don't see interrupts for completed selections, this status update could occur prior to us noticing the SELDO. The "select-out" queue logic will get confused by the MK_MESSAGE bit being set as this is used to catch packatized connections where we select-out with ATN. Since the abort pending feature doesn't work on any released controllers yet, this code was never executed. Add support for the AHD_EARLY_REQ_BUG. Don't ignore persistent REQ assertions just because they were asserted within the bus settle delay window. This allows us to tolerate devices like the GEM318 that violate the SCSI spec. Remove unintentional settnig of SG_CACHE_AVAIL. Writing this bit should have no effect, but who knows... On the Rev A, we must wait for HDMAENACK before loading additional segments to avoid clobbering the address of the first segment in the S/G FIFO. This resolves data-corruption issues with certain IBM (now Hitachi) and Fujitsu U320 drives. Rearrange calc_residual to avoid an extra jmp instruction. On RevA Silicon, if the target returns us to data-out after we have already trained for data-out, it is possible for us to transition the free running clock to data-valid before the required 100ns P1 setup time (8 P1 assertions in fast-160 mode). This will only happen if this L-Q is a continuation of a data transfer for which we have already prefetched data into our FIFO (LQ/Data followed by LQ/Data for the same write transaction). This can cause some target implementations to miss the first few data transfers on the bus. We detect this situation by noticing that this is the first data transfer after an LQ (LQIWORKONLQ true), that the data transfer is a continuation of a transfer already setup in our FIFO (SAVEPTRS interrupt), and that the transaction is a write (DIRECTION set in DFCNTRL). The delay is performed by disabling SCSIEN until we see the first REQ from the target. Only compile in snapshot savepointers handler for RevA silicon where it is enabled. Handle the cfg4icmd packetized interrupt. We just need to load the address and count, start the DMA, and CLRCHN once the transfer is complete. Fix an oversight in the overrun handler for packetized status operations. We need to wait for either CTXTDONE or an overrun when checking for an overrun. The previous code did not wait and thus could decide that no overrun had occurred even though an overrun will occur on the next data-valid req. Add some comment to this section for clarity. Use LAST_SEG_DONE instead of LASTSDONE for testing transfer completion in the packetized status case. LASTSDONE may come up more quickly since it only records completion on the SCSI side, but since LAST_SEG_DONE is used everywhere else (and needs to be), this is less confusing. Add a missing invalidation of the longjmp address in the non-pack handler. This code needs additional review. aic79xx_inline.h: Fix spelling error. aic79xx_osm.c: Set the cdb length for CDBs dma'ed from host memory. Add a comment indicating that, should CAM start supporting cdbs larger than 16bytes, the driver could store the CDB in the status buffer. aic79xx_pci.c: Add a table entry for the 39320A. Added a missing comma to an error string table. Fix spelling errors.
2003-02-27 23:23:16 +00:00
SET_SEQINTCODE(STATUS_OVERRUN)
Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-05-04 00:20:07 +00:00
/* SEQUENCER RESTARTED */
pkt_status_check_nonpackreq:
/*
* CTXTDONE may be held off if a NONPACKREQ is associated with
* the current context. If a NONPACKREQ is observed, decide
* if it is for the current context. If it is for the current
* context, we must defer NONPACKREQ processing until all data
* has transferred to the host.
*/
test SCSIPHASE, ~DATA_PHASE_MASK jz return;
test SCSISIGO, ATNO jnz . + 2;
test SSTAT2, NONPACKREQ jz return;
test SEQINTSRC, CTXTDONE jnz pkt_status_IU_done;
test DFSTATUS, FIFOEMP jz return;
/*
* The unexpected nonpkt phase handler assumes that any
* data channel use will have a FIFO reference count. It
* turns out that the status handler doesn't need a refernce
* count since the status received flag, and thus completion
* processing, cannot be set until the handler is finished.
* We increment the count here to make the nonpkt handler
* happy.
