freebsd-skq/sys/dev/aic7xxx/aic7xxx.seq

2175 lines
62 KiB
Plaintext
Raw Normal View History

/*
* Adaptec 274x/284x/294x device driver firmware for Linux and FreeBSD.
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
*
* Copyright (c) 1994-2001 Justin Gibbs.
* All rights reserved.
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
*
* 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. The name of the author may not be used to endorse or promote products
* derived from this software without specific prior written permission.
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
*
* Alternatively, this software may be distributed under the terms of the
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
* GNU Public License ("GPL").
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE FOR
* ANY DIRECT, INDIRECT, INCIDENTAL, 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 DAMAGE.
*
aic7xxx.c: Style nits. Make sure that our selection hardware is disabled as soon as possible after detecting a busfree and even go so far as to disable the selection hardware in advance of an event that will cause a busfree (ABORT or BUS DEVICE RESET message). The concern is that the selection hardware will select a target for which, after processing the bus free, there will be no commands pending. The sequencer idle loop will re-enable the selection should it still be necessary. In ahc_handle_scsiint(), clear SSTAT0 events several PCI transactions (most notably reads) prior to clearing SCSIINT. The newer chips seem to take a bit of time to see the change which can make the clearing of SCSIINT ineffective. Don't bother panicing at the end of ahc_handle_scsiint(). Getting to the final else just means we lost the race with clearing SCSIINT. In ahc_free(), handle init-level 0. This can happen when we fail the attach for RAID devices. While I'm here, also kill the parent dma tag. In ahc_match_scb(), consider initiator ccbs to be any that are not from the target mode group. This fixes a bug where an external target reset CCB was not getting cleaned up by the reset code. Don't bother freezing a ccb in any of our "abort" routines when the status is set to CAM_REQ_CMP. This can happen for a target reset ccb. aic7xxx.reg: Reserve space for a completion queue. This will be used to enhance performance in the near future. aic7xxx.seq: Remove an optimization for the 7890 autoflush bug that turned out to allow, in rare cases, some data to get lost. Implement a simpler, faster, fix for the PCI_2_1 retry bug that hangs the sequencer on an SCB dma for certain chips. Test against SAVED_SCSIID rather than SELID during target reselections. This is how we always did it in the past, but the code was modified while trying to work around an issue with the 7895. SAVED_SCSIID takes into account twin channel adapters such as the 2742T, whereas SELID does not have the channel bit. This caused invalid selection warnings and other strangeness on these cards. aic7xxx_pci.c Use the correct mask for checking the generic aic7892 entry.
2001-02-10 18:04:27 +00:00
* $Id: //depot/src/aic7xxx/aic7xxx.seq#19 $
*
1999-08-28 01:08:13 +00:00
* $FreeBSD$
*/
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
#include "aic7xxx.reg"
#include "scsi_message.h"
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
/*
* A few words on the waiting SCB list:
* After starting the selection hardware, we check for reconnecting targets
* as well as for our selection to complete just in case the reselection wins
* bus arbitration. The problem with this is that we must keep track of the
* SCB that we've already pulled from the QINFIFO and started the selection
* on just in case the reselection wins so that we can retry the selection at
* a later time. This problem cannot be resolved by holding a single entry
* in scratch ram since a reconnecting target can request sense and this will
* create yet another SCB waiting for selection. The solution used here is to
* use byte 27 of the SCB as a psuedo-next pointer and to thread a list
* of SCBs that are awaiting selection. Since 0-0xfe are valid SCB indexes,
* SCB_LIST_NULL is 0xff which is out of range. An entry is also added to
* this list everytime a request sense occurs or after completing a non-tagged
* command for which a second SCB has been queued. The sequencer will
* automatically consume the entries.
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
*/
ahc_eisa.c: Initialize rid to 0. This doesn't seem to make any difference (the driver doesn't care what rid it gets and no-one seems to check rid's value), but follows standard conventions. Pass in our device_t to ahc_alloc(). We now use device_T softc storage, so passing NULL results in a panic. Set the unit number in our softc so that the driver core can retrieve it. ahc_pci.c: Set the unit number in our softc so that the driver core can retrieve it. aic7770.c: Insert our softc into the list of softcs when initialization is successful. aic7xxx.c: Remove a workaround for an aic7895 bug we will never trigger. Add additional diagnostic info to ahc_dump_card_state(). Always panic the system if a sequencer assertion fails. AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. Replace a hard coded number with a constant. Guard against looping forever in ahc_pause_and_flushwork(). A hot eject or card failure may make the intstat register return 0xFF, so limit the number of interrupts we'll process. Correct the code in ahc_search_qinfifo() that guarantees that the sequencer will see an abort collision if the qinfifo is modified when a DMA is in progress. We now do this fixup after modifying the queue. This guarantees that the HSCB we place at the head of the queue is not the same as the old head. Using "next hscb" (guaranteed not to be the same as the first SCB) before clearing the queue could free up the original head hscb to be used during a remove operation placing it again at the head of the qinfifo. aic7xxx.h: Reduce the maximum number of outstanding commands to 253 from 254. To handle our output queue correctly on machines that only support 32bit stores, we must clear the array 4 bytes at a time. To avoid colliding with a DMA write from the sequencer, we must be sure that 4 slots are empty when we write to clear the queue. This reduces us to 253 SCBs: 1 that just completed and the known three additional empty slots in the queue that preceed it. Yahoo was able to force this race on one of their systems. Interrupts were disabled for such a time that the entire output queue was filled (254 entries complete without any processing), and our 32bit write to clear the status clobbered one entry. Add a feature tag for devices that are removable. aic7xxx.reg: Never use the sequencer interrupt value of 0xF0. We need to guanrantee that an INTSTAT value of 0xFF can only occur during card failure or a hot-eject. Align the busy targets table with the begining of scratch space. This seems to appease a chip bug in the aic7895. aic7xxx.seq: Be sure to disable select-out after a bus free event that occurs early in a selection. If we don't disable select-out, we will believe that it is enabled even though a new selection will never occur. Move the clearing of SELDI to just before a jump. This appeases another chip bug of the aic7895. Make the target mode command loop a bit more efficient. AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. Properly cleanup the last SCB we tested against should we fail to properly find an SCB for a reselection. Add some additional sequencer debugging code. aic7xxx_freebsd.c: Limit the driver to 253 outstanding commands per adapter. Guard against overflow in timeout handling. aic7xxx_inline.h: AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. aic7xxx_pci.c: Set the removable feature for the apa1480 cardbus and the 29160C Compact PCI card. Don't report high byte termination information for narrow cards. Use a PCI read rather than a questionable delay when fetching/setting termination settings.
2001-01-22 21:03:48 +00:00
bus_free_sel:
/*
* Turn off the selection hardware. We need to reset the
* selection request in order to perform a new selection.
*/
and SCSISEQ, TEMODE|ENSELI|ENRSELI|ENAUTOATNP, SCSISEQ;
and SIMODE1, ~ENBUSFREE;
1995-04-27 17:44:27 +00:00
poll_for_work:
call clear_target_state;
and SXFRCTL0, ~SPIOEN;
if ((ahc->features & AHC_ULTRA2) != 0) {
clr SCSIBUSL;
}
poll_for_work_loop:
test SSTAT0, SELDO|SELDI jnz selection;
o Correct the offsets into the syncrate table for paritcular negotiation features (DT, ULTRA2, ULTRA, FAST). The offsets where not properly updated when the DT entry was added and so the driver could attempt to negotiate a speed faster than that supported by the target device or even requested by the user via SCSI-Select settings. * o Update the target mode incoming command queue kernel index value ever 128 commands instead of 32. This means that the kernel will always try to keep its index (as seen on the card - the kernel may actually have cleared more space) 128 commands ahead of where the sequencer is adding entries. o Use the HS_MAILBOX register instead of the KERNEL_TQINPOS location in SRAM to indicate the kernel's target queue possition on Ultra2 cards. This avoids the "pause bug" on these cards and also turns out to be much more efficient. o When enabling or disabling a particular target id for target mode, make sure that the taret id in the SCSIID register does not reference an ID that is not to receive target selections. This is only an issue on chips that support the multiple target id feature where the value in SCSIID will still affect selection behavior regardless of the values in the target id bit field registers. o Remove some target mode debugging printfs. o Make sure that the sense length reported in ATIO commands is always zero. This driver does not, yet, report HBA generated sense information for accepted commands. o Honor the CAM_TIME_INFINITY and CAM_TIME_DEFAULT values for the CCB timeout field. o Make the driver compile with AHC_DEBUG again. * Noticed by: Andrew Gallatin<gallatin@cs.duke.edu>
2000-03-18 22:28:20 +00:00
test SCSISEQ, ENSELO jnz poll_for_work_loop;
if ((ahc->features & AHC_TWIN) != 0) {
/*
* Twin channel devices cannot handle things like SELTO
* interrupts on the "background" channel. So, if we
* are selecting, keep polling the current channel util
* either a selection or reselection occurs.
*/
xor SBLKCTL,SELBUSB; /* Toggle to the other bus */
test SSTAT0, SELDO|SELDI jnz selection;
xor SBLKCTL,SELBUSB; /* Toggle back */
}
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
cmp WAITING_SCBH,SCB_LIST_NULL jne start_waiting;
test_queue:
/* Has the driver posted any work for us? */
BEGIN_CRITICAL
if ((ahc->features & AHC_QUEUE_REGS) != 0) {
test QOFF_CTLSTA, SCB_AVAIL jz poll_for_work_loop;
} else {
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
mov A, QINPOS;
cmp KERNEL_QINPOS, A je poll_for_work_loop;
}
mov ARG_1, NEXT_QUEUED_SCB;
END_CRITICAL
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
/*
* We have at least one queued SCB now and we don't have any
* SCBs in the list of SCBs awaiting selection. Allocate a
* card SCB for the host's SCB and get to work on it.
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
*/
if ((ahc->flags & AHC_PAGESCBS) != 0) {
mov ALLZEROS call get_free_or_disc_scb;
} else {
/* In the non-paging case, the SCBID == hardware SCB index */
mov SCBPTR, ARG_1;
}
dma_queued_scb:
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
/*
* DMA the SCB from host ram into the current SCB location.
*/
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
mvi DMAPARAMS, HDMAEN|DIRECTION|FIFORESET;
mov ARG_1 call dma_scb;
/*
* Check one last time to see if this SCB was canceled
* before we completed the DMA operation. If it was,
* the QINFIFO next pointer will not match our saved
* value.
*/
mov A, ARG_1;
BEGIN_CRITICAL
cmp NEXT_QUEUED_SCB, A jne abort_qinscb;
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
if ((ahc->flags & AHC_SEQUENCER_DEBUG) != 0) {
cmp SCB_TAG, A je . + 2;
aic7xxx.c: When restarting the sequencer, ensure that the SCBCNT register is 0. A non-zero count will prevent the setting of the CCSCBDIR bit in any future dma operations. The only time CCSCBCNT would be non-zero is if we happened to halt the dma during a reset, but even that should never happen. Better safe than sorry. When a command completes before the target responds to an ATN for a recovery command, we now notify the kernel so that any recovery operation requeued in the qinfifo can be removed safely. In the past, we did this in ahc_done(), but ahc_done() may be called without the card paused. This also avoids a recursive call to ahc_search_qinifo() which could have occurred if ahc_search_qinififo() happened to be the routine to complete a recovery action. Fix 8bit math used for adjusting the qinfifo. The index must be wrapped properly within the 256 entry array. We rely on the fact that qinfifonext is a uint8_t in most cases to handle this wrap, but we missed a few spots where the resultant calculation was promoted to an int. Change the way that we deal with aborting the first or second entry from the qinfifo. We now swap the first entry in the qinfifo with the "next queued scb" to force the sequencer to see an abort collision if we ever touch the qinififo while the sequencer is mid SCB dma. aic7xxx.reg: Add new MKMSG_FAILED sequencer interrupt. This displaced the BOGUS_TAG interrupt used in some previous sequencer code debugging. aic7xxx.seq: Increment our position in the qinfifo only once the dma is complete and we have verified that the queue has not been changed during our DMA. This simplifies code in the kernel. Protect against "instruction creep" when issuing a pausing sequencer interrupt. On at least the 7890/91/96/97, the sequencer will coast after issuing the interrupt for up to two instructions. In the past we delt with this by using carefully placed nops. Now we call a routine to issue the interrupt followed by a nop and a ret. Tell the kernel should an SCB complete with the MK_MESSAGE flag still set. This means the target ignored our ATN request. Clear the channel twice as we exit the data phase. On the aic7890/91, the S/G preload logic may require the second clearing to get the last S/G out of the FIFO. aic7xxx_freebsd.c: Don't bother searching the qinfifo for a doubly queued recovery scb in ahc_done. This case is handled by the core driver now. Free the path used to issue async callbacks after the callback is complete. aic7xxx_inline.h: Split the SCB queue routine into a routine that swaps the SCB with the "next queued SCB" and a routine that calls the swapping routine and notifies the card of the new SCB. The swapping routine is now also used by ahc_search_qinfifo.
2000-11-06 20:05:38 +00:00
mvi SCB_MISMATCH call set_seqint;
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
}
mov NEXT_QUEUED_SCB, SCB_NEXT;
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
mov SCB_NEXT,WAITING_SCBH;
mov WAITING_SCBH, SCBPTR;
aic7xxx.c: When restarting the sequencer, ensure that the SCBCNT register is 0. A non-zero count will prevent the setting of the CCSCBDIR bit in any future dma operations. The only time CCSCBCNT would be non-zero is if we happened to halt the dma during a reset, but even that should never happen. Better safe than sorry. When a command completes before the target responds to an ATN for a recovery command, we now notify the kernel so that any recovery operation requeued in the qinfifo can be removed safely. In the past, we did this in ahc_done(), but ahc_done() may be called without the card paused. This also avoids a recursive call to ahc_search_qinifo() which could have occurred if ahc_search_qinififo() happened to be the routine to complete a recovery action. Fix 8bit math used for adjusting the qinfifo. The index must be wrapped properly within the 256 entry array. We rely on the fact that qinfifonext is a uint8_t in most cases to handle this wrap, but we missed a few spots where the resultant calculation was promoted to an int. Change the way that we deal with aborting the first or second entry from the qinfifo. We now swap the first entry in the qinfifo with the "next queued scb" to force the sequencer to see an abort collision if we ever touch the qinififo while the sequencer is mid SCB dma. aic7xxx.reg: Add new MKMSG_FAILED sequencer interrupt. This displaced the BOGUS_TAG interrupt used in some previous sequencer code debugging. aic7xxx.seq: Increment our position in the qinfifo only once the dma is complete and we have verified that the queue has not been changed during our DMA. This simplifies code in the kernel. Protect against "instruction creep" when issuing a pausing sequencer interrupt. On at least the 7890/91/96/97, the sequencer will coast after issuing the interrupt for up to two instructions. In the past we delt with this by using carefully placed nops. Now we call a routine to issue the interrupt followed by a nop and a ret. Tell the kernel should an SCB complete with the MK_MESSAGE flag still set. This means the target ignored our ATN request. Clear the channel twice as we exit the data phase. On the aic7890/91, the S/G preload logic may require the second clearing to get the last S/G out of the FIFO. aic7xxx_freebsd.c: Don't bother searching the qinfifo for a doubly queued recovery scb in ahc_done. This case is handled by the core driver now. Free the path used to issue async callbacks after the callback is complete. aic7xxx_inline.h: Split the SCB queue routine into a routine that swaps the SCB with the "next queued SCB" and a routine that calls the swapping routine and notifies the card of the new SCB. The swapping routine is now also used by ahc_search_qinfifo.
2000-11-06 20:05:38 +00:00
if ((ahc->features & AHC_QUEUE_REGS) != 0) {
mov NONE, SNSCB_QOFF;
} else {
inc QINPOS;
}
END_CRITICAL
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
start_waiting:
/*
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
* Start the first entry on the waiting SCB list.
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
*/
mov SCBPTR, WAITING_SCBH;
call start_selection;
jmp poll_for_work_loop;
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
abort_qinscb:
call add_scb_to_free_list;
jmp poll_for_work_loop;
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
start_selection:
Sync Perforce IDs, add tranceiver state change support, and correct numerous error recovery buglets. Many thanks to Tor Egge for his assistance in diagnosing problems with the error recovery code. aic7xxx.c: Report missed bus free events using their own sequencer interrupt code to avoid confusion with other "bad phase" interrupts. Remove a delay used in debugging. This delay could only be hit in certain, very extreme, error recovery scenarios. Handle transceiver state changes correctly. You can now plug an SE device into a hot-plug LVD bus without hanging the controller. When stepping through a critical section, panic if we step more than a reasonable number of times. After a bus reset, disable bus reset interupts until we either our first attempt to (re)select another device, or another device attemps to select us. This removes the need to busy wait in kernel for the scsi reset line to fall yet still ensures we see any reset events that impact the state of either our initiator or target roles. Before this change, we had the potential of servicing a "storm" of reset interrupts if the reset line was held for a significant amount of time. Indicate the current sequencer address whenever we dump the card's state. aic7xxx.reg: Transceiver state change register definitions. Add the missed bussfree sequencer interrupt code. Re-enable the scsi reset interrupt if it has been disabled before every attempt to (re)select a device and when we have been selected as a target. When being (re)selected, check to see if the selection dissappeared just after we enabled our bus free interrupt. If the bus has gone free again, go back to the idle loop and wait for another selection. Note two locations where we should change our behavior if ATN is still raised. If ATN is raised during the presentation of a command complete or disconnect message, we should ignore the message and expect the target to put us in msgout phase. We don't currently do this as it requires some code re-arrangement so that critical sections can be properly placed around our handling of these two events. Otherwise, we cannot guarantee that the check of ATN is atomic relative to our acking of the message in byte (the kernel could assert ATN). Only set the IDENTIFY_SEEN flag after we have settled on the SCB for this transaction. The kernel looks at this flag before assuming that SCB_TAG is valid. This avoids confusion during certain types of error recovery. Add a critical section around findSCB. We cannot allow the kernel to remove an entry from the disconnected list while we are traversing it. Ditto for get_free_or_disc_scb. aic7xxx_freebsd.c: Only assume that SCB_TAG is accurate if IDENTIFY_SEEN is set in SEQ_FLAGS. Fix a typo that caused us to execute some code for the non-SCB paging case when paging SCBs. This only occurred during error recovery.
2000-11-10 20:13:41 +00:00
/*
* If bus reset interrupts have been disabled (from a previous
* reset), re-enable them now. Resets are only of interest
* when we have outstanding transactions, so we can safely
* defer re-enabling the interrupt until, as an initiator,
* we start sending out transactions again.
*/
test SIMODE1, ENSCSIRST jnz . + 3;
mvi CLRSINT1, CLRSCSIRSTI;
or SIMODE1, ENSCSIRST;
if ((ahc->features & AHC_TWIN) != 0) {
and SINDEX,~SELBUSB,SBLKCTL;/* Clear channel select bit */
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
test SCB_SCSIID, TWIN_CHNLB jz . + 2;
or SINDEX, SELBUSB;
mov SBLKCTL,SINDEX; /* select channel */
}
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
initialize_scsiid:
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
if ((ahc->features & AHC_ULTRA2) != 0) {
mov SCSIID_ULTRA2, SCB_SCSIID;
} else if ((ahc->features & AHC_TWIN) != 0) {
and SCSIID, TWIN_TID|OID, SCB_SCSIID;
} else {
mov SCSIID, SCB_SCSIID;
}
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
if ((ahc->flags & AHC_TARGETROLE) != 0) {
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
mov SINDEX, SCSISEQ_TEMPLATE;
test SCB_CONTROL, TARGET_SCB jz . + 2;
or SINDEX, TEMODE;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
mov SCSISEQ, SINDEX ret;
} else {
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
mov SCSISEQ, SCSISEQ_TEMPLATE ret;
}
/*
* Initialize transfer settings and clear the SCSI channel.
* SINDEX should contain any additional bit's the client wants
* set in SXFRCTL0. We also assume that the current SCB is
* a valid SCB for the target we wish to talk to.
*/
initialize_channel:
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
or SXFRCTL0, SPIOEN|CLRSTCNT|CLRCHN;
set_transfer_settings:
if ((ahc->features & AHC_ULTRA) != 0) {
test SCB_CONTROL, ULTRAENB jz . + 2;
or SXFRCTL0, FAST20;
}
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
/*
* Initialize SCSIRATE with the appropriate value for this target.
*/
if ((ahc->features & AHC_ULTRA2) != 0) {
bmov SCSIRATE, SCB_SCSIRATE, 2 ret;
} else {
mov SCSIRATE, SCB_SCSIRATE ret;
}
selection:
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
/*
* We aren't expecting a bus free, so interrupt
* the kernel driver if it happens.
*/
mvi CLRSINT1,CLRBUSFREE;
or SIMODE1, ENBUSFREE;
Sync Perforce IDs, add tranceiver state change support, and correct numerous error recovery buglets. Many thanks to Tor Egge for his assistance in diagnosing problems with the error recovery code. aic7xxx.c: Report missed bus free events using their own sequencer interrupt code to avoid confusion with other "bad phase" interrupts. Remove a delay used in debugging. This delay could only be hit in certain, very extreme, error recovery scenarios. Handle transceiver state changes correctly. You can now plug an SE device into a hot-plug LVD bus without hanging the controller. When stepping through a critical section, panic if we step more than a reasonable number of times. After a bus reset, disable bus reset interupts until we either our first attempt to (re)select another device, or another device attemps to select us. This removes the need to busy wait in kernel for the scsi reset line to fall yet still ensures we see any reset events that impact the state of either our initiator or target roles. Before this change, we had the potential of servicing a "storm" of reset interrupts if the reset line was held for a significant amount of time. Indicate the current sequencer address whenever we dump the card's state. aic7xxx.reg: Transceiver state change register definitions. Add the missed bussfree sequencer interrupt code. Re-enable the scsi reset interrupt if it has been disabled before every attempt to (re)select a device and when we have been selected as a target. When being (re)selected, check to see if the selection dissappeared just after we enabled our bus free interrupt. If the bus has gone free again, go back to the idle loop and wait for another selection. Note two locations where we should change our behavior if ATN is still raised. If ATN is raised during the presentation of a command complete or disconnect message, we should ignore the message and expect the target to put us in msgout phase. We don't currently do this as it requires some code re-arrangement so that critical sections can be properly placed around our handling of these two events. Otherwise, we cannot guarantee that the check of ATN is atomic relative to our acking of the message in byte (the kernel could assert ATN). Only set the IDENTIFY_SEEN flag after we have settled on the SCB for this transaction. The kernel looks at this flag before assuming that SCB_TAG is valid. This avoids confusion during certain types of error recovery. Add a critical section around findSCB. We cannot allow the kernel to remove an entry from the disconnected list while we are traversing it. Ditto for get_free_or_disc_scb. aic7xxx_freebsd.c: Only assume that SCB_TAG is accurate if IDENTIFY_SEEN is set in SEQ_FLAGS. Fix a typo that caused us to execute some code for the non-SCB paging case when paging SCBs. This only occurred during error recovery.
2000-11-10 20:13:41 +00:00
/*
* Guard against a bus free after (re)selection
* but prior to enabling the busfree interrupt. SELDI
* and SELDO will be cleared in that case.
*/
ahc_eisa.c: Initialize rid to 0. This doesn't seem to make any difference (the driver doesn't care what rid it gets and no-one seems to check rid's value), but follows standard conventions. Pass in our device_t to ahc_alloc(). We now use device_T softc storage, so passing NULL results in a panic. Set the unit number in our softc so that the driver core can retrieve it. ahc_pci.c: Set the unit number in our softc so that the driver core can retrieve it. aic7770.c: Insert our softc into the list of softcs when initialization is successful. aic7xxx.c: Remove a workaround for an aic7895 bug we will never trigger. Add additional diagnostic info to ahc_dump_card_state(). Always panic the system if a sequencer assertion fails. AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. Replace a hard coded number with a constant. Guard against looping forever in ahc_pause_and_flushwork(). A hot eject or card failure may make the intstat register return 0xFF, so limit the number of interrupts we'll process. Correct the code in ahc_search_qinfifo() that guarantees that the sequencer will see an abort collision if the qinfifo is modified when a DMA is in progress. We now do this fixup after modifying the queue. This guarantees that the HSCB we place at the head of the queue is not the same as the old head. Using "next hscb" (guaranteed not to be the same as the first SCB) before clearing the queue could free up the original head hscb to be used during a remove operation placing it again at the head of the qinfifo. aic7xxx.h: Reduce the maximum number of outstanding commands to 253 from 254. To handle our output queue correctly on machines that only support 32bit stores, we must clear the array 4 bytes at a time. To avoid colliding with a DMA write from the sequencer, we must be sure that 4 slots are empty when we write to clear the queue. This reduces us to 253 SCBs: 1 that just completed and the known three additional empty slots in the queue that preceed it. Yahoo was able to force this race on one of their systems. Interrupts were disabled for such a time that the entire output queue was filled (254 entries complete without any processing), and our 32bit write to clear the status clobbered one entry. Add a feature tag for devices that are removable. aic7xxx.reg: Never use the sequencer interrupt value of 0xF0. We need to guanrantee that an INTSTAT value of 0xFF can only occur during card failure or a hot-eject. Align the busy targets table with the begining of scratch space. This seems to appease a chip bug in the aic7895. aic7xxx.seq: Be sure to disable select-out after a bus free event that occurs early in a selection. If we don't disable select-out, we will believe that it is enabled even though a new selection will never occur. Move the clearing of SELDI to just before a jump. This appeases another chip bug of the aic7895. Make the target mode command loop a bit more efficient. AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. Properly cleanup the last SCB we tested against should we fail to properly find an SCB for a reselection. Add some additional sequencer debugging code. aic7xxx_freebsd.c: Limit the driver to 253 outstanding commands per adapter. Guard against overflow in timeout handling. aic7xxx_inline.h: AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. aic7xxx_pci.c: Set the removable feature for the apa1480 cardbus and the 29160C Compact PCI card. Don't report high byte termination information for narrow cards. Use a PCI read rather than a questionable delay when fetching/setting termination settings.
2001-01-22 21:03:48 +00:00
test SSTAT0, SELDI|SELDO jz bus_free_sel;
test SSTAT0,SELDO jnz select_out;
select_in:
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
if ((ahc->flags & AHC_TARGETROLE) != 0) {
if ((ahc->flags & AHC_INITIATORROLE) != 0) {
test SSTAT0, TARGET jz initiator_reselect;
}
ahc_eisa.c: Initialize rid to 0. This doesn't seem to make any difference (the driver doesn't care what rid it gets and no-one seems to check rid's value), but follows standard conventions. Pass in our device_t to ahc_alloc(). We now use device_T softc storage, so passing NULL results in a panic. Set the unit number in our softc so that the driver core can retrieve it. ahc_pci.c: Set the unit number in our softc so that the driver core can retrieve it. aic7770.c: Insert our softc into the list of softcs when initialization is successful. aic7xxx.c: Remove a workaround for an aic7895 bug we will never trigger. Add additional diagnostic info to ahc_dump_card_state(). Always panic the system if a sequencer assertion fails. AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. Replace a hard coded number with a constant. Guard against looping forever in ahc_pause_and_flushwork(). A hot eject or card failure may make the intstat register return 0xFF, so limit the number of interrupts we'll process. Correct the code in ahc_search_qinfifo() that guarantees that the sequencer will see an abort collision if the qinfifo is modified when a DMA is in progress. We now do this fixup after modifying the queue. This guarantees that the HSCB we place at the head of the queue is not the same as the old head. Using "next hscb" (guaranteed not to be the same as the first SCB) before clearing the queue could free up the original head hscb to be used during a remove operation placing it again at the head of the qinfifo. aic7xxx.h: Reduce the maximum number of outstanding commands to 253 from 254. To handle our output queue correctly on machines that only support 32bit stores, we must clear the array 4 bytes at a time. To avoid colliding with a DMA write from the sequencer, we must be sure that 4 slots are empty when we write to clear the queue. This reduces us to 253 SCBs: 1 that just completed and the known three additional empty slots in the queue that preceed it. Yahoo was able to force this race on one of their systems. Interrupts were disabled for such a time that the entire output queue was filled (254 entries complete without any processing), and our 32bit write to clear the status clobbered one entry. Add a feature tag for devices that are removable. aic7xxx.reg: Never use the sequencer interrupt value of 0xF0. We need to guanrantee that an INTSTAT value of 0xFF can only occur during card failure or a hot-eject. Align the busy targets table with the begining of scratch space. This seems to appease a chip bug in the aic7895. aic7xxx.seq: Be sure to disable select-out after a bus free event that occurs early in a selection. If we don't disable select-out, we will believe that it is enabled even though a new selection will never occur. Move the clearing of SELDI to just before a jump. This appeases another chip bug of the aic7895. Make the target mode command loop a bit more efficient. AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. Properly cleanup the last SCB we tested against should we fail to properly find an SCB for a reselection. Add some additional sequencer debugging code. aic7xxx_freebsd.c: Limit the driver to 253 outstanding commands per adapter. Guard against overflow in timeout handling. aic7xxx_inline.h: AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. aic7xxx_pci.c: Set the removable feature for the apa1480 cardbus and the 29160C Compact PCI card. Don't report high byte termination information for narrow cards. Use a PCI read rather than a questionable delay when fetching/setting termination settings.
2001-01-22 21:03:48 +00:00
mvi CLRSINT0, CLRSELDI;
/*
* We've just been selected. Assert BSY and
* setup the phase for receiving messages
* from the target.
Sync Perforce IDs, add tranceiver state change support, and correct numerous error recovery buglets. Many thanks to Tor Egge for his assistance in diagnosing problems with the error recovery code. aic7xxx.c: Report missed bus free events using their own sequencer interrupt code to avoid confusion with other "bad phase" interrupts. Remove a delay used in debugging. This delay could only be hit in certain, very extreme, error recovery scenarios. Handle transceiver state changes correctly. You can now plug an SE device into a hot-plug LVD bus without hanging the controller. When stepping through a critical section, panic if we step more than a reasonable number of times. After a bus reset, disable bus reset interupts until we either our first attempt to (re)select another device, or another device attemps to select us. This removes the need to busy wait in kernel for the scsi reset line to fall yet still ensures we see any reset events that impact the state of either our initiator or target roles. Before this change, we had the potential of servicing a "storm" of reset interrupts if the reset line was held for a significant amount of time. Indicate the current sequencer address whenever we dump the card's state. aic7xxx.reg: Transceiver state change register definitions. Add the missed bussfree sequencer interrupt code. Re-enable the scsi reset interrupt if it has been disabled before every attempt to (re)select a device and when we have been selected as a target. When being (re)selected, check to see if the selection dissappeared just after we enabled our bus free interrupt. If the bus has gone free again, go back to the idle loop and wait for another selection. Note two locations where we should change our behavior if ATN is still raised. If ATN is raised during the presentation of a command complete or disconnect message, we should ignore the message and expect the target to put us in msgout phase. We don't currently do this as it requires some code re-arrangement so that critical sections can be properly placed around our handling of these two events. Otherwise, we cannot guarantee that the check of ATN is atomic relative to our acking of the message in byte (the kernel could assert ATN). Only set the IDENTIFY_SEEN flag after we have settled on the SCB for this transaction. The kernel looks at this flag before assuming that SCB_TAG is valid. This avoids confusion during certain types of error recovery. Add a critical section around findSCB. We cannot allow the kernel to remove an entry from the disconnected list while we are traversing it. Ditto for get_free_or_disc_scb. aic7xxx_freebsd.c: Only assume that SCB_TAG is accurate if IDENTIFY_SEEN is set in SEQ_FLAGS. Fix a typo that caused us to execute some code for the non-SCB paging case when paging SCBs. This only occurred during error recovery.
2000-11-10 20:13:41 +00:00
*
* If bus reset interrupts have been disabled (from a
* previous reset), re-enable them now. Resets are only
* of interest when we have outstanding transactions, so
* we can safely defer re-enabling the interrupt until,
* as a target, we start receiving transactions again.
*/
Sync Perforce IDs, add tranceiver state change support, and correct numerous error recovery buglets. Many thanks to Tor Egge for his assistance in diagnosing problems with the error recovery code. aic7xxx.c: Report missed bus free events using their own sequencer interrupt code to avoid confusion with other "bad phase" interrupts. Remove a delay used in debugging. This delay could only be hit in certain, very extreme, error recovery scenarios. Handle transceiver state changes correctly. You can now plug an SE device into a hot-plug LVD bus without hanging the controller. When stepping through a critical section, panic if we step more than a reasonable number of times. After a bus reset, disable bus reset interupts until we either our first attempt to (re)select another device, or another device attemps to select us. This removes the need to busy wait in kernel for the scsi reset line to fall yet still ensures we see any reset events that impact the state of either our initiator or target roles. Before this change, we had the potential of servicing a "storm" of reset interrupts if the reset line was held for a significant amount of time. Indicate the current sequencer address whenever we dump the card's state. aic7xxx.reg: Transceiver state change register definitions. Add the missed bussfree sequencer interrupt code. Re-enable the scsi reset interrupt if it has been disabled before every attempt to (re)select a device and when we have been selected as a target. When being (re)selected, check to see if the selection dissappeared just after we enabled our bus free interrupt. If the bus has gone free again, go back to the idle loop and wait for another selection. Note two locations where we should change our behavior if ATN is still raised. If ATN is raised during the presentation of a command complete or disconnect message, we should ignore the message and expect the target to put us in msgout phase. We don't currently do this as it requires some code re-arrangement so that critical sections can be properly placed around our handling of these two events. Otherwise, we cannot guarantee that the check of ATN is atomic relative to our acking of the message in byte (the kernel could assert ATN). Only set the IDENTIFY_SEEN flag after we have settled on the SCB for this transaction. The kernel looks at this flag before assuming that SCB_TAG is valid. This avoids confusion during certain types of error recovery. Add a critical section around findSCB. We cannot allow the kernel to remove an entry from the disconnected list while we are traversing it. Ditto for get_free_or_disc_scb. aic7xxx_freebsd.c: Only assume that SCB_TAG is accurate if IDENTIFY_SEEN is set in SEQ_FLAGS. Fix a typo that caused us to execute some code for the non-SCB paging case when paging SCBs. This only occurred during error recovery.
2000-11-10 20:13:41 +00:00
test SIMODE1, ENSCSIRST jnz . + 3;
mvi CLRSINT1, CLRSCSIRSTI;
or SIMODE1, ENSCSIRST;
mvi SCSISIGO, P_MESGOUT|BSYO;
/*
* Setup the DMA for sending the identify and
* command information.
*/
or SEQ_FLAGS, CMDPHASE_PENDING;
mov A, TQINPOS;
if ((ahc->features & AHC_CMD_CHAN) != 0) {
mvi DINDEX, CCHADDR;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
mvi SHARED_DATA_ADDR call set_32byte_addr;
mvi CCSCBCTL, CCSCBRESET;
} else {
mvi DINDEX, HADDR;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
mvi SHARED_DATA_ADDR call set_32byte_addr;
mvi DFCNTRL, FIFORESET;
}
/* Initiator that selected us */
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
and SAVED_SCSIID, SELID_MASK, SELID;
/* The Target ID we were selected at */
if ((ahc->features & AHC_MULTI_TID) != 0) {
and A, OID, TARGIDIN;
} else if ((ahc->features & AHC_ULTRA2) != 0) {
and A, OID, SCSIID_ULTRA2;
} else {
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
and A, OID, SCSIID;
}
or SAVED_SCSIID, A;
if ((ahc->features & AHC_TWIN) != 0) {
test SBLKCTL, SELBUSB jz . + 2;
or SAVED_SCSIID, TWIN_CHNLB;
}
if ((ahc->features & AHC_CMD_CHAN) != 0) {
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
mov CCSCBRAM, SAVED_SCSIID;
} else {
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
mov DFDAT, SAVED_SCSIID;
}
/*
* If ATN isn't asserted, the target isn't interested
* in talking to us. Go directly to bus free.
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
* XXX SCSI-1 may require us to assume lun 0 if
* ATN is false.
*/
test SCSISIGI, ATNI jz target_busfree;
/*
* Watch ATN closely now as we pull in messages from the
* initiator. We follow the guidlines from section 6.5
* of the SCSI-2 spec for what messages are allowed when.
*/
call target_inb;
/*
* Our first message must be one of IDENTIFY, ABORT, or
* BUS_DEVICE_RESET.
