d/freebsd-dev
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
761e526185
freebsd-dev/sys/dev/aac/aac.c

3804 lines
100 KiB
C
Raw Normal View History

A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
/*-
* Copyright (c) 2000 Michael Smith
Reformat for 80 columns. Sorry, but I had to do it.
2001-08-03 00:50:30 +00:00
* Copyright (c) 2001 Scott Long
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
* Copyright (c) 2000 BSDi
Reformat for 80 columns. Sorry, but I had to do it.
2001-08-03 00:50:30 +00:00
* Copyright (c) 2001 Adaptec, Inc.
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
*
* 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.
*/
Use __FBSDID(). Also some minor style cleanups.
2003-08-24 17:55:58 +00:00
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
/*
* Driver for the Adaptec 'FSA' family of PCI/SCSI RAID adapters.
*/
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
#define AAC_DRIVERNAME "aac"
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bugfixes. Close a race and logic bug in the timeout handling, don't call the interrupt handler from the upper half, etc. This fixes some serious stability problems that we were seeing on our production server. These patches have been tested for almost 6 months and are a highly recommended MFC candidate. Reviewed by: gibbs, merry, msmith MFC after: 4 days
2001-08-05 06:20:13 +00:00
#include "opt_aac.h"
Mega update to the aac driver. 1. Correctly handle commands initiated by the adapter. These commands are defered to a kthread responsible for their processing, then are properly returned to the controller. 2. Add/remove disk devices when notified by the card that a container was created/deleted/offline. 3. Implement crashdump functionality. 4. Support all ioctls needed for the management CLI to work. The linux version of this app can be found at the Dell or HP website. A native version will be forthcoming. MFC-after: 4.4-RELEASE
2001-08-29 23:34:05 +00:00
/* #include <stddef.h> */
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/malloc.h>
#include <sys/kernel.h>
Mega update to the aac driver. 1. Correctly handle commands initiated by the adapter. These commands are defered to a kthread responsible for their processing, then are properly returned to the controller. 2. Add/remove disk devices when notified by the card that a container was created/deleted/offline. 3. Implement crashdump functionality. 4. Support all ioctls needed for the management CLI to work. The linux version of this app can be found at the Dell or HP website. A native version will be forthcoming. MFC-after: 4.4-RELEASE
2001-08-29 23:34:05 +00:00
#include <sys/kthread.h>
Once again throw out the calculation of si_iosize_max and set it to 64K. Although it can go higher, it is not safe to so do on arrays with many members. Compromise by adding a tunable, "hw.aac.iosize_max" that can be set at boottime. Also document in the aac(4) manpage. MFC after: 4 weeks
2001-08-05 07:26:31 +00:00
#include <sys/sysctl.h>
Add functionality and fix bugs so the driver will work with soon-to-be released management apps. 1. Implement poll(). This will check for queued aif's so that a subsequent ioctl call to retrieve the next aif will not block. 2. Don't catch signals when sleeping on a fib sent from userland. This causes a race and panic due to the pthread context switcher waking up the tsleep at inopportune times. 3. Fix some whitespace nits. MFC after: 3 days
2001-12-02 03:47:33 +00:00
#include <sys/poll.h>
Use bioq_flush() to drain a bio queue with a specific error code. Retain the mistake of not updating the devstat API for now. Spell bioq_disksort() consistently with the remaining bioq_*(). #include <geom/geom_disk.h> where this is more appropriate.
2003-04-01 15:06:26 +00:00
#include <sys/ioccom.h>
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
#include <sys/bus.h>
#include <sys/conf.h>
#include <sys/signalvar.h>
Major bugfix and minor update. This should resolve the current issues with the driver locking up under load. - Restructure so that we use a static pool of commands/FIBs, rather than allocating them in clusters. The cluster allocation just made things more complicated, and allowed us to waste more memory in peak load situations. - Make queueing macros more like my other drivers. This adds queue stats for free. Add some debugging to take advantage of this. - Reimplement the periodic timeout scan. Kick the interrupt handler and the start routine every scan as well, just to be safe. Track busy commands properly. - Bring resource cleanup into line with resource allocation. We should now clean up correctly after a failed probe/unload/etc. - Try to start new commands when old ones are completed. We weren't doing this before, which could lead to deadlock when the controller was full. - Don't try to build a new command if we have found a deferred command. This could cause us to lose the deferred command. - Use diskerr() to report I/O errors. - Don't bail if the AdapterInfo structure is the wrong size. Some variation seems to be normal. We need to improve our handing of 2.x firmware sets. - Improve some comments in an attempt to try to make things clearer. - Restructure to avoid some warnings.
2000-12-27 13:14:56 +00:00
#include <sys/time.h>
Mega update to the aac driver. 1. Correctly handle commands initiated by the adapter. These commands are defered to a kthread responsible for their processing, then are properly returned to the controller. 2. Add/remove disk devices when notified by the card that a container was created/deleted/offline. 3. Implement crashdump functionality. 4. Support all ioctls needed for the management CLI to work. The linux version of this app can be found at the Dell or HP website. A native version will be forthcoming. MFC-after: 4.4-RELEASE
2001-08-29 23:34:05 +00:00
#include <sys/eventhandler.h>
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
#include <sys/rman.h>
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
#include <machine/bus.h>
Handle failures better in the passthrough bus creation code. Submitted by: Coverity Prevent analysis tool
2005-03-19 06:29:32 +00:00
#include <sys/bus_dma.h>
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
#include <machine/resource.h>
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
#include <dev/pci/pcireg.h>
#include <dev/pci/pcivar.h>
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
#include <dev/aac/aacreg.h>
Complete the repo-copy of aac_ioctl.h from sys/dev/aac/to sys/sys.
2004-12-09 22:20:25 +00:00
#include <sys/aac_ioctl.h>
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
#include <dev/aac/aacvar.h>
#include <dev/aac/aac_tables.h>
static void aac_startup(void *arg);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
static void aac_add_container(struct aac_softc *sc,
Prepare for a major update to the aac driver: Update the aac driver with the new crashdump api. Protect sync fibs with a mutex. Align all DMA buffers on a PAGE_SIZE boundary. MFC after: 3 days
2002-04-24 05:12:50 +00:00
struct aac_mntinforesp *mir, int f);
Add a CAM interface to the aac driver. This is useful in case you should ever connect a SCSI Cdrom/Tape/Jukebox/Scanner/Printer/kitty-litter-scooper to your high-end RAID controller. The interface to the arrays is still via the block interface; this merely provides a way to circumvent the RAID functionality and access the SCSI buses directly. Note that for somewhat obvious reasons, hard drives are not exposed to the da driver through this interface, though you can still talk to them via the pass driver. Be the first on your block to low-level format unsuspecting drives that are part of an array! To enable this, add the 'aacp' device to your kernel config. MFC after: 3 days
2002-04-27 01:31:17 +00:00
static void aac_get_bus_info(struct aac_softc *sc);
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
static void aac_daemon(void *arg);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
/* Command Processing */
Major bugfix and minor update. This should resolve the current issues with the driver locking up under load. - Restructure so that we use a static pool of commands/FIBs, rather than allocating them in clusters. The cluster allocation just made things more complicated, and allowed us to waste more memory in peak load situations. - Make queueing macros more like my other drivers. This adds queue stats for free. Add some debugging to take advantage of this. - Reimplement the periodic timeout scan. Kick the interrupt handler and the start routine every scan as well, just to be safe. Track busy commands properly. - Bring resource cleanup into line with resource allocation. We should now clean up correctly after a failed probe/unload/etc. - Try to start new commands when old ones are completed. We weren't doing this before, which could lead to deadlock when the controller was full. - Don't try to build a new command if we have found a deferred command. This could cause us to lose the deferred command. - Use diskerr() to report I/O errors. - Don't bail if the AdapterInfo structure is the wrong size. Some variation seems to be normal. We need to improve our handing of 2.x firmware sets. - Improve some comments in an attempt to try to make things clearer. - Restructure to avoid some warnings.
2000-12-27 13:14:56 +00:00
static void aac_timeout(struct aac_softc *sc);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
static void aac_complete(void *context, int pending);
static int aac_bio_command(struct aac_softc *sc, struct aac_command **cmp);
static void aac_bio_complete(struct aac_command *cm);
Remove the 'timeout' argument from aac_wait_command() as it isn't used and never will be. Update the XXX comment for this function to accurately reflect why things are the way they are.
2004-04-14 19:11:29 +00:00
static int aac_wait_command(struct aac_command *cm);
Various cleanups: - Move the command timeout check from a separate repeating timeout to the kthread since the kthread is already running periodically. - Move printing the hardware print buffer to the kthread. - Properly shut down the kernel thread on detach. - Detach the child array devices on detach. - Don't issue a controller halt command on detach. Doing so requires a PCI reset to wake the controller back up. The driver can now be unloaded as long as CAM support is not enabled.
2003-02-05 23:24:25 +00:00
static void aac_command_thread(struct aac_softc *sc);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
/* Command Buffer Management */
Handle the EINPROGRESS case of bus_dmamap_load() for data buffers.
2003-07-09 21:16:23 +00:00
static void aac_map_command_sg(void *arg, bus_dma_segment_t *segs,
int nseg, int error);
Reformat for 80 columns. Sorry, but I had to do it.
2001-08-03 00:50:30 +00:00
static void aac_map_command_helper(void *arg, bus_dma_segment_t *segs,
int nseg, int error);
Major bugfix and minor update. This should resolve the current issues with the driver locking up under load. - Restructure so that we use a static pool of commands/FIBs, rather than allocating them in clusters. The cluster allocation just made things more complicated, and allowed us to waste more memory in peak load situations. - Make queueing macros more like my other drivers. This adds queue stats for free. Add some debugging to take advantage of this. - Reimplement the periodic timeout scan. Kick the interrupt handler and the start routine every scan as well, just to be safe. Track busy commands properly. - Bring resource cleanup into line with resource allocation. We should now clean up correctly after a failed probe/unload/etc. - Try to start new commands when old ones are completed. We weren't doing this before, which could lead to deadlock when the controller was full. - Don't try to build a new command if we have found a deferred command. This could cause us to lose the deferred command. - Use diskerr() to report I/O errors. - Don't bail if the AdapterInfo structure is the wrong size. Some variation seems to be normal. We need to improve our handing of 2.x firmware sets. - Improve some comments in an attempt to try to make things clearer. - Restructure to avoid some warnings.
2000-12-27 13:14:56 +00:00
static int aac_alloc_commands(struct aac_softc *sc);
o Move the cleanup of the fib maps into aac_free_commands() so as to retain symetry with aac_alloc_commans(). Since aac_alloc_commands() allocates fib maps and places them onto the fib lists, aac_free_commands() should reverse those operations. o Combine two ifs with the same body with an ||. o Switch from uintptr_t to uint32_t for fib map load operations. The target is a uint32_t so using this type for the map load call avoids an extra cast. uintptr_t should only be used when you need an "int sized the same as the machine's poitner size" which is not the case here. o Removed the commented out M_WAITOK flag in the allocation in aac_alloc_commands(). The kernel will only block in the allocator if it can grow the size of the kernel. This usually results in a page-out which could involve this aac device. Thus, sleeping here could deadlock the machine. Assuming this operation is occurring outside of attach time, we have enough fibs to operate anyway, so waiting for fibs to free up is okay if not optimal. o In aac_alloc_commands(), if we cannot dmamem_alloc additional fib space, free the fib map. o In aac_alloc_commands(), if we cannot create per-command dmamaps, don't lose track of the fib map that is mapping all of the commands that we have already released into the free pool. Instead, just cut out of the loop and modify aac_free_commands to not attempt to free maps that have not been allocated. o Don't use a magic number when pre-allocating fibs. o Use PAGE_SIZE to allocate in page sized chunks instead of an architecture specific constant. Submitted by: gibbs
2003-02-19 21:38:29 +00:00
static void aac_free_commands(struct aac_softc *sc);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
static void aac_unmap_command(struct aac_command *cm);
/* Hardware Interface */
Diff reduction to Adaptec driver build 15317 (refactoring and code shuffling): - Resource allocation in aac_alloc (moved from from aac_init) - Interrupt setup in aac_setup_intr (from aac_attach) - Container probing in aac_get_container_info (from aac_startup and aac_handle_aif) - Firmware status check moved to aac_check_firmware from aac_init
2008-03-24 16:38:47 +00:00
static int aac_alloc(struct aac_softc *sc);
Reformat for 80 columns. Sorry, but I had to do it.