*/
inc SCB_FIFO_USE_COUNT;
/* FALLTHROUGH */
aic79xx.c: Clear the LQICRC_NLQ status should it pop up after we have already handled the SCSIPERR. During some streaming operations this status can be delayed until the stream ends. Without this change, the driver would complain about a "Missing case in ahd_handle_scsiint". In the LQOBUSFREE handler... Don't return the LQOMGR back to the idle state until after we have cleaned up ENSELO and any status related to this selection. The last thing we need is the LQO manager starting another select-out before we have updated the execution queue. It is not clear whether the LQOMGR would, or would not start a new selection early. Make sure ENSELO is off prior to clearing SELDO by flushing device writes. Move assignment of the next target SCB pointer inside of an if to make the code clearer. The effect is the same. Dump card state in both "Unexpected PKT busfree" paths. In ahd_reset(), set the chip to SCSI mode before reading SXFRCTL1. That register only exists in the SCSI mode. Also set the mode explicitly to the SCSI mode after chip reset due to paranoia. Re-arrange code so that SXFRCTL1 is restored as quickly after the chip reset as possible. S/G structurs must be 8byte aligned. Make this official by saying so in our DMA tag. Disable CIO bus stretch on MDFFSTAT if SHVALID is about to come true. This can cause a CIO bus lockup if a PCI or PCI-X error occurs while the stretch is occurring - the host cannot service the PCI-X error since the CIO bus is locked out and SHVALID will never resolve. The stretch was added in the Rev B to simplify the wait for SHVALID to resolve, but the code to do this in the open source sequencer is so simple it was never removed. Consistently use MAX_OFFSET for the user max syncrate set from non-volatile storage. This ensures that the offset does not conflict with AH?_OFFSET_UNKNOWN. Have ahd_pause_and_flushwork set the mode to ensure that it has access to the registers it checks. Also modify the checking of intstat so that the check against 0xFF can actually succeed if the INT_PEND mask is something other than 0xFF. Although there are no cardbus U320 controllers, this check may be needed to recover from a hot-plug PCI removal that occurs without informing the driver. Fix a typo. sg_prefetch_cnt -> sg_prefetch_align. This fixes an infinite loop at card initialization if the cacheline size is 0. aic79xx.h: Add AHD_EARLY_REQ_BUG bug flag. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. Change AH?_OFFSET_UNKNOWN to 0xFF. This is a value that the curr->offset can never be, unlike '0' which we previously used. This fixes code that only checks for a non-zero offset to determine if a sync negotiation is required since it will fire in the unknown case even if the goal is async. aic79xx.reg: Add comments for LQISTAT bits indicating their names in the 7902 data book. We use slightly different and more descriptive names in the firmware. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. aic79xx.seq: Update comments regarding rundown of the GSFIFO to reflect reality. Fix spelling errors. Since we use an 8byte address and 1 byte length, shorten the size of a block move for the legacy DMA'ed CDB case from 11 to 9 bytes. Remove code that, assuming the abort pending feature worked, would set MK_MESSAGE in the SCB's control byte on completion to catch invalid reselections. Since we don't see interrupts for completed selections, this status update could occur prior to us noticing the SELDO. The "select-out" queue logic will get confused by the MK_MESSAGE bit being set as this is used to catch packatized connections where we select-out with ATN. Since the abort pending feature doesn't work on any released controllers yet, this code was never executed. Add support for the AHD_EARLY_REQ_BUG. Don't ignore persistent REQ assertions just because they were asserted within the bus settle delay window. This allows us to tolerate devices like the GEM318 that violate the SCSI spec. Remove unintentional settnig of SG_CACHE_AVAIL. Writing this bit should have no effect, but who knows... On the Rev A, we must wait for HDMAENACK before loading additional segments to avoid clobbering the address of the first segment in the S/G FIFO. This resolves data-corruption issues with certain IBM (now Hitachi) and Fujitsu U320 drives. Rearrange calc_residual to avoid an extra jmp instruction. On RevA Silicon, if the target returns us to data-out after we have already trained for data-out, it is possible for us to transition the free running clock to data-valid before the required 100ns P1 setup time (8 P1 assertions in fast-160 mode). This will only happen if this L-Q is a continuation of a data transfer for which we have already prefetched data into our FIFO (LQ/Data followed by LQ/Data for the same write transaction). This can cause some target implementations to miss the first few data transfers on the bus. We detect this situation by noticing that this is the first data transfer after an LQ (LQIWORKONLQ true), that the data transfer is a continuation of a transfer already setup in our FIFO (SAVEPTRS interrupt), and that the transaction is a write (DIRECTION set in DFCNTRL). The delay is performed by disabling SCSIEN until we see the first REQ from the target. Only compile in snapshot savepointers handler for RevA silicon where it is enabled. Handle the cfg4icmd packetized interrupt. We just need to load the address and count, start the DMA, and CLRCHN once the transfer is complete. Fix an oversight in the overrun handler for packetized status operations. We need to wait for either CTXTDONE or an overrun when checking for an overrun. The previous code did not wait and thus could decide that no overrun had occurred even though an overrun will occur on the next data-valid req. Add some comment to this section for clarity. Use LAST_SEG_DONE instead of LASTSDONE for testing transfer completion in the packetized status case. LASTSDONE may come up more quickly since it only records completion on the SCSI side, but since LAST_SEG_DONE is used everywhere else (and needs to be), this is less confusing. Add a missing invalidation of the longjmp address in the non-pack handler. This code needs additional review. aic79xx_inline.h: Fix spelling error. aic79xx_osm.c: Set the cdb length for CDBs dma'ed from host memory. Add a comment indicating that, should CAM start supporting cdbs larger than 16bytes, the driver could store the CDB in the status buffer. aic79xx_pci.c: Add a table entry for the 39320A. Added a missing comma to an error string table. Fix spelling errors.