*/
test DINDEX, MSG_IDENTIFYFLAG jz host_target_message_loop;
/* Store for host */
if ((ahc->features & AHC_CMD_CHAN) != 0) {
mov CCSCBRAM, DINDEX;
} else {
mov DFDAT, DINDEX;
}
/* Remember for disconnection decision */
test DINDEX, MSG_IDENTIFY_DISCFLAG jnz . + 2;
/* XXX Honor per target settings too */
or SEQ_FLAGS, NO_DISCONNECT;
test SCSISIGI, ATNI jz ident_messages_done;
call target_inb;
/*
* If this is a tagged request, the tagged message must
* immediately follow the identify. We test for a valid
* tag message by seeing if it is >= MSG_SIMPLE_Q_TAG and
* < MSG_IGN_WIDE_RESIDUE.
*/
add A, -MSG_SIMPLE_Q_TAG, DINDEX;
jnc ident_messages_done;
add A, -MSG_IGN_WIDE_RESIDUE, DINDEX;
jc ident_messages_done;
/* Store for host */
if ((ahc->features & AHC_CMD_CHAN) != 0) {
mov CCSCBRAM, DINDEX;
} else {
mov DFDAT, DINDEX;
}
/*
* If the initiator doesn't feel like providing a tag number,
* we've got a failed selection and must transition to bus
* free.
*/
test SCSISIGI, ATNI jz target_busfree;
/*
* Store the tag for the host.
*/
call target_inb;
if ((ahc->features & AHC_CMD_CHAN) != 0) {
mov CCSCBRAM, DINDEX;
} else {
mov DFDAT, DINDEX;
}
mov INITIATOR_TAG, DINDEX;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
or SEQ_FLAGS, TARGET_CMD_IS_TAGGED;
test SCSISIGI, ATNI jz . + 2;
/* Initiator still wants to give us messages */
call target_inb;
jmp ident_messages_done;
/*
* Pushed message loop to allow the kernel to
* run it's own target mode message state engine.
*/
host_target_message_loop:
aic7xxx.c: When restarting the sequencer, ensure that the SCBCNT register is 0. A non-zero count will prevent the setting of the CCSCBDIR bit in any future dma operations. The only time CCSCBCNT would be non-zero is if we happened to halt the dma during a reset, but even that should never happen. Better safe than sorry. When a command completes before the target responds to an ATN for a recovery command, we now notify the kernel so that any recovery operation requeued in the qinfifo can be removed safely. In the past, we did this in ahc_done(), but ahc_done() may be called without the card paused. This also avoids a recursive call to ahc_search_qinifo() which could have occurred if ahc_search_qinififo() happened to be the routine to complete a recovery action. Fix 8bit math used for adjusting the qinfifo. The index must be wrapped properly within the 256 entry array. We rely on the fact that qinfifonext is a uint8_t in most cases to handle this wrap, but we missed a few spots where the resultant calculation was promoted to an int. Change the way that we deal with aborting the first or second entry from the qinfifo. We now swap the first entry in the qinfifo with the "next queued scb" to force the sequencer to see an abort collision if we ever touch the qinififo while the sequencer is mid SCB dma. aic7xxx.reg: Add new MKMSG_FAILED sequencer interrupt. This displaced the BOGUS_TAG interrupt used in some previous sequencer code debugging. aic7xxx.seq: Increment our position in the qinfifo only once the dma is complete and we have verified that the queue has not been changed during our DMA. This simplifies code in the kernel. Protect against "instruction creep" when issuing a pausing sequencer interrupt. On at least the 7890/91/96/97, the sequencer will coast after issuing the interrupt for up to two instructions. In the past we delt with this by using carefully placed nops. Now we call a routine to issue the interrupt followed by a nop and a ret. Tell the kernel should an SCB complete with the MK_MESSAGE flag still set. This means the target ignored our ATN request. Clear the channel twice as we exit the data phase. On the aic7890/91, the S/G preload logic may require the second clearing to get the last S/G out of the FIFO. aic7xxx_freebsd.c: Don't bother searching the qinfifo for a doubly queued recovery scb in ahc_done. This case is handled by the core driver now. Free the path used to issue async callbacks after the callback is complete. aic7xxx_inline.h: Split the SCB queue routine into a routine that swaps the SCB with the "next queued SCB" and a routine that calls the swapping routine and notifies the card of the new SCB. The swapping routine is now also used by ahc_search_qinfifo.
2000-11-06 20:05:38 +00:00
mvi HOST_MSG_LOOP call set_seqint;
cmp RETURN_1, EXIT_MSG_LOOP je target_ITloop;
test SSTAT0, SPIORDY jz .;
jmp host_target_message_loop;
ident_messages_done:
/* If ring buffer is full, return busy or queue full */
o Correct the offsets into the syncrate table for paritcular negotiation features (DT, ULTRA2, ULTRA, FAST). The offsets where not properly updated when the DT entry was added and so the driver could attempt to negotiate a speed faster than that supported by the target device or even requested by the user via SCSI-Select settings. * o Update the target mode incoming command queue kernel index value ever 128 commands instead of 32. This means that the kernel will always try to keep its index (as seen on the card - the kernel may actually have cleared more space) 128 commands ahead of where the sequencer is adding entries. o Use the HS_MAILBOX register instead of the KERNEL_TQINPOS location in SRAM to indicate the kernel's target queue possition on Ultra2 cards. This avoids the "pause bug" on these cards and also turns out to be much more efficient. o When enabling or disabling a particular target id for target mode, make sure that the taret id in the SCSIID register does not reference an ID that is not to receive target selections. This is only an issue on chips that support the multiple target id feature where the value in SCSIID will still affect selection behavior regardless of the values in the target id bit field registers. o Remove some target mode debugging printfs. o Make sure that the sense length reported in ATIO commands is always zero. This driver does not, yet, report HBA generated sense information for accepted commands. o Honor the CAM_TIME_INFINITY and CAM_TIME_DEFAULT values for the CCB timeout field. o Make the driver compile with AHC_DEBUG again. * Noticed by: Andrew Gallatin<gallatin@cs.duke.edu>
2000-03-18 22:28:20 +00:00
if ((ahc->features & AHC_HS_MAILBOX) != 0) {
and A, HOST_TQINPOS, HS_MAILBOX;
} else {
mov A, KERNEL_TQINPOS;
}
cmp TQINPOS, A jne tqinfifo_has_space;
mvi P_STATUS|BSYO call change_phase;
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
test SEQ_FLAGS, TARGET_CMD_IS_TAGGED jz . + 3;
mvi STATUS_QUEUE_FULL call target_outb;
jmp target_busfree_wait;
mvi STATUS_BUSY call target_outb;
jmp target_busfree_wait;
tqinfifo_has_space:
/* Terminate the ident list */
if ((ahc->features & AHC_CMD_CHAN) != 0) {
mvi CCSCBRAM, SCB_LIST_NULL;
} else {
mvi DFDAT, SCB_LIST_NULL;
}
or SEQ_FLAGS, TARG_CMD_PENDING|IDENTIFY_SEEN;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
test SCSISIGI, ATNI jnz target_mesgout_pending;
jmp target_ITloop;
/*
* We carefully toggle SPIOEN to allow us to return the
* message byte we receive so it can be checked prior to
* driving REQ on the bus for the next byte.
*/
target_inb:
/*
* Drive REQ on the bus by enabling SCSI PIO.
*/
or SXFRCTL0, SPIOEN;
/* Wait for the byte */
test SSTAT0, SPIORDY jz .;
/* Prevent our read from triggering another REQ */
and SXFRCTL0, ~SPIOEN;
/* Save latched contents */
mov DINDEX, SCSIDATL ret;
}
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
if ((ahc->flags & AHC_INITIATORROLE) != 0) {
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
/*
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
* Reselection has been initiated by a target. Make a note that we've been
* reselected, but haven't seen an IDENTIFY message from the target yet.
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
*/
initiator_reselect:
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
/* XXX test for and handle ONE BIT condition */
ahc_eisa.c: Initialize rid to 0. This doesn't seem to make any difference (the driver doesn't care what rid it gets and no-one seems to check rid's value), but follows standard conventions. Pass in our device_t to ahc_alloc(). We now use device_T softc storage, so passing NULL results in a panic. Set the unit number in our softc so that the driver core can retrieve it. ahc_pci.c: Set the unit number in our softc so that the driver core can retrieve it. aic7770.c: Insert our softc into the list of softcs when initialization is successful. aic7xxx.c: Remove a workaround for an aic7895 bug we will never trigger. Add additional diagnostic info to ahc_dump_card_state(). Always panic the system if a sequencer assertion fails. AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. Replace a hard coded number with a constant. Guard against looping forever in ahc_pause_and_flushwork(). A hot eject or card failure may make the intstat register return 0xFF, so limit the number of interrupts we'll process. Correct the code in ahc_search_qinfifo() that guarantees that the sequencer will see an abort collision if the qinfifo is modified when a DMA is in progress. We now do this fixup after modifying the queue. This guarantees that the HSCB we place at the head of the queue is not the same as the old head. Using "next hscb" (guaranteed not to be the same as the first SCB) before clearing the queue could free up the original head hscb to be used during a remove operation placing it again at the head of the qinfifo. aic7xxx.h: Reduce the maximum number of outstanding commands to 253 from 254. To handle our output queue correctly on machines that only support 32bit stores, we must clear the array 4 bytes at a time. To avoid colliding with a DMA write from the sequencer, we must be sure that 4 slots are empty when we write to clear the queue. This reduces us to 253 SCBs: 1 that just completed and the known three additional empty slots in the queue that preceed it. Yahoo was able to force this race on one of their systems. Interrupts were disabled for such a time that the entire output queue was filled (254 entries complete without any processing), and our 32bit write to clear the status clobbered one entry. Add a feature tag for devices that are removable. aic7xxx.reg: Never use the sequencer interrupt value of 0xF0. We need to guanrantee that an INTSTAT value of 0xFF can only occur during card failure or a hot-eject. Align the busy targets table with the begining of scratch space. This seems to appease a chip bug in the aic7895. aic7xxx.seq: Be sure to disable select-out after a bus free event that occurs early in a selection. If we don't disable select-out, we will believe that it is enabled even though a new selection will never occur. Move the clearing of SELDI to just before a jump. This appeases another chip bug of the aic7895. Make the target mode command loop a bit more efficient. AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. Properly cleanup the last SCB we tested against should we fail to properly find an SCB for a reselection. Add some additional sequencer debugging code. aic7xxx_freebsd.c: Limit the driver to 253 outstanding commands per adapter. Guard against overflow in timeout handling. aic7xxx_inline.h: AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. aic7xxx_pci.c: Set the removable feature for the apa1480 cardbus and the 29160C Compact PCI card. Don't report high byte termination information for narrow cards. Use a PCI read rather than a questionable delay when fetching/setting termination settings.
2001-01-22 21:03:48 +00:00
or SXFRCTL0, SPIOEN|CLRSTCNT|CLRCHN;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
and SAVED_SCSIID, SELID_MASK, SELID;
if ((ahc->features & AHC_ULTRA2) != 0) {
and A, OID, SCSIID_ULTRA2;
} else {
and A, OID, SCSIID;
}
or SAVED_SCSIID, A;
1998-09-21 16:46:13 +00:00
if ((ahc->features & AHC_TWIN) != 0) {
test SBLKCTL, SELBUSB jz . + 2;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
or SAVED_SCSIID, TWIN_CHNLB;
1998-09-21 16:46:13 +00:00
}
ahc_eisa.c: Initialize rid to 0. This doesn't seem to make any difference (the driver doesn't care what rid it gets and no-one seems to check rid's value), but follows standard conventions. Pass in our device_t to ahc_alloc(). We now use device_T softc storage, so passing NULL results in a panic. Set the unit number in our softc so that the driver core can retrieve it. ahc_pci.c: Set the unit number in our softc so that the driver core can retrieve it. aic7770.c: Insert our softc into the list of softcs when initialization is successful. aic7xxx.c: Remove a workaround for an aic7895 bug we will never trigger. Add additional diagnostic info to ahc_dump_card_state(). Always panic the system if a sequencer assertion fails. AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. Replace a hard coded number with a constant. Guard against looping forever in ahc_pause_and_flushwork(). A hot eject or card failure may make the intstat register return 0xFF, so limit the number of interrupts we'll process. Correct the code in ahc_search_qinfifo() that guarantees that the sequencer will see an abort collision if the qinfifo is modified when a DMA is in progress. We now do this fixup after modifying the queue. This guarantees that the HSCB we place at the head of the queue is not the same as the old head. Using "next hscb" (guaranteed not to be the same as the first SCB) before clearing the queue could free up the original head hscb to be used during a remove operation placing it again at the head of the qinfifo. aic7xxx.h: Reduce the maximum number of outstanding commands to 253 from 254. To handle our output queue correctly on machines that only support 32bit stores, we must clear the array 4 bytes at a time. To avoid colliding with a DMA write from the sequencer, we must be sure that 4 slots are empty when we write to clear the queue. This reduces us to 253 SCBs: 1 that just completed and the known three additional empty slots in the queue that preceed it. Yahoo was able to force this race on one of their systems. Interrupts were disabled for such a time that the entire output queue was filled (254 entries complete without any processing), and our 32bit write to clear the status clobbered one entry. Add a feature tag for devices that are removable. aic7xxx.reg: Never use the sequencer interrupt value of 0xF0. We need to guanrantee that an INTSTAT value of 0xFF can only occur during card failure or a hot-eject. Align the busy targets table with the begining of scratch space. This seems to appease a chip bug in the aic7895. aic7xxx.seq: Be sure to disable select-out after a bus free event that occurs early in a selection. If we don't disable select-out, we will believe that it is enabled even though a new selection will never occur. Move the clearing of SELDI to just before a jump. This appeases another chip bug of the aic7895. Make the target mode command loop a bit more efficient. AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. Properly cleanup the last SCB we tested against should we fail to properly find an SCB for a reselection. Add some additional sequencer debugging code. aic7xxx_freebsd.c: Limit the driver to 253 outstanding commands per adapter. Guard against overflow in timeout handling. aic7xxx_inline.h: AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. aic7xxx_pci.c: Set the removable feature for the apa1480 cardbus and the 29160C Compact PCI card. Don't report high byte termination information for narrow cards. Use a PCI read rather than a questionable delay when fetching/setting termination settings.
2001-01-22 21:03:48 +00:00
mvi CLRSINT0, CLRSELDI;
jmp ITloop;
}
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
/*
* After the selection, remove this SCB from the "waiting SCB"
* list. This is achieved by simply moving our "next" pointer into
* WAITING_SCBH. Our next pointer will be set to null the next time this
* SCB is used, so don't bother with it now.
*/
select_out:
/* Turn off the selection hardware */
o Correct the offsets into the syncrate table for paritcular negotiation features (DT, ULTRA2, ULTRA, FAST). The offsets where not properly updated when the DT entry was added and so the driver could attempt to negotiate a speed faster than that supported by the target device or even requested by the user via SCSI-Select settings. * o Update the target mode incoming command queue kernel index value ever 128 commands instead of 32. This means that the kernel will always try to keep its index (as seen on the card - the kernel may actually have cleared more space) 128 commands ahead of where the sequencer is adding entries. o Use the HS_MAILBOX register instead of the KERNEL_TQINPOS location in SRAM to indicate the kernel's target queue possition on Ultra2 cards. This avoids the "pause bug" on these cards and also turns out to be much more efficient. o When enabling or disabling a particular target id for target mode, make sure that the taret id in the SCSIID register does not reference an ID that is not to receive target selections. This is only an issue on chips that support the multiple target id feature where the value in SCSIID will still affect selection behavior regardless of the values in the target id bit field registers. o Remove some target mode debugging printfs. o Make sure that the sense length reported in ATIO commands is always zero. This driver does not, yet, report HBA generated sense information for accepted commands. o Honor the CAM_TIME_INFINITY and CAM_TIME_DEFAULT values for the CCB timeout field. o Make the driver compile with AHC_DEBUG again. * Noticed by: Andrew Gallatin<gallatin@cs.duke.edu>
2000-03-18 22:28:20 +00:00
and SCSISEQ, TEMODE|ENSELI|ENRSELI|ENAUTOATNP, SCSISEQ;
mvi CLRSINT0, CLRSELDO;
mov SCBPTR, WAITING_SCBH;
mov WAITING_SCBH,SCB_NEXT;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
mov SAVED_SCSIID, SCB_SCSIID;
mov SAVED_LUN, SCB_LUN;
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
call initialize_channel;
if ((ahc->flags & AHC_TARGETROLE) != 0) {
test SSTAT0, TARGET jz initiator_select;
/*
* We've just re-selected an initiator.
* Assert BSY and setup the phase for
* sending our identify messages.
*/
mvi P_MESGIN|BSYO call change_phase;
/*
* Start out with a simple identify message.
*/
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
or SCB_LUN, MSG_IDENTIFYFLAG call target_outb;
/*
* If we are the result of a tagged command, send
* a simple Q tag and the tag id.
*/
test SCB_CONTROL, TAG_ENB jz . + 3;
mvi MSG_SIMPLE_Q_TAG call target_outb;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
mov SCB_TARGET_INFO[SCB_INITIATOR_TAG] call target_outb;
target_synccmd:
/*
* Now determine what phases the host wants us
* to go through.
*/
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
mov SEQ_FLAGS, SCB_TARGET_INFO[SCB_TARGET_PHASES];
target_ITloop:
/*
* Start honoring ATN signals now that
* we properly identified ourselves.
*/
test SCSISIGI, ATNI jnz target_mesgout;
test SEQ_FLAGS, CMDPHASE_PENDING jnz target_cmdphase;
test SEQ_FLAGS, DPHASE_PENDING jnz target_dphase;
test SEQ_FLAGS, SPHASE_PENDING jnz target_sphase;
/*
* No more work to do. Either disconnect or not depending
* on the state of NO_DISCONNECT.
*/
test SEQ_FLAGS, NO_DISCONNECT jz target_disconnect;
if ((ahc->flags & AHC_PAGESCBS) != 0) {
mov ALLZEROS call get_free_or_disc_scb;
}
mov RETURN_1, ALLZEROS;
call complete_target_cmd;
cmp RETURN_1, CONT_MSG_LOOP jne .;
mvi DMAPARAMS, HDMAEN|DIRECTION|FIFORESET;
mov SCB_TAG call dma_scb;
jmp target_synccmd;
target_mesgout:
mvi SCSISIGO, P_MESGOUT|BSYO;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
target_mesgout_continue:
call target_inb;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
target_mesgout_pending:
/* Local Processing goes here... */
jmp host_target_message_loop;
target_disconnect:
mvi P_MESGIN|BSYO call change_phase;
test SEQ_FLAGS, DPHASE jz . + 2;
mvi MSG_SAVEDATAPOINTER call target_outb;
mvi MSG_DISCONNECT call target_outb;
target_busfree_wait:
/* Wait for preceding I/O session to complete. */
test SCSISIGI, ACKI jnz .;
target_busfree:
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
and SIMODE1, ~ENBUSFREE;
if ((ahc->features & AHC_ULTRA2) != 0) {
clr SCSIBUSL;
}
clr SCSISIGO;
Kill the "unpause_always" argument to unpause_sequencer(). The reasons for optimizing the unpause operation no-longer exist, and this is much safer. When restarting the sequencer, reconstitute the free SCB list on the card. This deals with a single instruction gap between marking the SCB as free and actually getting it onto the free list. Reduce the number of transfer negotiations that occur. In the past, we renegotiated after every reported check condition status. This ensures that we catch devices that have unexpectidly reset. In this situation, the target will always report the check condition before performing a data-phase. The new behavior is to renegotiate for any check-condition where the residual matches the orginal data-length of the command (including 0 length transffers). This avoids renegotiations during things like variable tape block reads, where the check condition is reported only to indicate the residual of the read. Revamp the parity error detection logic. We now properly report and handle injected parity errors in all phases. The old code used to hang on message-in parity errors. Correct the reporting of selection timeout errors to the XPT. When a selection timeout occurs, only the currently selecting command is flagged with SELTO status instead of aborting all currently active commands to that target. Fix flipped arguments in ahc_match_scb and in some of the callers of this routine. I wish that gcc allowed you to request warnings for enums passed as ints. Make ahc_find_msg generically handle all message types. Work around the target mode data-in wideodd bug in all non-U2 chips. We can now do sync-wide target mode transfers in target mode across the hole product line. Use lastphase exclusively for handling timeouts. The current phase doesn't take the bus free state into account. Fix a bug in the timeout handler that could cause corruption of the disconnected list. When sending an embedded cdb to a target, ensure that we start on a quad word boundary in the data-fifo. It seems that unaligned stores do not work correctly.
2000-02-09 21:25:00 +00:00
mvi LASTPHASE, P_BUSFREE;
call complete_target_cmd;
jmp poll_for_work;
target_cmdphase:
mvi P_COMMAND|BSYO call change_phase;
call target_inb;
mov A, DINDEX;
/* Store for host */
if ((ahc->features & AHC_CMD_CHAN) != 0) {
mov CCSCBRAM, A;
} else {
mov DFDAT, A;
}
/*
* Determine the number of bytes to read
* based on the command group code via table lookup.
* We reuse the first 8 bytes of the TARG_SCSIRATE
* BIOS array for this table. Count is one less than
* the total for the command since we've already fetched
* the first byte.
*/
shr A, CMD_GROUP_CODE_SHIFT;
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
add SINDEX, CMDSIZE_TABLE, A;
mov A, SINDIR;
test A, 0xFF jz command_phase_done;
or SXFRCTL0, SPIOEN;
ahc_eisa.c: Initialize rid to 0. This doesn't seem to make any difference (the driver doesn't care what rid it gets and no-one seems to check rid's value), but follows standard conventions. Pass in our device_t to ahc_alloc(). We now use device_T softc storage, so passing NULL results in a panic. Set the unit number in our softc so that the driver core can retrieve it. ahc_pci.c: Set the unit number in our softc so that the driver core can retrieve it. aic7770.c: Insert our softc into the list of softcs when initialization is successful. aic7xxx.c: Remove a workaround for an aic7895 bug we will never trigger. Add additional diagnostic info to ahc_dump_card_state(). Always panic the system if a sequencer assertion fails. AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. Replace a hard coded number with a constant. Guard against looping forever in ahc_pause_and_flushwork(). A hot eject or card failure may make the intstat register return 0xFF, so limit the number of interrupts we'll process. Correct the code in ahc_search_qinfifo() that guarantees that the sequencer will see an abort collision if the qinfifo is modified when a DMA is in progress. We now do this fixup after modifying the queue. This guarantees that the HSCB we place at the head of the queue is not the same as the old head. Using "next hscb" (guaranteed not to be the same as the first SCB) before clearing the queue could free up the original head hscb to be used during a remove operation placing it again at the head of the qinfifo. aic7xxx.h: Reduce the maximum number of outstanding commands to 253 from 254. To handle our output queue correctly on machines that only support 32bit stores, we must clear the array 4 bytes at a time. To avoid colliding with a DMA write from the sequencer, we must be sure that 4 slots are empty when we write to clear the queue. This reduces us to 253 SCBs: 1 that just completed and the known three additional empty slots in the queue that preceed it. Yahoo was able to force this race on one of their systems. Interrupts were disabled for such a time that the entire output queue was filled (254 entries complete without any processing), and our 32bit write to clear the status clobbered one entry. Add a feature tag for devices that are removable. aic7xxx.reg: Never use the sequencer interrupt value of 0xF0. We need to guanrantee that an INTSTAT value of 0xFF can only occur during card failure or a hot-eject. Align the busy targets table with the begining of scratch space. This seems to appease a chip bug in the aic7895. aic7xxx.seq: Be sure to disable select-out after a bus free event that occurs early in a selection. If we don't disable select-out, we will believe that it is enabled even though a new selection will never occur. Move the clearing of SELDI to just before a jump. This appeases another chip bug of the aic7895. Make the target mode command loop a bit more efficient. AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. Properly cleanup the last SCB we tested against should we fail to properly find an SCB for a reselection. Add some additional sequencer debugging code. aic7xxx_freebsd.c: Limit the driver to 253 outstanding commands per adapter. Guard against overflow in timeout handling. aic7xxx_inline.h: AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. aic7xxx_pci.c: Set the removable feature for the apa1480 cardbus and the 29160C Compact PCI card. Don't report high byte termination information for narrow cards. Use a PCI read rather than a questionable delay when fetching/setting termination settings.
2001-01-22 21:03:48 +00:00
command_loop:
test SSTAT0, SPIORDY jz .;
cmp A, 1 jne . + 2;
and SXFRCTL0, ~SPIOEN; /* Last Byte */
if ((ahc->features & AHC_CMD_CHAN) != 0) {
mov CCSCBRAM, SCSIDATL;
} else {
mov DFDAT, SCSIDATL;
}
dec A;
test A, 0xFF jnz command_loop;
command_phase_done:
and SEQ_FLAGS, ~CMDPHASE_PENDING;
jmp target_ITloop;
target_dphase:
/*
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
* Data phases on the bus are from the
* perspective of the initiator. The dma
* code looks at LASTPHASE to determine the
* data direction of the DMA. Toggle it for
* target transfers.
*/
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
xor LASTPHASE, IOI, SCB_TARGET_INFO[SCB_TARGET_DATA_DIR];
or SCB_TARGET_INFO[SCB_TARGET_DATA_DIR], BSYO
call change_phase;
jmp p_data;
target_sphase:
mvi P_STATUS|BSYO call change_phase;
mvi LASTPHASE, P_STATUS;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
mov SCB_TARGET_INFO[SCB_TARGET_STATUS] call target_outb;
/* XXX Watch for ATN or parity errors??? */
mvi SCSISIGO, P_MESGIN|BSYO;
/* MSG_CMDCMPLT is 0, but we can't do an immediate of 0 */
mov ALLZEROS call target_outb;
jmp target_busfree_wait;
complete_target_cmd:
test SEQ_FLAGS, TARG_CMD_PENDING jnz . + 2;
mov SCB_TAG jmp complete_post;
if ((ahc->features & AHC_CMD_CHAN) != 0) {
/* Set the valid byte */
mvi CCSCBADDR, 24;
mov CCSCBRAM, ALLONES;
mvi CCHCNT, 28;
or CCSCBCTL, CCSCBEN|CCSCBRESET;
test CCSCBCTL, CCSCBDONE jz .;
clr CCSCBCTL;
} else {
/* Set the valid byte */
or DFCNTRL, FIFORESET;
mvi DFWADDR, 3; /* Third 64bit word or byte 24 */
mov DFDAT, ALLONES;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
mvi 28 call set_hcnt;
or DFCNTRL, HDMAEN|FIFOFLUSH;
call dma_finish;
}
inc TQINPOS;
mvi INTSTAT,CMDCMPLT ret;
}
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
if ((ahc->flags & AHC_INITIATORROLE) != 0) {
initiator_select:
/*
* 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;
or SEQ_FLAGS, IDENTIFY_SEEN;
/*
* 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.
*/
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
mesgin_phasemis:
ITloop:
call phase_lock;
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
mov A, LASTPHASE;
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
test A, ~P_DATAIN jz p_data;
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
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;
aic7xxx.c: When restarting the sequencer, ensure that the SCBCNT register is 0. A non-zero count will prevent the setting of the CCSCBDIR bit in any future dma operations. The only time CCSCBCNT would be non-zero is if we happened to halt the dma during a reset, but even that should never happen. Better safe than sorry. When a command completes before the target responds to an ATN for a recovery command, we now notify the kernel so that any recovery operation requeued in the qinfifo can be removed safely. In the past, we did this in ahc_done(), but ahc_done() may be called without the card paused. This also avoids a recursive call to ahc_search_qinifo() which could have occurred if ahc_search_qinififo() happened to be the routine to complete a recovery action. Fix 8bit math used for adjusting the qinfifo. The index must be wrapped properly within the 256 entry array. We rely on the fact that qinfifonext is a uint8_t in most cases to handle this wrap, but we missed a few spots where the resultant calculation was promoted to an int. Change the way that we deal with aborting the first or second entry from the qinfifo. We now swap the first entry in the qinfifo with the "next queued scb" to force the sequencer to see an abort collision if we ever touch the qinififo while the sequencer is mid SCB dma. aic7xxx.reg: Add new MKMSG_FAILED sequencer interrupt. This displaced the BOGUS_TAG interrupt used in some previous sequencer code debugging. aic7xxx.seq: Increment our position in the qinfifo only once the dma is complete and we have verified that the queue has not been changed during our DMA. This simplifies code in the kernel. Protect against "instruction creep" when issuing a pausing sequencer interrupt. On at least the 7890/91/96/97, the sequencer will coast after issuing the interrupt for up to two instructions. In the past we delt with this by using carefully placed nops. Now we call a routine to issue the interrupt followed by a nop and a ret. Tell the kernel should an SCB complete with the MK_MESSAGE flag still set. This means the target ignored our ATN request. Clear the channel twice as we exit the data phase. On the aic7890/91, the S/G preload logic may require the second clearing to get the last S/G out of the FIFO. aic7xxx_freebsd.c: Don't bother searching the qinfifo for a doubly queued recovery scb in ahc_done. This case is handled by the core driver now. Free the path used to issue async callbacks after the callback is complete. aic7xxx_inline.h: Split the SCB queue routine into a routine that swaps the SCB with the "next queued SCB" and a routine that calls the swapping routine and notifies the card of the new SCB. The swapping routine is now also used by ahc_search_qinfifo.
2000-11-06 20:05:38 +00:00
mvi BAD_PHASE call set_seqint;
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
jmp ITloop; /* Try reading the bus again. */
await_busfree:
and SIMODE1, ~ENBUSFREE;
mov NONE, SCSIDATL; /* Ack the last byte */
if ((ahc->features & AHC_ULTRA2) != 0) {
clr SCSIBUSL; /* Prevent bit leakage durint SELTO */
}
and SXFRCTL0, ~SPIOEN;
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
test SSTAT1,REQINIT|BUSFREE jz .;
test SSTAT1, BUSFREE jnz poll_for_work;
Sync Perforce IDs, add tranceiver state change support, and correct numerous error recovery buglets. Many thanks to Tor Egge for his assistance in diagnosing problems with the error recovery code. aic7xxx.c: Report missed bus free events using their own sequencer interrupt code to avoid confusion with other "bad phase" interrupts. Remove a delay used in debugging. This delay could only be hit in certain, very extreme, error recovery scenarios. Handle transceiver state changes correctly. You can now plug an SE device into a hot-plug LVD bus without hanging the controller. When stepping through a critical section, panic if we step more than a reasonable number of times. After a bus reset, disable bus reset interupts until we either our first attempt to (re)select another device, or another device attemps to select us. This removes the need to busy wait in kernel for the scsi reset line to fall yet still ensures we see any reset events that impact the state of either our initiator or target roles. Before this change, we had the potential of servicing a "storm" of reset interrupts if the reset line was held for a significant amount of time. Indicate the current sequencer address whenever we dump the card's state. aic7xxx.reg: Transceiver state change register definitions. Add the missed bussfree sequencer interrupt code. Re-enable the scsi reset interrupt if it has been disabled before every attempt to (re)select a device and when we have been selected as a target. When being (re)selected, check to see if the selection dissappeared just after we enabled our bus free interrupt. If the bus has gone free again, go back to the idle loop and wait for another selection. Note two locations where we should change our behavior if ATN is still raised. If ATN is raised during the presentation of a command complete or disconnect message, we should ignore the message and expect the target to put us in msgout phase. We don't currently do this as it requires some code re-arrangement so that critical sections can be properly placed around our handling of these two events. Otherwise, we cannot guarantee that the check of ATN is atomic relative to our acking of the message in byte (the kernel could assert ATN). Only set the IDENTIFY_SEEN flag after we have settled on the SCB for this transaction. The kernel looks at this flag before assuming that SCB_TAG is valid. This avoids confusion during certain types of error recovery. Add a critical section around findSCB. We cannot allow the kernel to remove an entry from the disconnected list while we are traversing it. Ditto for get_free_or_disc_scb. aic7xxx_freebsd.c: Only assume that SCB_TAG is accurate if IDENTIFY_SEEN is set in SEQ_FLAGS. Fix a typo that caused us to execute some code for the non-SCB paging case when paging SCBs. This only occurred during error recovery.
2000-11-10 20:13:41 +00:00
mvi MISSED_BUSFREE call set_seqint;
}
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
clear_target_state:
/*
* We assume that the kernel driver may reset us
* at any time, even in the middle of a DMA, so
* clear DFCNTRL too.
*/
clr DFCNTRL;
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
or SXFRCTL0, CLRSTCNT|CLRCHN;
/*
* We don't know the target we will connect to,
* so default to narrow transfers to avoid
* parity problems.
*/
if ((ahc->features & AHC_ULTRA2) != 0) {
bmov SCSIRATE, ALLZEROS, 2;
} else {
clr SCSIRATE;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
if ((ahc->features & AHC_ULTRA) != 0) {
and SXFRCTL0, ~(FAST20);
}
}
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
mvi LASTPHASE, P_BUSFREE;
/* clear target specific flags */
clr SEQ_FLAGS ret;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
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 ret;
idle_loop:
if ((ahc->features & AHC_CMD_CHAN) != 0) {
/* Did we just finish fetching segs? */
cmp CCSGCTL, CCSGEN|CCSGDONE je idle_sgfetch_complete;
/* Are we actively fetching segments? */
test CCSGCTL, CCSGEN jnz return;
/*
* Do we need any more segments?
*/
test SCB_RESIDUAL_DATACNT[3], SG_LAST_SEG jnz return;
/*
* Do we have any prefetch left???
*/
cmp CCSGADDR, SG_PREFETCH_CNT jne idle_sg_avail;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
/*
* Need to fetch segments, but we can only do that
* if the command channel is completely idle. Make
* sure we don't have an SCB prefetch going on.
*/
test CCSCBCTL, CCSCBEN jnz return;
/*
* 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.
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
*/
mvi CCHCNT, SG_PREFETCH_CNT;
and CCHADDR[0], SG_PREFETCH_ALIGN_MASK, SCB_RESIDUAL_SGPTR;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
bmov CCHADDR[1], SCB_RESIDUAL_SGPTR[1], 3;
mvi CCSGCTL, CCSGEN|CCSGRESET ret;
idle_sgfetch_complete:
ahc_pci.c: Disable "cache line streaming" for aic7890/91 Rev A chips. I have never seen these chips fail using this feature, but some of Adaptec's regression tests have. Explicitly set "cache line streaming" to on for aic7896/97 chips. This was happening before, but this documents the fact that these chips will not function correctly without CACHETHEEN set. aic7xxx.h: Add new bug types. Fix a typo in a comment. aic7xxx.reg: Add a definition for the SHVALID bit in SSTAT3 for Ultra2/3 chips. This bit inicates whether the bottom most (current) element in the S/G fifo has exhausted its data count. aic7xxx.seq: Be more careful in how we turn off the secondary DMA channel. Being less careful may hang the PCI bus arbitor that negotiates between the two DMA engines. Remove an unecessary and incorrect flag set operation in the overrun case. On Ultra2/3 controllers, clear the dma FIFO before starting to handle an overrun. We don't want any residual bytes from the beginning of the overrun to cause the code that shuts down the DMA engine from hanging because the FIFO is not (and never will be) empty. If the data fifo is empty by the time we notice that a read transaction has completed, there is no need to hit the flush bit on aic7890/91 hardware that will not perform an auto-flush. Skip some cycles by short circuiting the manual flush code in this case. When transitioning out of data phase, make sure that we have the next S/G element loaded for the following reconnect if there is more work to do. The code would do this in most cases before, but there was a small window where the current S/G element could be exhausted before our fetch of the next S/G element completed. Since the S/G fetch is already initiated at this point, it makes sense to just wait for the segment to arrive instead of incuring even more latency by canceling the fetch and initiating it later. Fast path the end of data phase handling for the last S/G segment. In the general case, we might have worked ahead a bit by stuffing the S/G FIFO with additional segments. If we stop before using them all, we need to fixup our location in the S/G stream. Since we can't work past the last S/G segment, no fixups are ever required if we stop somewhere in that final segment. Fix a little buglet in the target mode dma bug handler. We were employing the workaround in all cases instead of only for the chips that require it. Fix the cause of SCB timeouts and possible "lost data" during read operations on the aic7890. When sending a data on any Ultra2/3 controller, the final segment must be marked as such so the FIFO will be flushed and cleaned up correctly when the transfer is ended. We failed to do this for the CDB transfer and so, if the target immediately transfered from command to data phase without an intervening disconnection, the first segment transferred would be any residual bytes from the cdb transfer. The Ultra160 controllers for some reason were not affected by this problem. Many Thanks to Tor Egge for bringing the aic7890 problem to my attention, providing analysis, as well as a mechanism to reproduce the problem.