2001-08-03 00:50:30 +00:00
static void aac_common_map(void *arg, bus_dma_segment_t *segs, int nseg,
int error);
Don't attach to Dell PERC2/QC cards that have a firmware rev of 1.x. This check is complicated by the fact that the Adaptec 5400S cards claim to use 1.x firmware also. PERC2/QC 1.x firmware is not compatible with this driver and will cause a system hang. MFC after: 3 days
2002-02-06 01:34:09 +00:00
static int aac_check_firmware(struct aac_softc *sc);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
static int aac_init(struct aac_softc *sc);
static int aac_sync_command(struct aac_softc *sc, u_int32_t command,
Reformat for 80 columns. Sorry, but I had to do it.
2001-08-03 00:50:30 +00:00
u_int32_t arg0, u_int32_t arg1, u_int32_t arg2,
u_int32_t arg3, u_int32_t *sp);
Diff reduction to Adaptec driver build 15317 (refactoring and code shuffling): - Resource allocation in aac_alloc (moved from from aac_init) - Interrupt setup in aac_setup_intr (from aac_attach) - Container probing in aac_get_container_info (from aac_startup and aac_handle_aif) - Firmware status check moved to aac_check_firmware from aac_init
2008-03-24 16:38:47 +00:00
static int aac_setup_intr(struct aac_softc *sc);
Reformat for 80 columns. Sorry, but I had to do it.
2001-08-03 00:50:30 +00:00
static int aac_enqueue_fib(struct aac_softc *sc, int queue,
Bugfixes. Close a race and logic bug in the timeout handling, don't call the interrupt handler from the upper half, etc. This fixes some serious stability problems that we were seeing on our production server. These patches have been tested for almost 6 months and are a highly recommended MFC candidate. Reviewed by: gibbs, merry, msmith MFC after: 4 days
2001-08-05 06:20:13 +00:00
struct aac_command *cm);
Reformat for 80 columns. Sorry, but I had to do it.
2001-08-03 00:50:30 +00:00
static int aac_dequeue_fib(struct aac_softc *sc, int queue,
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
u_int32_t *fib_size, struct aac_fib **fib_addr);
Mega update to the aac driver. 1. Correctly handle commands initiated by the adapter. These commands are defered to a kthread responsible for their processing, then are properly returned to the controller. 2. Add/remove disk devices when notified by the card that a container was created/deleted/offline. 3. Implement crashdump functionality. 4. Support all ioctls needed for the management CLI to work. The linux version of this app can be found at the Dell or HP website. A native version will be forthcoming. MFC-after: 4.4-RELEASE
2001-08-29 23:34:05 +00:00
static int aac_enqueue_response(struct aac_softc *sc, int queue,
struct aac_fib *fib);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
/* StrongARM interface */
static int aac_sa_get_fwstatus(struct aac_softc *sc);
static void aac_sa_qnotify(struct aac_softc *sc, int qbit);
static int aac_sa_get_istatus(struct aac_softc *sc);
static void aac_sa_clear_istatus(struct aac_softc *sc, int mask);
static void aac_sa_set_mailbox(struct aac_softc *sc, u_int32_t command,
Reformat for 80 columns. Sorry, but I had to do it.
2001-08-03 00:50:30 +00:00
u_int32_t arg0, u_int32_t arg1,
u_int32_t arg2, u_int32_t arg3);
Begin support for 64-bit address support and workarounds for newer cards: - Add data structuress for doing 64-bit scatter/gather - Move busdma tag creations around so that only the parent is created in aac_pci.c. - Retrieve the capabilities word from the firmware before setting up command structures and tags. This allows the driver to decide whether to do 64-bit commands, and if work-arounds are needed for systems with >2GB of RAM. - Only enable the SCSI passthrough if it's enabled in the capabilities word in the firmware. This should fix problems with the 2120S and 2200S cards in systems with more than 2GB of RAM. Full 64-bit support is forthcoming. MFC-After: 1 week
2003-03-26 17:50:11 +00:00
static int aac_sa_get_mailbox(struct aac_softc *sc, int mb);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
static void aac_sa_set_interrupts(struct aac_softc *sc, int enable);
- Make tables, device ID strings etc const. This includes #ifdef'ing 0 aac_command_status_table, which is actually unused since r111532. While at it, make aac_if a pointer to the now const interface tables instead of copying them over to the softc (this alone already reduces the size of aac.ko on amd64 by ~1 KiB). - Remove redundant softc members. - Use DEVMETHOD_END. - Use NULL instead of 0 for pointers. - Remove redundant bzero(9)'ing of the softc. - Use pci_enable_busmaster(9) instead of duplicating it. - Remove redundant checking for PCIM_CMD_MEMEN (resource allocation will just fail). - Canonicalize the error messages in case of resource allocation failures. - Add support for using MSI instead of INTx, controllable via the tunable hw.aac.enable_msi (defaulting to on). MFC after: 1 month
2013-03-01 19:55:10 +00:00
const struct aac_interface aac_sa_interface = {
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
aac_sa_get_fwstatus,
aac_sa_qnotify,
aac_sa_get_istatus,
aac_sa_clear_istatus,
aac_sa_set_mailbox,
Begin support for 64-bit address support and workarounds for newer cards: - Add data structuress for doing 64-bit scatter/gather - Move busdma tag creations around so that only the parent is created in aac_pci.c. - Retrieve the capabilities word from the firmware before setting up command structures and tags. This allows the driver to decide whether to do 64-bit commands, and if work-arounds are needed for systems with >2GB of RAM. - Only enable the SCSI passthrough if it's enabled in the capabilities word in the firmware. This should fix problems with the 2120S and 2200S cards in systems with more than 2GB of RAM. Full 64-bit support is forthcoming. MFC-After: 1 week
2003-03-26 17:50:11 +00:00
aac_sa_get_mailbox,
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
aac_sa_set_interrupts,
NULL, NULL, NULL
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
};
Remove the unused AAC_DRIVER_BUILD_DATE macro. Nuke whitespace at EOL while I'm here. Approved by: scottl (MAINTAINER)
2005-11-13 17:26:36 +00:00
/* i960Rx interface */
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
static int aac_rx_get_fwstatus(struct aac_softc *sc);
static void aac_rx_qnotify(struct aac_softc *sc, int qbit);
static int aac_rx_get_istatus(struct aac_softc *sc);
static void aac_rx_clear_istatus(struct aac_softc *sc, int mask);
static void aac_rx_set_mailbox(struct aac_softc *sc, u_int32_t command,
Reformat for 80 columns. Sorry, but I had to do it.
2001-08-03 00:50:30 +00:00
u_int32_t arg0, u_int32_t arg1,
u_int32_t arg2, u_int32_t arg3);
Begin support for 64-bit address support and workarounds for newer cards: - Add data structuress for doing 64-bit scatter/gather - Move busdma tag creations around so that only the parent is created in aac_pci.c. - Retrieve the capabilities word from the firmware before setting up command structures and tags. This allows the driver to decide whether to do 64-bit commands, and if work-arounds are needed for systems with >2GB of RAM. - Only enable the SCSI passthrough if it's enabled in the capabilities word in the firmware. This should fix problems with the 2120S and 2200S cards in systems with more than 2GB of RAM. Full 64-bit support is forthcoming. MFC-After: 1 week
2003-03-26 17:50:11 +00:00
static int aac_rx_get_mailbox(struct aac_softc *sc, int mb);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
static void aac_rx_set_interrupts(struct aac_softc *sc, int enable);
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
static int aac_rx_send_command(struct aac_softc *sc, struct aac_command *cm);
static int aac_rx_get_outb_queue(struct aac_softc *sc);
static void aac_rx_set_outb_queue(struct aac_softc *sc, int index);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
- Make tables, device ID strings etc const. This includes #ifdef'ing 0 aac_command_status_table, which is actually unused since r111532. While at it, make aac_if a pointer to the now const interface tables instead of copying them over to the softc (this alone already reduces the size of aac.ko on amd64 by ~1 KiB). - Remove redundant softc members. - Use DEVMETHOD_END. - Use NULL instead of 0 for pointers. - Remove redundant bzero(9)'ing of the softc. - Use pci_enable_busmaster(9) instead of duplicating it. - Remove redundant checking for PCIM_CMD_MEMEN (resource allocation will just fail). - Canonicalize the error messages in case of resource allocation failures. - Add support for using MSI instead of INTx, controllable via the tunable hw.aac.enable_msi (defaulting to on). MFC after: 1 month
2013-03-01 19:55:10 +00:00
const struct aac_interface aac_rx_interface = {
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
aac_rx_get_fwstatus,
aac_rx_qnotify,
aac_rx_get_istatus,
aac_rx_clear_istatus,
aac_rx_set_mailbox,
Begin support for 64-bit address support and workarounds for newer cards: - Add data structuress for doing 64-bit scatter/gather - Move busdma tag creations around so that only the parent is created in aac_pci.c. - Retrieve the capabilities word from the firmware before setting up command structures and tags. This allows the driver to decide whether to do 64-bit commands, and if work-arounds are needed for systems with >2GB of RAM. - Only enable the SCSI passthrough if it's enabled in the capabilities word in the firmware. This should fix problems with the 2120S and 2200S cards in systems with more than 2GB of RAM. Full 64-bit support is forthcoming. MFC-After: 1 week
2003-03-26 17:50:11 +00:00
aac_rx_get_mailbox,
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
aac_rx_set_interrupts,
aac_rx_send_command,
aac_rx_get_outb_queue,
aac_rx_set_outb_queue
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
};
Remove the unused AAC_DRIVER_BUILD_DATE macro. Nuke whitespace at EOL while I'm here. Approved by: scottl (MAINTAINER)
2005-11-13 17:26:36 +00:00
/* Rocket/MIPS interface */
Add support for the Adaptec RAID-On-Chip architecture. This in turn provides support for the Adaptec 2130S adapter. Thanks to Adaptec for providing hardware for this.
2004-08-13 01:44:09 +00:00
static int aac_rkt_get_fwstatus(struct aac_softc *sc);
static void aac_rkt_qnotify(struct aac_softc *sc, int qbit);
static int aac_rkt_get_istatus(struct aac_softc *sc);
static void aac_rkt_clear_istatus(struct aac_softc *sc, int mask);
static void aac_rkt_set_mailbox(struct aac_softc *sc, u_int32_t command,
u_int32_t arg0, u_int32_t arg1,
u_int32_t arg2, u_int32_t arg3);
static int aac_rkt_get_mailbox(struct aac_softc *sc, int mb);
static void aac_rkt_set_interrupts(struct aac_softc *sc, int enable);
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
static int aac_rkt_send_command(struct aac_softc *sc, struct aac_command *cm);
static int aac_rkt_get_outb_queue(struct aac_softc *sc);
static void aac_rkt_set_outb_queue(struct aac_softc *sc, int index);
Add support for the Adaptec RAID-On-Chip architecture. This in turn provides support for the Adaptec 2130S adapter. Thanks to Adaptec for providing hardware for this.
2004-08-13 01:44:09 +00:00
- Make tables, device ID strings etc const. This includes #ifdef'ing 0 aac_command_status_table, which is actually unused since r111532. While at it, make aac_if a pointer to the now const interface tables instead of copying them over to the softc (this alone already reduces the size of aac.ko on amd64 by ~1 KiB). - Remove redundant softc members. - Use DEVMETHOD_END. - Use NULL instead of 0 for pointers. - Remove redundant bzero(9)'ing of the softc. - Use pci_enable_busmaster(9) instead of duplicating it. - Remove redundant checking for PCIM_CMD_MEMEN (resource allocation will just fail). - Canonicalize the error messages in case of resource allocation failures. - Add support for using MSI instead of INTx, controllable via the tunable hw.aac.enable_msi (defaulting to on). MFC after: 1 month
2013-03-01 19:55:10 +00:00
const struct aac_interface aac_rkt_interface = {
Add support for the Adaptec RAID-On-Chip architecture. This in turn provides support for the Adaptec 2130S adapter. Thanks to Adaptec for providing hardware for this.
2004-08-13 01:44:09 +00:00
aac_rkt_get_fwstatus,
aac_rkt_qnotify,
aac_rkt_get_istatus,
aac_rkt_clear_istatus,
aac_rkt_set_mailbox,
aac_rkt_get_mailbox,
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
aac_rkt_set_interrupts,
aac_rkt_send_command,
aac_rkt_get_outb_queue,
aac_rkt_set_outb_queue
Add support for the Adaptec RAID-On-Chip architecture. This in turn provides support for the Adaptec 2130S adapter. Thanks to Adaptec for providing hardware for this.