2003-02-27 23:23:16 +00:00
/*
* Nonpackreq is a polled status. It can come true in three situations:
* we have received an L_Q, we have sent one or more L_Qs, or there is no
* L_Q context associated with this REQ (REQ occurs immediately after a
* (re)selection). Routines that know that the context responsible for this
* nonpackreq call directly into unexpected_nonpkt_phase. In the case of the
* top level idle loop, we exhaust all active contexts prior to determining that
* we simply do not have the full I_T_L_Q for this phase.
*/
unexpected_nonpkt_phase_find_ctxt:
/*
* This nonpackreq is most likely associated with one of the tags
* in a FIFO or an outgoing LQ. Only treat it as an I_T only
* nonpackreq if we've cleared out the FIFOs and handled any
* pending SELDO.
*/
SET_SRC_MODE M_SCSI;
SET_DST_MODE M_SCSI;
and A, FIFO1FREE|FIFO0FREE, DFFSTAT;
cmp A, FIFO1FREE|FIFO0FREE jne return;
test SSTAT0, SELDO jnz return;
mvi SCBPTR[1], SCB_LIST_NULL;
unexpected_nonpkt_phase:
test MODE_PTR, ~(MK_MODE(M_DFF1, M_DFF1))
jnz unexpected_nonpkt_mode_cleared;
SET_SRC_MODE M_DFF0;
SET_DST_MODE M_DFF0;
or LONGJMP_ADDR[1], INVALID_ADDR;
Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-05-04 00:20:07 +00:00
dec SCB_FIFO_USE_COUNT;
mvi DFFSXFRCTL, CLRCHN;
unexpected_nonpkt_mode_cleared:
mvi CLRSINT2, CLRNONPACKREQ;
Correct a very rare case where command ordering could be compromised by a transaction performing a driver handled message sequence (an scb with the MK_MESSAGE flag set). SCBs that perform host managed messaging must always be at the head of their per-target selection queue so that the firmware knows to manually assert ATN if the current negotiation agreement is packetized. In the past we guaranteed this by queuing these SCBs separarately in the execution queue. This exposes the system to potential command reordering in two cases: 1) Another SCB for the same ITL nexus is queued that does not have the MK_MESSAGE flag set. This SCB will be queued to the per-target list which can be serviced before the MK_MESSAGE scb that preceeded it. 2) If the target cannot accept all of the commands in the per-target selection queue in one selection, the remainder is queued to the tail of the selection queues so as to effect round-robin scheduling. This could allow the MK_MESSAGE scb to be sent to the target before the requeued commands. This commit changes the firmware policy to defer queuing MK_MESSAGE SCBs into the selection queues until this can be done without affecting order. This means that the target's selection queue is either empty, or the last SCB on the execution queue is also a MK_MESSAGE SCB. During any wait, the firmware halts the download of new SCBs so only a single "holding location" is required. Luckily, MK_MESSAGE SCBs are rare and typically occur only during CAM's bus probe where only one command is outstanding at a time. However, during some recovery scenarios, the reordering *could* occur. aic79xx.c: Update ahd_search_qinfifo() and helper routines to search for pending MK_MESSAGE scbs and properly restitch the execution queue if either the MK_MESSAGE SCB is being aborted, or the MK_MESSAGE SCB can be queued due to the execution queue draining due to aborts. Enable LQOBUSFREE status to assert an interrupt. This should be redundant since a BUSFREE interrupt should always occur along with an LQOBUSFREE event, but on the Rev A, this doesn't seem to be guaranteed. When a PPR request is rejected when a previously existing packetized agreement is in place, assume that the target has been reset without our knowledge and revert to async/narrow transfers. This corrects two issues: the stale ENATNO setting that was used to send the PPR is cleared so the firmware is not confused by a future packetized selection with ATN asserted but no MK_MESSAGE flag in the SCB and it speeds up recovery by aborting any pending packetized transactions that by definition are now dead. When re-queueing SCBs after a failed negotiation attempt, ensure command ordering by freezing the device queue first. Traverse the list of pending SCBs rather than the whole SCB array on the controller when pushing MK_MESSAGE flag changes out to the controller. The original code was optimized for the aic7xxx controllers where there are fewer controller slots then pending SCBs and the firmware picks SCB slots. For the U320 controller, the hope is that we have fewer pending SCBs then the 512 slots on the controller. Enhance some diagnostics. Factor out some common code. aic79xx.h: Add prototype for new ahd_done_with_status() that is used to factor out some commone code. aic79xx.reg: Add definisions for the pending MK_MESSAGE SCB. aic79xx.seq: Defer MK_MESSAGE SCB queing to the execution queue so as to preserve command ordering. Re-arrange some of the selection processing code so the above change had no performance impact on the common code path. Close a few critical section holes. When entering a non-packetized phase, manually enable busfree interrupts, since the controller hardware does not do this automatically. aic79xx_inline.h: Enhance logging for queued SCBs. aic79xx_osm.c: Add new a new DDB ahd command, ahd_dump, which invokes the ahd_dump_card_state() routine on the unit specified with the ahd_sunit DDB command. aic79xx_pci.c: Turn on the BUSFREEREV bug for the Rev B. controller. This is required to close the busfree during non-packetized phase hole.