2000-07-27 23:17:52 +00:00
clr CCSGCTL;
test CCSGCTL, CCSGEN jnz .;
and CCSGADDR, SG_PREFETCH_ADDR_MASK, SCB_RESIDUAL_SGPTR;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
idle_sg_avail:
if ((ahc->features & AHC_ULTRA2) != 0) {
/* Does the hardware have space for another SG entry? */
test DFSTATUS, PRELOAD_AVAIL jz return;
bmov HADDR, CCSGRAM, 4;
bmov SINDEX, CCSGRAM, 1;
test SINDEX, 0x1 jz . + 2;
xor DATA_COUNT_ODD, 0x1;
bmov HCNT[0], SINDEX, 1;
bmov HCNT[1], CCSGRAM, 2;
bmov SCB_RESIDUAL_DATACNT[3], CCSGRAM, 1;
call sg_advance;
mov SINDEX, SCB_RESIDUAL_SGPTR[0];
test DATA_COUNT_ODD, 0x1 jz . + 2;
or SINDEX, ODD_SEG;
test SCB_RESIDUAL_DATACNT[3], SG_LAST_SEG jz . + 2;
or SINDEX, LAST_SEG;
mov SG_CACHE_PRE, SINDEX;
/* Load the segment by writing DFCNTRL again */
mov DFCNTRL, DMAPARAMS;
}
ret;
}
if ((ahc->bugs & AHC_PCI_MWI_BUG) != 0 && ahc->pci_cachesize != 0) {
/*
* Calculate the trailing portion of this S/G segment that cannot
* be transferred using memory write and invalidate PCI transactions.
* XXX Can we optimize this for PCI writes only???
*/
calc_mwi_residual:
/*
* If the ending address is on a cacheline boundary,
* there is no need for an extra segment.
*/
mov A, HCNT[0];
add A, A, HADDR[0];
and A, CACHESIZE_MASK;
test A, 0xFF jz return;
/*
* If the transfer is less than a cachline,
* there is no need for an extra segment.
*/
test HCNT[1], 0xFF jnz calc_mwi_residual_final;
test HCNT[2], 0xFF jnz calc_mwi_residual_final;
add NONE, INVERTED_CACHESIZE_MASK, HCNT[0];
jnc return;
calc_mwi_residual_final:
mov MWI_RESIDUAL, A;
not A;
inc A;
add HCNT[0], A;
adc HCNT[1], -1;
adc HCNT[2], -1 ret;
}
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
/*
* If we re-enter the data phase after going through another phase, the
* STCNT may have been cleared, so restore it from the residual field.
*/
data_phase_reinit:
if ((ahc->features & AHC_ULTRA2) != 0) {
/*
* The preload circuitry requires us to
* reload the address too, so pull it from
* the shaddow address.
*/
bmov HADDR, SHADDR, 4;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
bmov HCNT, SCB_RESIDUAL_DATACNT, 3;
} else if ((ahc->features & AHC_CMD_CHAN) != 0) {
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
bmov STCNT, SCB_RESIDUAL_DATACNT, 3;
} else {
mvi DINDEX, STCNT;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
mvi SCB_RESIDUAL_DATACNT call bcopy_3;
}
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
and DATA_COUNT_ODD, 0x1, SCB_RESIDUAL_DATACNT[0];
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
jmp data_phase_loop;
p_data:
if ((ahc->features & AHC_ULTRA2) != 0) {
mvi DMAPARAMS, PRELOADEN|SCSIEN|HDMAEN;
} else {
mvi DMAPARAMS, WIDEODD|SCSIEN|SDMAEN|HDMAEN|FIFORESET;
}
test LASTPHASE, IOI jnz . + 2;
or DMAPARAMS, DIRECTION;
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
call assert; /*
* Ensure entering a data
* phase is okay - seen identify, etc.
*/
if ((ahc->features & AHC_CMD_CHAN) != 0) {
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
/* We don't have any valid S/G elements */
mvi CCSGADDR, SG_PREFETCH_CNT;
}
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
test SEQ_FLAGS, DPHASE jnz data_phase_reinit;
/* We have seen a data phase */
or SEQ_FLAGS, DPHASE;
/*
* Initialize the DMA address and counter from the SCB.
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
* Also set SCB_RESIDUAL_SGPTR, including the LAST_SEG
* flag in the highest byte of the data count. We cannot
* modify the saved values in the SCB until we see a save
* data pointers message.
*/
if ((ahc->features & AHC_CMD_CHAN) != 0) {
bmov HADDR, SCB_DATAPTR, 7;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
bmov SCB_RESIDUAL_DATACNT[3], SCB_DATACNT[3], 5;
} else {
mvi DINDEX, HADDR;
mvi SCB_DATAPTR call bcopy_7;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
mvi DINDEX, SCB_RESIDUAL_DATACNT + 3;
mvi SCB_DATACNT + 3 call bcopy_5;
}
if ((ahc->bugs & AHC_PCI_MWI_BUG) != 0 && ahc->pci_cachesize != 0) {
call calc_mwi_residual;
}
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
and SCB_RESIDUAL_SGPTR[0], ~SG_FULL_RESID;
and DATA_COUNT_ODD, 0x1, HCNT[0];
if ((ahc->features & AHC_ULTRA2) == 0) {
if ((ahc->features & AHC_CMD_CHAN) != 0) {
bmov STCNT, HCNT, 3;
} else {
call set_stcnt_from_hcnt;
}
}
data_phase_loop:
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
/* Guard against overruns */
test SCB_RESIDUAL_SGPTR[0], SG_LIST_NULL jz data_phase_inbounds;
/*
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
* Turn on `Bit Bucket' mode, wait until the target takes
* us to another phase, and then notify the host.
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
*/
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
and DMAPARAMS, DIRECTION;
mov DFCNTRL, DMAPARAMS;
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
or SXFRCTL1,BITBUCKET;
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
test SSTAT1,PHASEMIS jz .;
and SXFRCTL1, ~BITBUCKET;
aic7xxx.c: When restarting the sequencer, ensure that the SCBCNT register is 0. A non-zero count will prevent the setting of the CCSCBDIR bit in any future dma operations. The only time CCSCBCNT would be non-zero is if we happened to halt the dma during a reset, but even that should never happen. Better safe than sorry. When a command completes before the target responds to an ATN for a recovery command, we now notify the kernel so that any recovery operation requeued in the qinfifo can be removed safely. In the past, we did this in ahc_done(), but ahc_done() may be called without the card paused. This also avoids a recursive call to ahc_search_qinifo() which could have occurred if ahc_search_qinififo() happened to be the routine to complete a recovery action. Fix 8bit math used for adjusting the qinfifo. The index must be wrapped properly within the 256 entry array. We rely on the fact that qinfifonext is a uint8_t in most cases to handle this wrap, but we missed a few spots where the resultant calculation was promoted to an int. Change the way that we deal with aborting the first or second entry from the qinfifo. We now swap the first entry in the qinfifo with the "next queued scb" to force the sequencer to see an abort collision if we ever touch the qinififo while the sequencer is mid SCB dma. aic7xxx.reg: Add new MKMSG_FAILED sequencer interrupt. This displaced the BOGUS_TAG interrupt used in some previous sequencer code debugging. aic7xxx.seq: Increment our position in the qinfifo only once the dma is complete and we have verified that the queue has not been changed during our DMA. This simplifies code in the kernel. Protect against "instruction creep" when issuing a pausing sequencer interrupt. On at least the 7890/91/96/97, the sequencer will coast after issuing the interrupt for up to two instructions. In the past we delt with this by using carefully placed nops. Now we call a routine to issue the interrupt followed by a nop and a ret. Tell the kernel should an SCB complete with the MK_MESSAGE flag still set. This means the target ignored our ATN request. Clear the channel twice as we exit the data phase. On the aic7890/91, the S/G preload logic may require the second clearing to get the last S/G out of the FIFO. aic7xxx_freebsd.c: Don't bother searching the qinfifo for a doubly queued recovery scb in ahc_done. This case is handled by the core driver now. Free the path used to issue async callbacks after the callback is complete. aic7xxx_inline.h: Split the SCB queue routine into a routine that swaps the SCB with the "next queued SCB" and a routine that calls the swapping routine and notifies the card of the new SCB. The swapping routine is now also used by ahc_search_qinfifo.
2000-11-06 20:05:38 +00:00
mvi DATA_OVERRUN call set_seqint;
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
jmp ITloop;
data_phase_inbounds:
if ((ahc->features & AHC_ULTRA2) != 0) {
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
mov SINDEX, SCB_RESIDUAL_SGPTR[0];
test SCB_RESIDUAL_DATACNT[3], SG_LAST_SEG jz . + 2;
or SINDEX, LAST_SEG;
test DATA_COUNT_ODD, 0x1 jz . + 2;
or SINDEX, ODD_SEG;
mov SG_CACHE_PRE, SINDEX;
mov DFCNTRL, DMAPARAMS;
ultra2_dma_loop:
call idle_loop;
/*
* The transfer is complete if either the last segment
* completes or the target changes phase.
*/
test SG_CACHE_SHADOW, LAST_SEG_DONE jnz ultra2_dmafinish;
ahc_eisa.c: ahc_pci.c: Add detach support. Make use of soft allocated on our behalf by newbus. For PCI devices, disable the mapping type we aren't using for extra protection from rogue code. aic7xxx_93cx6.c: aic7xxx_93cx6.h: Sync perforce IDs. aic7xxx_freebsd.c: Capture the eventhandle returned by EVENTHANDER_REGISTER so we can kill the handler off during detach. Use AHC_* constants instead of hard coded numbers in a few more places. Test PPR option state when deciding to "really" negotiate when the CAM_NEGOTIATE flag is passed in a CCB. Make use of core "ahc_pause_and_flushwork" routine in our timeout handler rather than re-inventing this code. Cleanup all of our resources (really!) in ahc_platform_free(). We should be all set to become a module now. Implement the core ahc_detach() routine shared by all of the FreeBSD front-ends. aic7xxx_freebsd.h: Softc storage for our event handler. Null implementation for the ahc_platform_flushwork() OSM callback. FreeBSD doesn't need this as XPT callbacks are safe from all contexts and are done directly in ahc_done(). aic7xxx_inline.h: Implement new lazy interrupt scheme. To avoid an extra PCI bus read, we first check our completion queues to see if any work has completed. If work is available, we assume that this is the source of the interrupt and skip reading INTSTAT. Any remaining interrupt status will be cleared by a second call to the interrupt handler should the interrupt line still be asserted. This drops the interrupt handler down to a single PCI bus read in the common case of I/O completion. This is the same overhead as in the not so distant past, but the extra sanity of perforning a PCI read after clearing the command complete interrupt and before running the completion queue to avoid missing command complete interrupts added a cycle. aic7xxx.c: During initialization, be sure to initialize all scratch ram locations before they are read to avoid parity errors. In this case, we use a new function, ahc_unbusy_tcl() to initialize the scratch ram busy target table. Replace instances of ahc_index_busy_tcl() used to unbusy a tcl without looking at the old value with ahc_unbusy_tcl(). Modify ahc_sent_msg so that it can find single byte messages. ahc_sent_msg is now used to determine if a transfer negotiation attempt resulted in a bus free. Be more careful in filtering out only the SCSI interrupts of interest in ahc_handle_scsiint. Rearrange interrupt clearing code to ensure that at least one PCI transaction occurrs after hitting CLRSINT1 and writting to CLRINT. CLRSINT1 writes take a bit to take effect, and the re-arrangement provides sufficient delay to ensure the write to CLRINT is effective. The old code might report a spurious interrupt on some "fast" chipsets. export ahc-update_target_msg_request for use by OSM code. If a target does not respond to our ATN request, clear it once we move to a non-message phase. This avoids sending a MSG_NOOP in some later message out phase. Use max lun and max target constants instead of hard-coded values. Use softc storage built into our device_t under FreeBSD. Fix a bug in ahc_free() that caused us to delete resources that were not allocated. Clean up any tstate/lstate info in ahc_free(). Clear the powerdown state in ahc_reset() so that registers can be accessed. Add a preliminary function for pausing the chip and processing any posted work. Add a preliminary suspend and resume functions. aic7xxx.h: Limit the number of supported luns to 64. We don't support information unit transfers, so this is the maximum that makes sense for these chips. Add a new flag AHC_ALL_INTERRUPTS that forces the processing of all interrupt state in a single invokation of ahc_intr(). When the flag is not set, we use the lazy interrupt handling scheme. Add data structures to store controller state while we are suspended. Use constants instead of hard coded values where appropriate. Correct some harmless "unsigned/signed" conflicts. aic7xxx.seq: Only perform the SCSIBUSL fix on ULTRA2 or newer controllers. Older controllers seem to be confused by this. In target mode, ignore PHASEMIS during data phases. This bit seems to be flakey on U160 controllers acting in target mode. aic7xxx_pci.c: Add support for the 29160C CPCI adapter. Add definitions for subvendor ID information available for devices with the "9005" vendor id. We currently use this information to determine if a multi-function device doesn't have the second channel hooked up on a board. Add rudimentary power mode code so we can put the controller into the D0 state. In the future this will be an OSM callback so that in FreeBSD we don't duplicate functionality provided by the PCI code. The powerstate code was added after I'd completed my regression tests on this code. Only capture "left over BIOS state" if the POWRDN setting is not set in HCNTRL. In target mode, don't bother sending incremental CRC data.
2000-12-20 01:11:37 +00:00
if ((ahc->flags & AHC_TARGETROLE) != 0) {
/*
* As a target, we control the phases,
* so ignore PHASEMIS.
*/
test SSTAT0, TARGET jnz ultra2_dma_loop;
}
if ((ahc->flags & AHC_INITIATORROLE) != 0) {
test SSTAT1,PHASEMIS jz ultra2_dma_loop;
}
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
ultra2_dmafinish:
test DFCNTRL, DIRECTION jnz ultra2_dmafifoempty;
aic7xxx.c: Style nits. Make sure that our selection hardware is disabled as soon as possible after detecting a busfree and even go so far as to disable the selection hardware in advance of an event that will cause a busfree (ABORT or BUS DEVICE RESET message). The concern is that the selection hardware will select a target for which, after processing the bus free, there will be no commands pending. The sequencer idle loop will re-enable the selection should it still be necessary. In ahc_handle_scsiint(), clear SSTAT0 events several PCI transactions (most notably reads) prior to clearing SCSIINT. The newer chips seem to take a bit of time to see the change which can make the clearing of SCSIINT ineffective. Don't bother panicing at the end of ahc_handle_scsiint(). Getting to the final else just means we lost the race with clearing SCSIINT. In ahc_free(), handle init-level 0. This can happen when we fail the attach for RAID devices. While I'm here, also kill the parent dma tag. In ahc_match_scb(), consider initiator ccbs to be any that are not from the target mode group. This fixes a bug where an external target reset CCB was not getting cleaned up by the reset code. Don't bother freezing a ccb in any of our "abort" routines when the status is set to CAM_REQ_CMP. This can happen for a target reset ccb. aic7xxx.reg: Reserve space for a completion queue. This will be used to enhance performance in the near future. aic7xxx.seq: Remove an optimization for the 7890 autoflush bug that turned out to allow, in rare cases, some data to get lost. Implement a simpler, faster, fix for the PCI_2_1 retry bug that hangs the sequencer on an SCB dma for certain chips. Test against SAVED_SCSIID rather than SELID during target reselections. This is how we always did it in the past, but the code was modified while trying to work around an issue with the 7895. SAVED_SCSIID takes into account twin channel adapters such as the 2742T, whereas SELID does not have the channel bit. This caused invalid selection warnings and other strangeness on these cards. aic7xxx_pci.c Use the correct mask for checking the generic aic7892 entry.
2001-02-10 18:04:27 +00:00
if ((ahc->features & AHC_DT) == 0) {
and DFCNTRL, ~SCSIEN;
test DFCNTRL, SCSIEN jnz .;
ahc_pci.c: Disable "cache line streaming" for aic7890/91 Rev A chips. I have never seen these chips fail using this feature, but some of Adaptec's regression tests have. Explicitly set "cache line streaming" to on for aic7896/97 chips. This was happening before, but this documents the fact that these chips will not function correctly without CACHETHEEN set. aic7xxx.h: Add new bug types. Fix a typo in a comment. aic7xxx.reg: Add a definition for the SHVALID bit in SSTAT3 for Ultra2/3 chips. This bit inicates whether the bottom most (current) element in the S/G fifo has exhausted its data count. aic7xxx.seq: Be more careful in how we turn off the secondary DMA channel. Being less careful may hang the PCI bus arbitor that negotiates between the two DMA engines. Remove an unecessary and incorrect flag set operation in the overrun case. On Ultra2/3 controllers, clear the dma FIFO before starting to handle an overrun. We don't want any residual bytes from the beginning of the overrun to cause the code that shuts down the DMA engine from hanging because the FIFO is not (and never will be) empty. If the data fifo is empty by the time we notice that a read transaction has completed, there is no need to hit the flush bit on aic7890/91 hardware that will not perform an auto-flush. Skip some cycles by short circuiting the manual flush code in this case. When transitioning out of data phase, make sure that we have the next S/G element loaded for the following reconnect if there is more work to do. The code would do this in most cases before, but there was a small window where the current S/G element could be exhausted before our fetch of the next S/G element completed. Since the S/G fetch is already initiated at this point, it makes sense to just wait for the segment to arrive instead of incuring even more latency by canceling the fetch and initiating it later. Fast path the end of data phase handling for the last S/G segment. In the general case, we might have worked ahead a bit by stuffing the S/G FIFO with additional segments. If we stop before using them all, we need to fixup our location in the S/G stream. Since we can't work past the last S/G segment, no fixups are ever required if we stop somewhere in that final segment. Fix a little buglet in the target mode dma bug handler. We were employing the workaround in all cases instead of only for the chips that require it. Fix the cause of SCB timeouts and possible "lost data" during read operations on the aic7890. When sending a data on any Ultra2/3 controller, the final segment must be marked as such so the FIFO will be flushed and cleaned up correctly when the transfer is ended. We failed to do this for the CDB transfer and so, if the target immediately transfered from command to data phase without an intervening disconnection, the first segment transferred would be any residual bytes from the cdb transfer. The Ultra160 controllers for some reason were not affected by this problem. Many Thanks to Tor Egge for bringing the aic7890 problem to my attention, providing analysis, as well as a mechanism to reproduce the problem.
2000-07-27 23:17:52 +00:00
}
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
ultra2_dmafifoflush:
if ((ahc->bugs & AHC_AUTOFLUSH_BUG) != 0) {
/*
* On Rev A of the aic7890, the autoflush
* features doesn't function correctly.
* Perform an explicit manual flush. During
* a manual flush, the FIFOEMP bit becomes
* true every time the PCI FIFO empties
* regardless of the state of the SCSI FIFO.
* It can take up to 4 clock cycles for the
* SCSI FIFO to get data into the PCI FIFO
* and for FIFOEMP to de-assert. Here we
* guard against this condition by making
* sure the FIFOEMP bit stays on for 5 full
* clock cycles.
*/
or DFCNTRL, FIFOFLUSH;
test DFSTATUS, FIFOEMP jz ultra2_dmafifoflush;
test DFSTATUS, FIFOEMP jz ultra2_dmafifoflush;
test DFSTATUS, FIFOEMP jz ultra2_dmafifoflush;
test DFSTATUS, FIFOEMP jz ultra2_dmafifoflush;
}
test DFSTATUS, FIFOEMP jz ultra2_dmafifoflush;
ultra2_dmafifoempty:
/* Don't clobber an inprogress host data transfer */
test DFSTATUS, MREQPEND jnz ultra2_dmafifoempty;
ultra2_dmahalt:
and DFCNTRL, ~(SCSIEN|HDMAEN);
test DFCNTRL, HDMAEN jnz .;
ahc_pci.c: Disable "cache line streaming" for aic7890/91 Rev A chips. I have never seen these chips fail using this feature, but some of Adaptec's regression tests have. Explicitly set "cache line streaming" to on for aic7896/97 chips. This was happening before, but this documents the fact that these chips will not function correctly without CACHETHEEN set. aic7xxx.h: Add new bug types. Fix a typo in a comment. aic7xxx.reg: Add a definition for the SHVALID bit in SSTAT3 for Ultra2/3 chips. This bit inicates whether the bottom most (current) element in the S/G fifo has exhausted its data count. aic7xxx.seq: Be more careful in how we turn off the secondary DMA channel. Being less careful may hang the PCI bus arbitor that negotiates between the two DMA engines. Remove an unecessary and incorrect flag set operation in the overrun case. On Ultra2/3 controllers, clear the dma FIFO before starting to handle an overrun. We don't want any residual bytes from the beginning of the overrun to cause the code that shuts down the DMA engine from hanging because the FIFO is not (and never will be) empty. If the data fifo is empty by the time we notice that a read transaction has completed, there is no need to hit the flush bit on aic7890/91 hardware that will not perform an auto-flush. Skip some cycles by short circuiting the manual flush code in this case. When transitioning out of data phase, make sure that we have the next S/G element loaded for the following reconnect if there is more work to do. The code would do this in most cases before, but there was a small window where the current S/G element could be exhausted before our fetch of the next S/G element completed. Since the S/G fetch is already initiated at this point, it makes sense to just wait for the segment to arrive instead of incuring even more latency by canceling the fetch and initiating it later. Fast path the end of data phase handling for the last S/G segment. In the general case, we might have worked ahead a bit by stuffing the S/G FIFO with additional segments. If we stop before using them all, we need to fixup our location in the S/G stream. Since we can't work past the last S/G segment, no fixups are ever required if we stop somewhere in that final segment. Fix a little buglet in the target mode dma bug handler. We were employing the workaround in all cases instead of only for the chips that require it. Fix the cause of SCB timeouts and possible "lost data" during read operations on the aic7890. When sending a data on any Ultra2/3 controller, the final segment must be marked as such so the FIFO will be flushed and cleaned up correctly when the transfer is ended. We failed to do this for the CDB transfer and so, if the target immediately transfered from command to data phase without an intervening disconnection, the first segment transferred would be any residual bytes from the cdb transfer. The Ultra160 controllers for some reason were not affected by this problem. Many Thanks to Tor Egge for bringing the aic7890 problem to my attention, providing analysis, as well as a mechanism to reproduce the problem.
2000-07-27 23:17:52 +00:00
/*
* 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. Just jump to data_phase_finish.
*/
ultra2_ensure_sg:
test SG_CACHE_SHADOW, LAST_SEG jz ultra2_shvalid;
/* Record if we've consumed all S/G entries */
test SG_CACHE_SHADOW, LAST_SEG_DONE jz data_phase_finish;
or SCB_RESIDUAL_SGPTR[0], SG_LIST_NULL;
jmp data_phase_finish;
ultra2_shvalid:
test SSTAT2, SHVALID jnz sgptr_fixup;
call idle_loop;
jmp ultra2_ensure_sg;
sgptr_fixup:
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
/*
* 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_CACEPTR 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 DATA_COUNT_ODD;
test SG_CACHE_SHADOW, ODD_SEG jz . + 2;
or DATA_COUNT_ODD, 0x1;
ahc_pci.c: Disable "cache line streaming" for aic7890/91 Rev A chips. I have never seen these chips fail using this feature, but some of Adaptec's regression tests have. Explicitly set "cache line streaming" to on for aic7896/97 chips. This was happening before, but this documents the fact that these chips will not function correctly without CACHETHEEN set. aic7xxx.h: Add new bug types. Fix a typo in a comment. aic7xxx.reg: Add a definition for the SHVALID bit in SSTAT3 for Ultra2/3 chips. This bit inicates whether the bottom most (current) element in the S/G fifo has exhausted its data count. aic7xxx.seq: Be more careful in how we turn off the secondary DMA channel. Being less careful may hang the PCI bus arbitor that negotiates between the two DMA engines. Remove an unecessary and incorrect flag set operation in the overrun case. On Ultra2/3 controllers, clear the dma FIFO before starting to handle an overrun. We don't want any residual bytes from the beginning of the overrun to cause the code that shuts down the DMA engine from hanging because the FIFO is not (and never will be) empty. If the data fifo is empty by the time we notice that a read transaction has completed, there is no need to hit the flush bit on aic7890/91 hardware that will not perform an auto-flush. Skip some cycles by short circuiting the manual flush code in this case. When transitioning out of data phase, make sure that we have the next S/G element loaded for the following reconnect if there is more work to do. The code would do this in most cases before, but there was a small window where the current S/G element could be exhausted before our fetch of the next S/G element completed. Since the S/G fetch is already initiated at this point, it makes sense to just wait for the segment to arrive instead of incuring even more latency by canceling the fetch and initiating it later. Fast path the end of data phase handling for the last S/G segment. In the general case, we might have worked ahead a bit by stuffing the S/G FIFO with additional segments. If we stop before using them all, we need to fixup our location in the S/G stream. Since we can't work past the last S/G segment, no fixups are ever required if we stop somewhere in that final segment. Fix a little buglet in the target mode dma bug handler. We were employing the workaround in all cases instead of only for the chips that require it. Fix the cause of SCB timeouts and possible "lost data" during read operations on the aic7890. When sending a data on any Ultra2/3 controller, the final segment must be marked as such so the FIFO will be flushed and cleaned up correctly when the transfer is ended. We failed to do this for the CDB transfer and so, if the target immediately transfered from command to data phase without an intervening disconnection, the first segment transferred would be any residual bytes from the cdb transfer. The Ultra160 controllers for some reason were not affected by this problem. Many Thanks to Tor Egge for bringing the aic7890 problem to my attention, providing analysis, as well as a mechanism to reproduce the problem.
2000-07-27 23:17:52 +00:00
clr SCB_RESIDUAL_DATACNT[3]; /* We are not the last seg */
} else {
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
/* If we are the last SG block, tell the hardware. */
if ((ahc->bugs & AHC_PCI_MWI_BUG) != 0
&& ahc->pci_cachesize != 0) {
test MWI_RESIDUAL, 0xFF jnz dma_mid_sg;
}
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
test SCB_RESIDUAL_DATACNT[3], SG_LAST_SEG jz dma_mid_sg;
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
if ((ahc->flags & AHC_TARGETROLE) != 0) {
ahc_pci.c: Disable "cache line streaming" for aic7890/91 Rev A chips. I have never seen these chips fail using this feature, but some of Adaptec's regression tests have. Explicitly set "cache line streaming" to on for aic7896/97 chips. This was happening before, but this documents the fact that these chips will not function correctly without CACHETHEEN set. aic7xxx.h: Add new bug types. Fix a typo in a comment. aic7xxx.reg: Add a definition for the SHVALID bit in SSTAT3 for Ultra2/3 chips. This bit inicates whether the bottom most (current) element in the S/G fifo has exhausted its data count. aic7xxx.seq: Be more careful in how we turn off the secondary DMA channel. Being less careful may hang the PCI bus arbitor that negotiates between the two DMA engines. Remove an unecessary and incorrect flag set operation in the overrun case. On Ultra2/3 controllers, clear the dma FIFO before starting to handle an overrun. We don't want any residual bytes from the beginning of the overrun to cause the code that shuts down the DMA engine from hanging because the FIFO is not (and never will be) empty. If the data fifo is empty by the time we notice that a read transaction has completed, there is no need to hit the flush bit on aic7890/91 hardware that will not perform an auto-flush. Skip some cycles by short circuiting the manual flush code in this case. When transitioning out of data phase, make sure that we have the next S/G element loaded for the following reconnect if there is more work to do. The code would do this in most cases before, but there was a small window where the current S/G element could be exhausted before our fetch of the next S/G element completed. Since the S/G fetch is already initiated at this point, it makes sense to just wait for the segment to arrive instead of incuring even more latency by canceling the fetch and initiating it later. Fast path the end of data phase handling for the last S/G segment. In the general case, we might have worked ahead a bit by stuffing the S/G FIFO with additional segments. If we stop before using them all, we need to fixup our location in the S/G stream. Since we can't work past the last S/G segment, no fixups are ever required if we stop somewhere in that final segment. Fix a little buglet in the target mode dma bug handler. We were employing the workaround in all cases instead of only for the chips that require it. Fix the cause of SCB timeouts and possible "lost data" during read operations on the aic7890. When sending a data on any Ultra2/3 controller, the final segment must be marked as such so the FIFO will be flushed and cleaned up correctly when the transfer is ended. We failed to do this for the CDB transfer and so, if the target immediately transfered from command to data phase without an intervening disconnection, the first segment transferred would be any residual bytes from the cdb transfer. The Ultra160 controllers for some reason were not affected by this problem. Many Thanks to Tor Egge for bringing the aic7890 problem to my attention, providing analysis, as well as a mechanism to reproduce the problem.
2000-07-27 23:17:52 +00:00
test SSTAT0, TARGET jz dma_last_sg;
if ((ahc->flags & AHC_TMODE_WIDEODD_BUG) != 0) {
test DMAPARAMS, DIRECTION jz dma_mid_sg;
}
Kill the "unpause_always" argument to unpause_sequencer(). The reasons for optimizing the unpause operation no-longer exist, and this is much safer. When restarting the sequencer, reconstitute the free SCB list on the card. This deals with a single instruction gap between marking the SCB as free and actually getting it onto the free list. Reduce the number of transfer negotiations that occur. In the past, we renegotiated after every reported check condition status. This ensures that we catch devices that have unexpectidly reset. In this situation, the target will always report the check condition before performing a data-phase. The new behavior is to renegotiate for any check-condition where the residual matches the orginal data-length of the command (including 0 length transffers). This avoids renegotiations during things like variable tape block reads, where the check condition is reported only to indicate the residual of the read. Revamp the parity error detection logic. We now properly report and handle injected parity errors in all phases. The old code used to hang on message-in parity errors. Correct the reporting of selection timeout errors to the XPT. When a selection timeout occurs, only the currently selecting command is flagged with SELTO status instead of aborting all currently active commands to that target. Fix flipped arguments in ahc_match_scb and in some of the callers of this routine. I wish that gcc allowed you to request warnings for enums passed as ints. Make ahc_find_msg generically handle all message types. Work around the target mode data-in wideodd bug in all non-U2 chips. We can now do sync-wide target mode transfers in target mode across the hole product line. Use lastphase exclusively for handling timeouts. The current phase doesn't take the bus free state into account. Fix a bug in the timeout handler that could cause corruption of the disconnected list. When sending an embedded cdb to a target, ensure that we start on a quad word boundary in the data-fifo. It seems that unaligned stores do not work correctly.
2000-02-09 21:25:00 +00:00
}
ahc_pci.c: Disable "cache line streaming" for aic7890/91 Rev A chips. I have never seen these chips fail using this feature, but some of Adaptec's regression tests have. Explicitly set "cache line streaming" to on for aic7896/97 chips. This was happening before, but this documents the fact that these chips will not function correctly without CACHETHEEN set. aic7xxx.h: Add new bug types. Fix a typo in a comment. aic7xxx.reg: Add a definition for the SHVALID bit in SSTAT3 for Ultra2/3 chips. This bit inicates whether the bottom most (current) element in the S/G fifo has exhausted its data count. aic7xxx.seq: Be more careful in how we turn off the secondary DMA channel. Being less careful may hang the PCI bus arbitor that negotiates between the two DMA engines. Remove an unecessary and incorrect flag set operation in the overrun case. On Ultra2/3 controllers, clear the dma FIFO before starting to handle an overrun. We don't want any residual bytes from the beginning of the overrun to cause the code that shuts down the DMA engine from hanging because the FIFO is not (and never will be) empty. If the data fifo is empty by the time we notice that a read transaction has completed, there is no need to hit the flush bit on aic7890/91 hardware that will not perform an auto-flush. Skip some cycles by short circuiting the manual flush code in this case. When transitioning out of data phase, make sure that we have the next S/G element loaded for the following reconnect if there is more work to do. The code would do this in most cases before, but there was a small window where the current S/G element could be exhausted before our fetch of the next S/G element completed. Since the S/G fetch is already initiated at this point, it makes sense to just wait for the segment to arrive instead of incuring even more latency by canceling the fetch and initiating it later. Fast path the end of data phase handling for the last S/G segment. In the general case, we might have worked ahead a bit by stuffing the S/G FIFO with additional segments. If we stop before using them all, we need to fixup our location in the S/G stream. Since we can't work past the last S/G segment, no fixups are ever required if we stop somewhere in that final segment. Fix a little buglet in the target mode dma bug handler. We were employing the workaround in all cases instead of only for the chips that require it. Fix the cause of SCB timeouts and possible "lost data" during read operations on the aic7890. When sending a data on any Ultra2/3 controller, the final segment must be marked as such so the FIFO will be flushed and cleaned up correctly when the transfer is ended. We failed to do this for the CDB transfer and so, if the target immediately transfered from command to data phase without an intervening disconnection, the first segment transferred would be any residual bytes from the cdb transfer. The Ultra160 controllers for some reason were not affected by this problem. Many Thanks to Tor Egge for bringing the aic7890 problem to my attention, providing analysis, as well as a mechanism to reproduce the problem.
2000-07-27 23:17:52 +00:00
dma_last_sg:
and DMAPARAMS, ~WIDEODD;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
dma_mid_sg:
/* Start DMA data transfer. */
mov DFCNTRL, DMAPARAMS;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
dma_loop:
if ((ahc->features & AHC_CMD_CHAN) != 0) {
call idle_loop;
}
test SSTAT0,DMADONE jnz dma_dmadone;
test SSTAT1,PHASEMIS jz dma_loop; /* ie. underrun */
dma_phasemis:
/*
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
* We will be "done" DMAing when the transfer count goes to
* zero, or the target changes the phase (in light of this,
* it makes sense that the DMA circuitry doesn't ACK when
* PHASEMIS is active). If we are doing a SCSI->Host transfer,
* the data FIFO should be flushed auto-magically on STCNT=0
* or a phase change, so just wait for FIFO empty status.
*/
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
dma_checkfifo:
test DFCNTRL,DIRECTION jnz dma_fifoempty;
dma_fifoflush:
test DFSTATUS,FIFOEMP jz dma_fifoflush;
dma_fifoempty:
/* Don't clobber an inprogress host data transfer */
test DFSTATUS, MREQPEND jnz dma_fifoempty;
/*
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
* Now shut off the DMA and make sure that the DMA
* hardware has actually stopped. Touching the DMA
* counters, etc. while a DMA is active will result
* in an ILLSADDR exception.
*/
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
dma_dmadone:
and DFCNTRL, ~(SCSIEN|SDMAEN|HDMAEN);
dma_halt:
/*
* Some revisions of the aic78XX have a problem where, if the
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
* data fifo is full, but the PCI input latch is not empty,
* HDMAEN cannot be cleared. The fix used here is to drain
* the prefetched but unused data from the data fifo until
* there is space for the input latch to drain.
*/
if ((ahc->bugs & AHC_PCI_2_1_RETRY_BUG) != 0) {
mov NONE, DFDAT;
}
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
test DFCNTRL, (SCSIEN|SDMAEN|HDMAEN) jnz dma_halt;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
/* See if we have completed this last segment */
test STCNT[0], 0xff jnz data_phase_finish;
test STCNT[1], 0xff jnz data_phase_finish;
test STCNT[2], 0xff jnz data_phase_finish;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
/*
* Advance the scatter-gather pointers if needed
*/
if ((ahc->bugs & AHC_PCI_MWI_BUG) != 0
&& ahc->pci_cachesize != 0) {
test MWI_RESIDUAL, 0xFF jz no_mwi_resid;
/*
* Reload HADDR from SHADDR and setup the
* count to be the size of our residual.
*/
if ((ahc->features & AHC_CMD_CHAN) != 0) {
bmov HADDR, SHADDR, 4;
mov HCNT, MWI_RESIDUAL;
bmov HCNT[1], ALLZEROS, 2;
} else {
mvi DINDEX, HADDR;
mvi SHADDR call bcopy_4;
mov MWI_RESIDUAL call set_hcnt;
}
clr MWI_RESIDUAL;
jmp sg_load_done;
no_mwi_resid:
}
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
test SCB_RESIDUAL_DATACNT[3], SG_LAST_SEG jz sg_load;
or SCB_RESIDUAL_SGPTR[0], SG_LIST_NULL;
jmp data_phase_finish;
sg_load:
/*
* Load the next SG element's data address and length
* into the DMA engine. If we don't have hardware
* to perform a prefetch, we'll have to fetch the
* segment from host memory first.