2004-08-13 01:44:09 +00:00
};
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
/* Debugging and Diagnostics */
- Make tables, device ID strings etc const. This includes #ifdef'ing 0 aac_command_status_table, which is actually unused since r111532. While at it, make aac_if a pointer to the now const interface tables instead of copying them over to the softc (this alone already reduces the size of aac.ko on amd64 by ~1 KiB). - Remove redundant softc members. - Use DEVMETHOD_END. - Use NULL instead of 0 for pointers. - Remove redundant bzero(9)'ing of the softc. - Use pci_enable_busmaster(9) instead of duplicating it. - Remove redundant checking for PCIM_CMD_MEMEN (resource allocation will just fail). - Canonicalize the error messages in case of resource allocation failures. - Add support for using MSI instead of INTx, controllable via the tunable hw.aac.enable_msi (defaulting to on). MFC after: 1 month
2013-03-01 19:55:10 +00:00
static void aac_describe_controller(struct aac_softc *sc);
static const char *aac_describe_code(const struct aac_code_lookup *table,
Reformat for 80 columns. Sorry, but I had to do it.
2001-08-03 00:50:30 +00:00
u_int32_t code);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
/* Management Interface */
static d_open_t aac_open;
static d_ioctl_t aac_ioctl;
Add functionality and fix bugs so the driver will work with soon-to-be released management apps. 1. Implement poll(). This will check for queued aif's so that a subsequent ioctl call to retrieve the next aif will not block. 2. Don't catch signals when sleeping on a fib sent from userland. This causes a race and panic due to the pthread context switcher waking up the tsleep at inopportune times. 3. Fix some whitespace nits. MFC after: 3 days
2001-12-02 03:47:33 +00:00
static d_poll_t aac_poll;
Implement device unbusying via a cdevpriv destructor. Suggested by: jhb Tested by: Mark Johnston <mjohnston at sandvine dot com> Reviewed by: emaste, jhb MFC after: 10 days X-MFC: r212661
2010-09-16 17:49:10 +00:00
static void aac_cdevpriv_dtor(void *arg);
Reformat for 80 columns. Sorry, but I had to do it.
2001-08-03 00:50:30 +00:00
static int aac_ioctl_sendfib(struct aac_softc *sc, caddr_t ufib);
Add ioctls FSACTL_SEND_LARGE_FIB, FSACTL_SEND_RAW_SRB, FSACTL_LNX_SEND_LARGE_FIB, and FSACTL_LNX_SEND_RAW_SRB, and correct size checks on FIBs passed in from userspace. Both changes were obtained from Adaptec's driver build 15317. Adaptec's commandline RAID tool arcconf uses these ioctls when creating a RAID-10 array (and probably other operations too).
2008-03-20 17:59:19 +00:00
static int aac_ioctl_send_raw_srb(struct aac_softc *sc, caddr_t arg);
Reformat for 80 columns. Sorry, but I had to do it.
2001-08-03 00:50:30 +00:00
static void aac_handle_aif(struct aac_softc *sc,
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
struct aac_fib *fib);
Put the linux compat ioctl interface on death row. Add FreeBSD definitions of all of the ioctls, plus round out all ioctl definitions to match what exists for linux. Allow ioctls to be called through either the native or linux interface, though no apps exist (yet) that can take advantage of native calling.
2001-08-06 04:34:47 +00:00
static int aac_rev_check(struct aac_softc *sc, caddr_t udata);
Allow simultaneous opens of the device for issuing commands to the controller. This is merged from Adaptec driver build 11669.
2007-12-07 00:22:23 +00:00
static int aac_open_aif(struct aac_softc *sc, caddr_t arg);
static int aac_close_aif(struct aac_softc *sc, caddr_t arg);
Put the linux compat ioctl interface on death row. Add FreeBSD definitions of all of the ioctls, plus round out all ioctl definitions to match what exists for linux. Allow ioctls to be called through either the native or linux interface, though no apps exist (yet) that can take advantage of native calling.
2001-08-06 04:34:47 +00:00
static int aac_getnext_aif(struct aac_softc *sc, caddr_t arg);
Allow simultaneous opens of the device for issuing commands to the controller. This is merged from Adaptec driver build 11669.
2007-12-07 00:22:23 +00:00
static int aac_return_aif(struct aac_softc *sc,
struct aac_fib_context *ctx, caddr_t uptr);
Mega update to the aac driver. 1. Correctly handle commands initiated by the adapter. These commands are defered to a kthread responsible for their processing, then are properly returned to the controller. 2. Add/remove disk devices when notified by the card that a container was created/deleted/offline. 3. Implement crashdump functionality. 4. Support all ioctls needed for the management CLI to work. The linux version of this app can be found at the Dell or HP website. A native version will be forthcoming. MFC-after: 4.4-RELEASE
2001-08-29 23:34:05 +00:00
static int aac_query_disk(struct aac_softc *sc, caddr_t uptr);
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
static int aac_get_pci_info(struct aac_softc *sc, caddr_t uptr);
Implement FSACTL_LNX_GET_FEATURES and FSACTL_GET_FEATURES ioctls. RAID tools (e.g. arcconf) need this to be able to create arrays larger than 2TB. Submitted by: Adaptec, via driver build 15317
2008-03-28 19:07:25 +00:00
static int aac_supported_features(struct aac_softc *sc, caddr_t uptr);
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
static void aac_ioctl_event(struct aac_softc *sc,
Whitespace fixup: 8 spaces -> tab
2009-10-27 19:37:37 +00:00
struct aac_event *event, void *arg);
Diff reduction to Adaptec driver build 15317 (refactoring and code shuffling): - Resource allocation in aac_alloc (moved from from aac_init) - Interrupt setup in aac_setup_intr (from aac_attach) - Container probing in aac_get_container_info (from aac_startup and aac_handle_aif) - Firmware status check moved to aac_check_firmware from aac_init
2008-03-24 16:38:47 +00:00
static struct aac_mntinforesp *
aac_get_container_info(struct aac_softc *sc, struct aac_fib *fib, int cid);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
static struct cdevsw aac_cdevsw = {
Device megapatch 4/6: Introduce d_version field in struct cdevsw, this must always be initialized to D_VERSION. Flip sense of D_NOGIANT flag to D_NEEDGIANT, this involves removing four D_NOGIANT flags and adding 145 D_NEEDGIANT flags.
2004-02-21 21:10:55 +00:00
.d_version = D_VERSION,
Implement device unbusying via a cdevpriv destructor. Suggested by: jhb Tested by: Mark Johnston <mjohnston at sandvine dot com> Reviewed by: emaste, jhb MFC after: 10 days X-MFC: r212661
2010-09-16 17:49:10 +00:00
.d_flags = D_NEEDGIANT,
Gigacommit to improve device-driver source compatibility between branches: Initialize struct cdevsw using C99 sparse initializtion and remove all initializations to default values. This patch is automatically generated and has been tested by compiling LINT with all the fields in struct cdevsw in reverse order on alpha, sparc64 and i386. Approved by: re(scottl)
2003-03-03 12:15:54 +00:00
.d_open = aac_open,
.d_ioctl = aac_ioctl,
.d_poll = aac_poll,
.d_name = "aac",
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
};
Mark MALLOC_DEFINEs static that have no corresponding MALLOC_DECLAREs. This means that their use is restricted to a single C file.
2011-11-07 06:44:47 +00:00
static MALLOC_DEFINE(M_AACBUF, "aacbuf", "Buffers for the AAC driver");
Mega update to the aac driver. 1. Correctly handle commands initiated by the adapter. These commands are defered to a kthread responsible for their processing, then are properly returned to the controller. 2. Add/remove disk devices when notified by the card that a container was created/deleted/offline. 3. Implement crashdump functionality. 4. Support all ioctls needed for the management CLI to work. The linux version of this app can be found at the Dell or HP website. A native version will be forthcoming. MFC-after: 4.4-RELEASE
2001-08-29 23:34:05 +00:00
Once again throw out the calculation of si_iosize_max and set it to 64K. Although it can go higher, it is not safe to so do on arrays with many members. Compromise by adding a tunable, "hw.aac.iosize_max" that can be set at boottime. Also document in the aac(4) manpage. MFC after: 4 weeks
2001-08-05 07:26:31 +00:00
/* sysctl node */
- Make tables, device ID strings etc const. This includes #ifdef'ing 0 aac_command_status_table, which is actually unused since r111532. While at it, make aac_if a pointer to the now const interface tables instead of copying them over to the softc (this alone already reduces the size of aac.ko on amd64 by ~1 KiB). - Remove redundant softc members. - Use DEVMETHOD_END. - Use NULL instead of 0 for pointers. - Remove redundant bzero(9)'ing of the softc. - Use pci_enable_busmaster(9) instead of duplicating it. - Remove redundant checking for PCIM_CMD_MEMEN (resource allocation will just fail). - Canonicalize the error messages in case of resource allocation failures. - Add support for using MSI instead of INTx, controllable via the tunable hw.aac.enable_msi (defaulting to on). MFC after: 1 month
2013-03-01 19:55:10 +00:00
SYSCTL_NODE(_hw, OID_AUTO, aac, CTLFLAG_RD, 0, "AAC driver parameters");
Once again throw out the calculation of si_iosize_max and set it to 64K. Although it can go higher, it is not safe to so do on arrays with many members. Compromise by adding a tunable, "hw.aac.iosize_max" that can be set at boottime. Also document in the aac(4) manpage. MFC after: 4 weeks
2001-08-05 07:26:31 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
* Device Interface
*/
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
Change spelling and eliminate a typo in comments to reduce diffs with Adaptec's vendor driver. I have some fixes to bring in and this makes ongoing review of the FreeBSD-Adaptec driver diffs easier.
2008-03-14 21:59:11 +00:00
* Initialize the controller and softc
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
*/
int
aac_attach(struct aac_softc *sc)
{
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
int error, unit;
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
Change spelling and eliminate a typo in comments to reduce diffs with Adaptec's vendor driver. I have some fixes to bring in and this makes ongoing review of the FreeBSD-Adaptec driver diffs easier.
2008-03-14 21:59:11 +00:00
* Initialize per-controller queues.
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
*/
aac_initq_free(sc);
aac_initq_ready(sc);
aac_initq_busy(sc);
aac_initq_bio(sc);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
Change spelling and eliminate a typo in comments to reduce diffs with Adaptec's vendor driver. I have some fixes to bring in and this makes ongoing review of the FreeBSD-Adaptec driver diffs easier.
2008-03-14 21:59:11 +00:00
* Initialize command-completion task.
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
*/
TASK_INIT(&sc->aac_task_complete, 0, aac_complete, sc);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/* mark controller as suspended until we get ourselves organised */
sc->aac_state |= AAC_STATE_SUSPEND;
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Don't attach to Dell PERC2/QC cards that have a firmware rev of 1.x. This check is complicated by the fact that the Adaptec 5400S cards claim to use 1.x firmware also. PERC2/QC 1.x firmware is not compatible with this driver and will cause a system hang. MFC after: 3 days
2002-02-06 01:34:09 +00:00
/*
* Check that the firmware on the card is supported.
*/
if ((error = aac_check_firmware(sc)) != 0)
return(error);
Mega busdma API commit. Add two new arguments to bus_dma_tag_create(): lockfunc and lockfuncarg. Lockfunc allows a driver to provide a function for managing its locking semantics while using busdma. At the moment, this is used for the asynchronous busdma_swi and callback mechanism. Two lockfunc implementations are provided: busdma_lock_mutex() performs standard mutex operations on the mutex that is specified from lockfuncarg. dftl_lock() is a panic implementation and is defaulted to when NULL, NULL are passed to bus_dma_tag_create(). The only time that NULL, NULL should ever be used is when the driver ensures that bus_dmamap_load() will not be deferred. Drivers that do not provide their own locking can pass busdma_lock_mutex,&Giant args in order to preserve the former behaviour. sparc64 and powerpc do not provide real busdma_swi functions, so this is largely a noop on those platforms. The busdma_swi on is64 is not properly locked yet, so warnings will be emitted on this platform when busdma callback deferrals happen. If anyone gets panics or warnings from dflt_lock() being called, please let me know right away. Reviewed by: tmm, gibbs
2003-07-01 15:52:06 +00:00
/*
* Initialize locks
*/
Remove the AAC_LOCK macros. They no longer abstract anything and only obfuscate the code. No functional differences.