2004-08-04 17:55:34 +00:00
if ((ahd->bugs & AHD_BUSFREEREV_BUG) != 0) {
/*
* Test to ensure that the bus has not
* already gone free prior to clearing
* any stale busfree status. This avoids
* a window whereby a busfree just after
* a selection could be missed.
*/
test SCSISIGI, BSYI jz . + 2;
mvi CLRSINT1,CLRBUSFREE;
or SIMODE1, ENBUSFREE;
}
test SCSIPHASE, ~(MSG_IN_PHASE|MSG_OUT_PHASE) jnz illegal_phase;
SET_SEQINTCODE(ENTERING_NONPACK)
jmp ITloop;
illegal_phase:
SET_SEQINTCODE(ILLEGAL_PHASE)
jmp ITloop;
/*
* We have entered an overrun situation. If we have working
* BITBUCKET, flip that on and let the hardware eat any overrun
* data. Otherwise use an overrun buffer in the host to simulate
* BITBUCKET.
*/
Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-05-04 00:20:07 +00:00
pkt_handle_overrun_inc_use_count:
inc SCB_FIFO_USE_COUNT;
pkt_handle_overrun:
SET_SEQINTCODE(CFG4OVERRUN)
call freeze_queue;
if ((ahd->bugs & AHD_PKT_BITBUCKET_BUG) == 0) {
or DFFSXFRCTL, DFFBITBUCKET;
SET_SRC_MODE M_DFF1;
SET_DST_MODE M_DFF1;
} else {
call load_overrun_buf;
mvi DFCNTRL, (HDMAEN|SCSIEN|PRELOADEN);
}
call setjmp;
if ((ahd->bugs & AHD_PKT_BITBUCKET_BUG) != 0) {
test DFSTATUS, PRELOAD_AVAIL jz overrun_load_done;
call load_overrun_buf;
or DFCNTRL, PRELOADEN;
overrun_load_done:
test SEQINTSRC, CTXTDONE jnz pkt_overrun_end;
} else {
test DFFSXFRCTL, DFFBITBUCKET jz pkt_overrun_end;
}
test SSTAT2, NONPACKREQ jz return;
pkt_overrun_end:
or SCB_RESIDUAL_SGPTR, SG_OVERRUN_RESID;
test SEQINTSRC, CTXTDONE jz unexpected_nonpkt_phase;
Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-05-04 00:20:07 +00:00
dec SCB_FIFO_USE_COUNT;
aic79xx.c: Clear the LQICRC_NLQ status should it pop up after we have already handled the SCSIPERR. During some streaming operations this status can be delayed until the stream ends. Without this change, the driver would complain about a "Missing case in ahd_handle_scsiint". In the LQOBUSFREE handler... Don't return the LQOMGR back to the idle state until after we have cleaned up ENSELO and any status related to this selection. The last thing we need is the LQO manager starting another select-out before we have updated the execution queue. It is not clear whether the LQOMGR would, or would not start a new selection early. Make sure ENSELO is off prior to clearing SELDO by flushing device writes. Move assignment of the next target SCB pointer inside of an if to make the code clearer. The effect is the same. Dump card state in both "Unexpected PKT busfree" paths. In ahd_reset(), set the chip to SCSI mode before reading SXFRCTL1. That register only exists in the SCSI mode. Also set the mode explicitly to the SCSI mode after chip reset due to paranoia. Re-arrange code so that SXFRCTL1 is restored as quickly after the chip reset as possible. S/G structurs must be 8byte aligned. Make this official by saying so in our DMA tag. Disable CIO bus stretch on MDFFSTAT if SHVALID is about to come true. This can cause a CIO bus lockup if a PCI or PCI-X error occurs while the stretch is occurring - the host cannot service the PCI-X error since the CIO bus is locked out and SHVALID will never resolve. The stretch was added in the Rev B to simplify the wait for SHVALID to resolve, but the code to do this in the open source sequencer is so simple it was never removed. Consistently use MAX_OFFSET for the user max syncrate set from non-volatile storage. This ensures that the offset does not conflict with AH?_OFFSET_UNKNOWN. Have ahd_pause_and_flushwork set the mode to ensure that it has access to the registers it checks. Also modify the checking of intstat so that the check against 0xFF can actually succeed if the INT_PEND mask is something other than 0xFF. Although there are no cardbus U320 controllers, this check may be needed to recover from a hot-plug PCI removal that occurs without informing the driver. Fix a typo. sg_prefetch_cnt -> sg_prefetch_align. This fixes an infinite loop at card initialization if the cacheline size is 0. aic79xx.h: Add AHD_EARLY_REQ_BUG bug flag. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. Change AH?