*/
if ((ahc->features & AHC_CMD_CHAN) != 0) {
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
/* Wait for the idle loop to complete */
test CCSGCTL, CCSGEN jz . + 3;
call idle_loop;
test CCSGCTL, CCSGEN jnz . - 1;
bmov HADDR, CCSGRAM, 7;
test CCSGRAM, SG_LAST_SEG jz . + 2;
or SCB_RESIDUAL_DATACNT[3], SG_LAST_SEG;
} else {
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
mvi DINDEX, HADDR;
mvi SCB_RESIDUAL_SGPTR call bcopy_4;
mvi SG_SIZEOF call set_hcnt;
or DFCNTRL, HDMAEN|DIRECTION|FIFORESET;
call dma_finish;
mvi DINDEX, HADDR;
call dfdat_in_7;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
mov SCB_RESIDUAL_DATACNT[3], DFDAT;
}
if ((ahc->bugs & AHC_PCI_MWI_BUG) != 0
&& ahc->pci_cachesize != 0) {
call calc_mwi_residual;
}
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
/* Point to the new next sg in memory */
call sg_advance;
sg_load_done:
if ((ahc->features & AHC_CMD_CHAN) != 0) {
bmov STCNT, HCNT, 3;
} else {
call set_stcnt_from_hcnt;
}
/* Track odd'ness */
test HCNT[0], 0x1 jz . + 2;
xor DATA_COUNT_ODD, 0x1;
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
if ((ahc->flags & AHC_TARGETROLE) != 0) {
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
test SSTAT0, TARGET jnz data_phase_loop;
}
}
data_phase_finish:
/*
* If the target has left us in data phase, loop through
* the dma code again. In the case of ULTRA2 adapters,
* we should only loop if there is a data overrun. For
* all other adapters, we'll loop after each S/G element
* is loaded as well as if there is an overrun.
*/
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
if ((ahc->flags & AHC_TARGETROLE) != 0) {
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
test SSTAT0, TARGET jnz data_phase_done;
}
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
if ((ahc->flags & AHC_INITIATORROLE) != 0) {
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
test SSTAT1, REQINIT jz .;
test SSTAT1,PHASEMIS jz data_phase_loop;
if ((ahc->features & AHC_CMD_CHAN) != 0) {
/* Kill off any pending prefetch */
clr CCSGCTL;
test CCSGCTL, CCSGEN jnz .;
}
}
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
data_phase_done:
/*
* After a DMA finishes, save the SG and STCNT residuals back into
* the SCB. We use STCNT instead of HCNT, since it's a reflection
* of how many bytes were transferred on the SCSI (as opposed to the
* host) bus.
*/
if ((ahc->features & AHC_CMD_CHAN) != 0) {
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
/* Kill off any pending prefetch */
clr CCSGCTL;
test CCSGCTL, CCSGEN jnz .;
}
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
if ((ahc->bugs & AHC_PCI_MWI_BUG) != 0
&& ahc->pci_cachesize != 0) {
if ((ahc->features & AHC_CMD_CHAN) != 0) {
test MWI_RESIDUAL, 0xFF jz bmov_resid;
}
mov A, MWI_RESIDUAL;
add SCB_RESIDUAL_DATACNT[0], A, STCNT[0];
clr A;
adc SCB_RESIDUAL_DATACNT[1], A, STCNT[1];
adc SCB_RESIDUAL_DATACNT[2], A, STCNT[2];
clr MWI_RESIDUAL;
if ((ahc->features & AHC_CMD_CHAN) != 0) {
jmp . + 2;
bmov_resid:
bmov SCB_RESIDUAL_DATACNT, STCNT, 3;
}
} else if ((ahc->features & AHC_CMD_CHAN) != 0) {
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
bmov SCB_RESIDUAL_DATACNT, STCNT, 3;
} else {
mov SCB_RESIDUAL_DATACNT[0], STCNT[0];
mov SCB_RESIDUAL_DATACNT[1], STCNT[1];
mov SCB_RESIDUAL_DATACNT[2], STCNT[2];
}
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
/*
* Since we've been through a data phase, the SCB_RESID* fields
* are now initialized. Clear the full residual flag.
*/
and SCB_SGPTR[0], ~SG_FULL_RESID;
if ((ahc->features & AHC_ULTRA2) != 0) {
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
/* Clear the channel in case we return to data phase later */
or SXFRCTL0, CLRSTCNT|CLRCHN;
aic7xxx.c: When restarting the sequencer, ensure that the SCBCNT register is 0. A non-zero count will prevent the setting of the CCSCBDIR bit in any future dma operations. The only time CCSCBCNT would be non-zero is if we happened to halt the dma during a reset, but even that should never happen. Better safe than sorry. When a command completes before the target responds to an ATN for a recovery command, we now notify the kernel so that any recovery operation requeued in the qinfifo can be removed safely. In the past, we did this in ahc_done(), but ahc_done() may be called without the card paused. This also avoids a recursive call to ahc_search_qinifo() which could have occurred if ahc_search_qinififo() happened to be the routine to complete a recovery action. Fix 8bit math used for adjusting the qinfifo. The index must be wrapped properly within the 256 entry array. We rely on the fact that qinfifonext is a uint8_t in most cases to handle this wrap, but we missed a few spots where the resultant calculation was promoted to an int. Change the way that we deal with aborting the first or second entry from the qinfifo. We now swap the first entry in the qinfifo with the "next queued scb" to force the sequencer to see an abort collision if we ever touch the qinififo while the sequencer is mid SCB dma. aic7xxx.reg: Add new MKMSG_FAILED sequencer interrupt. This displaced the BOGUS_TAG interrupt used in some previous sequencer code debugging. aic7xxx.seq: Increment our position in the qinfifo only once the dma is complete and we have verified that the queue has not been changed during our DMA. This simplifies code in the kernel. Protect against "instruction creep" when issuing a pausing sequencer interrupt. On at least the 7890/91/96/97, the sequencer will coast after issuing the interrupt for up to two instructions. In the past we delt with this by using carefully placed nops. Now we call a routine to issue the interrupt followed by a nop and a ret. Tell the kernel should an SCB complete with the MK_MESSAGE flag still set. This means the target ignored our ATN request. Clear the channel twice as we exit the data phase. On the aic7890/91, the S/G preload logic may require the second clearing to get the last S/G out of the FIFO. aic7xxx_freebsd.c: Don't bother searching the qinfifo for a doubly queued recovery scb in ahc_done. This case is handled by the core driver now. Free the path used to issue async callbacks after the callback is complete. aic7xxx_inline.h: Split the SCB queue routine into a routine that swaps the SCB with the "next queued SCB" and a routine that calls the swapping routine and notifies the card of the new SCB. The swapping routine is now also used by ahc_search_qinfifo.
2000-11-06 20:05:38 +00:00
or SXFRCTL0, CLRSTCNT|CLRCHN;
}
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
if ((ahc->flags & AHC_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.
*/
test DFCNTRL, DIRECTION jz target_ITloop;
test SSTAT1, REQINIT jnz .;
jmp target_ITloop;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
} else {
jmp ITloop;
}
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
if ((ahc->flags & AHC_INITIATORROLE) != 0) {
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
/*
* Command phase. Set up the DMA registers and let 'er rip.
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
*/
p_command:
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
call assert;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
if ((ahc->features & AHC_ULTRA2) != 0) {
bmov HCNT[0], SCB_CDB_LEN, 1;
bmov HCNT[1], ALLZEROS, 2;
ahc_pci.c: Disable "cache line streaming" for aic7890/91 Rev A chips. I have never seen these chips fail using this feature, but some of Adaptec's regression tests have. Explicitly set "cache line streaming" to on for aic7896/97 chips. This was happening before, but this documents the fact that these chips will not function correctly without CACHETHEEN set. aic7xxx.h: Add new bug types. Fix a typo in a comment. aic7xxx.reg: Add a definition for the SHVALID bit in SSTAT3 for Ultra2/3 chips. This bit inicates whether the bottom most (current) element in the S/G fifo has exhausted its data count. aic7xxx.seq: Be more careful in how we turn off the secondary DMA channel. Being less careful may hang the PCI bus arbitor that negotiates between the two DMA engines. Remove an unecessary and incorrect flag set operation in the overrun case. On Ultra2/3 controllers, clear the dma FIFO before starting to handle an overrun. We don't want any residual bytes from the beginning of the overrun to cause the code that shuts down the DMA engine from hanging because the FIFO is not (and never will be) empty. If the data fifo is empty by the time we notice that a read transaction has completed, there is no need to hit the flush bit on aic7890/91 hardware that will not perform an auto-flush. Skip some cycles by short circuiting the manual flush code in this case. When transitioning out of data phase, make sure that we have the next S/G element loaded for the following reconnect if there is more work to do. The code would do this in most cases before, but there was a small window where the current S/G element could be exhausted before our fetch of the next S/G element completed. Since the S/G fetch is already initiated at this point, it makes sense to just wait for the segment to arrive instead of incuring even more latency by canceling the fetch and initiating it later. Fast path the end of data phase handling for the last S/G segment. In the general case, we might have worked ahead a bit by stuffing the S/G FIFO with additional segments. If we stop before using them all, we need to fixup our location in the S/G stream. Since we can't work past the last S/G segment, no fixups are ever required if we stop somewhere in that final segment. Fix a little buglet in the target mode dma bug handler. We were employing the workaround in all cases instead of only for the chips that require it. Fix the cause of SCB timeouts and possible "lost data" during read operations on the aic7890. When sending a data on any Ultra2/3 controller, the final segment must be marked as such so the FIFO will be flushed and cleaned up correctly when the transfer is ended. We failed to do this for the CDB transfer and so, if the target immediately transfered from command to data phase without an intervening disconnection, the first segment transferred would be any residual bytes from the cdb transfer. The Ultra160 controllers for some reason were not affected by this problem. Many Thanks to Tor Egge for bringing the aic7890 problem to my attention, providing analysis, as well as a mechanism to reproduce the problem.
2000-07-27 23:17:52 +00:00
mvi SG_CACHE_PRE, LAST_SEG;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
} else if ((ahc->features & AHC_CMD_CHAN) != 0) {
bmov STCNT[0], SCB_CDB_LEN, 1;
bmov STCNT[1], ALLZEROS, 2;
} else {
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
mov STCNT[0], SCB_CDB_LEN;
clr STCNT[1];
clr STCNT[2];
}
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
add NONE, -13, SCB_CDB_LEN;
mvi SCB_CDB_STORE jnc p_command_embedded;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
p_command_from_host:
if ((ahc->features & AHC_ULTRA2) != 0) {
bmov HADDR[0], SCB_CDB_PTR, 4;
mvi DFCNTRL, (PRELOADEN|SCSIEN|HDMAEN|DIRECTION);
} else {
if ((ahc->features & AHC_CMD_CHAN) != 0) {
bmov HADDR[0], SCB_CDB_PTR, 4;
bmov HCNT, STCNT, 3;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
} else {
mvi DINDEX, HADDR;
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
mvi SCB_CDB_PTR call bcopy_4;
mov SCB_CDB_LEN call set_hcnt;
}
mvi DFCNTRL, (SCSIEN|SDMAEN|HDMAEN|DIRECTION|FIFORESET);
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
}
jmp p_command_loop;
p_command_embedded:
/*
* The data fifo seems to require 4 byte alligned
* transfers from the sequencer. Force this to
* be the case by clearing HADDR[0] even though
* we aren't going to touch host memeory.
*/
clr HADDR[0];
if ((ahc->features & AHC_ULTRA2) != 0) {
mvi DFCNTRL, (PRELOADEN|SCSIEN|DIRECTION);
bmov DFDAT, SCB_CDB_STORE, 12;
} else if ((ahc->features & AHC_CMD_CHAN) != 0) {
ahc_eisa.c: Initialize rid to 0. This doesn't seem to make any difference (the driver doesn't care what rid it gets and no-one seems to check rid's value), but follows standard conventions. Pass in our device_t to ahc_alloc(). We now use device_T softc storage, so passing NULL results in a panic. Set the unit number in our softc so that the driver core can retrieve it. ahc_pci.c: Set the unit number in our softc so that the driver core can retrieve it. aic7770.c: Insert our softc into the list of softcs when initialization is successful. aic7xxx.c: Remove a workaround for an aic7895 bug we will never trigger. Add additional diagnostic info to ahc_dump_card_state(). Always panic the system if a sequencer assertion fails. AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. Replace a hard coded number with a constant. Guard against looping forever in ahc_pause_and_flushwork(). A hot eject or card failure may make the intstat register return 0xFF, so limit the number of interrupts we'll process. Correct the code in ahc_search_qinfifo() that guarantees that the sequencer will see an abort collision if the qinfifo is modified when a DMA is in progress. We now do this fixup after modifying the queue. This guarantees that the HSCB we place at the head of the queue is not the same as the old head. Using "next hscb" (guaranteed not to be the same as the first SCB) before clearing the queue could free up the original head hscb to be used during a remove operation placing it again at the head of the qinfifo. aic7xxx.h: Reduce the maximum number of outstanding commands to 253 from 254. To handle our output queue correctly on machines that only support 32bit stores, we must clear the array 4 bytes at a time. To avoid colliding with a DMA write from the sequencer, we must be sure that 4 slots are empty when we write to clear the queue. This reduces us to 253 SCBs: 1 that just completed and the known three additional empty slots in the queue that preceed it. Yahoo was able to force this race on one of their systems. Interrupts were disabled for such a time that the entire output queue was filled (254 entries complete without any processing), and our 32bit write to clear the status clobbered one entry. Add a feature tag for devices that are removable. aic7xxx.reg: Never use the sequencer interrupt value of 0xF0. We need to guanrantee that an INTSTAT value of 0xFF can only occur during card failure or a hot-eject. Align the busy targets table with the begining of scratch space. This seems to appease a chip bug in the aic7895. aic7xxx.seq: Be sure to disable select-out after a bus free event that occurs early in a selection. If we don't disable select-out, we will believe that it is enabled even though a new selection will never occur. Move the clearing of SELDI to just before a jump. This appeases another chip bug of the aic7895. Make the target mode command loop a bit more efficient. AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. Properly cleanup the last SCB we tested against should we fail to properly find an SCB for a reselection. Add some additional sequencer debugging code. aic7xxx_freebsd.c: Limit the driver to 253 outstanding commands per adapter. Guard against overflow in timeout handling. aic7xxx_inline.h: AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. aic7xxx_pci.c: Set the removable feature for the apa1480 cardbus and the 29160C Compact PCI card. Don't report high byte termination information for narrow cards. Use a PCI read rather than a questionable delay when fetching/setting termination settings.
2001-01-22 21:03:48 +00:00
if ((ahc->flags & AHC_SCB_BTT) != 0) {
/*
* On the 7895 the data FIFO will
* get corrupted if you try to dump
* data from external SCB memory into
* the FIFO while it is enabled. So,
* fill the fifo and then enable SCSI
* transfers.
*/
mvi DFCNTRL, (DIRECTION|FIFORESET);
} else {
mvi DFCNTRL, (SCSIEN|SDMAEN|DIRECTION|FIFORESET);
}
bmov DFDAT, SCB_CDB_STORE, 12;
ahc_eisa.c: Initialize rid to 0. This doesn't seem to make any difference (the driver doesn't care what rid it gets and no-one seems to check rid's value), but follows standard conventions. Pass in our device_t to ahc_alloc(). We now use device_T softc storage, so passing NULL results in a panic. Set the unit number in our softc so that the driver core can retrieve it. ahc_pci.c: Set the unit number in our softc so that the driver core can retrieve it. aic7770.c: Insert our softc into the list of softcs when initialization is successful. aic7xxx.c: Remove a workaround for an aic7895 bug we will never trigger. Add additional diagnostic info to ahc_dump_card_state(). Always panic the system if a sequencer assertion fails. AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. Replace a hard coded number with a constant. Guard against looping forever in ahc_pause_and_flushwork(). A hot eject or card failure may make the intstat register return 0xFF, so limit the number of interrupts we'll process. Correct the code in ahc_search_qinfifo() that guarantees that the sequencer will see an abort collision if the qinfifo is modified when a DMA is in progress. We now do this fixup after modifying the queue. This guarantees that the HSCB we place at the head of the queue is not the same as the old head. Using "next hscb" (guaranteed not to be the same as the first SCB) before clearing the queue could free up the original head hscb to be used during a remove operation placing it again at the head of the qinfifo. aic7xxx.h: Reduce the maximum number of outstanding commands to 253 from 254. To handle our output queue correctly on machines that only support 32bit stores, we must clear the array 4 bytes at a time. To avoid colliding with a DMA write from the sequencer, we must be sure that 4 slots are empty when we write to clear the queue. This reduces us to 253 SCBs: 1 that just completed and the known three additional empty slots in the queue that preceed it. Yahoo was able to force this race on one of their systems. Interrupts were disabled for such a time that the entire output queue was filled (254 entries complete without any processing), and our 32bit write to clear the status clobbered one entry. Add a feature tag for devices that are removable. aic7xxx.reg: Never use the sequencer interrupt value of 0xF0. We need to guanrantee that an INTSTAT value of 0xFF can only occur during card failure or a hot-eject. Align the busy targets table with the begining of scratch space. This seems to appease a chip bug in the aic7895. aic7xxx.seq: Be sure to disable select-out after a bus free event that occurs early in a selection. If we don't disable select-out, we will believe that it is enabled even though a new selection will never occur. Move the clearing of SELDI to just before a jump. This appeases another chip bug of the aic7895. Make the target mode command loop a bit more efficient. AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. Properly cleanup the last SCB we tested against should we fail to properly find an SCB for a reselection. Add some additional sequencer debugging code. aic7xxx_freebsd.c: Limit the driver to 253 outstanding commands per adapter. Guard against overflow in timeout handling. aic7xxx_inline.h: AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. aic7xxx_pci.c: Set the removable feature for the apa1480 cardbus and the 29160C Compact PCI card. Don't report high byte termination information for narrow cards. Use a PCI read rather than a questionable delay when fetching/setting termination settings.
2001-01-22 21:03:48 +00:00
if ((ahc->flags & AHC_SCB_BTT) != 0) {
mvi DFCNTRL, (SCSIEN|SDMAEN|DIRECTION|FIFOFLUSH);
} else {
aic7xxx.c: Correct the BUILD_TCL macro. It was placing the target id in the wrong bits. This was only an issue for adapters that do not perform SCB paging (aha-3940AUW for instance). Don't bother inlining ahc_index_busy_tcl. It is never used in a performance critical path and is a bit chunky. Correct ahc_index_busy_tcl to deal with "busy target tables" embedded in the latter half of 64byte SCBs. Don't initialize the busy target table to its empty state until after we have finished extracting configuration information from chip SRAM. In the common case of using 16 bytes of chip SRAM to do untagged target lookups, we were trashing the last 8 targets configuration data. (actually only target 8 because of the bug in the BUILD_TCL macro). Cram the "bus reset delivered" message back under bootverbose. Fix the cleanup of the SCB busy target table when aborting commands. If the lun is wildcarded, we must loop through all possible luns. aic7xxx.h: Only bother supporting 64 luns right now. It doesn't seem like either this driver or any peripherals will be doing information unit transfers (where the lun number is a 32 bit integer) any time soon. aic7xxx.seq: Fix support for the aic7895. We must flush the data FIFO if performing a manual transfer that is not a multiple of 8 bytes. We were doing this quite regularly for embedded cdbs. Manaually flush the fifo on earlier adapters when dealing with embedded cdbs too. We were stuffing the FIFO with 16 bytes instead, but triggering the flush is more efficient and allows us to remove two instructions from the "copy_to_fifo" routine.
2000-07-24 22:27:40 +00:00
or DFCNTRL, FIFOFLUSH;
}
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
} else {
mvi DFCNTRL, (SCSIEN|SDMAEN|DIRECTION|FIFORESET);
aic7xxx.c: Correct the BUILD_TCL macro. It was placing the target id in the wrong bits. This was only an issue for adapters that do not perform SCB paging (aha-3940AUW for instance). Don't bother inlining ahc_index_busy_tcl. It is never used in a performance critical path and is a bit chunky. Correct ahc_index_busy_tcl to deal with "busy target tables" embedded in the latter half of 64byte SCBs. Don't initialize the busy target table to its empty state until after we have finished extracting configuration information from chip SRAM. In the common case of using 16 bytes of chip SRAM to do untagged target lookups, we were trashing the last 8 targets configuration data. (actually only target 8 because of the bug in the BUILD_TCL macro). Cram the "bus reset delivered" message back under bootverbose. Fix the cleanup of the SCB busy target table when aborting commands. If the lun is wildcarded, we must loop through all possible luns. aic7xxx.h: Only bother supporting 64 luns right now. It doesn't seem like either this driver or any peripherals will be doing information unit transfers (where the lun number is a 32 bit integer) any time soon. aic7xxx.seq: Fix support for the aic7895. We must flush the data FIFO if performing a manual transfer that is not a multiple of 8 bytes. We were doing this quite regularly for embedded cdbs. Manaually flush the fifo on earlier adapters when dealing with embedded cdbs too. We were stuffing the FIFO with 16 bytes instead, but triggering the flush is more efficient and allows us to remove two instructions from the "copy_to_fifo" routine.
2000-07-24 22:27:40 +00:00
call copy_to_fifo_6;
call copy_to_fifo_6;
or DFCNTRL, FIFOFLUSH;
}
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
p_command_loop:
test SSTAT0, SDONE jnz . + 2;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
test SSTAT1, PHASEMIS jz p_command_loop;
Update copyrights to Y2K. 93cx6.c: Make the SRAM dump output a little prettier. aic7xxx.c: Store all SG entries into our SG array in kernel space. This makes data-overrun and other error reporting more useful as we can dump all SG entries. In the past, we only stored the SG entries that the sequencer might need to access, which meant we skipped the first element that is embedded into the SCB. Add a table of chip strings and replace ugly switch statements with table lookups. Add a table with bus phase strings and message reponses to parity errors in those phases. Use the table to pretty print bus phase messages as well as collapse another switch statement. Fix a bug in target mode that could cause us to unpause the sequencer early in bus reset processing. Add the 80MHz/DT mode into our syncrate table. This rate is not yet used or enabled. Correct some comments, clean up some code... aic7xxx.h: Add U160 controller feature information. Add some more bit fields for various SEEPROM formats. aic7xxx.reg: Add U160 register and register bit definitions. aic7xxx.seq: Make phasemis state tracking more straight forward. This avoids the consumption of SINDEX which is a very useful register. For the U160 chips, you must use the 'mov' instruction to update DFCNTRL. Using 'or' to set the PRELOADED bit is completely ineffective. At the end of the command phase, wair for our ACK signal to de-assert before disabling the SCSI dma engine. For slow devices, this avoids clearing the ACK before the other end has had a chance to see it and lower REQ.
2000-01-07 23:08:20 +00:00
/*
* Wait for our ACK to go-away on it's own
* instead of being killed by SCSIEN getting cleared.
*/
test SCSISIGI, ACKI jnz .;
and DFCNTRL, ~(SCSIEN|SDMAEN|HDMAEN);
test DFCNTRL, (SCSIEN|SDMAEN|HDMAEN) jnz .;
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
if ((ahc->features & AHC_ULTRA2) != 0) {
/* Drop any residual from the S/G Preload queue */
or SXFRCTL0, CLRSTCNT;
}
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
jmp ITloop;
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
/*
* Status phase. Wait for the data byte to appear, then read it
* and store it into the SCB.
*/
p_status:
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
call assert;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
mov SCB_SCSI_STATUS, SCSIDATL;
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
jmp ITloop;
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
/*
* 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.
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
*/
p_mesgout_retry:
or SCSISIGO,ATNO,LASTPHASE;/* turn on ATN for the retry */
p_mesgout:
mov SINDEX, MSG_OUT;
cmp SINDEX, MSG_IDENTIFYFLAG jne p_mesgout_from_host;
test SCB_CONTROL,MK_MESSAGE jnz host_message_loop;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
mov FUNCTION1, SCB_SCSIID;
mov A, FUNCTION1;
mov SINDEX, TARGET_MSG_REQUEST[0];
if ((ahc->features & AHC_TWIN) != 0) {
/* Second Channel uses high byte bits */
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
test SCB_SCSIID, TWIN_CHNLB jz . + 2;
mov SINDEX, TARGET_MSG_REQUEST[1];
} else if ((ahc->features & AHC_WIDE) != 0) {
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
test SCB_SCSIID, 0x80 jz . + 2; /* target > 7 */
mov SINDEX, TARGET_MSG_REQUEST[1];
}
test SINDEX, A jnz host_message_loop;
p_mesgout_identify:
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
or SINDEX, MSG_IDENTIFYFLAG|DISCENB, SCB_LUN;
test SCB_CONTROL, DISCENB jnz . + 2;
and SINDEX, ~DISCENB;
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
/*
* Send a tag message if TAG_ENB is set in the SCB control block.
* Use SCB_TAG (the position in the kernel's SCB array) as the tag value.
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
*/
p_mesgout_tag:
test SCB_CONTROL,TAG_ENB jz p_mesgout_onebyte;
mov SCSIDATL, SINDEX; /* Send the identify message */
call phase_lock;
cmp LASTPHASE, P_MESGOUT jne p_mesgout_done;
and SCSIDATL,TAG_ENB|SCB_TAG_TYPE,SCB_CONTROL;
call phase_lock;
cmp LASTPHASE, P_MESGOUT jne p_mesgout_done;
mov SCB_TAG jmp p_mesgout_onebyte;
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
/*
* Interrupt the driver, and allow it to handle this message
* phase and any required retries.
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
*/
p_mesgout_from_host:
cmp SINDEX, HOST_MSG jne p_mesgout_onebyte;
jmp host_message_loop;
p_mesgout_onebyte:
mvi CLRSINT1, CLRATNO;
mov SCSIDATL, 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;
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
p_mesgout_done:
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
mvi CLRSINT1,CLRATNO; /* Be sure to turn ATNO off */
mov LAST_MSG, MSG_OUT;
mvi MSG_OUT, MSG_NOOP; /* No message left */
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
jmp ITloop;
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
/*
* Message in phase. Bytes are read using Automatic PIO mode.
*/
p_mesgin:
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
mvi ACCUM call inb_first; /* read the 1st message byte */
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;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
cmp A,MSG_IGN_WIDE_RESIDUE je mesgin_ign_wide_residue;
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
cmp A,MSG_NOOP je mesgin_done;
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
/*
* Pushed message loop to allow the kernel to
Kill the "unpause_always" argument to unpause_sequencer(). The reasons for optimizing the unpause operation no-longer exist, and this is much safer. When restarting the sequencer, reconstitute the free SCB list on the card. This deals with a single instruction gap between marking the SCB as free and actually getting it onto the free list. Reduce the number of transfer negotiations that occur. In the past, we renegotiated after every reported check condition status. This ensures that we catch devices that have unexpectidly reset. In this situation, the target will always report the check condition before performing a data-phase. The new behavior is to renegotiate for any check-condition where the residual matches the orginal data-length of the command (including 0 length transffers). This avoids renegotiations during things like variable tape block reads, where the check condition is reported only to indicate the residual of the read. Revamp the parity error detection logic. We now properly report and handle injected parity errors in all phases. The old code used to hang on message-in parity errors. Correct the reporting of selection timeout errors to the XPT. When a selection timeout occurs, only the currently selecting command is flagged with SELTO status instead of aborting all currently active commands to that target. Fix flipped arguments in ahc_match_scb and in some of the callers of this routine. I wish that gcc allowed you to request warnings for enums passed as ints. Make ahc_find_msg generically handle all message types. Work around the target mode data-in wideodd bug in all non-U2 chips. We can now do sync-wide target mode transfers in target mode across the hole product line. Use lastphase exclusively for handling timeouts. The current phase doesn't take the bus free state into account. Fix a bug in the timeout handler that could cause corruption of the disconnected list. When sending an embedded cdb to a target, ensure that we start on a quad word boundary in the data-fifo. It seems that unaligned stores do not work correctly.
2000-02-09 21:25:00 +00:00
* 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...
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
*/
host_message_loop:
aic7xxx.c: When restarting the sequencer, ensure that the SCBCNT register is 0. A non-zero count will prevent the setting of the CCSCBDIR bit in any future dma operations. The only time CCSCBCNT would be non-zero is if we happened to halt the dma during a reset, but even that should never happen. Better safe than sorry. When a command completes before the target responds to an ATN for a recovery command, we now notify the kernel so that any recovery operation requeued in the qinfifo can be removed safely. In the past, we did this in ahc_done(), but ahc_done() may be called without the card paused. This also avoids a recursive call to ahc_search_qinifo() which could have occurred if ahc_search_qinififo() happened to be the routine to complete a recovery action. Fix 8bit math used for adjusting the qinfifo. The index must be wrapped properly within the 256 entry array. We rely on the fact that qinfifonext is a uint8_t in most cases to handle this wrap, but we missed a few spots where the resultant calculation was promoted to an int. Change the way that we deal with aborting the first or second entry from the qinfifo. We now swap the first entry in the qinfifo with the "next queued scb" to force the sequencer to see an abort collision if we ever touch the qinififo while the sequencer is mid SCB dma. aic7xxx.reg: Add new MKMSG_FAILED sequencer interrupt. This displaced the BOGUS_TAG interrupt used in some previous sequencer code debugging. aic7xxx.seq: Increment our position in the qinfifo only once the dma is complete and we have verified that the queue has not been changed during our DMA. This simplifies code in the kernel. Protect against "instruction creep" when issuing a pausing sequencer interrupt. On at least the 7890/91/96/97, the sequencer will coast after issuing the interrupt for up to two instructions. In the past we delt with this by using carefully placed nops. Now we call a routine to issue the interrupt followed by a nop and a ret. Tell the kernel should an SCB complete with the MK_MESSAGE flag still set. This means the target ignored our ATN request. Clear the channel twice as we exit the data phase. On the aic7890/91, the S/G preload logic may require the second clearing to get the last S/G out of the FIFO. aic7xxx_freebsd.c: Don't bother searching the qinfifo for a doubly queued recovery scb in ahc_done. This case is handled by the core driver now. Free the path used to issue async callbacks after the callback is complete. aic7xxx_inline.h: Split the SCB queue routine into a routine that swaps the SCB with the "next queued SCB" and a routine that calls the swapping routine and notifies the card of the new SCB. The swapping routine is now also used by ahc_search_qinfifo.
2000-11-06 20:05:38 +00:00
mvi HOST_MSG_LOOP call set_seqint;
call phase_lock;
cmp RETURN_1, EXIT_MSG_LOOP je ITloop + 1;
jmp host_message_loop;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
mesgin_ign_wide_residue:
if ((ahc->features & AHC_WIDE) != 0) {
test SCSIRATE, WIDEXFER jz mesgin_reject;
/* Pull the residue byte */
mvi ARG_1 call inb_next;
cmp ARG_1, 0x01 jne mesgin_reject;
test SCB_RESIDUAL_SGPTR[0], SG_LIST_NULL jz . + 2;
test DATA_COUNT_ODD, 0x1 jz mesgin_done;
aic7xxx.c: When restarting the sequencer, ensure that the SCBCNT register is 0. A non-zero count will prevent the setting of the CCSCBDIR bit in any future dma operations. The only time CCSCBCNT would be non-zero is if we happened to halt the dma during a reset, but even that should never happen. Better safe than sorry. When a command completes before the target responds to an ATN for a recovery command, we now notify the kernel so that any recovery operation requeued in the qinfifo can be removed safely. In the past, we did this in ahc_done(), but ahc_done() may be called without the card paused. This also avoids a recursive call to ahc_search_qinifo() which could have occurred if ahc_search_qinififo() happened to be the routine to complete a recovery action. Fix 8bit math used for adjusting the qinfifo. The index must be wrapped properly within the 256 entry array. We rely on the fact that qinfifonext is a uint8_t in most cases to handle this wrap, but we missed a few spots where the resultant calculation was promoted to an int. Change the way that we deal with aborting the first or second entry from the qinfifo. We now swap the first entry in the qinfifo with the "next queued scb" to force the sequencer to see an abort collision if we ever touch the qinififo while the sequencer is mid SCB dma. aic7xxx.reg: Add new MKMSG_FAILED sequencer interrupt. This displaced the BOGUS_TAG interrupt used in some previous sequencer code debugging. aic7xxx.seq: Increment our position in the qinfifo only once the dma is complete and we have verified that the queue has not been changed during our DMA. This simplifies code in the kernel. Protect against "instruction creep" when issuing a pausing sequencer interrupt. On at least the 7890/91/96/97, the sequencer will coast after issuing the interrupt for up to two instructions. In the past we delt with this by using carefully placed nops. Now we call a routine to issue the interrupt followed by a nop and a ret. Tell the kernel should an SCB complete with the MK_MESSAGE flag still set. This means the target ignored our ATN request. Clear the channel twice as we exit the data phase. On the aic7890/91, the S/G preload logic may require the second clearing to get the last S/G out of the FIFO. aic7xxx_freebsd.c: Don't bother searching the qinfifo for a doubly queued recovery scb in ahc_done. This case is handled by the core driver now. Free the path used to issue async callbacks after the callback is complete. aic7xxx_inline.h: Split the SCB queue routine into a routine that swaps the SCB with the "next queued SCB" and a routine that calls the swapping routine and notifies the card of the new SCB. The swapping routine is now also used by ahc_search_qinfifo.
2000-11-06 20:05:38 +00:00
mvi IGN_WIDE_RES call set_seqint;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
jmp mesgin_done;
}
mesgin_reject:
mvi MSG_MESSAGE_REJECT call mk_mesg;
mesgin_done:
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
mov NONE,SCSIDATL; /*dummy read from latch to ACK*/
jmp ITloop;
mesgin_complete:
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
/*
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
* We received a "command complete" message. Put the SCB_TAG into the QOUTFIFO,
* and trigger a completion interrupt. Before doing 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. In the case of a non zero status byte, we
* additionally interrupt the kernel driver synchronously, allowing it to
* decide if sense should be retrieved. If the kernel driver wishes to request
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
* sense, it will fill the kernel SCB with a request sense command, requeue
* it to the QINFIFO and tell us not to post to the QOUTFIFO by setting
* RETURN_1 to SEND_SENSE.
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
*/
Sync Perforce IDs, add tranceiver state change support, and correct numerous error recovery buglets. Many thanks to Tor Egge for his assistance in diagnosing problems with the error recovery code. aic7xxx.c: Report missed bus free events using their own sequencer interrupt code to avoid confusion with other "bad phase" interrupts. Remove a delay used in debugging. This delay could only be hit in certain, very extreme, error recovery scenarios. Handle transceiver state changes correctly. You can now plug an SE device into a hot-plug LVD bus without hanging the controller. When stepping through a critical section, panic if we step more than a reasonable number of times. After a bus reset, disable bus reset interupts until we either our first attempt to (re)select another device, or another device attemps to select us. This removes the need to busy wait in kernel for the scsi reset line to fall yet still ensures we see any reset events that impact the state of either our initiator or target roles. Before this change, we had the potential of servicing a "storm" of reset interrupts if the reset line was held for a significant amount of time. Indicate the current sequencer address whenever we dump the card's state. aic7xxx.reg: Transceiver state change register definitions. Add the missed bussfree sequencer interrupt code. Re-enable the scsi reset interrupt if it has been disabled before every attempt to (re)select a device and when we have been selected as a target. When being (re)selected, check to see if the selection dissappeared just after we enabled our bus free interrupt. If the bus has gone free again, go back to the idle loop and wait for another selection. Note two locations where we should change our behavior if ATN is still raised. If ATN is raised during the presentation of a command complete or disconnect message, we should ignore the message and expect the target to put us in msgout phase. We don't currently do this as it requires some code re-arrangement so that critical sections can be properly placed around our handling of these two events. Otherwise, we cannot guarantee that the check of ATN is atomic relative to our acking of the message in byte (the kernel could assert ATN). Only set the IDENTIFY_SEEN flag after we have settled on the SCB for this transaction. The kernel looks at this flag before assuming that SCB_TAG is valid. This avoids confusion during certain types of error recovery. Add a critical section around findSCB. We cannot allow the kernel to remove an entry from the disconnected list while we are traversing it. Ditto for get_free_or_disc_scb. aic7xxx_freebsd.c: Only assume that SCB_TAG is accurate if IDENTIFY_SEEN is set in SEQ_FLAGS. Fix a typo that caused us to execute some code for the non-SCB paging case when paging SCBs. This only occurred during error recovery.