2004-08-12 05:05:06 +00:00
mtx_init(&sc->aac_aifq_lock, "AAC AIF lock", NULL, MTX_DEF);
mtx_init(&sc->aac_io_lock, "AAC I/O lock", NULL, MTX_DEF);
mtx_init(&sc->aac_container_lock, "AAC container lock", NULL, MTX_DEF);
Mega busdma API commit. Add two new arguments to bus_dma_tag_create(): lockfunc and lockfuncarg. Lockfunc allows a driver to provide a function for managing its locking semantics while using busdma. At the moment, this is used for the asynchronous busdma_swi and callback mechanism. Two lockfunc implementations are provided: busdma_lock_mutex() performs standard mutex operations on the mutex that is specified from lockfuncarg. dftl_lock() is a panic implementation and is defaulted to when NULL, NULL are passed to bus_dma_tag_create(). The only time that NULL, NULL should ever be used is when the driver ensures that bus_dmamap_load() will not be deferred. Drivers that do not provide their own locking can pass busdma_lock_mutex,&Giant args in order to preserve the former behaviour. sparc64 and powerpc do not provide real busdma_swi functions, so this is largely a noop on those platforms. The busdma_swi on is64 is not properly locked yet, so warnings will be emitted on this platform when busdma callback deferrals happen. If anyone gets panics or warnings from dflt_lock() being called, please let me know right away. Reviewed by: tmm, gibbs
2003-07-01 15:52:06 +00:00
TAILQ_INIT(&sc->aac_container_tqh);
Initialize the event tailq. Submitted by: Frank Mayhar PR: kern/90882 MFC After: 1 day
2005-12-28 21:18:55 +00:00
TAILQ_INIT(&sc->aac_ev_cmfree);
Mega busdma API commit. Add two new arguments to bus_dma_tag_create(): lockfunc and lockfuncarg. Lockfunc allows a driver to provide a function for managing its locking semantics while using busdma. At the moment, this is used for the asynchronous busdma_swi and callback mechanism. Two lockfunc implementations are provided: busdma_lock_mutex() performs standard mutex operations on the mutex that is specified from lockfuncarg. dftl_lock() is a panic implementation and is defaulted to when NULL, NULL are passed to bus_dma_tag_create(). The only time that NULL, NULL should ever be used is when the driver ensures that bus_dmamap_load() will not be deferred. Drivers that do not provide their own locking can pass busdma_lock_mutex,&Giant args in order to preserve the former behaviour. sparc64 and powerpc do not provide real busdma_swi functions, so this is largely a noop on those platforms. The busdma_swi on is64 is not properly locked yet, so warnings will be emitted on this platform when busdma callback deferrals happen. If anyone gets panics or warnings from dflt_lock() being called, please let me know right away. Reviewed by: tmm, gibbs
2003-07-01 15:52:06 +00:00
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
/* Initialize the clock daemon callout. */
callout_init_mtx(&sc->aac_daemontime, &sc->aac_io_lock, 0);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
Change spelling and eliminate a typo in comments to reduce diffs with Adaptec's vendor driver. I have some fixes to bring in and this makes ongoing review of the FreeBSD-Adaptec driver diffs easier.
2008-03-14 21:59:11 +00:00
* Initialize the adapter.
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
*/
Diff reduction to Adaptec driver build 15317 (refactoring and code shuffling): - Resource allocation in aac_alloc (moved from from aac_init) - Interrupt setup in aac_setup_intr (from aac_attach) - Container probing in aac_get_container_info (from aac_startup and aac_handle_aif) - Firmware status check moved to aac_check_firmware from aac_init
2008-03-24 16:38:47 +00:00
if ((error = aac_alloc(sc)) != 0)
return(error);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
if ((error = aac_init(sc)) != 0)
return(error);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Remove the unused AAC_DRIVER_BUILD_DATE macro. Nuke whitespace at EOL while I'm here. Approved by: scottl (MAINTAINER)
2005-11-13 17:26:36 +00:00
/*
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
* Allocate and connect our interrupt.
*/
Diff reduction to Adaptec driver build 15317 (refactoring and code shuffling): - Resource allocation in aac_alloc (moved from from aac_init) - Interrupt setup in aac_setup_intr (from aac_attach) - Container probing in aac_get_container_info (from aac_startup and aac_handle_aif) - Firmware status check moved to aac_check_firmware from aac_init
2008-03-24 16:38:47 +00:00
if ((error = aac_setup_intr(sc)) != 0)
return(error);
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
Remove the unused AAC_DRIVER_BUILD_DATE macro. Nuke whitespace at EOL while I'm here. Approved by: scottl (MAINTAINER)
2005-11-13 17:26:36 +00:00
/*
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
* Print a little information about the controller.
*/
aac_describe_controller(sc);
Add a sysctl to report the firmware build number. Some older firmware versions have issues that can be worked around by avoiding certain operations. Add a sysctl dev.aac.#.firmware_build to make it easy for scripts or userland tools to detect the firmware version.
2012-02-13 16:48:49 +00:00
/*
* Add sysctls.
*/
SYSCTL_ADD_INT(device_get_sysctl_ctx(sc->aac_dev),
SYSCTL_CHILDREN(device_get_sysctl_tree(sc->aac_dev)),
OID_AUTO, "firmware_build", CTLFLAG_RD,
&sc->aac_revision.buildNumber, 0,
"firmware build number");
Add functionality and fix bugs so the driver will work with soon-to-be released management apps. 1. Implement poll(). This will check for queued aif's so that a subsequent ioctl call to retrieve the next aif will not block. 2. Don't catch signals when sleeping on a fib sent from userland. This causes a race and panic due to the pthread context switcher waking up the tsleep at inopportune times. 3. Fix some whitespace nits. MFC after: 3 days
2001-12-02 03:47:33 +00:00
/*
Bring aac out from under Giant: - the mutex aac_io_lock protects the main codepaths which handle queues and hardware registers. Only one acquire/release is done in the top-half and the taskqueue. This mutex also applies to the userland command path and CAM data path. - Move the taskqueue to the new Giant-free version. - Register the disk device with DISKFLAG_NOGIANT so the top-half processing runs without Giant. - Move the dynamic command allocator to the worker thread to avoid locking issues with bus_dmamem_alloc(). This gives about 20% improvement in most of my benchmarks.
2003-02-26 04:46:21 +00:00
* Register to probe our containers later.
Add functionality and fix bugs so the driver will work with soon-to-be released management apps. 1. Implement poll(). This will check for queued aif's so that a subsequent ioctl call to retrieve the next aif will not block. 2. Don't catch signals when sleeping on a fib sent from userland. This causes a race and panic due to the pthread context switcher waking up the tsleep at inopportune times. 3. Fix some whitespace nits. MFC after: 3 days
2001-12-02 03:47:33 +00:00
*/
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
sc->aac_ich.ich_func = aac_startup;
sc->aac_ich.ich_arg = sc;
if (config_intrhook_establish(&sc->aac_ich) != 0) {
device_printf(sc->aac_dev,
"can't establish configuration hook\n");
return(ENXIO);
}
/*
* Make the control device.
*/
unit = device_get_unit(sc->aac_dev);
Use OPERATOR instead of WHEEL for make_dev(), use 0640 instead of 0644 for disk devices to be consistent with other storage devices. Submitted by: kris Reviewed by: scottl
2002-12-27 17:52:16 +00:00
sc->aac_dev_t = make_dev(&aac_cdevsw, unit, UID_ROOT, GID_OPERATOR,
0640, "aac%d", unit);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
(void)make_dev_alias(sc->aac_dev_t, "afa%d", unit);
(void)make_dev_alias(sc->aac_dev_t, "hpn%d", unit);
sc->aac_dev_t->si_drv1 = sc;
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/* Create the AIF thread */
Rename the kthread_xxx (e.g. kthread_create()) calls to kproc_xxx as they actually make whole processes. Thos makes way for us to add REAL kthread_create() and friends that actually make theads. it turns out that most of these calls actually end up being moved back to the thread version when it's added. but we need to make this cosmetic change first. I'd LOVE to do this rename in 7.0 so that we can eventually MFC the new kthread_xxx() calls.
2007-10-20 23:23:23 +00:00
if (kproc_create((void(*)(void *))aac_command_thread, sc,
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
&sc->aifthread, 0, 0, "aac%daif", unit))
Add \n to the end of a printf string and remove it from panic strings.
2008-03-26 15:28:56 +00:00
panic("Could not create AIF thread");
Mega update to the aac driver. 1. Correctly handle commands initiated by the adapter. These commands are defered to a kthread responsible for their processing, then are properly returned to the controller. 2. Add/remove disk devices when notified by the card that a container was created/deleted/offline. 3. Implement crashdump functionality. 4. Support all ioctls needed for the management CLI to work. The linux version of this app can be found at the Dell or HP website. A native version will be forthcoming. MFC-after: 4.4-RELEASE
2001-08-29 23:34:05 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/* Register the shutdown method to only be called post-dump */
Don't include aac_cam.h, since it was nuked in the last commit. Deregister the shutdown eventhandler on unload.
2003-02-06 00:07:12 +00:00
if ((sc->eh = EVENTHANDLER_REGISTER(shutdown_final, aac_shutdown,
sc->aac_dev, SHUTDOWN_PRI_DEFAULT)) == NULL)
device_printf(sc->aac_dev,
"shutdown event registration failed\n");
Mega update to the aac driver. 1. Correctly handle commands initiated by the adapter. These commands are defered to a kthread responsible for their processing, then are properly returned to the controller. 2. Add/remove disk devices when notified by the card that a container was created/deleted/offline. 3. Implement crashdump functionality. 4. Support all ioctls needed for the management CLI to work. The linux version of this app can be found at the Dell or HP website. A native version will be forthcoming. MFC-after: 4.4-RELEASE
2001-08-29 23:34:05 +00:00
Add a CAM interface to the aac driver. This is useful in case you should ever connect a SCSI Cdrom/Tape/Jukebox/Scanner/Printer/kitty-litter-scooper to your high-end RAID controller. The interface to the arrays is still via the block interface; this merely provides a way to circumvent the RAID functionality and access the SCSI buses directly. Note that for somewhat obvious reasons, hard drives are not exposed to the da driver through this interface, though you can still talk to them via the pass driver. Be the first on your block to low-level format unsuspecting drives that are part of an array! To enable this, add the 'aacp' device to your kernel config. MFC after: 3 days
2002-04-27 01:31:17 +00:00
/* Register with CAM for the non-DASD devices */
Begin support for 64-bit address support and workarounds for newer cards: - Add data structuress for doing 64-bit scatter/gather - Move busdma tag creations around so that only the parent is created in aac_pci.c. - Retrieve the capabilities word from the firmware before setting up command structures and tags. This allows the driver to decide whether to do 64-bit commands, and if work-arounds are needed for systems with >2GB of RAM. - Only enable the SCSI passthrough if it's enabled in the capabilities word in the firmware. This should fix problems with the 2120S and 2200S cards in systems with more than 2GB of RAM. Full 64-bit support is forthcoming. MFC-After: 1 week
2003-03-26 17:50:11 +00:00
if ((sc->flags & AAC_FLAGS_ENABLE_CAM) != 0) {
Various cleanups: - Move the command timeout check from a separate repeating timeout to the kthread since the kthread is already running periodically. - Move printing the hardware print buffer to the kthread. - Properly shut down the kernel thread on detach. - Detach the child array devices on detach. - Don't issue a controller halt command on detach. Doing so requires a PCI reset to wake the controller back up. The driver can now be unloaded as long as CAM support is not enabled.
2003-02-05 23:24:25 +00:00
TAILQ_INIT(&sc->aac_sim_tqh);
Add a CAM interface to the aac driver. This is useful in case you should ever connect a SCSI Cdrom/Tape/Jukebox/Scanner/Printer/kitty-litter-scooper to your high-end RAID controller. The interface to the arrays is still via the block interface; this merely provides a way to circumvent the RAID functionality and access the SCSI buses directly. Note that for somewhat obvious reasons, hard drives are not exposed to the da driver through this interface, though you can still talk to them via the pass driver. Be the first on your block to low-level format unsuspecting drives that are part of an array! To enable this, add the 'aacp' device to your kernel config. MFC after: 3 days
2002-04-27 01:31:17 +00:00
aac_get_bus_info(sc);
Various cleanups: - Move the command timeout check from a separate repeating timeout to the kthread since the kthread is already running periodically. - Move printing the hardware print buffer to the kthread. - Properly shut down the kernel thread on detach. - Detach the child array devices on detach. - Don't issue a controller halt command on detach. Doing so requires a PCI reset to wake the controller back up. The driver can now be unloaded as long as CAM support is not enabled.
2003-02-05 23:24:25 +00:00
}
Add a CAM interface to the aac driver. This is useful in case you should ever connect a SCSI Cdrom/Tape/Jukebox/Scanner/Printer/kitty-litter-scooper to your high-end RAID controller. The interface to the arrays is still via the block interface; this merely provides a way to circumvent the RAID functionality and access the SCSI buses directly. Note that for somewhat obvious reasons, hard drives are not exposed to the da driver through this interface, though you can still talk to them via the pass driver. Be the first on your block to low-level format unsuspecting drives that are part of an array! To enable this, add the 'aacp' device to your kernel config. MFC after: 3 days
2002-04-27 01:31:17 +00:00
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
mtx_lock(&sc->aac_io_lock);
Do first controller time sync after 1 minute, as in Adaptec's vendor driver.