_OFFSET_UNKNOWN to 0xFF. This is a value that the curr->offset can never be, unlike '0' which we previously used. This fixes code that only checks for a non-zero offset to determine if a sync negotiation is required since it will fire in the unknown case even if the goal is async. aic79xx.reg: Add comments for LQISTAT bits indicating their names in the 7902 data book. We use slightly different and more descriptive names in the firmware. Fix spelling errors. Include the CDB's length just after the CDB pointer in the DMA'ed CDB case. aic79xx.seq: Update comments regarding rundown of the GSFIFO to reflect reality. Fix spelling errors. Since we use an 8byte address and 1 byte length, shorten the size of a block move for the legacy DMA'ed CDB case from 11 to 9 bytes. Remove code that, assuming the abort pending feature worked, would set MK_MESSAGE in the SCB's control byte on completion to catch invalid reselections. Since we don't see interrupts for completed selections, this status update could occur prior to us noticing the SELDO. The "select-out" queue logic will get confused by the MK_MESSAGE bit being set as this is used to catch packatized connections where we select-out with ATN. Since the abort pending feature doesn't work on any released controllers yet, this code was never executed. Add support for the AHD_EARLY_REQ_BUG. Don't ignore persistent REQ assertions just because they were asserted within the bus settle delay window. This allows us to tolerate devices like the GEM318 that violate the SCSI spec. Remove unintentional settnig of SG_CACHE_AVAIL. Writing this bit should have no effect, but who knows... On the Rev A, we must wait for HDMAENACK before loading additional segments to avoid clobbering the address of the first segment in the S/G FIFO. This resolves data-corruption issues with certain IBM (now Hitachi) and Fujitsu U320 drives. Rearrange calc_residual to avoid an extra jmp instruction. On RevA Silicon, if the target returns us to data-out after we have already trained for data-out, it is possible for us to transition the free running clock to data-valid before the required 100ns P1 setup time (8 P1 assertions in fast-160 mode). This will only happen if this L-Q is a continuation of a data transfer for which we have already prefetched data into our FIFO (LQ/Data followed by LQ/Data for the same write transaction). This can cause some target implementations to miss the first few data transfers on the bus. We detect this situation by noticing that this is the first data transfer after an LQ (LQIWORKONLQ true), that the data transfer is a continuation of a transfer already setup in our FIFO (SAVEPTRS interrupt), and that the transaction is a write (DIRECTION set in DFCNTRL). The delay is performed by disabling SCSIEN until we see the first REQ from the target. Only compile in snapshot savepointers handler for RevA silicon where it is enabled. Handle the cfg4icmd packetized interrupt. We just need to load the address and count, start the DMA, and CLRCHN once the transfer is complete. Fix an oversight in the overrun handler for packetized status operations. We need to wait for either CTXTDONE or an overrun when checking for an overrun. The previous code did not wait and thus could decide that no overrun had occurred even though an overrun will occur on the next data-valid req. Add some comment to this section for clarity. Use LAST_SEG_DONE instead of LASTSDONE for testing transfer completion in the packetized status case. LASTSDONE may come up more quickly since it only records completion on the SCSI side, but since LAST_SEG_DONE is used everywhere else (and needs to be), this is less confusing. Add a missing invalidation of the longjmp address in the non-pack handler. This code needs additional review. aic79xx_inline.h: Fix spelling error. aic79xx_osm.c: Set the cdb length for CDBs dma'ed from host memory. Add a comment indicating that, should CAM start supporting cdbs larger than 16bytes, the driver could store the CDB in the status buffer. aic79xx_pci.c: Add a table entry for the 39320A. Added a missing comma to an error string table. Fix spelling errors.