2000-11-10 20:13:41 +00:00
/*
* 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
aic7xxx.c: Style nits. Make sure that our selection hardware is disabled as soon as possible after detecting a busfree and even go so far as to disable the selection hardware in advance of an event that will cause a busfree (ABORT or BUS DEVICE RESET message). The concern is that the selection hardware will select a target for which, after processing the bus free, there will be no commands pending. The sequencer idle loop will re-enable the selection should it still be necessary. In ahc_handle_scsiint(), clear SSTAT0 events several PCI transactions (most notably reads) prior to clearing SCSIINT. The newer chips seem to take a bit of time to see the change which can make the clearing of SCSIINT ineffective. Don't bother panicing at the end of ahc_handle_scsiint(). Getting to the final else just means we lost the race with clearing SCSIINT. In ahc_free(), handle init-level 0. This can happen when we fail the attach for RAID devices. While I'm here, also kill the parent dma tag. In ahc_match_scb(), consider initiator ccbs to be any that are not from the target mode group. This fixes a bug where an external target reset CCB was not getting cleaned up by the reset code. Don't bother freezing a ccb in any of our "abort" routines when the status is set to CAM_REQ_CMP. This can happen for a target reset ccb. aic7xxx.reg: Reserve space for a completion queue. This will be used to enhance performance in the near future. aic7xxx.seq: Remove an optimization for the 7890 autoflush bug that turned out to allow, in rare cases, some data to get lost. Implement a simpler, faster, fix for the PCI_2_1 retry bug that hangs the sequencer on an SCB dma for certain chips. Test against SAVED_SCSIID rather than SELID during target reselections. This is how we always did it in the past, but the code was modified while trying to work around an issue with the 7895. SAVED_SCSIID takes into account twin channel adapters such as the 2742T, whereas SELID does not have the channel bit. This caused invalid selection warnings and other strangeness on these cards. aic7xxx_pci.c Use the correct mask for checking the generic aic7892 entry.
2001-02-10 18:04:27 +00:00
* 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.
Sync Perforce IDs, add tranceiver state change support, and correct numerous error recovery buglets. Many thanks to Tor Egge for his assistance in diagnosing problems with the error recovery code. aic7xxx.c: Report missed bus free events using their own sequencer interrupt code to avoid confusion with other "bad phase" interrupts. Remove a delay used in debugging. This delay could only be hit in certain, very extreme, error recovery scenarios. Handle transceiver state changes correctly. You can now plug an SE device into a hot-plug LVD bus without hanging the controller. When stepping through a critical section, panic if we step more than a reasonable number of times. After a bus reset, disable bus reset interupts until we either our first attempt to (re)select another device, or another device attemps to select us. This removes the need to busy wait in kernel for the scsi reset line to fall yet still ensures we see any reset events that impact the state of either our initiator or target roles. Before this change, we had the potential of servicing a "storm" of reset interrupts if the reset line was held for a significant amount of time. Indicate the current sequencer address whenever we dump the card's state. aic7xxx.reg: Transceiver state change register definitions. Add the missed bussfree sequencer interrupt code. Re-enable the scsi reset interrupt if it has been disabled before every attempt to (re)select a device and when we have been selected as a target. When being (re)selected, check to see if the selection dissappeared just after we enabled our bus free interrupt. If the bus has gone free again, go back to the idle loop and wait for another selection. Note two locations where we should change our behavior if ATN is still raised. If ATN is raised during the presentation of a command complete or disconnect message, we should ignore the message and expect the target to put us in msgout phase. We don't currently do this as it requires some code re-arrangement so that critical sections can be properly placed around our handling of these two events. Otherwise, we cannot guarantee that the check of ATN is atomic relative to our acking of the message in byte (the kernel could assert ATN). Only set the IDENTIFY_SEEN flag after we have settled on the SCB for this transaction. The kernel looks at this flag before assuming that SCB_TAG is valid. This avoids confusion during certain types of error recovery. Add a critical section around findSCB. We cannot allow the kernel to remove an entry from the disconnected list while we are traversing it. Ditto for get_free_or_disc_scb. aic7xxx_freebsd.c: Only assume that SCB_TAG is accurate if IDENTIFY_SEEN is set in SEQ_FLAGS. Fix a typo that caused us to execute some code for the non-SCB paging case when paging SCBs. This only occurred during error recovery.
2000-11-10 20:13:41 +00:00
test SCSISIGI, ATNI jnz mesgin_done;
*/
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
/*
aic7xxx.c: When restarting the sequencer, ensure that the SCBCNT register is 0. A non-zero count will prevent the setting of the CCSCBDIR bit in any future dma operations. The only time CCSCBCNT would be non-zero is if we happened to halt the dma during a reset, but even that should never happen. Better safe than sorry. When a command completes before the target responds to an ATN for a recovery command, we now notify the kernel so that any recovery operation requeued in the qinfifo can be removed safely. In the past, we did this in ahc_done(), but ahc_done() may be called without the card paused. This also avoids a recursive call to ahc_search_qinifo() which could have occurred if ahc_search_qinififo() happened to be the routine to complete a recovery action. Fix 8bit math used for adjusting the qinfifo. The index must be wrapped properly within the 256 entry array. We rely on the fact that qinfifonext is a uint8_t in most cases to handle this wrap, but we missed a few spots where the resultant calculation was promoted to an int. Change the way that we deal with aborting the first or second entry from the qinfifo. We now swap the first entry in the qinfifo with the "next queued scb" to force the sequencer to see an abort collision if we ever touch the qinififo while the sequencer is mid SCB dma. aic7xxx.reg: Add new MKMSG_FAILED sequencer interrupt. This displaced the BOGUS_TAG interrupt used in some previous sequencer code debugging. aic7xxx.seq: Increment our position in the qinfifo only once the dma is complete and we have verified that the queue has not been changed during our DMA. This simplifies code in the kernel. Protect against "instruction creep" when issuing a pausing sequencer interrupt. On at least the 7890/91/96/97, the sequencer will coast after issuing the interrupt for up to two instructions. In the past we delt with this by using carefully placed nops. Now we call a routine to issue the interrupt followed by a nop and a ret. Tell the kernel should an SCB complete with the MK_MESSAGE flag still set. This means the target ignored our ATN request. Clear the channel twice as we exit the data phase. On the aic7890/91, the S/G preload logic may require the second clearing to get the last S/G out of the FIFO. aic7xxx_freebsd.c: Don't bother searching the qinfifo for a doubly queued recovery scb in ahc_done. This case is handled by the core driver now. Free the path used to issue async callbacks after the callback is complete. aic7xxx_inline.h: Split the SCB queue routine into a routine that swaps the SCB with the "next queued SCB" and a routine that calls the swapping routine and notifies the card of the new SCB. The swapping routine is now also used by ahc_search_qinfifo.
2000-11-06 20:05:38 +00:00
* See if we attempted to deliver a message but the target ingnored us.
*/
test SCB_CONTROL, MK_MESSAGE jz . + 2;
mvi MKMSG_FAILED call set_seqint;
/*
* Check for residuals
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
*/
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
test SCB_SGPTR, SG_LIST_NULL jnz check_status;/* No xfer */
test SCB_SGPTR, SG_FULL_RESID jnz upload_scb;/* Never xfered */
test SCB_RESIDUAL_SGPTR, SG_LIST_NULL jz upload_scb;
check_status:
test SCB_SCSI_STATUS,0xff jz complete; /* Good Status? */
upload_scb:
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
or SCB_SGPTR, SG_RESID_VALID;
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
mvi DMAPARAMS, FIFORESET;
mov SCB_TAG call dma_scb;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
test SCB_SCSI_STATUS, 0xff jz complete; /* Just a residual? */
aic7xxx.c: When restarting the sequencer, ensure that the SCBCNT register is 0. A non-zero count will prevent the setting of the CCSCBDIR bit in any future dma operations. The only time CCSCBCNT would be non-zero is if we happened to halt the dma during a reset, but even that should never happen. Better safe than sorry. When a command completes before the target responds to an ATN for a recovery command, we now notify the kernel so that any recovery operation requeued in the qinfifo can be removed safely. In the past, we did this in ahc_done(), but ahc_done() may be called without the card paused. This also avoids a recursive call to ahc_search_qinifo() which could have occurred if ahc_search_qinififo() happened to be the routine to complete a recovery action. Fix 8bit math used for adjusting the qinfifo. The index must be wrapped properly within the 256 entry array. We rely on the fact that qinfifonext is a uint8_t in most cases to handle this wrap, but we missed a few spots where the resultant calculation was promoted to an int. Change the way that we deal with aborting the first or second entry from the qinfifo. We now swap the first entry in the qinfifo with the "next queued scb" to force the sequencer to see an abort collision if we ever touch the qinififo while the sequencer is mid SCB dma. aic7xxx.reg: Add new MKMSG_FAILED sequencer interrupt. This displaced the BOGUS_TAG interrupt used in some previous sequencer code debugging. aic7xxx.seq: Increment our position in the qinfifo only once the dma is complete and we have verified that the queue has not been changed during our DMA. This simplifies code in the kernel. Protect against "instruction creep" when issuing a pausing sequencer interrupt. On at least the 7890/91/96/97, the sequencer will coast after issuing the interrupt for up to two instructions. In the past we delt with this by using carefully placed nops. Now we call a routine to issue the interrupt followed by a nop and a ret. Tell the kernel should an SCB complete with the MK_MESSAGE flag still set. This means the target ignored our ATN request. Clear the channel twice as we exit the data phase. On the aic7890/91, the S/G preload logic may require the second clearing to get the last S/G out of the FIFO. aic7xxx_freebsd.c: Don't bother searching the qinfifo for a doubly queued recovery scb in ahc_done. This case is handled by the core driver now. Free the path used to issue async callbacks after the callback is complete. aic7xxx_inline.h: Split the SCB queue routine into a routine that swaps the SCB with the "next queued SCB" and a routine that calls the swapping routine and notifies the card of the new SCB. The swapping routine is now also used by ahc_search_qinfifo.
2000-11-06 20:05:38 +00:00
mvi BAD_STATUS call set_seqint; /* let driver know */
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
cmp RETURN_1, SEND_SENSE jne complete;
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
call add_scb_to_free_list;
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
jmp await_busfree;
complete:
mov SCB_TAG call complete_post;
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
jmp await_busfree;
}
complete_post:
/* Post the SCBID in SINDEX and issue an interrupt */
call add_scb_to_free_list;
mov ARG_1, SINDEX;
if ((ahc->features & AHC_QUEUE_REGS) != 0) {
mov A, SDSCB_QOFF;
} else {
mov A, QOUTPOS;
}
mvi QOUTFIFO_OFFSET call post_byte_setup;
mov ARG_1 call post_byte;
if ((ahc->features & AHC_QUEUE_REGS) == 0) {
inc QOUTPOS;
}
mvi INTSTAT,CMDCMPLT ret;
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
if ((ahc->flags & AHC_INITIATORROLE) != 0) {
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
/*
* Is it a disconnect message? Set a flag in the SCB to remind us
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
* 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.
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
*/
mesgin_disconnect:
Sync Perforce IDs, add tranceiver state change support, and correct numerous error recovery buglets. Many thanks to Tor Egge for his assistance in diagnosing problems with the error recovery code. aic7xxx.c: Report missed bus free events using their own sequencer interrupt code to avoid confusion with other "bad phase" interrupts. Remove a delay used in debugging. This delay could only be hit in certain, very extreme, error recovery scenarios. Handle transceiver state changes correctly. You can now plug an SE device into a hot-plug LVD bus without hanging the controller. When stepping through a critical section, panic if we step more than a reasonable number of times. After a bus reset, disable bus reset interupts until we either our first attempt to (re)select another device, or another device attemps to select us. This removes the need to busy wait in kernel for the scsi reset line to fall yet still ensures we see any reset events that impact the state of either our initiator or target roles. Before this change, we had the potential of servicing a "storm" of reset interrupts if the reset line was held for a significant amount of time. Indicate the current sequencer address whenever we dump the card's state. aic7xxx.reg: Transceiver state change register definitions. Add the missed bussfree sequencer interrupt code. Re-enable the scsi reset interrupt if it has been disabled before every attempt to (re)select a device and when we have been selected as a target. When being (re)selected, check to see if the selection dissappeared just after we enabled our bus free interrupt. If the bus has gone free again, go back to the idle loop and wait for another selection. Note two locations where we should change our behavior if ATN is still raised. If ATN is raised during the presentation of a command complete or disconnect message, we should ignore the message and expect the target to put us in msgout phase. We don't currently do this as it requires some code re-arrangement so that critical sections can be properly placed around our handling of these two events. Otherwise, we cannot guarantee that the check of ATN is atomic relative to our acking of the message in byte (the kernel could assert ATN). Only set the IDENTIFY_SEEN flag after we have settled on the SCB for this transaction. The kernel looks at this flag before assuming that SCB_TAG is valid. This avoids confusion during certain types of error recovery. Add a critical section around findSCB. We cannot allow the kernel to remove an entry from the disconnected list while we are traversing it. Ditto for get_free_or_disc_scb. aic7xxx_freebsd.c: Only assume that SCB_TAG is accurate if IDENTIFY_SEEN is set in SEQ_FLAGS. Fix a typo that caused us to execute some code for the non-SCB paging case when paging SCBs. This only occurred during error recovery.
2000-11-10 20:13:41 +00:00
/*
* 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;
*/
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
or SCB_CONTROL,DISCONNECTED;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
if ((ahc->flags & AHC_PAGESCBS) != 0) {
call add_scb_to_disc_list;
}
test SCB_CONTROL, TAG_ENB jnz await_busfree;
mov ARG_1, SCB_TAG;
mov SAVED_LUN, SCB_LUN;
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
mov SCB_SCSIID call set_busy_target;
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
jmp await_busfree;
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
/*
* 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.
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
*/
mesgin_sdptrs:
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
test SEQ_FLAGS, DPHASE jz mesgin_done;
/*
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
* The SCB_SGPTR becomes the next one we'll download,
* and the SCB_DATAPTR becomes the current SHADDR.
* Use the residual number since STCNT is corrupted by
* any message transfer.
*/
if ((ahc->features & AHC_CMD_CHAN) != 0) {
bmov SCB_DATAPTR, SHADDR, 4;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
bmov SCB_DATACNT, SCB_RESIDUAL_DATACNT, 8;
} else {
mvi DINDEX, SCB_DATAPTR;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
mvi SHADDR call bcopy_4;
mvi SCB_RESIDUAL_DATACNT call bcopy_8;
}
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
jmp mesgin_done;
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
/*
* 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.
*/
mesgin_rdptrs:
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
and SEQ_FLAGS, ~DPHASE; /*
* We'll reload them
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
* the next time through
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
* the dataphase.
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
*/
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
jmp mesgin_done;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
/*
* Index into our Busy Target table. SINDEX and DINDEX are modified
* upon return. SCBPTR may be modified by this action.
*/
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
set_busy_target:
shr DINDEX, 4, SINDEX;
ahc_eisa.c: Initialize rid to 0. This doesn't seem to make any difference (the driver doesn't care what rid it gets and no-one seems to check rid's value), but follows standard conventions. Pass in our device_t to ahc_alloc(). We now use device_T softc storage, so passing NULL results in a panic. Set the unit number in our softc so that the driver core can retrieve it. ahc_pci.c: Set the unit number in our softc so that the driver core can retrieve it. aic7770.c: Insert our softc into the list of softcs when initialization is successful. aic7xxx.c: Remove a workaround for an aic7895 bug we will never trigger. Add additional diagnostic info to ahc_dump_card_state(). Always panic the system if a sequencer assertion fails. AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. Replace a hard coded number with a constant. Guard against looping forever in ahc_pause_and_flushwork(). A hot eject or card failure may make the intstat register return 0xFF, so limit the number of interrupts we'll process. Correct the code in ahc_search_qinfifo() that guarantees that the sequencer will see an abort collision if the qinfifo is modified when a DMA is in progress. We now do this fixup after modifying the queue. This guarantees that the HSCB we place at the head of the queue is not the same as the old head. Using "next hscb" (guaranteed not to be the same as the first SCB) before clearing the queue could free up the original head hscb to be used during a remove operation placing it again at the head of the qinfifo. aic7xxx.h: Reduce the maximum number of outstanding commands to 253 from 254. To handle our output queue correctly on machines that only support 32bit stores, we must clear the array 4 bytes at a time. To avoid colliding with a DMA write from the sequencer, we must be sure that 4 slots are empty when we write to clear the queue. This reduces us to 253 SCBs: 1 that just completed and the known three additional empty slots in the queue that preceed it. Yahoo was able to force this race on one of their systems. Interrupts were disabled for such a time that the entire output queue was filled (254 entries complete without any processing), and our 32bit write to clear the status clobbered one entry. Add a feature tag for devices that are removable. aic7xxx.reg: Never use the sequencer interrupt value of 0xF0. We need to guanrantee that an INTSTAT value of 0xFF can only occur during card failure or a hot-eject. Align the busy targets table with the begining of scratch space. This seems to appease a chip bug in the aic7895. aic7xxx.seq: Be sure to disable select-out after a bus free event that occurs early in a selection. If we don't disable select-out, we will believe that it is enabled even though a new selection will never occur. Move the clearing of SELDI to just before a jump. This appeases another chip bug of the aic7895. Make the target mode command loop a bit more efficient. AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. Properly cleanup the last SCB we tested against should we fail to properly find an SCB for a reselection. Add some additional sequencer debugging code. aic7xxx_freebsd.c: Limit the driver to 253 outstanding commands per adapter. Guard against overflow in timeout handling. aic7xxx_inline.h: AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. aic7xxx_pci.c: Set the removable feature for the apa1480 cardbus and the 29160C Compact PCI card. Don't report high byte termination information for narrow cards. Use a PCI read rather than a questionable delay when fetching/setting termination settings.
2001-01-22 21:03:48 +00:00
if ((ahc->flags & AHC_SCB_BTT) != 0) {
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
mov SCBPTR, SAVED_LUN;
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
add DINDEX, SCB_64_BTT;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
} else {
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
add DINDEX, BUSY_TARGETS;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
}
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
mov DINDIR, ARG_1 ret;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
/*
* Identify message? For a reconnecting target, this tells us the lun
* that the reconnection is for - find the correct SCB and switch to it,
* clearing the "disconnected" bit so we don't "find" it by accident later.
*/
mesgin_identify:
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
/*
* Determine whether a target is using tagged or non-tagged
* transactions by first looking at the transaction stored in
* the busy target array. If there is no untagged transaction
* for this target or the transaction is for a different lun, then
* this must be an untagged transaction.
*/
aic7xxx.c: Style nits. Make sure that our selection hardware is disabled as soon as possible after detecting a busfree and even go so far as to disable the selection hardware in advance of an event that will cause a busfree (ABORT or BUS DEVICE RESET message). The concern is that the selection hardware will select a target for which, after processing the bus free, there will be no commands pending. The sequencer idle loop will re-enable the selection should it still be necessary. In ahc_handle_scsiint(), clear SSTAT0 events several PCI transactions (most notably reads) prior to clearing SCSIINT. The newer chips seem to take a bit of time to see the change which can make the clearing of SCSIINT ineffective. Don't bother panicing at the end of ahc_handle_scsiint(). Getting to the final else just means we lost the race with clearing SCSIINT. In ahc_free(), handle init-level 0. This can happen when we fail the attach for RAID devices. While I'm here, also kill the parent dma tag. In ahc_match_scb(), consider initiator ccbs to be any that are not from the target mode group. This fixes a bug where an external target reset CCB was not getting cleaned up by the reset code. Don't bother freezing a ccb in any of our "abort" routines when the status is set to CAM_REQ_CMP. This can happen for a target reset ccb. aic7xxx.reg: Reserve space for a completion queue. This will be used to enhance performance in the near future. aic7xxx.seq: Remove an optimization for the 7890 autoflush bug that turned out to allow, in rare cases, some data to get lost. Implement a simpler, faster, fix for the PCI_2_1 retry bug that hangs the sequencer on an SCB dma for certain chips. Test against SAVED_SCSIID rather than SELID during target reselections. This is how we always did it in the past, but the code was modified while trying to work around an issue with the 7895. SAVED_SCSIID takes into account twin channel adapters such as the 2742T, whereas SELID does not have the channel bit. This caused invalid selection warnings and other strangeness on these cards. aic7xxx_pci.c Use the correct mask for checking the generic aic7892 entry.
2001-02-10 18:04:27 +00:00
shr SINDEX, 4, SAVED_SCSIID;
ahc_eisa.c: Initialize rid to 0. This doesn't seem to make any difference (the driver doesn't care what rid it gets and no-one seems to check rid's value), but follows standard conventions. Pass in our device_t to ahc_alloc(). We now use device_T softc storage, so passing NULL results in a panic. Set the unit number in our softc so that the driver core can retrieve it. ahc_pci.c: Set the unit number in our softc so that the driver core can retrieve it. aic7770.c: Insert our softc into the list of softcs when initialization is successful. aic7xxx.c: Remove a workaround for an aic7895 bug we will never trigger. Add additional diagnostic info to ahc_dump_card_state(). Always panic the system if a sequencer assertion fails. AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. Replace a hard coded number with a constant. Guard against looping forever in ahc_pause_and_flushwork(). A hot eject or card failure may make the intstat register return 0xFF, so limit the number of interrupts we'll process. Correct the code in ahc_search_qinfifo() that guarantees that the sequencer will see an abort collision if the qinfifo is modified when a DMA is in progress. We now do this fixup after modifying the queue. This guarantees that the HSCB we place at the head of the queue is not the same as the old head. Using "next hscb" (guaranteed not to be the same as the first SCB) before clearing the queue could free up the original head hscb to be used during a remove operation placing it again at the head of the qinfifo. aic7xxx.h: Reduce the maximum number of outstanding commands to 253 from 254. To handle our output queue correctly on machines that only support 32bit stores, we must clear the array 4 bytes at a time. To avoid colliding with a DMA write from the sequencer, we must be sure that 4 slots are empty when we write to clear the queue. This reduces us to 253 SCBs: 1 that just completed and the known three additional empty slots in the queue that preceed it. Yahoo was able to force this race on one of their systems. Interrupts were disabled for such a time that the entire output queue was filled (254 entries complete without any processing), and our 32bit write to clear the status clobbered one entry. Add a feature tag for devices that are removable. aic7xxx.reg: Never use the sequencer interrupt value of 0xF0. We need to guanrantee that an INTSTAT value of 0xFF can only occur during card failure or a hot-eject. Align the busy targets table with the begining of scratch space. This seems to appease a chip bug in the aic7895. aic7xxx.seq: Be sure to disable select-out after a bus free event that occurs early in a selection. If we don't disable select-out, we will believe that it is enabled even though a new selection will never occur. Move the clearing of SELDI to just before a jump. This appeases another chip bug of the aic7895. Make the target mode command loop a bit more efficient. AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. Properly cleanup the last SCB we tested against should we fail to properly find an SCB for a reselection. Add some additional sequencer debugging code. aic7xxx_freebsd.c: Limit the driver to 253 outstanding commands per adapter. Guard against overflow in timeout handling. aic7xxx_inline.h: AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. aic7xxx_pci.c: Set the removable feature for the apa1480 cardbus and the 29160C Compact PCI card. Don't report high byte termination information for narrow cards. Use a PCI read rather than a questionable delay when fetching/setting termination settings.
2001-01-22 21:03:48 +00:00
and SAVED_LUN, MSG_IDENTIFY_LUNMASK, A;
if ((ahc->flags & AHC_SCB_BTT) != 0) {
add SINDEX, SCB_64_BTT;
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
mov SCBPTR, SAVED_LUN;
ahc_eisa.c: Initialize rid to 0. This doesn't seem to make any difference (the driver doesn't care what rid it gets and no-one seems to check rid's value), but follows standard conventions. Pass in our device_t to ahc_alloc(). We now use device_T softc storage, so passing NULL results in a panic. Set the unit number in our softc so that the driver core can retrieve it. ahc_pci.c: Set the unit number in our softc so that the driver core can retrieve it. aic7770.c: Insert our softc into the list of softcs when initialization is successful. aic7xxx.c: Remove a workaround for an aic7895 bug we will never trigger. Add additional diagnostic info to ahc_dump_card_state(). Always panic the system if a sequencer assertion fails. AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. Replace a hard coded number with a constant. Guard against looping forever in ahc_pause_and_flushwork(). A hot eject or card failure may make the intstat register return 0xFF, so limit the number of interrupts we'll process. Correct the code in ahc_search_qinfifo() that guarantees that the sequencer will see an abort collision if the qinfifo is modified when a DMA is in progress. We now do this fixup after modifying the queue. This guarantees that the HSCB we place at the head of the queue is not the same as the old head. Using "next hscb" (guaranteed not to be the same as the first SCB) before clearing the queue could free up the original head hscb to be used during a remove operation placing it again at the head of the qinfifo. aic7xxx.h: Reduce the maximum number of outstanding commands to 253 from 254. To handle our output queue correctly on machines that only support 32bit stores, we must clear the array 4 bytes at a time. To avoid colliding with a DMA write from the sequencer, we must be sure that 4 slots are empty when we write to clear the queue. This reduces us to 253 SCBs: 1 that just completed and the known three additional empty slots in the queue that preceed it. Yahoo was able to force this race on one of their systems. Interrupts were disabled for such a time that the entire output queue was filled (254 entries complete without any processing), and our 32bit write to clear the status clobbered one entry. Add a feature tag for devices that are removable. aic7xxx.reg: Never use the sequencer interrupt value of 0xF0. We need to guanrantee that an INTSTAT value of 0xFF can only occur during card failure or a hot-eject. Align the busy targets table with the begining of scratch space. This seems to appease a chip bug in the aic7895. aic7xxx.seq: Be sure to disable select-out after a bus free event that occurs early in a selection. If we don't disable select-out, we will believe that it is enabled even though a new selection will never occur. Move the clearing of SELDI to just before a jump. This appeases another chip bug of the aic7895. Make the target mode command loop a bit more efficient. AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. Properly cleanup the last SCB we tested against should we fail to properly find an SCB for a reselection. Add some additional sequencer debugging code. aic7xxx_freebsd.c: Limit the driver to 253 outstanding commands per adapter. Guard against overflow in timeout handling. aic7xxx_inline.h: AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. aic7xxx_pci.c: Set the removable feature for the apa1480 cardbus and the 29160C Compact PCI card. Don't report high byte termination information for narrow cards. Use a PCI read rather than a questionable delay when fetching/setting termination settings.
2001-01-22 21:03:48 +00:00
if ((ahc->flags & AHC_SEQUENCER_DEBUG) != 0) {
add NONE, -SCB_64_BTT, SINDEX;
jc . + 2;
mvi INTSTAT, OUT_OF_RANGE;
nop;
add NONE, -(SCB_64_BTT + 16), SINDEX;
jnc . + 2;
mvi INTSTAT, OUT_OF_RANGE;
nop;
}
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
} else {
ahc_eisa.c: Initialize rid to 0. This doesn't seem to make any difference (the driver doesn't care what rid it gets and no-one seems to check rid's value), but follows standard conventions. Pass in our device_t to ahc_alloc(). We now use device_T softc storage, so passing NULL results in a panic. Set the unit number in our softc so that the driver core can retrieve it. ahc_pci.c: Set the unit number in our softc so that the driver core can retrieve it. aic7770.c: Insert our softc into the list of softcs when initialization is successful. aic7xxx.c: Remove a workaround for an aic7895 bug we will never trigger. Add additional diagnostic info to ahc_dump_card_state(). Always panic the system if a sequencer assertion fails. AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. Replace a hard coded number with a constant. Guard against looping forever in ahc_pause_and_flushwork(). A hot eject or card failure may make the intstat register return 0xFF, so limit the number of interrupts we'll process. Correct the code in ahc_search_qinfifo() that guarantees that the sequencer will see an abort collision if the qinfifo is modified when a DMA is in progress. We now do this fixup after modifying the queue. This guarantees that the HSCB we place at the head of the queue is not the same as the old head. Using "next hscb" (guaranteed not to be the same as the first SCB) before clearing the queue could free up the original head hscb to be used during a remove operation placing it again at the head of the qinfifo. aic7xxx.h: Reduce the maximum number of outstanding commands to 253 from 254. To handle our output queue correctly on machines that only support 32bit stores, we must clear the array 4 bytes at a time. To avoid colliding with a DMA write from the sequencer, we must be sure that 4 slots are empty when we write to clear the queue. This reduces us to 253 SCBs: 1 that just completed and the known three additional empty slots in the queue that preceed it. Yahoo was able to force this race on one of their systems. Interrupts were disabled for such a time that the entire output queue was filled (254 entries complete without any processing), and our 32bit write to clear the status clobbered one entry. Add a feature tag for devices that are removable. aic7xxx.reg: Never use the sequencer interrupt value of 0xF0. We need to guanrantee that an INTSTAT value of 0xFF can only occur during card failure or a hot-eject. Align the busy targets table with the begining of scratch space. This seems to appease a chip bug in the aic7895. aic7xxx.seq: Be sure to disable select-out after a bus free event that occurs early in a selection. If we don't disable select-out, we will believe that it is enabled even though a new selection will never occur. Move the clearing of SELDI to just before a jump. This appeases another chip bug of the aic7895. Make the target mode command loop a bit more efficient. AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. Properly cleanup the last SCB we tested against should we fail to properly find an SCB for a reselection. Add some additional sequencer debugging code. aic7xxx_freebsd.c: Limit the driver to 253 outstanding commands per adapter. Guard against overflow in timeout handling. aic7xxx_inline.h: AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. aic7xxx_pci.c: Set the removable feature for the apa1480 cardbus and the 29160C Compact PCI card. Don't report high byte termination information for narrow cards. Use a PCI read rather than a questionable delay when fetching/setting termination settings.
2001-01-22 21:03:48 +00:00
add SINDEX, BUSY_TARGETS;
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
if ((ahc->flags & AHC_SEQUENCER_DEBUG) != 0) {
ahc_eisa.c: Initialize rid to 0. This doesn't seem to make any difference (the driver doesn't care what rid it gets and no-one seems to check rid's value), but follows standard conventions. Pass in our device_t to ahc_alloc(). We now use device_T softc storage, so passing NULL results in a panic. Set the unit number in our softc so that the driver core can retrieve it. ahc_pci.c: Set the unit number in our softc so that the driver core can retrieve it. aic7770.c: Insert our softc into the list of softcs when initialization is successful. aic7xxx.c: Remove a workaround for an aic7895 bug we will never trigger. Add additional diagnostic info to ahc_dump_card_state(). Always panic the system if a sequencer assertion fails. AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. Replace a hard coded number with a constant. Guard against looping forever in ahc_pause_and_flushwork(). A hot eject or card failure may make the intstat register return 0xFF, so limit the number of interrupts we'll process. Correct the code in ahc_search_qinfifo() that guarantees that the sequencer will see an abort collision if the qinfifo is modified when a DMA is in progress. We now do this fixup after modifying the queue. This guarantees that the HSCB we place at the head of the queue is not the same as the old head. Using "next hscb" (guaranteed not to be the same as the first SCB) before clearing the queue could free up the original head hscb to be used during a remove operation placing it again at the head of the qinfifo. aic7xxx.h: Reduce the maximum number of outstanding commands to 253 from 254. To handle our output queue correctly on machines that only support 32bit stores, we must clear the array 4 bytes at a time. To avoid colliding with a DMA write from the sequencer, we must be sure that 4 slots are empty when we write to clear the queue. This reduces us to 253 SCBs: 1 that just completed and the known three additional empty slots in the queue that preceed it. Yahoo was able to force this race on one of their systems. Interrupts were disabled for such a time that the entire output queue was filled (254 entries complete without any processing), and our 32bit write to clear the status clobbered one entry. Add a feature tag for devices that are removable. aic7xxx.reg: Never use the sequencer interrupt value of 0xF0. We need to guanrantee that an INTSTAT value of 0xFF can only occur during card failure or a hot-eject. Align the busy targets table with the begining of scratch space. This seems to appease a chip bug in the aic7895. aic7xxx.seq: Be sure to disable select-out after a bus free event that occurs early in a selection. If we don't disable select-out, we will believe that it is enabled even though a new selection will never occur. Move the clearing of SELDI to just before a jump. This appeases another chip bug of the aic7895. Make the target mode command loop a bit more efficient. AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. Properly cleanup the last SCB we tested against should we fail to properly find an SCB for a reselection. Add some additional sequencer debugging code. aic7xxx_freebsd.c: Limit the driver to 253 outstanding commands per adapter. Guard against overflow in timeout handling. aic7xxx_inline.h: AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. aic7xxx_pci.c: Set the removable feature for the apa1480 cardbus and the 29160C Compact PCI card. Don't report high byte termination information for narrow cards. Use a PCI read rather than a questionable delay when fetching/setting termination settings.
2001-01-22 21:03:48 +00:00
add NONE, -BUSY_TARGETS, SINDEX;
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
jc . + 2;
mvi INTSTAT, OUT_OF_RANGE;
aic7xxx.c: When restarting the sequencer, ensure that the SCBCNT register is 0. A non-zero count will prevent the setting of the CCSCBDIR bit in any future dma operations. The only time CCSCBCNT would be non-zero is if we happened to halt the dma during a reset, but even that should never happen. Better safe than sorry. When a command completes before the target responds to an ATN for a recovery command, we now notify the kernel so that any recovery operation requeued in the qinfifo can be removed safely. In the past, we did this in ahc_done(), but ahc_done() may be called without the card paused. This also avoids a recursive call to ahc_search_qinifo() which could have occurred if ahc_search_qinififo() happened to be the routine to complete a recovery action. Fix 8bit math used for adjusting the qinfifo. The index must be wrapped properly within the 256 entry array. We rely on the fact that qinfifonext is a uint8_t in most cases to handle this wrap, but we missed a few spots where the resultant calculation was promoted to an int. Change the way that we deal with aborting the first or second entry from the qinfifo. We now swap the first entry in the qinfifo with the "next queued scb" to force the sequencer to see an abort collision if we ever touch the qinififo while the sequencer is mid SCB dma. aic7xxx.reg: Add new MKMSG_FAILED sequencer interrupt. This displaced the BOGUS_TAG interrupt used in some previous sequencer code debugging. aic7xxx.seq: Increment our position in the qinfifo only once the dma is complete and we have verified that the queue has not been changed during our DMA. This simplifies code in the kernel. Protect against "instruction creep" when issuing a pausing sequencer interrupt. On at least the 7890/91/96/97, the sequencer will coast after issuing the interrupt for up to two instructions. In the past we delt with this by using carefully placed nops. Now we call a routine to issue the interrupt followed by a nop and a ret. Tell the kernel should an SCB complete with the MK_MESSAGE flag still set. This means the target ignored our ATN request. Clear the channel twice as we exit the data phase. On the aic7890/91, the S/G preload logic may require the second clearing to get the last S/G out of the FIFO. aic7xxx_freebsd.c: Don't bother searching the qinfifo for a doubly queued recovery scb in ahc_done. This case is handled by the core driver now. Free the path used to issue async callbacks after the callback is complete. aic7xxx_inline.h: Split the SCB queue routine into a routine that swaps the SCB with the "next queued SCB" and a routine that calls the swapping routine and notifies the card of the new SCB. The swapping routine is now also used by ahc_search_qinfifo.
2000-11-06 20:05:38 +00:00
nop;
ahc_eisa.c: Initialize rid to 0. This doesn't seem to make any difference (the driver doesn't care what rid it gets and no-one seems to check rid's value), but follows standard conventions. Pass in our device_t to ahc_alloc(). We now use device_T softc storage, so passing NULL results in a panic. Set the unit number in our softc so that the driver core can retrieve it. ahc_pci.c: Set the unit number in our softc so that the driver core can retrieve it. aic7770.c: Insert our softc into the list of softcs when initialization is successful. aic7xxx.c: Remove a workaround for an aic7895 bug we will never trigger. Add additional diagnostic info to ahc_dump_card_state(). Always panic the system if a sequencer assertion fails. AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. Replace a hard coded number with a constant. Guard against looping forever in ahc_pause_and_flushwork(). A hot eject or card failure may make the intstat register return 0xFF, so limit the number of interrupts we'll process. Correct the code in ahc_search_qinfifo() that guarantees that the sequencer will see an abort collision if the qinfifo is modified when a DMA is in progress. We now do this fixup after modifying the queue. This guarantees that the HSCB we place at the head of the queue is not the same as the old head. Using "next hscb" (guaranteed not to be the same as the first SCB) before clearing the queue could free up the original head hscb to be used during a remove operation placing it again at the head of the qinfifo. aic7xxx.h: Reduce the maximum number of outstanding commands to 253 from 254. To handle our output queue correctly on machines that only support 32bit stores, we must clear the array 4 bytes at a time. To avoid colliding with a DMA write from the sequencer, we must be sure that 4 slots are empty when we write to clear the queue. This reduces us to 253 SCBs: 1 that just completed and the known three additional empty slots in the queue that preceed it. Yahoo was able to force this race on one of their systems. Interrupts were disabled for such a time that the entire output queue was filled (254 entries complete without any processing), and our 32bit write to clear the status clobbered one entry. Add a feature tag for devices that are removable. aic7xxx.reg: Never use the sequencer interrupt value of 0xF0. We need to guanrantee that an INTSTAT value of 0xFF can only occur during card failure or a hot-eject. Align the busy targets table with the begining of scratch space. This seems to appease a chip bug in the aic7895. aic7xxx.seq: Be sure to disable select-out after a bus free event that occurs early in a selection. If we don't disable select-out, we will believe that it is enabled even though a new selection will never occur. Move the clearing of SELDI to just before a jump. This appeases another chip bug of the aic7895. Make the target mode command loop a bit more efficient. AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. Properly cleanup the last SCB we tested against should we fail to properly find an SCB for a reselection. Add some additional sequencer debugging code. aic7xxx_freebsd.c: Limit the driver to 253 outstanding commands per adapter. Guard against overflow in timeout handling. aic7xxx_inline.h: AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. aic7xxx_pci.c: Set the removable feature for the apa1480 cardbus and the 29160C Compact PCI card. Don't report high byte termination information for narrow cards. Use a PCI read rather than a questionable delay when fetching/setting termination settings.