2009-10-28 13:50:28 +00:00
callout_reset(&sc->aac_daemontime, 60 * hz, aac_daemon, sc);
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
mtx_unlock(&sc->aac_io_lock);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
return(0);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
static void
aac_daemon(void *arg)
{
struct timeval tv;
struct aac_softc *sc;
struct aac_fib *fib;
sc = arg;
mtx_assert(&sc->aac_io_lock, MA_OWNED);
if (callout_pending(&sc->aac_daemontime) ||
callout_active(&sc->aac_daemontime) == 0)
return;
getmicrotime(&tv);
aac_alloc_sync_fib(sc, &fib);
*(uint32_t *)fib->data = tv.tv_sec;
aac_sync_fib(sc, SendHostTime, 0, fib, sizeof(uint32_t));
aac_release_sync_fib(sc);
callout_schedule(&sc->aac_daemontime, 30 * 60 * hz);
}
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
void
aac_add_event(struct aac_softc *sc, struct aac_event *event)
{
switch (event->ev_type & AAC_EVENT_MASK) {
case AAC_EVENT_CMFREE:
TAILQ_INSERT_TAIL(&sc->aac_ev_cmfree, event, ev_links);
break;
default:
device_printf(sc->aac_dev, "aac_add event: unknown event %d\n",
event->ev_type);
break;
}
}
Diff reduction to Adaptec driver build 15317 (refactoring and code shuffling): - Resource allocation in aac_alloc (moved from from aac_init) - Interrupt setup in aac_setup_intr (from aac_attach) - Container probing in aac_get_container_info (from aac_startup and aac_handle_aif) - Firmware status check moved to aac_check_firmware from aac_init
2008-03-24 16:38:47 +00:00
/*
* Request information of container #cid
*/
static struct aac_mntinforesp *
aac_get_container_info(struct aac_softc *sc, struct aac_fib *fib, int cid)
{
struct aac_mntinfo *mi;
mi = (struct aac_mntinfo *)&fib->data[0];
Add 64-bit array support for RAIDs > 2TB. This corresponds to ~ Adaptec driver build 15317. Tested on: Adaptec 2230S, Firmware 4.2-0 (8205) ICP ICP5085BL, Firmware 5.2-0 (12814) Submitted by: Adaptec
2008-03-25 21:39:06 +00:00
/* use 64-bit LBA if enabled */
mi->Command = (sc->flags & AAC_FLAGS_LBA_64BIT) ?
VM_NameServe64 : VM_NameServe;
Diff reduction to Adaptec driver build 15317 (refactoring and code shuffling): - Resource allocation in aac_alloc (moved from from aac_init) - Interrupt setup in aac_setup_intr (from aac_attach) - Container probing in aac_get_container_info (from aac_startup and aac_handle_aif) - Firmware status check moved to aac_check_firmware from aac_init
2008-03-24 16:38:47 +00:00
mi->MntType = FT_FILESYS;
mi->MntCount = cid;
if (aac_sync_fib(sc, ContainerCommand, 0, fib,
sizeof(struct aac_mntinfo))) {
Use device_printf where possible, and otherwise at least include the driver name in printf strings. Reported by: Mark Johnston
2010-09-16 23:33:24 +00:00
device_printf(sc->aac_dev, "Error probing container %d\n", cid);
Diff reduction to Adaptec driver build 15317 (refactoring and code shuffling): - Resource allocation in aac_alloc (moved from from aac_init) - Interrupt setup in aac_setup_intr (from aac_attach) - Container probing in aac_get_container_info (from aac_startup and aac_handle_aif) - Firmware status check moved to aac_check_firmware from aac_init
2008-03-24 16:38:47 +00:00
return (NULL);
}
return ((struct aac_mntinforesp *)&fib->data[0]);
}
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
* Probe for containers, create disks.
*/
static void
aac_startup(void *arg)
{
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
struct aac_softc *sc;
Prepare for a major update to the aac driver: Update the aac driver with the new crashdump api. Protect sync fibs with a mutex. Align all DMA buffers on a PAGE_SIZE boundary. MFC after: 3 days
2002-04-24 05:12:50 +00:00
struct aac_fib *fib;
Diff reduction to Adaptec driver build 15317 (refactoring and code shuffling): - Resource allocation in aac_alloc (moved from from aac_init) - Interrupt setup in aac_setup_intr (from aac_attach) - Container probing in aac_get_container_info (from aac_startup and aac_handle_aif) - Firmware status check moved to aac_check_firmware from aac_init
2008-03-24 16:38:47 +00:00
struct aac_mntinforesp *mir;
When scanning for changed containers, don't assume that the controller will respond in a sane manner. Thanks to Petri Helenius <pete@he.iki.fi> for spotting this and pestering me to fix it.
2003-06-03 02:10:55 +00:00
int count = 0, i = 0;
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
sc = (struct aac_softc *)arg;
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/* disconnect ourselves from the intrhook chain */
config_intrhook_disestablish(&sc->aac_ich);
Major bugfix and minor update. This should resolve the current issues with the driver locking up under load. - Restructure so that we use a static pool of commands/FIBs, rather than allocating them in clusters. The cluster allocation just made things more complicated, and allowed us to waste more memory in peak load situations. - Make queueing macros more like my other drivers. This adds queue stats for free. Add some debugging to take advantage of this. - Reimplement the periodic timeout scan. Kick the interrupt handler and the start routine every scan as well, just to be safe. Track busy commands properly. - Bring resource cleanup into line with resource allocation. We should now clean up correctly after a failed probe/unload/etc. - Try to start new commands when old ones are completed. We weren't doing this before, which could lead to deadlock when the controller was full. - Don't try to build a new command if we have found a deferred command. This could cause us to lose the deferred command. - Use diskerr() to report I/O errors. - Don't bail if the AdapterInfo structure is the wrong size. Some variation seems to be normal. We need to improve our handing of 2.x firmware sets. - Improve some comments in an attempt to try to make things clearer. - Restructure to avoid some warnings.
2000-12-27 13:14:56 +00:00
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
mtx_lock(&sc->aac_io_lock);
Collapse sync fib locking into normal i/o locking. The former didn't protect the registers so it was trivially possible for a sync command and i/o command to fight each other and confuse the controller. Make the sync fib alloc/release functions inline and remove the somewhat worthless AAC_SYNC_LOCK_FORCE flag. Thanks to Adil Katchi for helping me to track this down in RELENG_4.
2004-06-02 18:15:48 +00:00
aac_alloc_sync_fib(sc, &fib);
Prepare for a major update to the aac driver: Update the aac driver with the new crashdump api. Protect sync fibs with a mutex. Align all DMA buffers on a PAGE_SIZE boundary. MFC after: 3 days
2002-04-24 05:12:50 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/* loop over possible containers */
do {
Diff reduction to Adaptec driver build 15317 (refactoring and code shuffling): - Resource allocation in aac_alloc (moved from from aac_init) - Interrupt setup in aac_setup_intr (from aac_attach) - Container probing in aac_get_container_info (from aac_startup and aac_handle_aif) - Firmware status check moved to aac_check_firmware from aac_init
2008-03-24 16:38:47 +00:00
if ((mir = aac_get_container_info(sc, fib, i)) == NULL)
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
continue;
Diff reduction to Adaptec driver build 15317 (refactoring and code shuffling): - Resource allocation in aac_alloc (moved from from aac_init) - Interrupt setup in aac_setup_intr (from aac_attach) - Container probing in aac_get_container_info (from aac_startup and aac_handle_aif) - Firmware status check moved to aac_check_firmware from aac_init
2008-03-24 16:38:47 +00:00
if (i == 0)
count = mir->MntRespCount;
Prepare for a major update to the aac driver: Update the aac driver with the new crashdump api. Protect sync fibs with a mutex. Align all DMA buffers on a PAGE_SIZE boundary. MFC after: 3 days
2002-04-24 05:12:50 +00:00
aac_add_container(sc, mir, 0);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
i++;
When scanning for changed containers, don't assume that the controller will respond in a sane manner. Thanks to Petri Helenius <pete@he.iki.fi> for spotting this and pestering me to fix it.
2003-06-03 02:10:55 +00:00
} while ((i < count) && (i < AAC_MAX_CONTAINERS));
Prepare for a major update to the aac driver: Update the aac driver with the new crashdump api. Protect sync fibs with a mutex. Align all DMA buffers on a PAGE_SIZE boundary. MFC after: 3 days
2002-04-24 05:12:50 +00:00
aac_release_sync_fib(sc);
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
mtx_unlock(&sc->aac_io_lock);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/* poke the bus to actually attach the child devices */
if (bus_generic_attach(sc->aac_dev))
device_printf(sc->aac_dev, "bus_generic_attach failed\n");
/* mark the controller up */
sc->aac_state &= ~AAC_STATE_SUSPEND;
/* enable interrupts now */
AAC_UNMASK_INTERRUPTS(sc);
}
/*
Change spelling and eliminate a typo in comments to reduce diffs with Adaptec's vendor driver. I have some fixes to bring in and this makes ongoing review of the FreeBSD-Adaptec driver diffs easier.
2008-03-14 21:59:11 +00:00
* Create a device to represent a new container
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
*/
static void
Prepare for a major update to the aac driver: Update the aac driver with the new crashdump api. Protect sync fibs with a mutex. Align all DMA buffers on a PAGE_SIZE boundary. MFC after: 3 days
2002-04-24 05:12:50 +00:00
aac_add_container(struct aac_softc *sc, struct aac_mntinforesp *mir, int f)
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
{
struct aac_container *co;
device_t child;
Remove the unused AAC_DRIVER_BUILD_DATE macro. Nuke whitespace at EOL while I'm here. Approved by: scottl (MAINTAINER)
2005-11-13 17:26:36 +00:00
/*
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
* Check container volume type for validity. Note that many of
* the possible types may never show up.
*/
if ((mir->Status == ST_OK) && (mir->MntTable[0].VolType != CT_NONE)) {
Free the container and sim objects on detach. Convert several uses of MALLOC to malloc.
2003-02-06 00:37:37 +00:00
co = (struct aac_container *)malloc(sizeof *co, M_AACBUF,
M_NOWAIT | M_ZERO);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
if (co == NULL)
Add \n to the end of a printf string and remove it from panic strings.
2008-03-26 15:28:56 +00:00
panic("Out of memory?!");
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_INIT_B, "id %x name '%.16s' size %u type %d",
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
mir->MntTable[0].ObjectId,
mir->MntTable[0].FileSystemName,
mir->MntTable[0].Capacity, mir->MntTable[0].VolType);
Remove the unused AAC_DRIVER_BUILD_DATE macro. Nuke whitespace at EOL while I'm here. Approved by: scottl (MAINTAINER)
2005-11-13 17:26:36 +00:00
Add a CAM interface to the aac driver. This is useful in case you should ever connect a SCSI Cdrom/Tape/Jukebox/Scanner/Printer/kitty-litter-scooper to your high-end RAID controller. The interface to the arrays is still via the block interface; this merely provides a way to circumvent the RAID functionality and access the SCSI buses directly. Note that for somewhat obvious reasons, hard drives are not exposed to the da driver through this interface, though you can still talk to them via the pass driver. Be the first on your block to low-level format unsuspecting drives that are part of an array! To enable this, add the 'aacp' device to your kernel config. MFC after: 3 days
2002-04-27 01:31:17 +00:00
if ((child = device_add_child(sc->aac_dev, "aacd", -1)) == NULL)
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
device_printf(sc->aac_dev, "device_add_child failed\n");
else
device_set_ivars(child, co);
device_set_desc(child, aac_describe_code(aac_container_types,
mir->MntTable[0].VolType));
co->co_disk = child;
co->co_found = f;
bcopy(&mir->MntTable[0], &co->co_mntobj,
sizeof(struct aac_mntobj));
Remove the AAC_LOCK macros. They no longer abstract anything and only obfuscate the code. No functional differences.
2004-08-12 05:05:06 +00:00
mtx_lock(&sc->aac_container_lock);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
TAILQ_INSERT_TAIL(&sc->aac_container_tqh, co, co_link);
Remove the AAC_LOCK macros. They no longer abstract anything and only obfuscate the code. No functional differences.