2003-02-27 23:23:16 +00:00
or LONGJMP_ADDR[1], INVALID_ADDR;
Correct spelling errors. Switch to handling bad SCSI status as a sequencer interrupt instead of having the kernel proccess these failures via the completion queue. This is done because: o The old scheme required us to pause the sequencer and clear critical sections for each SCB. It seems that these pause actions, if coincident with a sequencer FIFO interrupt, would result in a FIFO interrupt getting lost or directing to the wrong FIFO. This caused hangs when the driver was stressed under high "queue full" loads. o The completion code assumed that it was always called with the sequencer running. This may not be the case in timeout processing where completions occur manually via ahd_pause_and_flushwork(). o With this scheme, the extra expense of clearing critical sections is avoided since the sequencer will only self pause once all pending selections have cleared and it is not in a critical section. aic79xx.c Add code to handle the new BAD_SCB_STATUS sequencer interrupt code. This just redirects the SCB through the already existing ahd_complete_scb() code path. Remove code in ahd_handle_scsi_status() that paused the sequencer, made sure that no selections where pending, and cleared critical sections. Bad status SCBs are now only processed when all of these conditions are true. aic79xx.reg: Add the BAD_SCB_STATUS sequencer interrupt code. aic79xx.seq: When completing an SCB upload to the host, if we are doing this because the SCB contains non-zero SCSI status, defer completing the SCB until there are no pending selection events. When completing these SCBs, use the new BAD_SCB_STATUS sequencer interrupt. For all other uploaded SCBs (currently only for underruns), the SCB is completed via the normal done queue. Additionally, keep the SCB that is currently being uploaded on the COMPLETE_DMA_SCB list until the dma is completed, not just until the DMA is started. This ensures that the DMA is restarted properly should the host disable the DMA transfer for some reason. In our RevA workaround for Maxtor drives, guard against the host pausing us while trying to pause I/O until the first data-valid REQ by clearing the current snapshot so that we can tell if the transfer has completed prior to us noticing the REQINIT status. In cfg4data_intr, shave off an instruction before getting the data path running by adding an entrypoint to the overrun handler to also increment the FIFO use count. In the overrun handler, be sure to clear our LONGJMP address in both exit paths. Perform a few sequencer optimizations. aic79xx.c: Print the full path from the SCB when a packetized status overrun occurs. Remove references to LONGJMP_SCB which is being removed from firmware usage. Print the new SCB_FIFO_USE_COUNT field in the per-SCB section of ahd_dump_card_state(). The SCB_TAG field is now re-used by the sequencer, so it no longer makes sense to reference this field in the kernel driver. aic79xx.h: Re-arrange fields in the hardware SCB from largest size type to smallest. This makes it easier to move fields without changing field alignment. The hardware scb tag field is now down near the "spare" portion of the SCB to facilitate reuse by the sequencer. aic79xx.reg: Remove LONGJMP_ADDR. Rearrange SCB fields to match aic79xx.h. Add SCB_FIFO_USE_COUNT as the first byte of the SCB_TAG field. aic79xx.seq: Add a per-SCB "Fifos in use count" field and use it to determine when it is safe (all data posted) to deliver status back to the host. The old method involved polling one or both FIFOs to verify that the current task did not have pending data. This makes running down the GSFIFO very cheap, so we will empty the GSFIFO in one idle loop pass in all cases. Use this simplification of the completion process to prune down the data FIFO teardown sequencer for packetized transfers. Much more code is now shared between the data residual and transfer complete cases. Correct some issues in the packetized status handler. It used to be possible to CLRCHN our FIFO before status had fully transferred to the host. We also failed to handle NONPACKREQ phases that could occur should a CRC error occur during transmission of the status data packet. Correct a few big endian issues: aic79xx.c: aic79xx_inline.h: aic79xx_pci.c: aic79xx_osm.c: o Always get the SCB's tag via the SCB_GET_TAG acccessor o Add missing use of byte swapping macros when touching hscb fields. o Don't double swap SEEPROM data when it is printed. Correct a big-endian bug. We cannot assign a o When assigning a 32bit LE variable to a 64bit LE variable, we must be explict about how the words of the 64bit LE variable are initialized. Cast to (uint32_t*) to do this. aic79xx.c: In ahd_clear_critical_section(), hit CRLSCSIINT after restoring the interrupt masks to avoid what appears to be a glitch on SCSIINT. Any real SCSIINT status will be persistent and will immidiately reset SCSIINT. This clear should only get rid of spurious SCSIINTs. This glitch was the cause of the "Unexpected PKT busfree" status that occurred under high queue full loads Call ahd_fini_scbdata() after shutdown so that any ahd_chip_init() routine that might access SCB data will not access free'd memory. Reset the bus on an IOERR since the chip doesn't seem to reset to the new voltage level without this. Change offset calculation for scatter gather maps so that the calculation is correct if an integral multiple of sg lists does not fit in the allocation size. Adjust bus dma tag for data buffers based on 39BIT addressing flag in our softc. Use the QFREEZE count to simplify ahd_pause_and_flushworkd(). We can thus rely on the sequencer eventually clearing ENSELO. In ahd_abort_scbs(), fix a bug that could potentially corrupt sequencer state. The saved SCB was being restored in the SCSI mode instead of the saved mode. It turns out that the SCB did not need to be saved at all as the scbptr is already restored by all subroutines called during this function that modify that register. aic79xx.c: aic79xx.h: aic79xx_pci.c: Add support for parsing the seeprom vital product data. The VPD data are currently unused. aic79xx.h: aic79xx.seq: aic79xx_pci.c: Add a firmware workaround to make the LED blink brighter during packetized operations on the H2A. aic79xx_inline.h: The host does not use timer interrupts, so don't gate our decision on whether or not to unpause the sequencer on whether or not a timer interrupt is pending.