2001-01-22 21:03:48 +00:00
add NONE, -(BUSY_TARGETS + 16), SINDEX;
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
jnc . + 2;
mvi INTSTAT, OUT_OF_RANGE;
aic7xxx.c: When restarting the sequencer, ensure that the SCBCNT register is 0. A non-zero count will prevent the setting of the CCSCBDIR bit in any future dma operations. The only time CCSCBCNT would be non-zero is if we happened to halt the dma during a reset, but even that should never happen. Better safe than sorry. When a command completes before the target responds to an ATN for a recovery command, we now notify the kernel so that any recovery operation requeued in the qinfifo can be removed safely. In the past, we did this in ahc_done(), but ahc_done() may be called without the card paused. This also avoids a recursive call to ahc_search_qinifo() which could have occurred if ahc_search_qinififo() happened to be the routine to complete a recovery action. Fix 8bit math used for adjusting the qinfifo. The index must be wrapped properly within the 256 entry array. We rely on the fact that qinfifonext is a uint8_t in most cases to handle this wrap, but we missed a few spots where the resultant calculation was promoted to an int. Change the way that we deal with aborting the first or second entry from the qinfifo. We now swap the first entry in the qinfifo with the "next queued scb" to force the sequencer to see an abort collision if we ever touch the qinififo while the sequencer is mid SCB dma. aic7xxx.reg: Add new MKMSG_FAILED sequencer interrupt. This displaced the BOGUS_TAG interrupt used in some previous sequencer code debugging. aic7xxx.seq: Increment our position in the qinfifo only once the dma is complete and we have verified that the queue has not been changed during our DMA. This simplifies code in the kernel. Protect against "instruction creep" when issuing a pausing sequencer interrupt. On at least the 7890/91/96/97, the sequencer will coast after issuing the interrupt for up to two instructions. In the past we delt with this by using carefully placed nops. Now we call a routine to issue the interrupt followed by a nop and a ret. Tell the kernel should an SCB complete with the MK_MESSAGE flag still set. This means the target ignored our ATN request. Clear the channel twice as we exit the data phase. On the aic7890/91, the S/G preload logic may require the second clearing to get the last S/G out of the FIFO. aic7xxx_freebsd.c: Don't bother searching the qinfifo for a doubly queued recovery scb in ahc_done. This case is handled by the core driver now. Free the path used to issue async callbacks after the callback is complete. aic7xxx_inline.h: Split the SCB queue routine into a routine that swaps the SCB with the "next queued SCB" and a routine that calls the swapping routine and notifies the card of the new SCB. The swapping routine is now also used by ahc_search_qinfifo.
2000-11-06 20:05:38 +00:00
nop;
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
}
}
mov ARG_1, SINDIR;
cmp ARG_1, SCB_LIST_NULL je snoop_tag;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
if ((ahc->flags & AHC_PAGESCBS) != 0) {
Sync Perforce IDs, add tranceiver state change support, and correct numerous error recovery buglets. Many thanks to Tor Egge for his assistance in diagnosing problems with the error recovery code. aic7xxx.c: Report missed bus free events using their own sequencer interrupt code to avoid confusion with other "bad phase" interrupts. Remove a delay used in debugging. This delay could only be hit in certain, very extreme, error recovery scenarios. Handle transceiver state changes correctly. You can now plug an SE device into a hot-plug LVD bus without hanging the controller. When stepping through a critical section, panic if we step more than a reasonable number of times. After a bus reset, disable bus reset interupts until we either our first attempt to (re)select another device, or another device attemps to select us. This removes the need to busy wait in kernel for the scsi reset line to fall yet still ensures we see any reset events that impact the state of either our initiator or target roles. Before this change, we had the potential of servicing a "storm" of reset interrupts if the reset line was held for a significant amount of time. Indicate the current sequencer address whenever we dump the card's state. aic7xxx.reg: Transceiver state change register definitions. Add the missed bussfree sequencer interrupt code. Re-enable the scsi reset interrupt if it has been disabled before every attempt to (re)select a device and when we have been selected as a target. When being (re)selected, check to see if the selection dissappeared just after we enabled our bus free interrupt. If the bus has gone free again, go back to the idle loop and wait for another selection. Note two locations where we should change our behavior if ATN is still raised. If ATN is raised during the presentation of a command complete or disconnect message, we should ignore the message and expect the target to put us in msgout phase. We don't currently do this as it requires some code re-arrangement so that critical sections can be properly placed around our handling of these two events. Otherwise, we cannot guarantee that the check of ATN is atomic relative to our acking of the message in byte (the kernel could assert ATN). Only set the IDENTIFY_SEEN flag after we have settled on the SCB for this transaction. The kernel looks at this flag before assuming that SCB_TAG is valid. This avoids confusion during certain types of error recovery. Add a critical section around findSCB. We cannot allow the kernel to remove an entry from the disconnected list while we are traversing it. Ditto for get_free_or_disc_scb. aic7xxx_freebsd.c: Only assume that SCB_TAG is accurate if IDENTIFY_SEEN is set in SEQ_FLAGS. Fix a typo that caused us to execute some code for the non-SCB paging case when paging SCBs. This only occurred during error recovery.
2000-11-10 20:13:41 +00:00
mov ARG_1 call findSCB;
} else {
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
mov SCBPTR, RETURN_1;
}
ahc_eisa.c: Initialize rid to 0. This doesn't seem to make any difference (the driver doesn't care what rid it gets and no-one seems to check rid's value), but follows standard conventions. Pass in our device_t to ahc_alloc(). We now use device_T softc storage, so passing NULL results in a panic. Set the unit number in our softc so that the driver core can retrieve it. ahc_pci.c: Set the unit number in our softc so that the driver core can retrieve it. aic7770.c: Insert our softc into the list of softcs when initialization is successful. aic7xxx.c: Remove a workaround for an aic7895 bug we will never trigger. Add additional diagnostic info to ahc_dump_card_state(). Always panic the system if a sequencer assertion fails. AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. Replace a hard coded number with a constant. Guard against looping forever in ahc_pause_and_flushwork(). A hot eject or card failure may make the intstat register return 0xFF, so limit the number of interrupts we'll process. Correct the code in ahc_search_qinfifo() that guarantees that the sequencer will see an abort collision if the qinfifo is modified when a DMA is in progress. We now do this fixup after modifying the queue. This guarantees that the HSCB we place at the head of the queue is not the same as the old head. Using "next hscb" (guaranteed not to be the same as the first SCB) before clearing the queue could free up the original head hscb to be used during a remove operation placing it again at the head of the qinfifo. aic7xxx.h: Reduce the maximum number of outstanding commands to 253 from 254. To handle our output queue correctly on machines that only support 32bit stores, we must clear the array 4 bytes at a time. To avoid colliding with a DMA write from the sequencer, we must be sure that 4 slots are empty when we write to clear the queue. This reduces us to 253 SCBs: 1 that just completed and the known three additional empty slots in the queue that preceed it. Yahoo was able to force this race on one of their systems. Interrupts were disabled for such a time that the entire output queue was filled (254 entries complete without any processing), and our 32bit write to clear the status clobbered one entry. Add a feature tag for devices that are removable. aic7xxx.reg: Never use the sequencer interrupt value of 0xF0. We need to guanrantee that an INTSTAT value of 0xFF can only occur during card failure or a hot-eject. Align the busy targets table with the begining of scratch space. This seems to appease a chip bug in the aic7895. aic7xxx.seq: Be sure to disable select-out after a bus free event that occurs early in a selection. If we don't disable select-out, we will believe that it is enabled even though a new selection will never occur. Move the clearing of SELDI to just before a jump. This appeases another chip bug of the aic7895. Make the target mode command loop a bit more efficient. AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. Properly cleanup the last SCB we tested against should we fail to properly find an SCB for a reselection. Add some additional sequencer debugging code. aic7xxx_freebsd.c: Limit the driver to 253 outstanding commands per adapter. Guard against overflow in timeout handling. aic7xxx_inline.h: AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. aic7xxx_pci.c: Set the removable feature for the apa1480 cardbus and the 29160C Compact PCI card. Don't report high byte termination information for narrow cards. Use a PCI read rather than a questionable delay when fetching/setting termination settings.
2001-01-22 21:03:48 +00:00
if ((ahc->flags & AHC_SCB_BTT) != 0) {
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
jmp setup_SCB_id_lun_okay;
} else {
ahc_eisa.c: Initialize rid to 0. This doesn't seem to make any difference (the driver doesn't care what rid it gets and no-one seems to check rid's value), but follows standard conventions. Pass in our device_t to ahc_alloc(). We now use device_T softc storage, so passing NULL results in a panic. Set the unit number in our softc so that the driver core can retrieve it. ahc_pci.c: Set the unit number in our softc so that the driver core can retrieve it. aic7770.c: Insert our softc into the list of softcs when initialization is successful. aic7xxx.c: Remove a workaround for an aic7895 bug we will never trigger. Add additional diagnostic info to ahc_dump_card_state(). Always panic the system if a sequencer assertion fails. AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. Replace a hard coded number with a constant. Guard against looping forever in ahc_pause_and_flushwork(). A hot eject or card failure may make the intstat register return 0xFF, so limit the number of interrupts we'll process. Correct the code in ahc_search_qinfifo() that guarantees that the sequencer will see an abort collision if the qinfifo is modified when a DMA is in progress. We now do this fixup after modifying the queue. This guarantees that the HSCB we place at the head of the queue is not the same as the old head. Using "next hscb" (guaranteed not to be the same as the first SCB) before clearing the queue could free up the original head hscb to be used during a remove operation placing it again at the head of the qinfifo. aic7xxx.h: Reduce the maximum number of outstanding commands to 253 from 254. To handle our output queue correctly on machines that only support 32bit stores, we must clear the array 4 bytes at a time. To avoid colliding with a DMA write from the sequencer, we must be sure that 4 slots are empty when we write to clear the queue. This reduces us to 253 SCBs: 1 that just completed and the known three additional empty slots in the queue that preceed it. Yahoo was able to force this race on one of their systems. Interrupts were disabled for such a time that the entire output queue was filled (254 entries complete without any processing), and our 32bit write to clear the status clobbered one entry. Add a feature tag for devices that are removable. aic7xxx.reg: Never use the sequencer interrupt value of 0xF0. We need to guanrantee that an INTSTAT value of 0xFF can only occur during card failure or a hot-eject. Align the busy targets table with the begining of scratch space. This seems to appease a chip bug in the aic7895. aic7xxx.seq: Be sure to disable select-out after a bus free event that occurs early in a selection. If we don't disable select-out, we will believe that it is enabled even though a new selection will never occur. Move the clearing of SELDI to just before a jump. This appeases another chip bug of the aic7895. Make the target mode command loop a bit more efficient. AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. Properly cleanup the last SCB we tested against should we fail to properly find an SCB for a reselection. Add some additional sequencer debugging code. aic7xxx_freebsd.c: Limit the driver to 253 outstanding commands per adapter. Guard against overflow in timeout handling. aic7xxx_inline.h: AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. aic7xxx_pci.c: Set the removable feature for the apa1480 cardbus and the 29160C Compact PCI card. Don't report high byte termination information for narrow cards. Use a PCI read rather than a questionable delay when fetching/setting termination settings.
2001-01-22 21:03:48 +00:00
jmp setup_SCB_id_okay;
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
}
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
/*
* 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
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
* SCB. With SCB paging, we must search for non-tagged
* transactions since the SCB may exist in any slot. If we're not
* using SCB paging, we can use the tag as the direct index to the
* SCB.
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
*/
snoop_tag:
ahc_eisa.c: Initialize rid to 0. This doesn't seem to make any difference (the driver doesn't care what rid it gets and no-one seems to check rid's value), but follows standard conventions. Pass in our device_t to ahc_alloc(). We now use device_T softc storage, so passing NULL results in a panic. Set the unit number in our softc so that the driver core can retrieve it. ahc_pci.c: Set the unit number in our softc so that the driver core can retrieve it. aic7770.c: Insert our softc into the list of softcs when initialization is successful. aic7xxx.c: Remove a workaround for an aic7895 bug we will never trigger. Add additional diagnostic info to ahc_dump_card_state(). Always panic the system if a sequencer assertion fails. AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. Replace a hard coded number with a constant. Guard against looping forever in ahc_pause_and_flushwork(). A hot eject or card failure may make the intstat register return 0xFF, so limit the number of interrupts we'll process. Correct the code in ahc_search_qinfifo() that guarantees that the sequencer will see an abort collision if the qinfifo is modified when a DMA is in progress. We now do this fixup after modifying the queue. This guarantees that the HSCB we place at the head of the queue is not the same as the old head. Using "next hscb" (guaranteed not to be the same as the first SCB) before clearing the queue could free up the original head hscb to be used during a remove operation placing it again at the head of the qinfifo. aic7xxx.h: Reduce the maximum number of outstanding commands to 253 from 254. To handle our output queue correctly on machines that only support 32bit stores, we must clear the array 4 bytes at a time. To avoid colliding with a DMA write from the sequencer, we must be sure that 4 slots are empty when we write to clear the queue. This reduces us to 253 SCBs: 1 that just completed and the known three additional empty slots in the queue that preceed it. Yahoo was able to force this race on one of their systems. Interrupts were disabled for such a time that the entire output queue was filled (254 entries complete without any processing), and our 32bit write to clear the status clobbered one entry. Add a feature tag for devices that are removable. aic7xxx.reg: Never use the sequencer interrupt value of 0xF0. We need to guanrantee that an INTSTAT value of 0xFF can only occur during card failure or a hot-eject. Align the busy targets table with the begining of scratch space. This seems to appease a chip bug in the aic7895. aic7xxx.seq: Be sure to disable select-out after a bus free event that occurs early in a selection. If we don't disable select-out, we will believe that it is enabled even though a new selection will never occur. Move the clearing of SELDI to just before a jump. This appeases another chip bug of the aic7895. Make the target mode command loop a bit more efficient. AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. Properly cleanup the last SCB we tested against should we fail to properly find an SCB for a reselection. Add some additional sequencer debugging code. aic7xxx_freebsd.c: Limit the driver to 253 outstanding commands per adapter. Guard against overflow in timeout handling. aic7xxx_inline.h: AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. aic7xxx_pci.c: Set the removable feature for the apa1480 cardbus and the 29160C Compact PCI card. Don't report high byte termination information for narrow cards. Use a PCI read rather than a questionable delay when fetching/setting termination settings.
2001-01-22 21:03:48 +00:00
if ((ahc->flags & AHC_SEQUENCER_DEBUG) != 0) {
or SEQ_FLAGS, 0x80;
}
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
mov NONE,SCSIDATL; /* ACK Identify MSG */
call phase_lock;
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
if ((ahc->flags & AHC_SEQUENCER_DEBUG) != 0) {
or SEQ_FLAGS, 0x1;
}
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
cmp LASTPHASE, P_MESGIN jne not_found;
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
if ((ahc->flags & AHC_SEQUENCER_DEBUG) != 0) {
or SEQ_FLAGS, 0x2;
}
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
cmp SCSIBUSL,MSG_SIMPLE_Q_TAG jne not_found;
get_tag:
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
if ((ahc->flags & AHC_PAGESCBS) != 0) {
mvi ARG_1 call inb_next; /* tag value */
mov ARG_1 call findSCB;
} else {
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
mvi ARG_1 call inb_next; /* tag value */
mov SCBPTR, ARG_1;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
}
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
/*
* Ensure that the SCB the tag points to is for
* an SCB transaction to the reconnecting target.
*/
setup_SCB:
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
if ((ahc->flags & AHC_SEQUENCER_DEBUG) != 0) {
or SEQ_FLAGS, 0x4;
}
ahc_eisa.c: Initialize rid to 0. This doesn't seem to make any difference (the driver doesn't care what rid it gets and no-one seems to check rid's value), but follows standard conventions. Pass in our device_t to ahc_alloc(). We now use device_T softc storage, so passing NULL results in a panic. Set the unit number in our softc so that the driver core can retrieve it. ahc_pci.c: Set the unit number in our softc so that the driver core can retrieve it. aic7770.c: Insert our softc into the list of softcs when initialization is successful. aic7xxx.c: Remove a workaround for an aic7895 bug we will never trigger. Add additional diagnostic info to ahc_dump_card_state(). Always panic the system if a sequencer assertion fails. AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. Replace a hard coded number with a constant. Guard against looping forever in ahc_pause_and_flushwork(). A hot eject or card failure may make the intstat register return 0xFF, so limit the number of interrupts we'll process. Correct the code in ahc_search_qinfifo() that guarantees that the sequencer will see an abort collision if the qinfifo is modified when a DMA is in progress. We now do this fixup after modifying the queue. This guarantees that the HSCB we place at the head of the queue is not the same as the old head. Using "next hscb" (guaranteed not to be the same as the first SCB) before clearing the queue could free up the original head hscb to be used during a remove operation placing it again at the head of the qinfifo. aic7xxx.h: Reduce the maximum number of outstanding commands to 253 from 254. To handle our output queue correctly on machines that only support 32bit stores, we must clear the array 4 bytes at a time. To avoid colliding with a DMA write from the sequencer, we must be sure that 4 slots are empty when we write to clear the queue. This reduces us to 253 SCBs: 1 that just completed and the known three additional empty slots in the queue that preceed it. Yahoo was able to force this race on one of their systems. Interrupts were disabled for such a time that the entire output queue was filled (254 entries complete without any processing), and our 32bit write to clear the status clobbered one entry. Add a feature tag for devices that are removable. aic7xxx.reg: Never use the sequencer interrupt value of 0xF0. We need to guanrantee that an INTSTAT value of 0xFF can only occur during card failure or a hot-eject. Align the busy targets table with the begining of scratch space. This seems to appease a chip bug in the aic7895. aic7xxx.seq: Be sure to disable select-out after a bus free event that occurs early in a selection. If we don't disable select-out, we will believe that it is enabled even though a new selection will never occur. Move the clearing of SELDI to just before a jump. This appeases another chip bug of the aic7895. Make the target mode command loop a bit more efficient. AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. Properly cleanup the last SCB we tested against should we fail to properly find an SCB for a reselection. Add some additional sequencer debugging code. aic7xxx_freebsd.c: Limit the driver to 253 outstanding commands per adapter. Guard against overflow in timeout handling. aic7xxx_inline.h: AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. aic7xxx_pci.c: Set the removable feature for the apa1480 cardbus and the 29160C Compact PCI card. Don't report high byte termination information for narrow cards. Use a PCI read rather than a questionable delay when fetching/setting termination settings.
2001-01-22 21:03:48 +00:00
mov A, SCB_SCSIID;
cmp SAVED_SCSIID, A jne not_found_cleanup_scb;
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
if ((ahc->flags & AHC_SEQUENCER_DEBUG) != 0) {
or SEQ_FLAGS, 0x8;
}
ahc_eisa.c: Initialize rid to 0. This doesn't seem to make any difference (the driver doesn't care what rid it gets and no-one seems to check rid's value), but follows standard conventions. Pass in our device_t to ahc_alloc(). We now use device_T softc storage, so passing NULL results in a panic. Set the unit number in our softc so that the driver core can retrieve it. ahc_pci.c: Set the unit number in our softc so that the driver core can retrieve it. aic7770.c: Insert our softc into the list of softcs when initialization is successful. aic7xxx.c: Remove a workaround for an aic7895 bug we will never trigger. Add additional diagnostic info to ahc_dump_card_state(). Always panic the system if a sequencer assertion fails. AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. Replace a hard coded number with a constant. Guard against looping forever in ahc_pause_and_flushwork(). A hot eject or card failure may make the intstat register return 0xFF, so limit the number of interrupts we'll process. Correct the code in ahc_search_qinfifo() that guarantees that the sequencer will see an abort collision if the qinfifo is modified when a DMA is in progress. We now do this fixup after modifying the queue. This guarantees that the HSCB we place at the head of the queue is not the same as the old head. Using "next hscb" (guaranteed not to be the same as the first SCB) before clearing the queue could free up the original head hscb to be used during a remove operation placing it again at the head of the qinfifo. aic7xxx.h: Reduce the maximum number of outstanding commands to 253 from 254. To handle our output queue correctly on machines that only support 32bit stores, we must clear the array 4 bytes at a time. To avoid colliding with a DMA write from the sequencer, we must be sure that 4 slots are empty when we write to clear the queue. This reduces us to 253 SCBs: 1 that just completed and the known three additional empty slots in the queue that preceed it. Yahoo was able to force this race on one of their systems. Interrupts were disabled for such a time that the entire output queue was filled (254 entries complete without any processing), and our 32bit write to clear the status clobbered one entry. Add a feature tag for devices that are removable. aic7xxx.reg: Never use the sequencer interrupt value of 0xF0. We need to guanrantee that an INTSTAT value of 0xFF can only occur during card failure or a hot-eject. Align the busy targets table with the begining of scratch space. This seems to appease a chip bug in the aic7895. aic7xxx.seq: Be sure to disable select-out after a bus free event that occurs early in a selection. If we don't disable select-out, we will believe that it is enabled even though a new selection will never occur. Move the clearing of SELDI to just before a jump. This appeases another chip bug of the aic7895. Make the target mode command loop a bit more efficient. AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. Properly cleanup the last SCB we tested against should we fail to properly find an SCB for a reselection. Add some additional sequencer debugging code. aic7xxx_freebsd.c: Limit the driver to 253 outstanding commands per adapter. Guard against overflow in timeout handling. aic7xxx_inline.h: AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. aic7xxx_pci.c: Set the removable feature for the apa1480 cardbus and the 29160C Compact PCI card. Don't report high byte termination information for narrow cards. Use a PCI read rather than a questionable delay when fetching/setting termination settings.
2001-01-22 21:03:48 +00:00
setup_SCB_id_okay:
mov A, SCB_LUN;
cmp SAVED_LUN, A jne not_found_cleanup_scb;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
setup_SCB_id_lun_okay:
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
if ((ahc->flags & AHC_SEQUENCER_DEBUG) != 0) {
or SEQ_FLAGS, 0x10;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
}
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
test SCB_CONTROL,DISCONNECTED jz not_found_cleanup_scb;
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
and SCB_CONTROL,~DISCONNECTED;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
test SCB_CONTROL, TAG_ENB jnz setup_SCB_tagged;
mov A, SCBPTR;
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
mvi ARG_1, SCB_LIST_NULL;
mov SAVED_SCSIID call set_busy_target;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
mov SCBPTR, A;
setup_SCB_tagged:
Sync Perforce IDs, add tranceiver state change support, and correct numerous error recovery buglets. Many thanks to Tor Egge for his assistance in diagnosing problems with the error recovery code. aic7xxx.c: Report missed bus free events using their own sequencer interrupt code to avoid confusion with other "bad phase" interrupts. Remove a delay used in debugging. This delay could only be hit in certain, very extreme, error recovery scenarios. Handle transceiver state changes correctly. You can now plug an SE device into a hot-plug LVD bus without hanging the controller. When stepping through a critical section, panic if we step more than a reasonable number of times. After a bus reset, disable bus reset interupts until we either our first attempt to (re)select another device, or another device attemps to select us. This removes the need to busy wait in kernel for the scsi reset line to fall yet still ensures we see any reset events that impact the state of either our initiator or target roles. Before this change, we had the potential of servicing a "storm" of reset interrupts if the reset line was held for a significant amount of time. Indicate the current sequencer address whenever we dump the card's state. aic7xxx.reg: Transceiver state change register definitions. Add the missed bussfree sequencer interrupt code. Re-enable the scsi reset interrupt if it has been disabled before every attempt to (re)select a device and when we have been selected as a target. When being (re)selected, check to see if the selection dissappeared just after we enabled our bus free interrupt. If the bus has gone free again, go back to the idle loop and wait for another selection. Note two locations where we should change our behavior if ATN is still raised. If ATN is raised during the presentation of a command complete or disconnect message, we should ignore the message and expect the target to put us in msgout phase. We don't currently do this as it requires some code re-arrangement so that critical sections can be properly placed around our handling of these two events. Otherwise, we cannot guarantee that the check of ATN is atomic relative to our acking of the message in byte (the kernel could assert ATN). Only set the IDENTIFY_SEEN flag after we have settled on the SCB for this transaction. The kernel looks at this flag before assuming that SCB_TAG is valid. This avoids confusion during certain types of error recovery. Add a critical section around findSCB. We cannot allow the kernel to remove an entry from the disconnected list while we are traversing it. Ditto for get_free_or_disc_scb. aic7xxx_freebsd.c: Only assume that SCB_TAG is accurate if IDENTIFY_SEEN is set in SEQ_FLAGS. Fix a typo that caused us to execute some code for the non-SCB paging case when paging SCBs. This only occurred during error recovery.
2000-11-10 20:13:41 +00:00
mvi SEQ_FLAGS,IDENTIFY_SEEN; /* make note of IDENTIFY */
call set_transfer_settings;
/* 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;
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
jmp mesgin_done;
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
not_found_cleanup_scb:
if ((ahc->flags & AHC_PAGESCBS) != 0) {
call add_scb_to_free_list;
}
not_found:
aic7xxx.c: When restarting the sequencer, ensure that the SCBCNT register is 0. A non-zero count will prevent the setting of the CCSCBDIR bit in any future dma operations. The only time CCSCBCNT would be non-zero is if we happened to halt the dma during a reset, but even that should never happen. Better safe than sorry. When a command completes before the target responds to an ATN for a recovery command, we now notify the kernel so that any recovery operation requeued in the qinfifo can be removed safely. In the past, we did this in ahc_done(), but ahc_done() may be called without the card paused. This also avoids a recursive call to ahc_search_qinifo() which could have occurred if ahc_search_qinififo() happened to be the routine to complete a recovery action. Fix 8bit math used for adjusting the qinfifo. The index must be wrapped properly within the 256 entry array. We rely on the fact that qinfifonext is a uint8_t in most cases to handle this wrap, but we missed a few spots where the resultant calculation was promoted to an int. Change the way that we deal with aborting the first or second entry from the qinfifo. We now swap the first entry in the qinfifo with the "next queued scb" to force the sequencer to see an abort collision if we ever touch the qinififo while the sequencer is mid SCB dma. aic7xxx.reg: Add new MKMSG_FAILED sequencer interrupt. This displaced the BOGUS_TAG interrupt used in some previous sequencer code debugging. aic7xxx.seq: Increment our position in the qinfifo only once the dma is complete and we have verified that the queue has not been changed during our DMA. This simplifies code in the kernel. Protect against "instruction creep" when issuing a pausing sequencer interrupt. On at least the 7890/91/96/97, the sequencer will coast after issuing the interrupt for up to two instructions. In the past we delt with this by using carefully placed nops. Now we call a routine to issue the interrupt followed by a nop and a ret. Tell the kernel should an SCB complete with the MK_MESSAGE flag still set. This means the target ignored our ATN request. Clear the channel twice as we exit the data phase. On the aic7890/91, the S/G preload logic may require the second clearing to get the last S/G out of the FIFO. aic7xxx_freebsd.c: Don't bother searching the qinfifo for a doubly queued recovery scb in ahc_done. This case is handled by the core driver now. Free the path used to issue async callbacks after the callback is complete. aic7xxx_inline.h: Split the SCB queue routine into a routine that swaps the SCB with the "next queued SCB" and a routine that calls the swapping routine and notifies the card of the new SCB. The swapping routine is now also used by ahc_search_qinfifo.
2000-11-06 20:05:38 +00:00
mvi NO_MATCH call set_seqint;
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
jmp mesgin_done;
mk_mesg:
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
or SCSISIGO,ATNO,LASTPHASE;/* turn on ATNO */
mov MSG_OUT,SINDEX ret;
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
/*
* Functions to read data in Automatic PIO mode.
*
* According to Adaptec's documentation, 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.
*/
Kill the "unpause_always" argument to unpause_sequencer(). The reasons for optimizing the unpause operation no-longer exist, and this is much safer. When restarting the sequencer, reconstitute the free SCB list on the card. This deals with a single instruction gap between marking the SCB as free and actually getting it onto the free list. Reduce the number of transfer negotiations that occur. In the past, we renegotiated after every reported check condition status. This ensures that we catch devices that have unexpectidly reset. In this situation, the target will always report the check condition before performing a data-phase. The new behavior is to renegotiate for any check-condition where the residual matches the orginal data-length of the command (including 0 length transffers). This avoids renegotiations during things like variable tape block reads, where the check condition is reported only to indicate the residual of the read. Revamp the parity error detection logic. We now properly report and handle injected parity errors in all phases. The old code used to hang on message-in parity errors. Correct the reporting of selection timeout errors to the XPT. When a selection timeout occurs, only the currently selecting command is flagged with SELTO status instead of aborting all currently active commands to that target. Fix flipped arguments in ahc_match_scb and in some of the callers of this routine. I wish that gcc allowed you to request warnings for enums passed as ints. Make ahc_find_msg generically handle all message types. Work around the target mode data-in wideodd bug in all non-U2 chips. We can now do sync-wide target mode transfers in target mode across the hole product line. Use lastphase exclusively for handling timeouts. The current phase doesn't take the bus free state into account. Fix a bug in the timeout handler that could cause corruption of the disconnected list. When sending an embedded cdb to a target, ensure that we start on a quad word boundary in the data-fifo. It seems that unaligned stores do not work correctly.
2000-02-09 21:25:00 +00:00
inb_next_wait_perr:
aic7xxx.c: When restarting the sequencer, ensure that the SCBCNT register is 0. A non-zero count will prevent the setting of the CCSCBDIR bit in any future dma operations. The only time CCSCBCNT would be non-zero is if we happened to halt the dma during a reset, but even that should never happen. Better safe than sorry. When a command completes before the target responds to an ATN for a recovery command, we now notify the kernel so that any recovery operation requeued in the qinfifo can be removed safely. In the past, we did this in ahc_done(), but ahc_done() may be called without the card paused. This also avoids a recursive call to ahc_search_qinifo() which could have occurred if ahc_search_qinififo() happened to be the routine to complete a recovery action. Fix 8bit math used for adjusting the qinfifo. The index must be wrapped properly within the 256 entry array. We rely on the fact that qinfifonext is a uint8_t in most cases to handle this wrap, but we missed a few spots where the resultant calculation was promoted to an int. Change the way that we deal with aborting the first or second entry from the qinfifo. We now swap the first entry in the qinfifo with the "next queued scb" to force the sequencer to see an abort collision if we ever touch the qinififo while the sequencer is mid SCB dma. aic7xxx.reg: Add new MKMSG_FAILED sequencer interrupt. This displaced the BOGUS_TAG interrupt used in some previous sequencer code debugging. aic7xxx.seq: Increment our position in the qinfifo only once the dma is complete and we have verified that the queue has not been changed during our DMA. This simplifies code in the kernel. Protect against "instruction creep" when issuing a pausing sequencer interrupt. On at least the 7890/91/96/97, the sequencer will coast after issuing the interrupt for up to two instructions. In the past we delt with this by using carefully placed nops. Now we call a routine to issue the interrupt followed by a nop and a ret. Tell the kernel should an SCB complete with the MK_MESSAGE flag still set. This means the target ignored our ATN request. Clear the channel twice as we exit the data phase. On the aic7890/91, the S/G preload logic may require the second clearing to get the last S/G out of the FIFO. aic7xxx_freebsd.c: Don't bother searching the qinfifo for a doubly queued recovery scb in ahc_done. This case is handled by the core driver now. Free the path used to issue async callbacks after the callback is complete. aic7xxx_inline.h: Split the SCB queue routine into a routine that swaps the SCB with the "next queued SCB" and a routine that calls the swapping routine and notifies the card of the new SCB. The swapping routine is now also used by ahc_search_qinfifo.
2000-11-06 20:05:38 +00:00
mvi PERR_DETECTED call set_seqint;
Kill the "unpause_always" argument to unpause_sequencer(). The reasons for optimizing the unpause operation no-longer exist, and this is much safer. When restarting the sequencer, reconstitute the free SCB list on the card. This deals with a single instruction gap between marking the SCB as free and actually getting it onto the free list. Reduce the number of transfer negotiations that occur. In the past, we renegotiated after every reported check condition status. This ensures that we catch devices that have unexpectidly reset. In this situation, the target will always report the check condition before performing a data-phase. The new behavior is to renegotiate for any check-condition where the residual matches the orginal data-length of the command (including 0 length transffers). This avoids renegotiations during things like variable tape block reads, where the check condition is reported only to indicate the residual of the read. Revamp the parity error detection logic. We now properly report and handle injected parity errors in all phases. The old code used to hang on message-in parity errors. Correct the reporting of selection timeout errors to the XPT. When a selection timeout occurs, only the currently selecting command is flagged with SELTO status instead of aborting all currently active commands to that target. Fix flipped arguments in ahc_match_scb and in some of the callers of this routine. I wish that gcc allowed you to request warnings for enums passed as ints. Make ahc_find_msg generically handle all message types. Work around the target mode data-in wideodd bug in all non-U2 chips. We can now do sync-wide target mode transfers in target mode across the hole product line. Use lastphase exclusively for handling timeouts. The current phase doesn't take the bus free state into account. Fix a bug in the timeout handler that could cause corruption of the disconnected list. When sending an embedded cdb to a target, ensure that we start on a quad word boundary in the data-fifo. It seems that unaligned stores do not work correctly.
2000-02-09 21:25:00 +00:00
jmp inb_next_wait;
inb_next:
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
mov NONE,SCSIDATL; /*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.
*/
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
test SSTAT1, REQINIT jz inb_next_wait;
Kill the "unpause_always" argument to unpause_sequencer(). The reasons for optimizing the unpause operation no-longer exist, and this is much safer. When restarting the sequencer, reconstitute the free SCB list on the card. This deals with a single instruction gap between marking the SCB as free and actually getting it onto the free list. Reduce the number of transfer negotiations that occur. In the past, we renegotiated after every reported check condition status. This ensures that we catch devices that have unexpectidly reset. In this situation, the target will always report the check condition before performing a data-phase. The new behavior is to renegotiate for any check-condition where the residual matches the orginal data-length of the command (including 0 length transffers). This avoids renegotiations during things like variable tape block reads, where the check condition is reported only to indicate the residual of the read. Revamp the parity error detection logic. We now properly report and handle injected parity errors in all phases. The old code used to hang on message-in parity errors. Correct the reporting of selection timeout errors to the XPT. When a selection timeout occurs, only the currently selecting command is flagged with SELTO status instead of aborting all currently active commands to that target. Fix flipped arguments in ahc_match_scb and in some of the callers of this routine. I wish that gcc allowed you to request warnings for enums passed as ints. Make ahc_find_msg generically handle all message types. Work around the target mode data-in wideodd bug in all non-U2 chips. We can now do sync-wide target mode transfers in target mode across the hole product line. Use lastphase exclusively for handling timeouts. The current phase doesn't take the bus free state into account. Fix a bug in the timeout handler that could cause corruption of the disconnected list. When sending an embedded cdb to a target, ensure that we start on a quad word boundary in the data-fifo. It seems that unaligned stores do not work correctly.