2004-08-12 05:05:06 +00:00
mtx_unlock(&sc->aac_container_lock);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
}
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
Diff reduction to Adaptec driver build 15317 (refactoring and code shuffling): - Resource allocation in aac_alloc (moved from from aac_init) - Interrupt setup in aac_setup_intr (from aac_attach) - Container probing in aac_get_container_info (from aac_startup and aac_handle_aif) - Firmware status check moved to aac_check_firmware from aac_init
2008-03-24 16:38:47 +00:00
/*
* Allocate resources associated with (sc)
*/
static int
aac_alloc(struct aac_softc *sc)
{
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
Diff reduction to Adaptec driver build 15317 (refactoring and code shuffling): - Resource allocation in aac_alloc (moved from from aac_init) - Interrupt setup in aac_setup_intr (from aac_attach) - Container probing in aac_get_container_info (from aac_startup and aac_handle_aif) - Firmware status check moved to aac_check_firmware from aac_init
2008-03-24 16:38:47 +00:00
/*
* Create DMA tag for mapping buffers into controller-addressable space.
*/
if (bus_dma_tag_create(sc->aac_parent_dmat, /* parent */
1, 0, /* algnmnt, boundary */
(sc->flags & AAC_FLAGS_SG_64BIT) ?
BUS_SPACE_MAXADDR :
BUS_SPACE_MAXADDR_32BIT, /* lowaddr */
BUS_SPACE_MAXADDR, /* highaddr */
NULL, NULL, /* filter, filterarg */
Remove MAXBSIZE use from drivers where it has nothing to do. In some cases limits are just not needed, in others -- DFLTPHYS is the right constant to use instead. MFC after: 1 month
2015-03-22 16:10:28 +00:00
sc->aac_max_sectors << 9, /* maxsize */
Diff reduction to Adaptec driver build 15317 (refactoring and code shuffling): - Resource allocation in aac_alloc (moved from from aac_init) - Interrupt setup in aac_setup_intr (from aac_attach) - Container probing in aac_get_container_info (from aac_startup and aac_handle_aif) - Firmware status check moved to aac_check_firmware from aac_init
2008-03-24 16:38:47 +00:00
sc->aac_sg_tablesize, /* nsegments */
Remove MAXBSIZE use from drivers where it has nothing to do. In some cases limits are just not needed, in others -- DFLTPHYS is the right constant to use instead. MFC after: 1 month
2015-03-22 16:10:28 +00:00
BUS_SPACE_MAXSIZE_32BIT, /* maxsegsize */
Diff reduction to Adaptec driver build 15317 (refactoring and code shuffling): - Resource allocation in aac_alloc (moved from from aac_init) - Interrupt setup in aac_setup_intr (from aac_attach) - Container probing in aac_get_container_info (from aac_startup and aac_handle_aif) - Firmware status check moved to aac_check_firmware from aac_init
2008-03-24 16:38:47 +00:00
BUS_DMA_ALLOCNOW, /* flags */
busdma_lock_mutex, /* lockfunc */
&sc->aac_io_lock, /* lockfuncarg */
&sc->aac_buffer_dmat)) {
device_printf(sc->aac_dev, "can't allocate buffer DMA tag\n");
return (ENOMEM);
}
/*
* Create DMA tag for mapping FIBs into controller-addressable space..
*/
if (bus_dma_tag_create(sc->aac_parent_dmat, /* parent */
1, 0, /* algnmnt, boundary */
(sc->flags & AAC_FLAGS_4GB_WINDOW) ?
BUS_SPACE_MAXADDR_32BIT :
0x7fffffff, /* lowaddr */
BUS_SPACE_MAXADDR, /* highaddr */
NULL, NULL, /* filter, filterarg */
sc->aac_max_fibs_alloc *
sc->aac_max_fib_size, /* maxsize */
1, /* nsegments */
sc->aac_max_fibs_alloc *
sc->aac_max_fib_size, /* maxsize */
0, /* flags */
NULL, NULL, /* No locking needed */
&sc->aac_fib_dmat)) {
Remove extraneous semicolons, no functional changes. Submitted by: Marc Balmer <marc@msys.ch> MFC after: 1 week
2010-01-07 21:01:37 +00:00
device_printf(sc->aac_dev, "can't allocate FIB DMA tag\n");
Diff reduction to Adaptec driver build 15317 (refactoring and code shuffling): - Resource allocation in aac_alloc (moved from from aac_init) - Interrupt setup in aac_setup_intr (from aac_attach) - Container probing in aac_get_container_info (from aac_startup and aac_handle_aif) - Firmware status check moved to aac_check_firmware from aac_init
2008-03-24 16:38:47 +00:00
return (ENOMEM);
}
/*
* Create DMA tag for the common structure and allocate it.
*/
if (bus_dma_tag_create(sc->aac_parent_dmat, /* parent */
1, 0, /* algnmnt, boundary */
(sc->flags & AAC_FLAGS_4GB_WINDOW) ?
BUS_SPACE_MAXADDR_32BIT :
0x7fffffff, /* lowaddr */
BUS_SPACE_MAXADDR, /* highaddr */
NULL, NULL, /* filter, filterarg */
8192 + sizeof(struct aac_common), /* maxsize */
1, /* nsegments */
BUS_SPACE_MAXSIZE_32BIT, /* maxsegsize */
0, /* flags */
NULL, NULL, /* No locking needed */
&sc->aac_common_dmat)) {
device_printf(sc->aac_dev,
"can't allocate common structure DMA tag\n");
return (ENOMEM);
}
if (bus_dmamem_alloc(sc->aac_common_dmat, (void **)&sc->aac_common,
BUS_DMA_NOWAIT, &sc->aac_common_dmamap)) {
device_printf(sc->aac_dev, "can't allocate common structure\n");
return (ENOMEM);
}
/*
* Work around a bug in the 2120 and 2200 that cannot DMA commands
* below address 8192 in physical memory.
* XXX If the padding is not needed, can it be put to use instead
* of ignored?
*/
(void)bus_dmamap_load(sc->aac_common_dmat, sc->aac_common_dmamap,
sc->aac_common, 8192 + sizeof(*sc->aac_common),
aac_common_map, sc, 0);
if (sc->aac_common_busaddr < 8192) {
sc->aac_common = (struct aac_common *)
((uint8_t *)sc->aac_common + 8192);
sc->aac_common_busaddr += 8192;
}
bzero(sc->aac_common, sizeof(*sc->aac_common));
/* Allocate some FIBs and associated command structs */
TAILQ_INIT(&sc->aac_fibmap_tqh);
sc->aac_commands = malloc(sc->aac_max_fibs * sizeof(struct aac_command),
M_AACBUF, M_WAITOK|M_ZERO);
- Try pre-allocating all FIBs upfront. Previously we tried pre-allocating 128 FIBs first and allocated more later if necessary. Remove now unused definitions from the header file[1]. - Force sequential bus scanning. It seems parallel scanning is in fact slower and causes more harm than good[1]. Adjust a comment to reflect that. PR: kern/141269 Submitted by: Alexander Sack (asack at niksun dot com)[1] Reviewed by: scottl
2009-12-08 05:35:51 +00:00
while (sc->total_fibs < sc->aac_max_fibs) {
Diff reduction to Adaptec driver build 15317 (refactoring and code shuffling): - Resource allocation in aac_alloc (moved from from aac_init) - Interrupt setup in aac_setup_intr (from aac_attach) - Container probing in aac_get_container_info (from aac_startup and aac_handle_aif) - Firmware status check moved to aac_check_firmware from aac_init
2008-03-24 16:38:47 +00:00
if (aac_alloc_commands(sc) != 0)
break;
}
if (sc->total_fibs == 0)
return (ENOMEM);
return (0);
}
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
* Free all of the resources associated with (sc)
*
* Should not be called if the controller is active.
*/
void
aac_free(struct aac_softc *sc)
{
Implement a new dynamic command allocator. FIBs are allocated in 1 page blocks now, which should eliminate problems with the driver failing to attach due to insufficient contiguous RAM. Allow the FIB pool to grow from the default of 128 to the max of 512 as demand grows. Also pad the adapter init struct to work around the 2120/2200 DMA bug now that there is no longer a FIB slab.
2003-02-10 00:34:24 +00:00
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/* remove the control device */
if (sc->aac_dev_t != NULL)
destroy_dev(sc->aac_dev_t);
/* throw away any FIB buffers, discard the FIB DMA tag */
o Move the cleanup of the fib maps into aac_free_commands() so as to retain symetry with aac_alloc_commans(). Since aac_alloc_commands() allocates fib maps and places them onto the fib lists, aac_free_commands() should reverse those operations. o Combine two ifs with the same body with an ||. o Switch from uintptr_t to uint32_t for fib map load operations. The target is a uint32_t so using this type for the map load call avoids an extra cast. uintptr_t should only be used when you need an "int sized the same as the machine's poitner size" which is not the case here. o Removed the commented out M_WAITOK flag in the allocation in aac_alloc_commands(). The kernel will only block in the allocator if it can grow the size of the kernel. This usually results in a page-out which could involve this aac device. Thus, sleeping here could deadlock the machine. Assuming this operation is occurring outside of attach time, we have enough fibs to operate anyway, so waiting for fibs to free up is okay if not optimal. o In aac_alloc_commands(), if we cannot dmamem_alloc additional fib space, free the fib map. o In aac_alloc_commands(), if we cannot create per-command dmamaps, don't lose track of the fib map that is mapping all of the commands that we have already released into the free pool. Instead, just cut out of the loop and modify aac_free_commands to not attempt to free maps that have not been allocated. o Don't use a magic number when pre-allocating fibs. o Use PAGE_SIZE to allocate in page sized chunks instead of an architecture specific constant. Submitted by: gibbs
2003-02-19 21:38:29 +00:00
aac_free_commands(sc);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
if (sc->aac_fib_dmat)
bus_dma_tag_destroy(sc->aac_fib_dmat);
Implement a new dynamic command allocator. FIBs are allocated in 1 page blocks now, which should eliminate problems with the driver failing to attach due to insufficient contiguous RAM. Allow the FIB pool to grow from the default of 128 to the max of 512 as demand grows. Also pad the adapter init struct to work around the 2120/2200 DMA bug now that there is no longer a FIB slab.
2003-02-10 00:34:24 +00:00
free(sc->aac_commands, M_AACBUF);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/* destroy the common area */
if (sc->aac_common) {
bus_dmamap_unload(sc->aac_common_dmat, sc->aac_common_dmamap);
bus_dmamem_free(sc->aac_common_dmat, sc->aac_common,
sc->aac_common_dmamap);
}
if (sc->aac_common_dmat)
bus_dma_tag_destroy(sc->aac_common_dmat);
/* disconnect the interrupt handler */
if (sc->aac_intr)
bus_teardown_intr(sc->aac_dev, sc->aac_irq, sc->aac_intr);
Free the MSI again on detach if allocated. Arguably, this code would be better off living in aac_pci.c, but it doesn't seem worth creating a aac_pci_detach() and it's also not the first PCI-specific bit in aac.c MFC after: 3 days
2013-12-29 17:37:32 +00:00
if (sc->aac_irq != NULL) {
- Make tables, device ID strings etc const. This includes #ifdef'ing 0 aac_command_status_table, which is actually unused since r111532. While at it, make aac_if a pointer to the now const interface tables instead of copying them over to the softc (this alone already reduces the size of aac.ko on amd64 by ~1 KiB). - Remove redundant softc members. - Use DEVMETHOD_END. - Use NULL instead of 0 for pointers. - Remove redundant bzero(9)'ing of the softc. - Use pci_enable_busmaster(9) instead of duplicating it. - Remove redundant checking for PCIM_CMD_MEMEN (resource allocation will just fail). - Canonicalize the error messages in case of resource allocation failures. - Add support for using MSI instead of INTx, controllable via the tunable hw.aac.enable_msi (defaulting to on). MFC after: 1 month
2013-03-01 19:55:10 +00:00
bus_release_resource(sc->aac_dev, SYS_RES_IRQ,
rman_get_rid(sc->aac_irq), sc->aac_irq);
Free the MSI again on detach if allocated. Arguably, this code would be better off living in aac_pci.c, but it doesn't seem worth creating a aac_pci_detach() and it's also not the first PCI-specific bit in aac.c MFC after: 3 days
2013-12-29 17:37:32 +00:00
pci_release_msi(sc->aac_dev);
}
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/* destroy data-transfer DMA tag */
if (sc->aac_buffer_dmat)
bus_dma_tag_destroy(sc->aac_buffer_dmat);
/* destroy the parent DMA tag */
if (sc->aac_parent_dmat)
bus_dma_tag_destroy(sc->aac_parent_dmat);
/* release the register window mapping */
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
if (sc->aac_regs_res0 != NULL)
bus_release_resource(sc->aac_dev, SYS_RES_MEMORY,
- Make tables, device ID strings etc const. This includes #ifdef'ing 0 aac_command_status_table, which is actually unused since r111532. While at it, make aac_if a pointer to the now const interface tables instead of copying them over to the softc (this alone already reduces the size of aac.ko on amd64 by ~1 KiB). - Remove redundant softc members. - Use DEVMETHOD_END. - Use NULL instead of 0 for pointers. - Remove redundant bzero(9)'ing of the softc. - Use pci_enable_busmaster(9) instead of duplicating it. - Remove redundant checking for PCIM_CMD_MEMEN (resource allocation will just fail). - Canonicalize the error messages in case of resource allocation failures. - Add support for using MSI instead of INTx, controllable via the tunable hw.aac.enable_msi (defaulting to on). MFC after: 1 month
2013-03-01 19:55:10 +00:00
rman_get_rid(sc->aac_regs_res0), sc->aac_regs_res0);
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
if (sc->aac_hwif == AAC_HWIF_NARK && sc->aac_regs_res1 != NULL)
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
bus_release_resource(sc->aac_dev, SYS_RES_MEMORY,
- Make tables, device ID strings etc const. This includes #ifdef'ing 0 aac_command_status_table, which is actually unused since r111532. While at it, make aac_if a pointer to the now const interface tables instead of copying them over to the softc (this alone already reduces the size of aac.ko on amd64 by ~1 KiB). - Remove redundant softc members. - Use DEVMETHOD_END. - Use NULL instead of 0 for pointers. - Remove redundant bzero(9)'ing of the softc. - Use pci_enable_busmaster(9) instead of duplicating it. - Remove redundant checking for PCIM_CMD_MEMEN (resource allocation will just fail). - Canonicalize the error messages in case of resource allocation failures. - Add support for using MSI instead of INTx, controllable via the tunable hw.aac.enable_msi (defaulting to on). MFC after: 1 month
2013-03-01 19:55:10 +00:00
rman_get_rid(sc->aac_regs_res1), sc->aac_regs_res1);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
* Disconnect from the controller completely, in preparation for unload.