2003-05-04 00:20:07 +00:00
test SCB_CONTROL, STATUS_RCVD jnz pkt_complete_scb_if_fifos_idle;
mvi DFFSXFRCTL, CLRCHN ret;
if ((ahd->bugs & AHD_PKT_BITBUCKET_BUG) != 0) {
load_overrun_buf:
/*
* Load a dummy segment if preload space is available.
*/
mov HADDR[0], SHARED_DATA_ADDR;
add HADDR[1], PKT_OVERRUN_BUFOFFSET, SHARED_DATA_ADDR[1];
mov ACCUM_SAVE, A;
clr A;
adc HADDR[2], A, SHARED_DATA_ADDR[2];
adc HADDR[3], A, SHARED_DATA_ADDR[3];
mov A, ACCUM_SAVE;
bmov HADDR[4], ALLZEROS, 4;
/* PKT_OVERRUN_BUFSIZE is a multiple of 256 */
clr HCNT[0];
mvi HCNT[1], ((PKT_OVERRUN_BUFSIZE >> 8) & 0xFF);
aic79xx.c: Implement the SCB_SILENT flag. This is useful for hushing up the driver during DV or other operations that we expect to cause transmission errors. The messages will still print if the SHOW_MASKED_ERRORS debug option is enabled. Save and restore the NEGOADDR address when setting new transfer settings. The sequencer performs lookups in the negotiation table too and it expects NEGOADDR to remain consistent across pause/unpause sessions. Consistently use "offset" instead of "period" to determine if we are running sync or not. Add a SHOW_MESSAGES diagnostic for when we assert ATN during message processing. Print out IU, QAS, and RTI features when showing transfer options. Limit the syncrate after all option conformance changes have taken place in ahd_devlimited_syncrate. Changes in options may change the final syncrate we accept. Keep a copy of the hs_mailbox in our softc so that we can perform read/modify/write operations on the hs_mailbox without having to pause the sequencer to read the last written value. Use the ENINT_COALESS flag in the hs_mailbox to toggle interrupt coalessing. Add entrypoints for enabling interrupt coalessing and setting both a timeout (how long to wait for commands to be coalessed) and a maximum commands to coaless value. Add a statistics timer that decides when to enable or disable interrupt coalessing based on load. Add a routine, ahd_reset_cmds_pending() which is used to update the CMDS_PENDING sequencer variable whenever error recovery compeltes SCBs without notifying the sequencer. Since ahd_reset_cmds_pending is called during ahd_unpause() only if we've aborted SCBs, its call to ahd_flush_qoutfifo should not cause recursion through ahd_run_qoutfifo(). A panic has been added to ensure that this recursion does not occur. In ahd_search_qinfifo, update the CMDS_PENDING sequencer variable directly. ahd_search_qinififo can be called in situations where using ahd_reset_cmds_pending() might cause recursion. Since we can safely determine the exact number to reduce CMDS_PENDING by in this scenario without running the qoutfifo, the manual update is sufficient. Clean up diagnostics. Add ahd_flush_qoutfifo() which will run the qoutfifo as well as complete any commands sitting on the sequencer's COMPLETE_SCB lists or the good status FIFO. Use this routine in several places that did similar things in an add-hoc, but incomplete, fashion. A call to this routine was also added to ahd_abort_scbs() to close a race. In ahd_pause_and_flushwork() only return once selections are safely disabled. Flush all completed commands via ahd_flush_qoutfifo(). Remove "Now packetized" diagnostic now that this information is incorperated into the actual negotiation messages that are displayed. When forcing renegotiation, don't clober the current ppr_options. Much of the driver uses this information to determine if we are currently packetized or not. Remove some stray spaces at column 1 in ahd_set_tags. When complaining about getting a host message loop request with no pending messages, print out the SCB_CONTROL register down on the card. Modify the ahd_sent_msg() routine to handle a search for an outgoing identify message. Use this to detect a msg reject on an identify message which typically indicates that the target thought we were packetized. Force a renegotiation in this case. In ahd_search_qinfifo(), wait more effectively for SCB DMA activities to cease. We also disable SCB fetch operations since we are about to change the qinfifo and any fetch in progress will likely be invalidated. In ahd_qinfifo_count(), fix the qinfifo empty case. In ahd_dump_card_state(), print out CCSCBCTL in the correct mode. If we are a narrow controller, don't set the current width to unknown when forcing a future