2000-02-09 21:25:00 +00:00
test SSTAT1, SCSIPERR jnz inb_next_wait_perr;
inb_next_check_phase:
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
and LASTPHASE, PHASE_MASK, SCSISIGI;
cmp LASTPHASE, P_MESGIN jne mesgin_phasemis;
inb_first:
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
mov DINDEX,SINDEX;
mov DINDIR,SCSIBUSL ret; /*read byte directly from bus*/
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
inb_last:
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
mov NONE,SCSIDATL ret; /*dummy read from latch to ACK*/
}
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
if ((ahc->flags & AHC_TARGETROLE) != 0) {
/*
* Change to a new phase. If we are changing the state of the I/O signal,
* from out to in, wait an additional data release delay before continuing.
*/
change_phase:
/* Wait for preceding I/O session to complete. */
test SCSISIGI, ACKI jnz .;
/* Change the phase */
and DINDEX, IOI, SCSISIGI;
mov SCSISIGO, SINDEX;
and A, IOI, SINDEX;
/*
* If the data direction has changed, from
* out (initiator driving) to in (target driving),
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
* we must wait at least a data release delay plus
* the normal bus settle delay. [SCSI III SPI 10.11.0]
*/
cmp DINDEX, A je change_phase_wait;
test SINDEX, IOI jz change_phase_wait;
call change_phase_wait;
change_phase_wait:
nop;
nop;
nop;
nop ret;
/*
* Send a byte to an initiator in Automatic PIO mode.
*/
target_outb:
or SXFRCTL0, SPIOEN;
test SSTAT0, SPIORDY jz .;
mov SCSIDATL, SINDEX;
test SSTAT0, SPIORDY jz .;
and SXFRCTL0, ~SPIOEN ret;
}
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
/*
* Assert that if we've been reselected, then we've seen an IDENTIFY
* message.
*/
assert:
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
test SEQ_FLAGS,IDENTIFY_SEEN jnz return; /* seen IDENTIFY? */
aic7xxx.c: When restarting the sequencer, ensure that the SCBCNT register is 0. A non-zero count will prevent the setting of the CCSCBDIR bit in any future dma operations. The only time CCSCBCNT would be non-zero is if we happened to halt the dma during a reset, but even that should never happen. Better safe than sorry. When a command completes before the target responds to an ATN for a recovery command, we now notify the kernel so that any recovery operation requeued in the qinfifo can be removed safely. In the past, we did this in ahc_done(), but ahc_done() may be called without the card paused. This also avoids a recursive call to ahc_search_qinifo() which could have occurred if ahc_search_qinififo() happened to be the routine to complete a recovery action. Fix 8bit math used for adjusting the qinfifo. The index must be wrapped properly within the 256 entry array. We rely on the fact that qinfifonext is a uint8_t in most cases to handle this wrap, but we missed a few spots where the resultant calculation was promoted to an int. Change the way that we deal with aborting the first or second entry from the qinfifo. We now swap the first entry in the qinfifo with the "next queued scb" to force the sequencer to see an abort collision if we ever touch the qinififo while the sequencer is mid SCB dma. aic7xxx.reg: Add new MKMSG_FAILED sequencer interrupt. This displaced the BOGUS_TAG interrupt used in some previous sequencer code debugging. aic7xxx.seq: Increment our position in the qinfifo only once the dma is complete and we have verified that the queue has not been changed during our DMA. This simplifies code in the kernel. Protect against "instruction creep" when issuing a pausing sequencer interrupt. On at least the 7890/91/96/97, the sequencer will coast after issuing the interrupt for up to two instructions. In the past we delt with this by using carefully placed nops. Now we call a routine to issue the interrupt followed by a nop and a ret. Tell the kernel should an SCB complete with the MK_MESSAGE flag still set. This means the target ignored our ATN request. Clear the channel twice as we exit the data phase. On the aic7890/91, the S/G preload logic may require the second clearing to get the last S/G out of the FIFO. aic7xxx_freebsd.c: Don't bother searching the qinfifo for a doubly queued recovery scb in ahc_done. This case is handled by the core driver now. Free the path used to issue async callbacks after the callback is complete. aic7xxx_inline.h: Split the SCB queue routine into a routine that swaps the SCB with the "next queued SCB" and a routine that calls the swapping routine and notifies the card of the new SCB. The swapping routine is now also used by ahc_search_qinfifo.
2000-11-06 20:05:38 +00:00
mvi NO_IDENT jmp set_seqint; /* no - tell the kernel */
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
/*
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
* Locate a disconnected SCB by SCBID. Upon return, SCBPTR and SINDEX will
* be set to the position of the SCB. If the SCB cannot be found locally,
* it will be paged in from host memory. RETURN_2 stores the address of the
* preceding SCB in the disconnected list which can be used to speed up
* removal of the found SCB from the disconnected list.
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
*/
if ((ahc->flags & AHC_PAGESCBS) != 0) {
Sync Perforce IDs, add tranceiver state change support, and correct numerous error recovery buglets. Many thanks to Tor Egge for his assistance in diagnosing problems with the error recovery code. aic7xxx.c: Report missed bus free events using their own sequencer interrupt code to avoid confusion with other "bad phase" interrupts. Remove a delay used in debugging. This delay could only be hit in certain, very extreme, error recovery scenarios. Handle transceiver state changes correctly. You can now plug an SE device into a hot-plug LVD bus without hanging the controller. When stepping through a critical section, panic if we step more than a reasonable number of times. After a bus reset, disable bus reset interupts until we either our first attempt to (re)select another device, or another device attemps to select us. This removes the need to busy wait in kernel for the scsi reset line to fall yet still ensures we see any reset events that impact the state of either our initiator or target roles. Before this change, we had the potential of servicing a "storm" of reset interrupts if the reset line was held for a significant amount of time. Indicate the current sequencer address whenever we dump the card's state. aic7xxx.reg: Transceiver state change register definitions. Add the missed bussfree sequencer interrupt code. Re-enable the scsi reset interrupt if it has been disabled before every attempt to (re)select a device and when we have been selected as a target. When being (re)selected, check to see if the selection dissappeared just after we enabled our bus free interrupt. If the bus has gone free again, go back to the idle loop and wait for another selection. Note two locations where we should change our behavior if ATN is still raised. If ATN is raised during the presentation of a command complete or disconnect message, we should ignore the message and expect the target to put us in msgout phase. We don't currently do this as it requires some code re-arrangement so that critical sections can be properly placed around our handling of these two events. Otherwise, we cannot guarantee that the check of ATN is atomic relative to our acking of the message in byte (the kernel could assert ATN). Only set the IDENTIFY_SEEN flag after we have settled on the SCB for this transaction. The kernel looks at this flag before assuming that SCB_TAG is valid. This avoids confusion during certain types of error recovery. Add a critical section around findSCB. We cannot allow the kernel to remove an entry from the disconnected list while we are traversing it. Ditto for get_free_or_disc_scb. aic7xxx_freebsd.c: Only assume that SCB_TAG is accurate if IDENTIFY_SEEN is set in SEQ_FLAGS. Fix a typo that caused us to execute some code for the non-SCB paging case when paging SCBs. This only occurred during error recovery.
2000-11-10 20:13:41 +00:00
BEGIN_CRITICAL
findSCB:
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
mov A, SINDEX; /* Tag passed in SINDEX */
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
cmp DISCONNECTED_SCBH, SCB_LIST_NULL je findSCB_notFound;
mov SCBPTR, DISCONNECTED_SCBH; /* Initialize SCBPTR */
mvi ARG_2, SCB_LIST_NULL; /* Head of list */
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
jmp findSCB_loop;
findSCB_next:
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
cmp SCB_NEXT, SCB_LIST_NULL je findSCB_notFound;
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
mov ARG_2, SCBPTR;
mov SCBPTR,SCB_NEXT;
findSCB_loop:
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
cmp SCB_TAG, A jne findSCB_next;
rem_scb_from_disc_list:
cmp ARG_2, SCB_LIST_NULL je rHead;
mov DINDEX, SCB_NEXT;
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
mov SINDEX, SCBPTR;
mov SCBPTR, ARG_2;
mov SCB_NEXT, DINDEX;
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
mov SCBPTR, SINDEX ret;
rHead:
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
mov DISCONNECTED_SCBH,SCB_NEXT ret;
Sync Perforce IDs, add tranceiver state change support, and correct numerous error recovery buglets. Many thanks to Tor Egge for his assistance in diagnosing problems with the error recovery code. aic7xxx.c: Report missed bus free events using their own sequencer interrupt code to avoid confusion with other "bad phase" interrupts. Remove a delay used in debugging. This delay could only be hit in certain, very extreme, error recovery scenarios. Handle transceiver state changes correctly. You can now plug an SE device into a hot-plug LVD bus without hanging the controller. When stepping through a critical section, panic if we step more than a reasonable number of times. After a bus reset, disable bus reset interupts until we either our first attempt to (re)select another device, or another device attemps to select us. This removes the need to busy wait in kernel for the scsi reset line to fall yet still ensures we see any reset events that impact the state of either our initiator or target roles. Before this change, we had the potential of servicing a "storm" of reset interrupts if the reset line was held for a significant amount of time. Indicate the current sequencer address whenever we dump the card's state. aic7xxx.reg: Transceiver state change register definitions. Add the missed bussfree sequencer interrupt code. Re-enable the scsi reset interrupt if it has been disabled before every attempt to (re)select a device and when we have been selected as a target. When being (re)selected, check to see if the selection dissappeared just after we enabled our bus free interrupt. If the bus has gone free again, go back to the idle loop and wait for another selection. Note two locations where we should change our behavior if ATN is still raised. If ATN is raised during the presentation of a command complete or disconnect message, we should ignore the message and expect the target to put us in msgout phase. We don't currently do this as it requires some code re-arrangement so that critical sections can be properly placed around our handling of these two events. Otherwise, we cannot guarantee that the check of ATN is atomic relative to our acking of the message in byte (the kernel could assert ATN). Only set the IDENTIFY_SEEN flag after we have settled on the SCB for this transaction. The kernel looks at this flag before assuming that SCB_TAG is valid. This avoids confusion during certain types of error recovery. Add a critical section around findSCB. We cannot allow the kernel to remove an entry from the disconnected list while we are traversing it. Ditto for get_free_or_disc_scb. aic7xxx_freebsd.c: Only assume that SCB_TAG is accurate if IDENTIFY_SEEN is set in SEQ_FLAGS. Fix a typo that caused us to execute some code for the non-SCB paging case when paging SCBs. This only occurred during error recovery.
2000-11-10 20:13:41 +00:00
END_CRITICAL
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
findSCB_notFound:
/*
* We didn't find it. Page in the SCB.
*/
mov ARG_1, A; /* Save tag */
mov ALLZEROS call get_free_or_disc_scb;
mvi DMAPARAMS, HDMAEN|DIRECTION|FIFORESET;
mov ARG_1 jmp dma_scb;
}
/*
* Prepare the hardware to post a byte to host memory given an
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
* index of (A + (256 * SINDEX)) and a base address of SHARED_DATA_ADDR.
*/
post_byte_setup:
mov ARG_2, SINDEX;
if ((ahc->features & AHC_CMD_CHAN) != 0) {
mvi DINDEX, CCHADDR;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
mvi SHARED_DATA_ADDR call set_1byte_addr;
mvi CCHCNT, 1;
mvi CCSCBCTL, CCSCBRESET ret;
} else {
mvi DINDEX, HADDR;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
mvi SHARED_DATA_ADDR call set_1byte_addr;
mvi 1 call set_hcnt;
mvi DFCNTRL, FIFORESET ret;
}
post_byte:
if ((ahc->features & AHC_CMD_CHAN) != 0) {
bmov CCSCBRAM, SINDEX, 1;
or CCSCBCTL, CCSCBEN|CCSCBRESET;
test CCSCBCTL, CCSCBDONE jz .;
clr CCSCBCTL ret;
} else {
mov DFDAT, SINDEX;
or DFCNTRL, HDMAEN|FIFOFLUSH;
jmp dma_finish;
}
Kill the "unpause_always" argument to unpause_sequencer(). The reasons for optimizing the unpause operation no-longer exist, and this is much safer. When restarting the sequencer, reconstitute the free SCB list on the card. This deals with a single instruction gap between marking the SCB as free and actually getting it onto the free list. Reduce the number of transfer negotiations that occur. In the past, we renegotiated after every reported check condition status. This ensures that we catch devices that have unexpectidly reset. In this situation, the target will always report the check condition before performing a data-phase. The new behavior is to renegotiate for any check-condition where the residual matches the orginal data-length of the command (including 0 length transffers). This avoids renegotiations during things like variable tape block reads, where the check condition is reported only to indicate the residual of the read. Revamp the parity error detection logic. We now properly report and handle injected parity errors in all phases. The old code used to hang on message-in parity errors. Correct the reporting of selection timeout errors to the XPT. When a selection timeout occurs, only the currently selecting command is flagged with SELTO status instead of aborting all currently active commands to that target. Fix flipped arguments in ahc_match_scb and in some of the callers of this routine. I wish that gcc allowed you to request warnings for enums passed as ints. Make ahc_find_msg generically handle all message types. Work around the target mode data-in wideodd bug in all non-U2 chips. We can now do sync-wide target mode transfers in target mode across the hole product line. Use lastphase exclusively for handling timeouts. The current phase doesn't take the bus free state into account. Fix a bug in the timeout handler that could cause corruption of the disconnected list. When sending an embedded cdb to a target, ensure that we start on a quad word boundary in the data-fifo. It seems that unaligned stores do not work correctly.
2000-02-09 21:25:00 +00:00
phase_lock_perr:
aic7xxx.c: When restarting the sequencer, ensure that the SCBCNT register is 0. A non-zero count will prevent the setting of the CCSCBDIR bit in any future dma operations. The only time CCSCBCNT would be non-zero is if we happened to halt the dma during a reset, but even that should never happen. Better safe than sorry. When a command completes before the target responds to an ATN for a recovery command, we now notify the kernel so that any recovery operation requeued in the qinfifo can be removed safely. In the past, we did this in ahc_done(), but ahc_done() may be called without the card paused. This also avoids a recursive call to ahc_search_qinifo() which could have occurred if ahc_search_qinififo() happened to be the routine to complete a recovery action. Fix 8bit math used for adjusting the qinfifo. The index must be wrapped properly within the 256 entry array. We rely on the fact that qinfifonext is a uint8_t in most cases to handle this wrap, but we missed a few spots where the resultant calculation was promoted to an int. Change the way that we deal with aborting the first or second entry from the qinfifo. We now swap the first entry in the qinfifo with the "next queued scb" to force the sequencer to see an abort collision if we ever touch the qinififo while the sequencer is mid SCB dma. aic7xxx.reg: Add new MKMSG_FAILED sequencer interrupt. This displaced the BOGUS_TAG interrupt used in some previous sequencer code debugging. aic7xxx.seq: Increment our position in the qinfifo only once the dma is complete and we have verified that the queue has not been changed during our DMA. This simplifies code in the kernel. Protect against "instruction creep" when issuing a pausing sequencer interrupt. On at least the 7890/91/96/97, the sequencer will coast after issuing the interrupt for up to two instructions. In the past we delt with this by using carefully placed nops. Now we call a routine to issue the interrupt followed by a nop and a ret. Tell the kernel should an SCB complete with the MK_MESSAGE flag still set. This means the target ignored our ATN request. Clear the channel twice as we exit the data phase. On the aic7890/91, the S/G preload logic may require the second clearing to get the last S/G out of the FIFO. aic7xxx_freebsd.c: Don't bother searching the qinfifo for a doubly queued recovery scb in ahc_done. This case is handled by the core driver now. Free the path used to issue async callbacks after the callback is complete. aic7xxx_inline.h: Split the SCB queue routine into a routine that swaps the SCB with the "next queued SCB" and a routine that calls the swapping routine and notifies the card of the new SCB. The swapping routine is now also used by ahc_search_qinfifo.
2000-11-06 20:05:38 +00:00
mvi PERR_DETECTED call set_seqint;
phase_lock:
Kill the "unpause_always" argument to unpause_sequencer(). The reasons for optimizing the unpause operation no-longer exist, and this is much safer. When restarting the sequencer, reconstitute the free SCB list on the card. This deals with a single instruction gap between marking the SCB as free and actually getting it onto the free list. Reduce the number of transfer negotiations that occur. In the past, we renegotiated after every reported check condition status. This ensures that we catch devices that have unexpectidly reset. In this situation, the target will always report the check condition before performing a data-phase. The new behavior is to renegotiate for any check-condition where the residual matches the orginal data-length of the command (including 0 length transffers). This avoids renegotiations during things like variable tape block reads, where the check condition is reported only to indicate the residual of the read. Revamp the parity error detection logic. We now properly report and handle injected parity errors in all phases. The old code used to hang on message-in parity errors. Correct the reporting of selection timeout errors to the XPT. When a selection timeout occurs, only the currently selecting command is flagged with SELTO status instead of aborting all currently active commands to that target. Fix flipped arguments in ahc_match_scb and in some of the callers of this routine. I wish that gcc allowed you to request warnings for enums passed as ints. Make ahc_find_msg generically handle all message types. Work around the target mode data-in wideodd bug in all non-U2 chips. We can now do sync-wide target mode transfers in target mode across the hole product line. Use lastphase exclusively for handling timeouts. The current phase doesn't take the bus free state into account. Fix a bug in the timeout handler that could cause corruption of the disconnected list. When sending an embedded cdb to a target, ensure that we start on a quad word boundary in the data-fifo. It seems that unaligned stores do not work correctly.
2000-02-09 21:25:00 +00:00
/*
* If there is a parity error, wait for the kernel to
* see the interrupt and prepare our message response
* before continuing.
*/
test SSTAT1, REQINIT jz phase_lock;
Kill the "unpause_always" argument to unpause_sequencer(). The reasons for optimizing the unpause operation no-longer exist, and this is much safer. When restarting the sequencer, reconstitute the free SCB list on the card. This deals with a single instruction gap between marking the SCB as free and actually getting it onto the free list. Reduce the number of transfer negotiations that occur. In the past, we renegotiated after every reported check condition status. This ensures that we catch devices that have unexpectidly reset. In this situation, the target will always report the check condition before performing a data-phase. The new behavior is to renegotiate for any check-condition where the residual matches the orginal data-length of the command (including 0 length transffers). This avoids renegotiations during things like variable tape block reads, where the check condition is reported only to indicate the residual of the read. Revamp the parity error detection logic. We now properly report and handle injected parity errors in all phases. The old code used to hang on message-in parity errors. Correct the reporting of selection timeout errors to the XPT. When a selection timeout occurs, only the currently selecting command is flagged with SELTO status instead of aborting all currently active commands to that target. Fix flipped arguments in ahc_match_scb and in some of the callers of this routine. I wish that gcc allowed you to request warnings for enums passed as ints. Make ahc_find_msg generically handle all message types. Work around the target mode data-in wideodd bug in all non-U2 chips. We can now do sync-wide target mode transfers in target mode across the hole product line. Use lastphase exclusively for handling timeouts. The current phase doesn't take the bus free state into account. Fix a bug in the timeout handler that could cause corruption of the disconnected list. When sending an embedded cdb to a target, ensure that we start on a quad word boundary in the data-fifo. It seems that unaligned stores do not work correctly.
2000-02-09 21:25:00 +00:00
test SSTAT1, SCSIPERR jnz phase_lock_perr;
phase_lock_latch_phase:
and SCSISIGO, PHASE_MASK, SCSISIGI;
and LASTPHASE, PHASE_MASK, SCSISIGI ret;
if ((ahc->features & AHC_CMD_CHAN) == 0) {
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
set_hcnt:
mov HCNT[0], SINDEX;
clear_hcnt:
clr HCNT[1];
clr HCNT[2] ret;
set_stcnt_from_hcnt:
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
mov STCNT[0], HCNT[0];
mov STCNT[1], HCNT[1];
mov STCNT[2], HCNT[2] ret;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
bcopy_8:
mov DINDIR, SINDIR;
bcopy_7:
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
mov DINDIR, SINDIR;
mov DINDIR, SINDIR;
bcopy_5:
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
mov DINDIR, SINDIR;
bcopy_4:
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
mov DINDIR, SINDIR;
bcopy_3:
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
mov DINDIR, SINDIR;
mov DINDIR, SINDIR;
mov DINDIR, SINDIR ret;
}
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
if ((ahc->flags & AHC_TARGETROLE) != 0) {
/*
* Setup addr assuming that A is an index into
* an array of 32byte objects, SINDEX contains
* the base address of that array, and DINDEX
* contains the base address of the location
* to store the indexed address.
*/
set_32byte_addr:
shr ARG_2, 3, A;
shl A, 5;
jmp set_1byte_addr;
}
/*
* Setup addr assuming that A is an index into
* an array of 64byte objects, SINDEX contains
* the base address of that array, and DINDEX
* contains the base address of the location
* to store the indexed address.
*/
set_64byte_addr:
shr ARG_2, 2, A;
shl A, 6;
/*
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
* Setup addr assuming that A + (ARG_2 * 256) is an
* index into an array of 1byte objects, SINDEX contains
* the base address of that array, and DINDEX contains
* the base address of the location to store the computed
* address.
*/
set_1byte_addr:
add DINDIR, A, SINDIR;
mov A, ARG_2;
adc DINDIR, A, SINDIR;
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
clr A;
adc DINDIR, A, SINDIR;
adc DINDIR, A, SINDIR ret;
/*
* Either post or fetch and SCB from host memory based on the
* DIRECTION bit in DMAPARAMS. The host SCB index is in SINDEX.
*/
dma_scb:
mov A, SINDEX;
if ((ahc->features & AHC_CMD_CHAN) != 0) {
mvi DINDEX, CCHADDR;
mvi HSCB_ADDR call set_64byte_addr;
mov CCSCBPTR, SCBPTR;
test DMAPARAMS, DIRECTION jz dma_scb_tohost;
ahc_eisa.c: Initialize rid to 0. This doesn't seem to make any difference (the driver doesn't care what rid it gets and no-one seems to check rid's value), but follows standard conventions. Pass in our device_t to ahc_alloc(). We now use device_T softc storage, so passing NULL results in a panic. Set the unit number in our softc so that the driver core can retrieve it. ahc_pci.c: Set the unit number in our softc so that the driver core can retrieve it. aic7770.c: Insert our softc into the list of softcs when initialization is successful. aic7xxx.c: Remove a workaround for an aic7895 bug we will never trigger. Add additional diagnostic info to ahc_dump_card_state(). Always panic the system if a sequencer assertion fails. AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. Replace a hard coded number with a constant. Guard against looping forever in ahc_pause_and_flushwork(). A hot eject or card failure may make the intstat register return 0xFF, so limit the number of interrupts we'll process. Correct the code in ahc_search_qinfifo() that guarantees that the sequencer will see an abort collision if the qinfifo is modified when a DMA is in progress. We now do this fixup after modifying the queue. This guarantees that the HSCB we place at the head of the queue is not the same as the old head. Using "next hscb" (guaranteed not to be the same as the first SCB) before clearing the queue could free up the original head hscb to be used during a remove operation placing it again at the head of the qinfifo. aic7xxx.h: Reduce the maximum number of outstanding commands to 253 from 254. To handle our output queue correctly on machines that only support 32bit stores, we must clear the array 4 bytes at a time. To avoid colliding with a DMA write from the sequencer, we must be sure that 4 slots are empty when we write to clear the queue. This reduces us to 253 SCBs: 1 that just completed and the known three additional empty slots in the queue that preceed it. Yahoo was able to force this race on one of their systems. Interrupts were disabled for such a time that the entire output queue was filled (254 entries complete without any processing), and our 32bit write to clear the status clobbered one entry. Add a feature tag for devices that are removable. aic7xxx.reg: Never use the sequencer interrupt value of 0xF0. We need to guanrantee that an INTSTAT value of 0xFF can only occur during card failure or a hot-eject. Align the busy targets table with the begining of scratch space. This seems to appease a chip bug in the aic7895. aic7xxx.seq: Be sure to disable select-out after a bus free event that occurs early in a selection. If we don't disable select-out, we will believe that it is enabled even though a new selection will never occur. Move the clearing of SELDI to just before a jump. This appeases another chip bug of the aic7895. Make the target mode command loop a bit more efficient. AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. Properly cleanup the last SCB we tested against should we fail to properly find an SCB for a reselection. Add some additional sequencer debugging code. aic7xxx_freebsd.c: Limit the driver to 253 outstanding commands per adapter. Guard against overflow in timeout handling. aic7xxx_inline.h: AHC_SCB_BTT is a "flag" not a "feature". Check the right field in the softc. aic7xxx_pci.c: Set the removable feature for the apa1480 cardbus and the 29160C Compact PCI card. Don't report high byte termination information for narrow cards. Use a PCI read rather than a questionable delay when fetching/setting termination settings.
2001-01-22 21:03:48 +00:00
if ((ahc->flags & AHC_SCB_BTT) != 0) {
mvi CCHCNT, SCB_DOWNLOAD_SIZE_64;
} else {
mvi CCHCNT, SCB_DOWNLOAD_SIZE;
}
mvi CCSCBCTL, CCARREN|CCSCBEN|CCSCBDIR|CCSCBRESET;
cmp CCSCBCTL, CCSCBDONE|ARRDONE|CCARREN|CCSCBEN|CCSCBDIR jne .;
jmp dma_scb_finish;
dma_scb_tohost:
mvi CCHCNT, SCB_UPLOAD_SIZE;
if ((ahc->features & AHC_ULTRA2) == 0) {
mvi CCSCBCTL, CCSCBRESET;
bmov CCSCBRAM, SCB_BASE, SCB_UPLOAD_SIZE;
or CCSCBCTL, CCSCBEN|CCSCBRESET;
Sync Perforce IDs, add tranceiver state change support, and correct numerous error recovery buglets. Many thanks to Tor Egge for his assistance in diagnosing problems with the error recovery code. aic7xxx.c: Report missed bus free events using their own sequencer interrupt code to avoid confusion with other "bad phase" interrupts. Remove a delay used in debugging. This delay could only be hit in certain, very extreme, error recovery scenarios. Handle transceiver state changes correctly. You can now plug an SE device into a hot-plug LVD bus without hanging the controller. When stepping through a critical section, panic if we step more than a reasonable number of times. After a bus reset, disable bus reset interupts until we either our first attempt to (re)select another device, or another device attemps to select us. This removes the need to busy wait in kernel for the scsi reset line to fall yet still ensures we see any reset events that impact the state of either our initiator or target roles. Before this change, we had the potential of servicing a "storm" of reset interrupts if the reset line was held for a significant amount of time. Indicate the current sequencer address whenever we dump the card's state. aic7xxx.reg: Transceiver state change register definitions. Add the missed bussfree sequencer interrupt code. Re-enable the scsi reset interrupt if it has been disabled before every attempt to (re)select a device and when we have been selected as a target. When being (re)selected, check to see if the selection dissappeared just after we enabled our bus free interrupt. If the bus has gone free again, go back to the idle loop and wait for another selection. Note two locations where we should change our behavior if ATN is still raised. If ATN is raised during the presentation of a command complete or disconnect message, we should ignore the message and expect the target to put us in msgout phase. We don't currently do this as it requires some code re-arrangement so that critical sections can be properly placed around our handling of these two events. Otherwise, we cannot guarantee that the check of ATN is atomic relative to our acking of the message in byte (the kernel could assert ATN). Only set the IDENTIFY_SEEN flag after we have settled on the SCB for this transaction. The kernel looks at this flag before assuming that SCB_TAG is valid. This avoids confusion during certain types of error recovery. Add a critical section around findSCB. We cannot allow the kernel to remove an entry from the disconnected list while we are traversing it. Ditto for get_free_or_disc_scb. aic7xxx_freebsd.c: Only assume that SCB_TAG is accurate if IDENTIFY_SEEN is set in SEQ_FLAGS. Fix a typo that caused us to execute some code for the non-SCB paging case when paging SCBs. This only occurred during error recovery.
2000-11-10 20:13:41 +00:00
test CCSCBCTL, CCSCBDONE jz .;
} else if ((ahc->bugs & AHC_SCBCHAN_UPLOAD_BUG) != 0) {
mvi CCSCBCTL, CCARREN|CCSCBRESET;
cmp CCSCBCTL, ARRDONE|CCARREN jne .;
mvi CCHCNT, SCB_UPLOAD_SIZE;
mvi CCSCBCTL, CCSCBEN|CCSCBRESET;
cmp CCSCBCTL, CCSCBDONE|CCSCBEN jne .;
} else {
mvi CCSCBCTL, CCARREN|CCSCBEN|CCSCBRESET;
cmp CCSCBCTL, CCSCBDONE|ARRDONE|CCARREN|CCSCBEN jne .;
}
dma_scb_finish:
clr CCSCBCTL;
test CCSCBCTL, CCARREN|CCSCBEN jnz .;
ret;
} else {
mvi DINDEX, HADDR;
mvi HSCB_ADDR call set_64byte_addr;
mvi SCB_DOWNLOAD_SIZE call set_hcnt;
mov DFCNTRL, DMAPARAMS;
test DMAPARAMS, DIRECTION jnz dma_scb_fromhost;
/* Fill it with the SCB data */
copy_scb_tofifo:
mvi SINDEX, SCB_BASE;
add A, SCB_DOWNLOAD_SIZE, SINDEX;
copy_scb_tofifo_loop:
call copy_to_fifo_8;
cmp SINDEX, A jne copy_scb_tofifo_loop;
or DFCNTRL, HDMAEN|FIFOFLUSH;
jmp dma_finish;
dma_scb_fromhost:
mvi DINDEX, SCB_BASE;
if ((ahc->bugs & AHC_PCI_2_1_RETRY_BUG) != 0) {
/*
* The PCI module will only issue a PCI
* retry if the data FIFO is empty. If the
* host disconnects in the middle of a
* transfer, we must empty the fifo of all
* available data to force the chip to
* continue the transfer. This does not
* happen for SCSI transfers as the SCSI module
* will drain the FIFO as data is made available.
aic7xxx.c: Style nits. Make sure that our selection hardware is disabled as soon as possible after detecting a busfree and even go so far as to disable the selection hardware in advance of an event that will cause a busfree (ABORT or BUS DEVICE RESET message). The concern is that the selection hardware will select a target for which, after processing the bus free, there will be no commands pending. The sequencer idle loop will re-enable the selection should it still be necessary. In ahc_handle_scsiint(), clear SSTAT0 events several PCI transactions (most notably reads) prior to clearing SCSIINT. The newer chips seem to take a bit of time to see the change which can make the clearing of SCSIINT ineffective. Don't bother panicing at the end of ahc_handle_scsiint(). Getting to the final else just means we lost the race with clearing SCSIINT. In ahc_free(), handle init-level 0. This can happen when we fail the attach for RAID devices. While I'm here, also kill the parent dma tag. In ahc_match_scb(), consider initiator ccbs to be any that are not from the target mode group. This fixes a bug where an external target reset CCB was not getting cleaned up by the reset code. Don't bother freezing a ccb in any of our "abort" routines when the status is set to CAM_REQ_CMP. This can happen for a target reset ccb. aic7xxx.reg: Reserve space for a completion queue. This will be used to enhance performance in the near future. aic7xxx.seq: Remove an optimization for the 7890 autoflush bug that turned out to allow, in rare cases, some data to get lost. Implement a simpler, faster, fix for the PCI_2_1 retry bug that hangs the sequencer on an SCB dma for certain chips. Test against SAVED_SCSIID rather than SELID during target reselections. This is how we always did it in the past, but the code was modified while trying to work around an issue with the 7895. SAVED_SCSIID takes into account twin channel adapters such as the 2742T, whereas SELID does not have the channel bit. This caused invalid selection warnings and other strangeness on these cards. aic7xxx_pci.c Use the correct mask for checking the generic aic7892 entry.
2001-02-10 18:04:27 +00:00
* When the hang occurs, we know that a multiple
* of 8 bytes are in the FIFO because the PCI
* module has an 8 byte input latch that only
* dumps to the FIFO when HCNT == 0 or the
* latch is full.
*/
aic7xxx.c: Style nits. Make sure that our selection hardware is disabled as soon as possible after detecting a busfree and even go so far as to disable the selection hardware in advance of an event that will cause a busfree (ABORT or BUS DEVICE RESET message). The concern is that the selection hardware will select a target for which, after processing the bus free, there will be no commands pending. The sequencer idle loop will re-enable the selection should it still be necessary. In ahc_handle_scsiint(), clear SSTAT0 events several PCI transactions (most notably reads) prior to clearing SCSIINT. The newer chips seem to take a bit of time to see the change which can make the clearing of SCSIINT ineffective. Don't bother panicing at the end of ahc_handle_scsiint(). Getting to the final else just means we lost the race with clearing SCSIINT. In ahc_free(), handle init-level 0. This can happen when we fail the attach for RAID devices. While I'm here, also kill the parent dma tag. In ahc_match_scb(), consider initiator ccbs to be any that are not from the target mode group. This fixes a bug where an external target reset CCB was not getting cleaned up by the reset code. Don't bother freezing a ccb in any of our "abort" routines when the status is set to CAM_REQ_CMP. This can happen for a target reset ccb. aic7xxx.reg: Reserve space for a completion queue. This will be used to enhance performance in the near future. aic7xxx.seq: Remove an optimization for the 7890 autoflush bug that turned out to allow, in rare cases, some data to get lost. Implement a simpler, faster, fix for the PCI_2_1 retry bug that hangs the sequencer on an SCB dma for certain chips. Test against SAVED_SCSIID rather than SELID during target reselections. This is how we always did it in the past, but the code was modified while trying to work around an issue with the 7895. SAVED_SCSIID takes into account twin channel adapters such as the 2742T, whereas SELID does not have the channel bit. This caused invalid selection warnings and other strangeness on these cards. aic7xxx_pci.c Use the correct mask for checking the generic aic7892 entry.
2001-02-10 18:04:27 +00:00
clr A;
/* Wait for some data to arrive. */
dma_scb_hang_fifo:
test DFSTATUS, FIFOEMP jnz dma_scb_hang_fifo;
dma_scb_hang_wait:
test DFSTATUS, MREQPEND jnz dma_scb_hang_wait;
test DFSTATUS, HDONE jnz dma_scb_hang_dma_done;
test DFSTATUS, HDONE jnz dma_scb_hang_dma_done;
test DFSTATUS, HDONE jnz dma_scb_hang_dma_done;
/*
aic7xxx.c: Style nits. Make sure that our selection hardware is disabled as soon as possible after detecting a busfree and even go so far as to disable the selection hardware in advance of an event that will cause a busfree (ABORT or BUS DEVICE RESET message). The concern is that the selection hardware will select a target for which, after processing the bus free, there will be no commands pending. The sequencer idle loop will re-enable the selection should it still be necessary. In ahc_handle_scsiint(), clear SSTAT0 events several PCI transactions (most notably reads) prior to clearing SCSIINT. The newer chips seem to take a bit of time to see the change which can make the clearing of SCSIINT ineffective. Don't bother panicing at the end of ahc_handle_scsiint(). Getting to the final else just means we lost the race with clearing SCSIINT. In ahc_free(), handle init-level 0. This can happen when we fail the attach for RAID devices. While I'm here, also kill the parent dma tag. In ahc_match_scb(), consider initiator ccbs to be any that are not from the target mode group. This fixes a bug where an external target reset CCB was not getting cleaned up by the reset code. Don't bother freezing a ccb in any of our "abort" routines when the status is set to CAM_REQ_CMP. This can happen for a target reset ccb. aic7xxx.reg: Reserve space for a completion queue. This will be used to enhance performance in the near future. aic7xxx.seq: Remove an optimization for the 7890 autoflush bug that turned out to allow, in rare cases, some data to get lost. Implement a simpler, faster, fix for the PCI_2_1 retry bug that hangs the sequencer on an SCB dma for certain chips. Test against SAVED_SCSIID rather than SELID during target reselections. This is how we always did it in the past, but the code was modified while trying to work around an issue with the 7895. SAVED_SCSIID takes into account twin channel adapters such as the 2742T, whereas SELID does not have the channel bit. This caused invalid selection warnings and other strangeness on these cards. aic7xxx_pci.c Use the correct mask for checking the generic aic7892 entry.
2001-02-10 18:04:27 +00:00
* The PCI module no longer intends to perform
* a PCI transaction and HDONE has not come true.
* We are hung. Drain the fifo.