*/
int
aac_detach(device_t dev)
{
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
struct aac_softc *sc;
Various cleanups: - Move the command timeout check from a separate repeating timeout to the kthread since the kthread is already running periodically. - Move printing the hardware print buffer to the kthread. - Properly shut down the kernel thread on detach. - Detach the child array devices on detach. - Don't issue a controller halt command on detach. Doing so requires a PCI reset to wake the controller back up. The driver can now be unloaded as long as CAM support is not enabled.
2003-02-05 23:24:25 +00:00
struct aac_container *co;
struct aac_sim *sim;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
int error;
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
sc = device_get_softc(dev);
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
callout_drain(&sc->aac_daemontime);
- Fix races on detach handling of AAC_IFFLAGS_* mask - Fix races on setting AAC_AIFFLAGS_ALLOCFIBS - Remove some unused AAC_IFFLAGS_* bits. Please note that the kthread still makes a difference between the total mask and AAC_AIFFLAGS_ALLOCFIBS because more flags may be added in the future to aifflags. Sponsored by: Sandvine Incorporated Reported and reviewed by: emaste MFC after: 2 weeks
2011-06-10 20:23:56 +00:00
mtx_lock(&sc->aac_io_lock);
while (sc->aifflags & AAC_AIFFLAGS_RUNNING) {
sc->aifflags |= AAC_AIFFLAGS_EXIT;
wakeup(sc->aifthread);
msleep(sc->aac_dev, &sc->aac_io_lock, PUSER, "aacdch", 0);
}
mtx_unlock(&sc->aac_io_lock);
KASSERT((sc->aifflags & AAC_AIFFLAGS_RUNNING) == 0,
("%s: invalid detach state", __func__));
Various cleanups: - Move the command timeout check from a separate repeating timeout to the kthread since the kthread is already running periodically. - Move printing the hardware print buffer to the kthread. - Properly shut down the kernel thread on detach. - Detach the child array devices on detach. - Don't issue a controller halt command on detach. Doing so requires a PCI reset to wake the controller back up. The driver can now be unloaded as long as CAM support is not enabled.
2003-02-05 23:24:25 +00:00
/* Remove the child containers */
Free the container and sim objects on detach. Convert several uses of MALLOC to malloc.
2003-02-06 00:37:37 +00:00
while ((co = TAILQ_FIRST(&sc->aac_container_tqh)) != NULL) {
Various cleanups: - Move the command timeout check from a separate repeating timeout to the kthread since the kthread is already running periodically. - Move printing the hardware print buffer to the kthread. - Properly shut down the kernel thread on detach. - Detach the child array devices on detach. - Don't issue a controller halt command on detach. Doing so requires a PCI reset to wake the controller back up. The driver can now be unloaded as long as CAM support is not enabled.
2003-02-05 23:24:25 +00:00
error = device_delete_child(dev, co->co_disk);
if (error)
return (error);
On detach, don't remove the child from our list of children unless it has successfully detached.
2003-02-20 23:13:48 +00:00
TAILQ_REMOVE(&sc->aac_container_tqh, co, co_link);
Free the container and sim objects on detach. Convert several uses of MALLOC to malloc.
2003-02-06 00:37:37 +00:00
free(co, M_AACBUF);
Various cleanups: - Move the command timeout check from a separate repeating timeout to the kthread since the kthread is already running periodically. - Move printing the hardware print buffer to the kthread. - Properly shut down the kernel thread on detach. - Detach the child array devices on detach. - Don't issue a controller halt command on detach. Doing so requires a PCI reset to wake the controller back up. The driver can now be unloaded as long as CAM support is not enabled.
2003-02-05 23:24:25 +00:00
}
/* Remove the CAM SIMs */
Free the container and sim objects on detach. Convert several uses of MALLOC to malloc.
2003-02-06 00:37:37 +00:00
while ((sim = TAILQ_FIRST(&sc->aac_sim_tqh)) != NULL) {
TAILQ_REMOVE(&sc->aac_sim_tqh, sim, sim_link);
Various cleanups: - Move the command timeout check from a separate repeating timeout to the kthread since the kthread is already running periodically. - Move printing the hardware print buffer to the kthread. - Properly shut down the kernel thread on detach. - Detach the child array devices on detach. - Don't issue a controller halt command on detach. Doing so requires a PCI reset to wake the controller back up. The driver can now be unloaded as long as CAM support is not enabled.
2003-02-05 23:24:25 +00:00
error = device_delete_child(dev, sim->sim_dev);
if (error)
return (error);
Free the container and sim objects on detach. Convert several uses of MALLOC to malloc.
2003-02-06 00:37:37 +00:00
free(sim, M_AACBUF);
Various cleanups: - Move the command timeout check from a separate repeating timeout to the kthread since the kthread is already running periodically. - Move printing the hardware print buffer to the kthread. - Properly shut down the kernel thread on detach. - Detach the child array devices on detach. - Don't issue a controller halt command on detach. Doing so requires a PCI reset to wake the controller back up. The driver can now be unloaded as long as CAM support is not enabled.
2003-02-05 23:24:25 +00:00
}
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
if ((error = aac_shutdown(dev)))
return(error);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Don't include aac_cam.h, since it was nuked in the last commit. Deregister the shutdown eventhandler on unload.
2003-02-06 00:07:12 +00:00
EVENTHANDLER_DEREGISTER(shutdown_final, sc->eh);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
aac_free(sc);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Destroy mutexes on detach.
2004-08-12 05:31:17 +00:00
mtx_destroy(&sc->aac_aifq_lock);
mtx_destroy(&sc->aac_io_lock);
mtx_destroy(&sc->aac_container_lock);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
return(0);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
* Bring the controller down to a dormant state and detach all child devices.
*
* This function is called before detach or system shutdown.
*
Major bugfix and minor update. This should resolve the current issues with the driver locking up under load. - Restructure so that we use a static pool of commands/FIBs, rather than allocating them in clusters. The cluster allocation just made things more complicated, and allowed us to waste more memory in peak load situations. - Make queueing macros more like my other drivers. This adds queue stats for free. Add some debugging to take advantage of this. - Reimplement the periodic timeout scan. Kick the interrupt handler and the start routine every scan as well, just to be safe. Track busy commands properly. - Bring resource cleanup into line with resource allocation. We should now clean up correctly after a failed probe/unload/etc. - Try to start new commands when old ones are completed. We weren't doing this before, which could lead to deadlock when the controller was full. - Don't try to build a new command if we have found a deferred command. This could cause us to lose the deferred command. - Use diskerr() to report I/O errors. - Don't bail if the AdapterInfo structure is the wrong size. Some variation seems to be normal. We need to improve our handing of 2.x firmware sets. - Improve some comments in an attempt to try to make things clearer. - Restructure to avoid some warnings.
2000-12-27 13:14:56 +00:00
* Note that we can assume that the bioq on the controller is empty, as we won't
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
* allow shutdown if any device is open.
*/
int
aac_shutdown(device_t dev)
{
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
struct aac_softc *sc;
Prepare for a major update to the aac driver: Update the aac driver with the new crashdump api. Protect sync fibs with a mutex. Align all DMA buffers on a PAGE_SIZE boundary. MFC after: 3 days
2002-04-24 05:12:50 +00:00
struct aac_fib *fib;
struct aac_close_command *cc;
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
sc = device_get_softc(dev);
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
sc->aac_state |= AAC_STATE_SUSPEND;
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Remove the unused AAC_DRIVER_BUILD_DATE macro. Nuke whitespace at EOL while I'm here. Approved by: scottl (MAINTAINER)
2005-11-13 17:26:36 +00:00
/*
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
* Send a Container shutdown followed by a HostShutdown FIB to the
* controller to convince it that we don't want to talk to it anymore.
* We've been closed and all I/O completed already
Mega update to the aac driver. 1. Correctly handle commands initiated by the adapter. These commands are defered to a kthread responsible for their processing, then are properly returned to the controller. 2. Add/remove disk devices when notified by the card that a container was created/deleted/offline. 3. Implement crashdump functionality. 4. Support all ioctls needed for the management CLI to work. The linux version of this app can be found at the Dell or HP website. A native version will be forthcoming. MFC-after: 4.4-RELEASE
2001-08-29 23:34:05 +00:00
*/
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
device_printf(sc->aac_dev, "shutting down controller...");
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
mtx_lock(&sc->aac_io_lock);
Collapse sync fib locking into normal i/o locking. The former didn't protect the registers so it was trivially possible for a sync command and i/o command to fight each other and confuse the controller. Make the sync fib alloc/release functions inline and remove the somewhat worthless AAC_SYNC_LOCK_FORCE flag. Thanks to Adil Katchi for helping me to track this down in RELENG_4.
2004-06-02 18:15:48 +00:00
aac_alloc_sync_fib(sc, &fib);
Prepare for a major update to the aac driver: Update the aac driver with the new crashdump api. Protect sync fibs with a mutex. Align all DMA buffers on a PAGE_SIZE boundary. MFC after: 3 days
2002-04-24 05:12:50 +00:00
cc = (struct aac_close_command *)&fib->data[0];
bzero out the sync command buffer when sending commands. This was causing problems when enumerating multiple arrays. This is an MFC candidate.
2002-05-03 00:07:50 +00:00
bzero(cc, sizeof(struct aac_close_command));
Prepare for a major update to the aac driver: Update the aac driver with the new crashdump api. Protect sync fibs with a mutex. Align all DMA buffers on a PAGE_SIZE boundary. MFC after: 3 days
2002-04-24 05:12:50 +00:00
cc->Command = VM_CloseAll;
cc->ContainerId = 0xffffffff;
if (aac_sync_fib(sc, ContainerCommand, 0, fib,
sizeof(struct aac_close_command)))
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
printf("FAILED.\n");
Various cleanups: - Move the command timeout check from a separate repeating timeout to the kthread since the kthread is already running periodically. - Move printing the hardware print buffer to the kthread. - Properly shut down the kernel thread on detach. - Detach the child array devices on detach. - Don't issue a controller halt command on detach. Doing so requires a PCI reset to wake the controller back up. The driver can now be unloaded as long as CAM support is not enabled.
2003-02-05 23:24:25 +00:00
else
printf("done\n");
#if 0
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
else {
Prepare for a major update to the aac driver: Update the aac driver with the new crashdump api. Protect sync fibs with a mutex. Align all DMA buffers on a PAGE_SIZE boundary. MFC after: 3 days
2002-04-24 05:12:50 +00:00
fib->data[0] = 0;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
* XXX Issuing this command to the controller makes it shut down
* but also keeps it from coming back up without a reset of the
* PCI bus. This is not desirable if you are just unloading the
* driver module with the intent to reload it later.