negotiation. This just confuses the code into attempting a wide negotiation on a narrow bus. Add support for task management function completions. Modify ahd_handle_devreset so that it can handle lun resets in addition to target resets. Use ahd_handle_devreset for lun and target reset task management functions. Handle the abort task TMF race case better. We now wait until any current selections are over and then set the TMF back to zero. This should cause the sequencer to ignore the abort TMF completion should it occur. Correct a bug in the illegal phase handler that caused us to drop down to narrow when handling the unexpected command phase case after 3rd party reset of a packetized device. Indicate the features, bugs, and flags set in the softc that are used to control firmware patch download when booting verbose. aic79xx.h: Add coalessing and HS_MAILBOX fields. Add per-softc variables for the stats "daemon". Add a debug option for interrupt coalessing activities. Add two new softc flags: o AHD_UPDATE_PEND_CMDS Run ahd_reset_cmds_pending() on the next unpause. o AHD_RUNNING_QOUTFIFO Used to catch recursion through ahd_run_qoutfifo(). aic79xx.reg: Correct register addresses related to the software timer and the DFDBCTL register. Add constants paramaterizing the software timer. Add scratch ram locations for storing interrupt coalessing tunables. Break INTMASK in SEQITNCTL out into INTMASK1 and INTMASK2. In at least the REV A, these are writable bits. We make use of that for a swtimer workaround in the sequencer. Since HS_MAILBOX autoclears, provide a sequencer variable to store its contents. Add SEQINT codes for handling task management completions. aic79xx.seq: Correct ignore wide residue processing check for a wide negotiation being in effect. We must be in the SCSI register window in order to access the negotiation table. Use the software timer and a commands completed count to implement interrupt coalessing. The command complete is deferred until either the maximum command threshold or a the expiration of a command deferral timer. If we have more SCBs to complete to the host (sitting in COMPLETE_SCB lists), always try to coaless them up to our coalessing limit. If coalessing is enabled, but we have fewer commands oustanting than the host's min coalessing limit, complete the command immediately. Add code to track the number of commands outstanding. Commands are outstanding from the time they are placed into the execution queue until the DMA to post completion is setup. Add a workaround for intvec_2 interrupts on the H2A4. In H2A4, the mode pointer is not saved for intvec2, but is restored on iret. This can lead to the restoration of a bogus mode ptr. Manually clear the intmask bits and do a normal return to compensate. We use intvec_2 to track interrupt coalessing timeouts. Since we cannot disable the swtimer's countdown, simply mask its interrupt once we no longer care about it firing. In idle_loop_cchan, update LOCAL_HS_MAILBOX everytime we are notified of an HS_MAILBOX update via the HS_MAILBOX_ACT bit in QOFF_CTLSTA. We have to use a local copy of persistant portions of the HS_MAILBOX as the mailbox auto-clears on any read. Move the test for the cfg4istat interrupt up an instruction to hopefully close a race between the next outgoing selection and our disabling of selections. Add a missing ret to the last instruction in load_overrun_buf. Add notifications to the host of task management completions as well as the completions for commands that completed successfully before their corresponding TMF could be sent. Hold a critical section during select-out processing until we have a fully identified connection. This removes a race condition with the legacy abort handler. Correct a few spelling errors in some comments. aic79xx_inline.h: Call ahd_reset_cmds_pending() in ahd_unpause if required. Update cmdcmplt interrupt statistics in our interrupt handler. Allow callers to ahd_send_scb() to set the task management function. aic79xx_pci.c: Disable SERR and pause the controller prior to performing our mmapped I/O test. The U320 controllers do not support "auto-access-pause". aic79xx_osm.c: Set the task management function now that ahd_send_scb() doesn't do it for us. We also perform a lun reset in response to BDR requests to packetized devices.
2003-01-20 20:17:35 +00:00
clr HCNT[2] ret;
}