*/
dma_scb_hang_empty_fifo:
/*
aic7xxx.c: Style nits. Make sure that our selection hardware is disabled as soon as possible after detecting a busfree and even go so far as to disable the selection hardware in advance of an event that will cause a busfree (ABORT or BUS DEVICE RESET message). The concern is that the selection hardware will select a target for which, after processing the bus free, there will be no commands pending. The sequencer idle loop will re-enable the selection should it still be necessary. In ahc_handle_scsiint(), clear SSTAT0 events several PCI transactions (most notably reads) prior to clearing SCSIINT. The newer chips seem to take a bit of time to see the change which can make the clearing of SCSIINT ineffective. Don't bother panicing at the end of ahc_handle_scsiint(). Getting to the final else just means we lost the race with clearing SCSIINT. In ahc_free(), handle init-level 0. This can happen when we fail the attach for RAID devices. While I'm here, also kill the parent dma tag. In ahc_match_scb(), consider initiator ccbs to be any that are not from the target mode group. This fixes a bug where an external target reset CCB was not getting cleaned up by the reset code. Don't bother freezing a ccb in any of our "abort" routines when the status is set to CAM_REQ_CMP. This can happen for a target reset ccb. aic7xxx.reg: Reserve space for a completion queue. This will be used to enhance performance in the near future. aic7xxx.seq: Remove an optimization for the 7890 autoflush bug that turned out to allow, in rare cases, some data to get lost. Implement a simpler, faster, fix for the PCI_2_1 retry bug that hangs the sequencer on an SCB dma for certain chips. Test against SAVED_SCSIID rather than SELID during target reselections. This is how we always did it in the past, but the code was modified while trying to work around an issue with the 7895. SAVED_SCSIID takes into account twin channel adapters such as the 2742T, whereas SELID does not have the channel bit. This caused invalid selection warnings and other strangeness on these cards. aic7xxx_pci.c Use the correct mask for checking the generic aic7892 entry.
2001-02-10 18:04:27 +00:00
* Skip lines not yet transfered into the FIFO.
*/
add SINDEX, 7, HCNT;
shr SINDEX, 3;
/*
* Skip lines already copied out of the FIFO.
*/
add A, A, SINDEX;
call dma_scb_hang_dma_drain_fifo;
/*
* Set the lines transferred to all but
* those yet to reach the FIFO.
*/
aic7xxx.c: Style nits. Make sure that our selection hardware is disabled as soon as possible after detecting a busfree and even go so far as to disable the selection hardware in advance of an event that will cause a busfree (ABORT or BUS DEVICE RESET message). The concern is that the selection hardware will select a target for which, after processing the bus free, there will be no commands pending. The sequencer idle loop will re-enable the selection should it still be necessary. In ahc_handle_scsiint(), clear SSTAT0 events several PCI transactions (most notably reads) prior to clearing SCSIINT. The newer chips seem to take a bit of time to see the change which can make the clearing of SCSIINT ineffective. Don't bother panicing at the end of ahc_handle_scsiint(). Getting to the final else just means we lost the race with clearing SCSIINT. In ahc_free(), handle init-level 0. This can happen when we fail the attach for RAID devices. While I'm here, also kill the parent dma tag. In ahc_match_scb(), consider initiator ccbs to be any that are not from the target mode group. This fixes a bug where an external target reset CCB was not getting cleaned up by the reset code. Don't bother freezing a ccb in any of our "abort" routines when the status is set to CAM_REQ_CMP. This can happen for a target reset ccb. aic7xxx.reg: Reserve space for a completion queue. This will be used to enhance performance in the near future. aic7xxx.seq: Remove an optimization for the 7890 autoflush bug that turned out to allow, in rare cases, some data to get lost. Implement a simpler, faster, fix for the PCI_2_1 retry bug that hangs the sequencer on an SCB dma for certain chips. Test against SAVED_SCSIID rather than SELID during target reselections. This is how we always did it in the past, but the code was modified while trying to work around an issue with the 7895. SAVED_SCSIID takes into account twin channel adapters such as the 2742T, whereas SELID does not have the channel bit. This caused invalid selection warnings and other strangeness on these cards. aic7xxx_pci.c Use the correct mask for checking the generic aic7892 entry.
2001-02-10 18:04:27 +00:00
not SINDEX;
add A, 5, SINDEX;
jmp dma_scb_hang_fifo;
dma_scb_hang_dma_done:
and DFCNTRL, ~HDMAEN;
test DFCNTRL, HDMAEN jnz .;
aic7xxx.c: Style nits. Make sure that our selection hardware is disabled as soon as possible after detecting a busfree and even go so far as to disable the selection hardware in advance of an event that will cause a busfree (ABORT or BUS DEVICE RESET message). The concern is that the selection hardware will select a target for which, after processing the bus free, there will be no commands pending. The sequencer idle loop will re-enable the selection should it still be necessary. In ahc_handle_scsiint(), clear SSTAT0 events several PCI transactions (most notably reads) prior to clearing SCSIINT. The newer chips seem to take a bit of time to see the change which can make the clearing of SCSIINT ineffective. Don't bother panicing at the end of ahc_handle_scsiint(). Getting to the final else just means we lost the race with clearing SCSIINT. In ahc_free(), handle init-level 0. This can happen when we fail the attach for RAID devices. While I'm here, also kill the parent dma tag. In ahc_match_scb(), consider initiator ccbs to be any that are not from the target mode group. This fixes a bug where an external target reset CCB was not getting cleaned up by the reset code. Don't bother freezing a ccb in any of our "abort" routines when the status is set to CAM_REQ_CMP. This can happen for a target reset ccb. aic7xxx.reg: Reserve space for a completion queue. This will be used to enhance performance in the near future. aic7xxx.seq: Remove an optimization for the 7890 autoflush bug that turned out to allow, in rare cases, some data to get lost. Implement a simpler, faster, fix for the PCI_2_1 retry bug that hangs the sequencer on an SCB dma for certain chips. Test against SAVED_SCSIID rather than SELID during target reselections. This is how we always did it in the past, but the code was modified while trying to work around an issue with the 7895. SAVED_SCSIID takes into account twin channel adapters such as the 2742T, whereas SELID does not have the channel bit. This caused invalid selection warnings and other strangeness on these cards. aic7xxx_pci.c Use the correct mask for checking the generic aic7892 entry.
2001-02-10 18:04:27 +00:00
dma_scb_hang_dma_drain_fifo:
add SEQADDR0, A;
} else {
call dma_finish;
}
aic7xxx.c: Style nits. Make sure that our selection hardware is disabled as soon as possible after detecting a busfree and even go so far as to disable the selection hardware in advance of an event that will cause a busfree (ABORT or BUS DEVICE RESET message). The concern is that the selection hardware will select a target for which, after processing the bus free, there will be no commands pending. The sequencer idle loop will re-enable the selection should it still be necessary. In ahc_handle_scsiint(), clear SSTAT0 events several PCI transactions (most notably reads) prior to clearing SCSIINT. The newer chips seem to take a bit of time to see the change which can make the clearing of SCSIINT ineffective. Don't bother panicing at the end of ahc_handle_scsiint(). Getting to the final else just means we lost the race with clearing SCSIINT. In ahc_free(), handle init-level 0. This can happen when we fail the attach for RAID devices. While I'm here, also kill the parent dma tag. In ahc_match_scb(), consider initiator ccbs to be any that are not from the target mode group. This fixes a bug where an external target reset CCB was not getting cleaned up by the reset code. Don't bother freezing a ccb in any of our "abort" routines when the status is set to CAM_REQ_CMP. This can happen for a target reset ccb. aic7xxx.reg: Reserve space for a completion queue. This will be used to enhance performance in the near future. aic7xxx.seq: Remove an optimization for the 7890 autoflush bug that turned out to allow, in rare cases, some data to get lost. Implement a simpler, faster, fix for the PCI_2_1 retry bug that hangs the sequencer on an SCB dma for certain chips. Test against SAVED_SCSIID rather than SELID during target reselections. This is how we always did it in the past, but the code was modified while trying to work around an issue with the 7895. SAVED_SCSIID takes into account twin channel adapters such as the 2742T, whereas SELID does not have the channel bit. This caused invalid selection warnings and other strangeness on these cards. aic7xxx_pci.c Use the correct mask for checking the generic aic7892 entry.
2001-02-10 18:04:27 +00:00
/* If we were putting the SCB, we are done */
call dfdat_in_8;
call dfdat_in_8;
call dfdat_in_8;
dfdat_in_8:
mov DINDIR,DFDAT;
dfdat_in_7:
mov DINDIR,DFDAT;
mov DINDIR,DFDAT;
mov DINDIR,DFDAT;
mov DINDIR,DFDAT;
mov DINDIR,DFDAT;
dfdat_in_2:
mov DINDIR,DFDAT;
mov DINDIR,DFDAT ret;
}
copy_to_fifo_8:
mov DFDAT,SINDIR;
mov DFDAT,SINDIR;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
copy_to_fifo_6:
mov DFDAT,SINDIR;
copy_to_fifo_5:
mov DFDAT,SINDIR;
copy_to_fifo_4:
mov DFDAT,SINDIR;
mov DFDAT,SINDIR;
mov DFDAT,SINDIR;
mov DFDAT,SINDIR ret;
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
/*
* Wait for DMA from host memory to data FIFO to complete, then disable
* DMA and wait for it to acknowledge that it's off.
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
*/
dma_finish:
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
test DFSTATUS,HDONE jz dma_finish;
/* Turn off DMA */
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
and DFCNTRL, ~HDMAEN;
test DFCNTRL, HDMAEN jnz .;
ret;
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
add_scb_to_free_list:
if ((ahc->flags & AHC_PAGESCBS) != 0) {
BEGIN_CRITICAL
mov SCB_NEXT, FREE_SCBH;
Kill the "unpause_always" argument to unpause_sequencer(). The reasons for optimizing the unpause operation no-longer exist, and this is much safer. When restarting the sequencer, reconstitute the free SCB list on the card. This deals with a single instruction gap between marking the SCB as free and actually getting it onto the free list. Reduce the number of transfer negotiations that occur. In the past, we renegotiated after every reported check condition status. This ensures that we catch devices that have unexpectidly reset. In this situation, the target will always report the check condition before performing a data-phase. The new behavior is to renegotiate for any check-condition where the residual matches the orginal data-length of the command (including 0 length transffers). This avoids renegotiations during things like variable tape block reads, where the check condition is reported only to indicate the residual of the read. Revamp the parity error detection logic. We now properly report and handle injected parity errors in all phases. The old code used to hang on message-in parity errors. Correct the reporting of selection timeout errors to the XPT. When a selection timeout occurs, only the currently selecting command is flagged with SELTO status instead of aborting all currently active commands to that target. Fix flipped arguments in ahc_match_scb and in some of the callers of this routine. I wish that gcc allowed you to request warnings for enums passed as ints. Make ahc_find_msg generically handle all message types. Work around the target mode data-in wideodd bug in all non-U2 chips. We can now do sync-wide target mode transfers in target mode across the hole product line. Use lastphase exclusively for handling timeouts. The current phase doesn't take the bus free state into account. Fix a bug in the timeout handler that could cause corruption of the disconnected list. When sending an embedded cdb to a target, ensure that we start on a quad word boundary in the data-fifo. It seems that unaligned stores do not work correctly.
2000-02-09 21:25:00 +00:00
mvi SCB_TAG, SCB_LIST_NULL;
mov FREE_SCBH, SCBPTR ret;
END_CRITICAL
Kill the "unpause_always" argument to unpause_sequencer(). The reasons for optimizing the unpause operation no-longer exist, and this is much safer. When restarting the sequencer, reconstitute the free SCB list on the card. This deals with a single instruction gap between marking the SCB as free and actually getting it onto the free list. Reduce the number of transfer negotiations that occur. In the past, we renegotiated after every reported check condition status. This ensures that we catch devices that have unexpectidly reset. In this situation, the target will always report the check condition before performing a data-phase. The new behavior is to renegotiate for any check-condition where the residual matches the orginal data-length of the command (including 0 length transffers). This avoids renegotiations during things like variable tape block reads, where the check condition is reported only to indicate the residual of the read. Revamp the parity error detection logic. We now properly report and handle injected parity errors in all phases. The old code used to hang on message-in parity errors. Correct the reporting of selection timeout errors to the XPT. When a selection timeout occurs, only the currently selecting command is flagged with SELTO status instead of aborting all currently active commands to that target. Fix flipped arguments in ahc_match_scb and in some of the callers of this routine. I wish that gcc allowed you to request warnings for enums passed as ints. Make ahc_find_msg generically handle all message types. Work around the target mode data-in wideodd bug in all non-U2 chips. We can now do sync-wide target mode transfers in target mode across the hole product line. Use lastphase exclusively for handling timeouts. The current phase doesn't take the bus free state into account. Fix a bug in the timeout handler that could cause corruption of the disconnected list. When sending an embedded cdb to a target, ensure that we start on a quad word boundary in the data-fifo. It seems that unaligned stores do not work correctly.
2000-02-09 21:25:00 +00:00
} else {
mvi SCB_TAG, SCB_LIST_NULL ret;
}
if ((ahc->flags & AHC_PAGESCBS) != 0) {
get_free_or_disc_scb:
Sync Perforce IDs, add tranceiver state change support, and correct numerous error recovery buglets. Many thanks to Tor Egge for his assistance in diagnosing problems with the error recovery code. aic7xxx.c: Report missed bus free events using their own sequencer interrupt code to avoid confusion with other "bad phase" interrupts. Remove a delay used in debugging. This delay could only be hit in certain, very extreme, error recovery scenarios. Handle transceiver state changes correctly. You can now plug an SE device into a hot-plug LVD bus without hanging the controller. When stepping through a critical section, panic if we step more than a reasonable number of times. After a bus reset, disable bus reset interupts until we either our first attempt to (re)select another device, or another device attemps to select us. This removes the need to busy wait in kernel for the scsi reset line to fall yet still ensures we see any reset events that impact the state of either our initiator or target roles. Before this change, we had the potential of servicing a "storm" of reset interrupts if the reset line was held for a significant amount of time. Indicate the current sequencer address whenever we dump the card's state. aic7xxx.reg: Transceiver state change register definitions. Add the missed bussfree sequencer interrupt code. Re-enable the scsi reset interrupt if it has been disabled before every attempt to (re)select a device and when we have been selected as a target. When being (re)selected, check to see if the selection dissappeared just after we enabled our bus free interrupt. If the bus has gone free again, go back to the idle loop and wait for another selection. Note two locations where we should change our behavior if ATN is still raised. If ATN is raised during the presentation of a command complete or disconnect message, we should ignore the message and expect the target to put us in msgout phase. We don't currently do this as it requires some code re-arrangement so that critical sections can be properly placed around our handling of these two events. Otherwise, we cannot guarantee that the check of ATN is atomic relative to our acking of the message in byte (the kernel could assert ATN). Only set the IDENTIFY_SEEN flag after we have settled on the SCB for this transaction. The kernel looks at this flag before assuming that SCB_TAG is valid. This avoids confusion during certain types of error recovery. Add a critical section around findSCB. We cannot allow the kernel to remove an entry from the disconnected list while we are traversing it. Ditto for get_free_or_disc_scb. aic7xxx_freebsd.c: Only assume that SCB_TAG is accurate if IDENTIFY_SEEN is set in SEQ_FLAGS. Fix a typo that caused us to execute some code for the non-SCB paging case when paging SCBs. This only occurred during error recovery.
2000-11-10 20:13:41 +00:00
BEGIN_CRITICAL
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
cmp FREE_SCBH, SCB_LIST_NULL jne dequeue_free_scb;
cmp DISCONNECTED_SCBH, SCB_LIST_NULL jne dequeue_disc_scb;
return_error:
aic7xxx.c: When restarting the sequencer, ensure that the SCBCNT register is 0. A non-zero count will prevent the setting of the CCSCBDIR bit in any future dma operations. The only time CCSCBCNT would be non-zero is if we happened to halt the dma during a reset, but even that should never happen. Better safe than sorry. When a command completes before the target responds to an ATN for a recovery command, we now notify the kernel so that any recovery operation requeued in the qinfifo can be removed safely. In the past, we did this in ahc_done(), but ahc_done() may be called without the card paused. This also avoids a recursive call to ahc_search_qinifo() which could have occurred if ahc_search_qinififo() happened to be the routine to complete a recovery action. Fix 8bit math used for adjusting the qinfifo. The index must be wrapped properly within the 256 entry array. We rely on the fact that qinfifonext is a uint8_t in most cases to handle this wrap, but we missed a few spots where the resultant calculation was promoted to an int. Change the way that we deal with aborting the first or second entry from the qinfifo. We now swap the first entry in the qinfifo with the "next queued scb" to force the sequencer to see an abort collision if we ever touch the qinififo while the sequencer is mid SCB dma. aic7xxx.reg: Add new MKMSG_FAILED sequencer interrupt. This displaced the BOGUS_TAG interrupt used in some previous sequencer code debugging. aic7xxx.seq: Increment our position in the qinfifo only once the dma is complete and we have verified that the queue has not been changed during our DMA. This simplifies code in the kernel. Protect against "instruction creep" when issuing a pausing sequencer interrupt. On at least the 7890/91/96/97, the sequencer will coast after issuing the interrupt for up to two instructions. In the past we delt with this by using carefully placed nops. Now we call a routine to issue the interrupt followed by a nop and a ret. Tell the kernel should an SCB complete with the MK_MESSAGE flag still set. This means the target ignored our ATN request. Clear the channel twice as we exit the data phase. On the aic7890/91, the S/G preload logic may require the second clearing to get the last S/G out of the FIFO. aic7xxx_freebsd.c: Don't bother searching the qinfifo for a doubly queued recovery scb in ahc_done. This case is handled by the core driver now. Free the path used to issue async callbacks after the callback is complete. aic7xxx_inline.h: Split the SCB queue routine into a routine that swaps the SCB with the "next queued SCB" and a routine that calls the swapping routine and notifies the card of the new SCB. The swapping routine is now also used by ahc_search_qinfifo.
2000-11-06 20:05:38 +00:00
mvi NO_FREE_SCB call set_seqint;
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
mvi SINDEX, SCB_LIST_NULL ret;
dequeue_disc_scb:
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
mov SCBPTR, DISCONNECTED_SCBH;
Sync Perforce IDs, add tranceiver state change support, and correct numerous error recovery buglets. Many thanks to Tor Egge for his assistance in diagnosing problems with the error recovery code. aic7xxx.c: Report missed bus free events using their own sequencer interrupt code to avoid confusion with other "bad phase" interrupts. Remove a delay used in debugging. This delay could only be hit in certain, very extreme, error recovery scenarios. Handle transceiver state changes correctly. You can now plug an SE device into a hot-plug LVD bus without hanging the controller. When stepping through a critical section, panic if we step more than a reasonable number of times. After a bus reset, disable bus reset interupts until we either our first attempt to (re)select another device, or another device attemps to select us. This removes the need to busy wait in kernel for the scsi reset line to fall yet still ensures we see any reset events that impact the state of either our initiator or target roles. Before this change, we had the potential of servicing a "storm" of reset interrupts if the reset line was held for a significant amount of time. Indicate the current sequencer address whenever we dump the card's state. aic7xxx.reg: Transceiver state change register definitions. Add the missed bussfree sequencer interrupt code. Re-enable the scsi reset interrupt if it has been disabled before every attempt to (re)select a device and when we have been selected as a target. When being (re)selected, check to see if the selection dissappeared just after we enabled our bus free interrupt. If the bus has gone free again, go back to the idle loop and wait for another selection. Note two locations where we should change our behavior if ATN is still raised. If ATN is raised during the presentation of a command complete or disconnect message, we should ignore the message and expect the target to put us in msgout phase. We don't currently do this as it requires some code re-arrangement so that critical sections can be properly placed around our handling of these two events. Otherwise, we cannot guarantee that the check of ATN is atomic relative to our acking of the message in byte (the kernel could assert ATN). Only set the IDENTIFY_SEEN flag after we have settled on the SCB for this transaction. The kernel looks at this flag before assuming that SCB_TAG is valid. This avoids confusion during certain types of error recovery. Add a critical section around findSCB. We cannot allow the kernel to remove an entry from the disconnected list while we are traversing it. Ditto for get_free_or_disc_scb. aic7xxx_freebsd.c: Only assume that SCB_TAG is accurate if IDENTIFY_SEEN is set in SEQ_FLAGS. Fix a typo that caused us to execute some code for the non-SCB paging case when paging SCBs. This only occurred during error recovery.
2000-11-10 20:13:41 +00:00
mov DISCONNECTED_SCBH, SCB_NEXT;
END_CRITICAL
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
mvi DMAPARAMS, FIFORESET;
Sync Perforce IDs, add tranceiver state change support, and correct numerous error recovery buglets. Many thanks to Tor Egge for his assistance in diagnosing problems with the error recovery code. aic7xxx.c: Report missed bus free events using their own sequencer interrupt code to avoid confusion with other "bad phase" interrupts. Remove a delay used in debugging. This delay could only be hit in certain, very extreme, error recovery scenarios. Handle transceiver state changes correctly. You can now plug an SE device into a hot-plug LVD bus without hanging the controller. When stepping through a critical section, panic if we step more than a reasonable number of times. After a bus reset, disable bus reset interupts until we either our first attempt to (re)select another device, or another device attemps to select us. This removes the need to busy wait in kernel for the scsi reset line to fall yet still ensures we see any reset events that impact the state of either our initiator or target roles. Before this change, we had the potential of servicing a "storm" of reset interrupts if the reset line was held for a significant amount of time. Indicate the current sequencer address whenever we dump the card's state. aic7xxx.reg: Transceiver state change register definitions. Add the missed bussfree sequencer interrupt code. Re-enable the scsi reset interrupt if it has been disabled before every attempt to (re)select a device and when we have been selected as a target. When being (re)selected, check to see if the selection dissappeared just after we enabled our bus free interrupt. If the bus has gone free again, go back to the idle loop and wait for another selection. Note two locations where we should change our behavior if ATN is still raised. If ATN is raised during the presentation of a command complete or disconnect message, we should ignore the message and expect the target to put us in msgout phase. We don't currently do this as it requires some code re-arrangement so that critical sections can be properly placed around our handling of these two events. Otherwise, we cannot guarantee that the check of ATN is atomic relative to our acking of the message in byte (the kernel could assert ATN). Only set the IDENTIFY_SEEN flag after we have settled on the SCB for this transaction. The kernel looks at this flag before assuming that SCB_TAG is valid. This avoids confusion during certain types of error recovery. Add a critical section around findSCB. We cannot allow the kernel to remove an entry from the disconnected list while we are traversing it. Ditto for get_free_or_disc_scb. aic7xxx_freebsd.c: Only assume that SCB_TAG is accurate if IDENTIFY_SEEN is set in SEQ_FLAGS. Fix a typo that caused us to execute some code for the non-SCB paging case when paging SCBs. This only occurred during error recovery.
2000-11-10 20:13:41 +00:00
mov SCB_TAG jmp dma_scb;
BEGIN_CRITICAL
dequeue_free_scb:
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
mov SCBPTR, FREE_SCBH;
mov FREE_SCBH, SCB_NEXT ret;
Sync Perforce IDs, add tranceiver state change support, and correct numerous error recovery buglets. Many thanks to Tor Egge for his assistance in diagnosing problems with the error recovery code. aic7xxx.c: Report missed bus free events using their own sequencer interrupt code to avoid confusion with other "bad phase" interrupts. Remove a delay used in debugging. This delay could only be hit in certain, very extreme, error recovery scenarios. Handle transceiver state changes correctly. You can now plug an SE device into a hot-plug LVD bus without hanging the controller. When stepping through a critical section, panic if we step more than a reasonable number of times. After a bus reset, disable bus reset interupts until we either our first attempt to (re)select another device, or another device attemps to select us. This removes the need to busy wait in kernel for the scsi reset line to fall yet still ensures we see any reset events that impact the state of either our initiator or target roles. Before this change, we had the potential of servicing a "storm" of reset interrupts if the reset line was held for a significant amount of time. Indicate the current sequencer address whenever we dump the card's state. aic7xxx.reg: Transceiver state change register definitions. Add the missed bussfree sequencer interrupt code. Re-enable the scsi reset interrupt if it has been disabled before every attempt to (re)select a device and when we have been selected as a target. When being (re)selected, check to see if the selection dissappeared just after we enabled our bus free interrupt. If the bus has gone free again, go back to the idle loop and wait for another selection. Note two locations where we should change our behavior if ATN is still raised. If ATN is raised during the presentation of a command complete or disconnect message, we should ignore the message and expect the target to put us in msgout phase. We don't currently do this as it requires some code re-arrangement so that critical sections can be properly placed around our handling of these two events. Otherwise, we cannot guarantee that the check of ATN is atomic relative to our acking of the message in byte (the kernel could assert ATN). Only set the IDENTIFY_SEEN flag after we have settled on the SCB for this transaction. The kernel looks at this flag before assuming that SCB_TAG is valid. This avoids confusion during certain types of error recovery. Add a critical section around findSCB. We cannot allow the kernel to remove an entry from the disconnected list while we are traversing it. Ditto for get_free_or_disc_scb. aic7xxx_freebsd.c: Only assume that SCB_TAG is accurate if IDENTIFY_SEEN is set in SEQ_FLAGS. Fix a typo that caused us to execute some code for the non-SCB paging case when paging SCBs. This only occurred during error recovery.
2000-11-10 20:13:41 +00:00
END_CRITICAL
add_scb_to_disc_list:
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
/*
* Link this SCB into the DISCONNECTED list. This list holds the
* candidates for paging out an SCB if one is needed for a new command.
* Modifying the disconnected list is a critical(pause dissabled) section.
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
*/
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
BEGIN_CRITICAL
Makefile gram.y scan.l sequencer.h symbol.c symbol.h aic7xxx_asm.c: New sequencer assembler for the aic7xxx adapters. This assembler performs some amount of register type checking, allows bit manipulation of symbolic constants, and generates "patch tables" for conditionalized downloading of portions of the program. This makes it easier to take full advantage of the different features of the aic7xxx cards without imposing run time penalies or being bound to the small memory footprints of the low end cards for features like target mode. aic7xxx.reg: New, assembler parsed, register definitions fo the aic7xxx cards. This was done primarily in anticipation of 7810 support which will have a different register layout, but should be able to use the same assembler. The kernel aic7xxx driver consumes a generated file in the compile directory to get the definitions of the register locations. aic7xxx.seq: Convert to the slighly different syntax of the new assembler. Conditionalize SCB_PAGING, ultra, and twin features which shaves quite a bit of space once the program is downloaded. Add code to leave the selection hardware enabled during reconnects that win bus arbitration. This ensures that we will rearbitrate as soon as the bus goes free instead of delaying for a bit. When we expect the bus to go free, perform all of the cleanup associated with that event "up front" and enter a loop awaiting bus free. If we see a REQ first, complain, but attempt to continue. This will hopefully address, or at least help diagnose, the "target didn't send identify" messages that have been reported. Spelling corrections obtained from NetBSD.
1997-03-16 07:08:19 +00:00
mov SCB_NEXT, DISCONNECTED_SCBH;
mov DISCONNECTED_SCBH, SCBPTR ret;
aic7xxx.c: Filter incoming transfer negotiation requests to ensure they never exceed the settings specified by the user. In restart sequencer attempt to deal with a bug in the aic7895. If a third party reset occurs at just the right time, the stack register can lock up. When restarting the sequencer after handling the SCSI reset, poke SEQADDR1 before resting the sequencers program counter. When something strange happens, dump the card's transaction state via ahc_dump_card_state(). This should aid in debugging. Handle request sense transactions via the QINFIFO instead of attaching them to the waiting queue directly. The waiting queue consumes card SCB resources and, in the pathological case of every target on the bus beating our selection attemps and issuing a check condition, could have caused us to run out of SCBs. I have never seen this happen, and only early cards with 3 or 4 SCBs had any real chance of ever getting into this state. Add additional sequencer interrupt codes to support firmware diagnostics. The diagnostic code is enabled with the AHC_DEBUG_SEQUENCER kernel option. Make it possible to switch into and out of target mode on the fly. The card comes up by default as an initiator but will switch into target mode as soon as an enable lun operation is performed. As always, target mode behavior is gated by the AHC_TMODE_ENABLE kernel option so most users will not be affected by this change. In ahc_update_target_msg_request(), also issue a new request if the ppr_options have changed. Never issue a PPR as a target. It is forbidden by the spec. Correct a bug in ahc_parse_msg() that prevented us from responding to PPR messages as a target. Mark SCBs that are on the untagged queue with a flag instead of checking several fields in the SCB to see if the SCB should be on the queue. This makes it easier for things like automatic request sense requests to be queued without touching the untagged queues even though they are untagged requests. When dealing with ignore wide residue messages that occur in the middle of a transfer, reset HADDR, not SHADDR for non-ultra2 chips. Although SHADDR is where the firmware fetches the ending transfer address for a save data pointers request, it is readonly. Setting HADDR has the side effect of also updating SHADDR. Cleanup the output of ahc_dump_card_state() by nulling out the free scb list in the non-paging case. The free list is only used if we must page SCBs. Correct the transmission of cdbs > 12 bytes in length. When swapping HSCBs prior to notifing the sequencer of the new transaction, the bus address pointer for the cdb must also be recalculated to reflect its new location. We now defer the calculation of the cdb address until just before queing it to the card. When pulling transfer negotiation settings out of scratch ram, convert 5MHz/clock doubled settings to 10MHz. Add a new function ahc_qinfifo_requeue_tail() for use by error recovery actions and auto-request sense operations. These operations always occur when the sequencer is paused, so we can avoid the extra expense incurred in the normal SCB queue method. Use the BMOV instruction for all single byte moves on controllers that support it. The bmov instruction is twice as fast as an AND with an immediate of 0xFF as is used on older controllers. Correct a few bugs in ahc_dump_card_state(). If we have hardware assisted queue registers, use them to get the sequencer's idea of the head of the queue. When enumerating the untagged queue, it helps to use the correct index for the queue. aic7xxx.h: Indicate via a feature flag, which controllers can take on both the target and the initiator role at the same time. Add the AHC_SEQUENCER_DEBUG flag. Add the SCB_CDB32_PTR flag used for dealing with cdbs with lengths between 13 and 32 bytes. Add new prototypes. aic7xxx.reg: Allow the SCSIBUSL register to be written to. This is required to fix a selection timeout problem on the 7892/99. Cleanup the sequencer interrupt codes so that all debugging codes are grouped at the end of the list. Correct the definition of the ULTRA_ENB and DISC_DSB locations in scratch ram. This prevented the driver from properly honoring these settings when no serial eeprom was available. Remove an unused sequencer flag. aic7xxx.seq: Just before a potential select-out, clear the SCSIBUSL register. Occasionally, during a selection timeout, the contents of the register may be presented on the bus, causing much confusion. Add sequencer diagnostic code to detect software and or hardware bugs. The code attempts to verify most list operations so any corruption is caught before it occurs. We also track information about why a particular reconnection request was rejected. Don't clobber the digital REQ/ACK filter setting in SXFRCTL0 when clearing the channel. Fix a target mode bug that would cause us to return busy status instead of queue full in respnse to a tagged transaction. Cleanup the overrun case. It turns out that by simply butting the chip in bitbucket mode, it will ack any bytes until the phase changes. This drasticaly simplifies things. Prior to leaving the data phase, make sure that the S/G preload queue is empty. Remove code to place a request sense request on the waiting queue. This is all handled by the kernel now. Change the semantics of "findSCB". In the past, findSCB ensured that a freshly paged in SCB appeared on the disconnected list. The problem with this is that there is no guarantee that the paged in SCB is for a disconnected transation. We now defer any list manipulation to the caller who usually discards the SCB via the free list. Inline some busy target table operations. Add a critical section to protect adding an SCB to the disconnected list. aic7xxx_freebsd.c: Handle changes in the transfer negotiation setting API to filter incoming requests. No filtering is necessary for "goal" requests from the XPT. Set the SCB_CDB32_PTR flag when queing a transaction with a large cdb. In ahc_timeout, only take action if the active SCB is the timedout SCB. This deals with the case of two transactions to the same device with different timeout values. Use ahc_qinfifo_requeu_tail() instead of home grown version. aic7xxx_inline.h: Honor SCB_CDB32_PTR when queuing a new request. aic7xxx_pci.c: Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
END_CRITICAL
}
aic7xxx.c: When restarting the sequencer, ensure that the SCBCNT register is 0. A non-zero count will prevent the setting of the CCSCBDIR bit in any future dma operations. The only time CCSCBCNT would be non-zero is if we happened to halt the dma during a reset, but even that should never happen. Better safe than sorry. When a command completes before the target responds to an ATN for a recovery command, we now notify the kernel so that any recovery operation requeued in the qinfifo can be removed safely. In the past, we did this in ahc_done(), but ahc_done() may be called without the card paused. This also avoids a recursive call to ahc_search_qinifo() which could have occurred if ahc_search_qinififo() happened to be the routine to complete a recovery action. Fix 8bit math used for adjusting the qinfifo. The index must be wrapped properly within the 256 entry array. We rely on the fact that qinfifonext is a uint8_t in most cases to handle this wrap, but we missed a few spots where the resultant calculation was promoted to an int. Change the way that we deal with aborting the first or second entry from the qinfifo. We now swap the first entry in the qinfifo with the "next queued scb" to force the sequencer to see an abort collision if we ever touch the qinififo while the sequencer is mid SCB dma. aic7xxx.reg: Add new MKMSG_FAILED sequencer interrupt. This displaced the BOGUS_TAG interrupt used in some previous sequencer code debugging. aic7xxx.seq: Increment our position in the qinfifo only once the dma is complete and we have verified that the queue has not been changed during our DMA. This simplifies code in the kernel. Protect against "instruction creep" when issuing a pausing sequencer interrupt. On at least the 7890/91/96/97, the sequencer will coast after issuing the interrupt for up to two instructions. In the past we delt with this by using carefully placed nops. Now we call a routine to issue the interrupt followed by a nop and a ret. Tell the kernel should an SCB complete with the MK_MESSAGE flag still set. This means the target ignored our ATN request. Clear the channel twice as we exit the data phase. On the aic7890/91, the S/G preload logic may require the second clearing to get the last S/G out of the FIFO. aic7xxx_freebsd.c: Don't bother searching the qinfifo for a doubly queued recovery scb in ahc_done. This case is handled by the core driver now. Free the path used to issue async callbacks after the callback is complete. aic7xxx_inline.h: Split the SCB queue routine into a routine that swaps the SCB with the "next queued SCB" and a routine that calls the swapping routine and notifies the card of the new SCB. The swapping routine is now also used by ahc_search_qinfifo.
2000-11-06 20:05:38 +00:00
set_seqint:
mov INTSTAT, SINDEX;
nop;
o Convert to <inttypes.h> style fixed sized types to facilitate porting to other systems. o Normalize copyright text. o Clean up probe code function interfaces by passing around a single structure of common arguments instead of passing "too many" args in each function call. o Add support for the AAA-131 as a SCSI adapter. o Add support for the AHA-4944 courtesy of "Matthew N. Dodd" <winter@jurai.net o Correct manual termination support for PCI cards. The bit definitions for manual termination control in the SEEPROM were incorrect. o Add support for extracting NVRAM information from SCB 2 for BIOSen that use this mechanism to pass this data to OS drivers. o Properly set the STPWLEVEL bit in PCI config space based on the setting in an SEEPROM. o Go back to useing 32byte SCBs for all controllers. The current firmware allows us to embed 12byte cdbs on all controllers in a 32byte SCB, and larger cdbs are rarely used, so it is a better use of this space to offer more SCBs (32). o Add support for U160 transfers. o Add an idle loop executed during data transfers that prefetches S/G segments on controllers that have a secondary DMA engine (aic789X). o Improve the performance of reselections by avoiding an extra one byte DMA in the case of an SCB lookup miss for the reselecting target. We now keep a 16byte "untagged target" array on the card for dealing with untagged reselections. If the controller has external SCB ram and can support 64byte SCBs, then we use an "untagged target/lun" array to maximize concurrency. Without external SCB ram, the controller is limited to one untagged transaction per target, auto-request sense operations excluded. o Correct the setup of the STPWEN bit in SXFRCTL1. This control line is tri-stated until set to one, so set it to one and then set it to the desired value. o Add tagged queuing support to our target role implementation. o Handle the common cases of the ignore wide residue message in firmware. o Add preliminary support for 39bit addressing. o Add support for assembling on big-endian machines. Big-endian support is not complete in the driver. o Correctly remove SCBs in the waiting for selection queue when freezing a device queue. o Now that we understand more about the autoflush bug on the aic7890, only use the workaround on devices that need it. o Add a workaround for the "aic7890 hangs the system when you attempt to pause it" problem. We can now pause the aic7890 safely regardless of what instruction it is executing.
2000-07-18 20:12:14 +00:00
return:
ret;