*/
Prepare for a major update to the aac driver: Update the aac driver with the new crashdump api. Protect sync fibs with a mutex. Align all DMA buffers on a PAGE_SIZE boundary. MFC after: 3 days
2002-04-24 05:12:50 +00:00
if (aac_sync_fib(sc, FsaHostShutdown, AAC_FIBSTATE_SHUTDOWN,
fib, 1)) {
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
printf("FAILED.\n");
} else {
printf("done.\n");
}
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
Various cleanups: - Move the command timeout check from a separate repeating timeout to the kthread since the kthread is already running periodically. - Move printing the hardware print buffer to the kthread. - Properly shut down the kernel thread on detach. - Detach the child array devices on detach. - Don't issue a controller halt command on detach. Doing so requires a PCI reset to wake the controller back up. The driver can now be unloaded as long as CAM support is not enabled.
2003-02-05 23:24:25 +00:00
#endif
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
AAC_MASK_INTERRUPTS(sc);
Release the sync fib after the controller has been shut down. This also releases the I/O lock instead of just leaking it.
2004-08-12 04:47:32 +00:00
aac_release_sync_fib(sc);
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
mtx_unlock(&sc->aac_io_lock);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
return(0);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
* Bring the controller to a quiescent state, ready for system suspend.
*/
int
aac_suspend(device_t dev)
{
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
struct aac_softc *sc;
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
sc = device_get_softc(dev);
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
sc->aac_state |= AAC_STATE_SUSPEND;
Remove the unused AAC_DRIVER_BUILD_DATE macro. Nuke whitespace at EOL while I'm here. Approved by: scottl (MAINTAINER)
2005-11-13 17:26:36 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
AAC_MASK_INTERRUPTS(sc);
return(0);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
* Bring the controller back to a state ready for operation.
*/
int
aac_resume(device_t dev)
{
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
struct aac_softc *sc;
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
sc = device_get_softc(dev);
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
sc->aac_state &= ~AAC_STATE_SUSPEND;
AAC_UNMASK_INTERRUPTS(sc);
return(0);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
* Interrupt handler for NEW_COMM interface.
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
*/
void
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
aac_new_intr(void *arg)
{
struct aac_softc *sc;
u_int32_t index, fast;
struct aac_command *cm;
struct aac_fib *fib;
int i;
sc = (struct aac_softc *)arg;
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
mtx_lock(&sc->aac_io_lock);
while (1) {
index = AAC_GET_OUTB_QUEUE(sc);
if (index == 0xffffffff)
index = AAC_GET_OUTB_QUEUE(sc);
if (index == 0xffffffff)
break;
if (index & 2) {
if (index == 0xfffffffe) {
/* XXX This means that the controller wants
* more work. Ignore it for now.
*/
continue;
}
/* AIF */
fib = (struct aac_fib *)malloc(sizeof *fib, M_AACBUF,
M_NOWAIT | M_ZERO);
if (fib == NULL) {
/* If we're really this short on memory,
* hopefully breaking out of the handler will
* allow something to get freed. This
* actually sucks a whole lot.
*/
break;
}
index &= ~2;
for (i = 0; i < sizeof(struct aac_fib)/4; ++i)
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
((u_int32_t *)fib)[i] = AAC_MEM1_GETREG4(sc, index + i*4);
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
aac_handle_aif(sc, fib);
free(fib, M_AACBUF);
/*
* AIF memory is owned by the adapter, so let it
* know that we are done with it.
*/
AAC_SET_OUTB_QUEUE(sc, index);
AAC_CLEAR_ISTATUS(sc, AAC_DB_RESPONSE_READY);
} else {
fast = index & 1;
cm = sc->aac_commands + (index >> 2);
fib = cm->cm_fib;
if (fast) {
fib->Header.XferState |= AAC_FIBSTATE_DONEADAP;
*((u_int32_t *)(fib->data)) = AAC_ERROR_NORMAL;
}
aac_remove_busy(cm);
aac_unmap_command(cm);
cm->cm_flags |= AAC_CMD_COMPLETED;
/* is there a completion handler? */
if (cm->cm_complete != NULL) {
cm->cm_complete(cm);
} else {
/* assume that someone is sleeping on this
* command
*/
wakeup(cm);
}
sc->flags &= ~AAC_QUEUE_FRZN;
}
}
/* see if we can start some more I/O */
if ((sc->flags & AAC_QUEUE_FRZN) == 0)
aac_startio(sc);
mtx_unlock(&sc->aac_io_lock);
}
Rename aac_fast_intr to aac_filter to reflect its current use. Eliminate the fallback of using the filter as an interrupt handler, as it is no longer needed. Discussed with: scottl, jhb
2009-10-29 14:53:45 +00:00
/*
* Interrupt filter for !NEW_COMM interface.
*/
o break newbus api: add a new argument of type driver_filter_t to bus_setup_intr() o add an int return code to all fast handlers o retire INTR_FAST/IH_FAST For more info: http://docs.freebsd.org/cgi/getmsg.cgi?fetch=465712+0+current/freebsd-current Reviewed by: many Approved by: re@
2007-02-23 12:19:07 +00:00
int
Rename aac_fast_intr to aac_filter to reflect its current use. Eliminate the fallback of using the filter as an interrupt handler, as it is no longer needed. Discussed with: scottl, jhb
2009-10-29 14:53:45 +00:00
aac_filter(void *arg)
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
{
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
struct aac_softc *sc;
Various cleanups: - Move the command timeout check from a separate repeating timeout to the kthread since the kthread is already running periodically. - Move printing the hardware print buffer to the kthread. - Properly shut down the kernel thread on detach. - Detach the child array devices on detach. - Don't issue a controller halt command on detach. Doing so requires a PCI reset to wake the controller back up. The driver can now be unloaded as long as CAM support is not enabled.
2003-02-05 23:24:25 +00:00
u_int16_t reason;
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
sc = (struct aac_softc *)arg;
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
Major bugfixes for large memory and fast systems. aac.c: Re-arrange the interrupt handler to optimize the common case of the adapter interrupting us because one or more commands are complete, and do a read across the pci bus to ensure that all posted status writes are flushed. This should close a race that could cause command completion interrupts to be lost. Follow the spec a bit closer when filling out command structures. Enable the Fast Response feature to eliminate the need for the card to DMA successfull command completions back into host memory. Tell the controller how much physical memory we have. Without this there was a chance that our DMA regions would collide with the memory window used by the cache on the controller. The result would be massive data corruption. This seemed to mainly affect systems with >2GB of memory. Fix a few whitespace problems. aac_debug.c: Add an extra diagnostic when printing out commands. aac_disk.c: Add extra sanity checks. aacreg.h: Prepare for making this 64-bit clean by reducing the use of enumeration types in structures. Many thanks to Justin Gibbs for helping track these down.
2003-01-11 01:59:21 +00:00
/*
Take the plunge and make this driver be INTR_FAST. This re-arranges the interrupt handler so that no locks are needed, and schedules the command completion routine with a taskqueue_fast. This also corrects the locking in the command thread and removes the need for operation flags. Simple load tests show that this is now considerably faster than FreeBSD 4.x in the SMP case when multiple i/o tasks are running.
2004-01-30 07:04:39 +00:00
* Read the status register directly. This is faster than taking the
* driver lock and reading the queues directly. It also saves having
* to turn parts of the driver lock into a spin mutex, which would be
* ugly.
Major bugfixes for large memory and fast systems. aac.c: Re-arrange the interrupt handler to optimize the common case of the adapter interrupting us because one or more commands are complete, and do a read across the pci bus to ensure that all posted status writes are flushed. This should close a race that could cause command completion interrupts to be lost. Follow the spec a bit closer when filling out command structures. Enable the Fast Response feature to eliminate the need for the card to DMA successfull command completions back into host memory. Tell the controller how much physical memory we have. Without this there was a chance that our DMA regions would collide with the memory window used by the cache on the controller. The result would be massive data corruption. This seemed to mainly affect systems with >2GB of memory. Fix a few whitespace problems. aac_debug.c: Add an extra diagnostic when printing out commands. aac_disk.c: Add extra sanity checks. aacreg.h: Prepare for making this 64-bit clean by reducing the use of enumeration types in structures. Many thanks to Justin Gibbs for helping track these down.
2003-01-11 01:59:21 +00:00
*/
Take the plunge and make this driver be INTR_FAST. This re-arranges the interrupt handler so that no locks are needed, and schedules the command completion routine with a taskqueue_fast. This also corrects the locking in the command thread and removes the need for operation flags. Simple load tests show that this is now considerably faster than FreeBSD 4.x in the SMP case when multiple i/o tasks are running.
2004-01-30 07:04:39 +00:00
reason = AAC_GET_ISTATUS(sc);
Major bugfixes for large memory and fast systems. aac.c: Re-arrange the interrupt handler to optimize the common case of the adapter interrupting us because one or more commands are complete, and do a read across the pci bus to ensure that all posted status writes are flushed. This should close a race that could cause command completion interrupts to be lost. Follow the spec a bit closer when filling out command structures. Enable the Fast Response feature to eliminate the need for the card to DMA successfull command completions back into host memory. Tell the controller how much physical memory we have. Without this there was a chance that our DMA regions would collide with the memory window used by the cache on the controller. The result would be massive data corruption. This seemed to mainly affect systems with >2GB of memory. Fix a few whitespace problems. aac_debug.c: Add an extra diagnostic when printing out commands. aac_disk.c: Add extra sanity checks. aacreg.h: Prepare for making this 64-bit clean by reducing the use of enumeration types in structures. Many thanks to Justin Gibbs for helping track these down.
2003-01-11 01:59:21 +00:00
AAC_CLEAR_ISTATUS(sc, reason);
Take the plunge and make this driver be INTR_FAST. This re-arranges the interrupt handler so that no locks are needed, and schedules the command completion routine with a taskqueue_fast. This also corrects the locking in the command thread and removes the need for operation flags. Simple load tests show that this is now considerably faster than FreeBSD 4.x in the SMP case when multiple i/o tasks are running.
2004-01-30 07:04:39 +00:00
/* handle completion processing */
Major bugfixes for large memory and fast systems. aac.c: Re-arrange the interrupt handler to optimize the common case of the adapter interrupting us because one or more commands are complete, and do a read across the pci bus to ensure that all posted status writes are flushed. This should close a race that could cause command completion interrupts to be lost. Follow the spec a bit closer when filling out command structures. Enable the Fast Response feature to eliminate the need for the card to DMA successfull command completions back into host memory. Tell the controller how much physical memory we have. Without this there was a chance that our DMA regions would collide with the memory window used by the cache on the controller. The result would be massive data corruption. This seemed to mainly affect systems with >2GB of memory. Fix a few whitespace problems. aac_debug.c: Add an extra diagnostic when printing out commands. aac_disk.c: Add extra sanity checks. aacreg.h: Prepare for making this 64-bit clean by reducing the use of enumeration types in structures. Many thanks to Justin Gibbs for helping track these down.
2003-01-11 01:59:21 +00:00
if (reason & AAC_DB_RESPONSE_READY)
Remove taskqueue_enqueue_fast(). taskqueue_enqueue() was changed to support both fast and non-fast taskqueues 10 years ago in r154167. It has been a compat shim ever since. It's time for the compat shim to go. Submitted by: Howard Su <howard0su@gmail.com> Reviewed by: sephe Differential Revision: https://reviews.freebsd.org/D5131
2016-03-01 17:47:32 +00:00
taskqueue_enqueue(taskqueue_fast, &sc->aac_task_complete);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
Take the plunge and make this driver be INTR_FAST. This re-arranges the interrupt handler so that no locks are needed, and schedules the command completion routine with a taskqueue_fast. This also corrects the locking in the command thread and removes the need for operation flags. Simple load tests show that this is now considerably faster than FreeBSD 4.x in the SMP case when multiple i/o tasks are running.
2004-01-30 07:04:39 +00:00
/* controller wants to talk to us */
if (reason & (AAC_DB_PRINTF | AAC_DB_COMMAND_READY)) {
/*
* XXX Make sure that we don't get fooled by strange messages
* that start with a NULL.
*/
if ((reason & AAC_DB_PRINTF) &&
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
(sc->aac_common->ac_printf[0] == 0))
Take the plunge and make this driver be INTR_FAST. This re-arranges the interrupt handler so that no locks are needed, and schedules the command completion routine with a taskqueue_fast. This also corrects the locking in the command thread and removes the need for operation flags. Simple load tests show that this is now considerably faster than FreeBSD 4.x in the SMP case when multiple i/o tasks are running.
2004-01-30 07:04:39 +00:00
sc->aac_common->ac_printf[0] = 32;
Various cleanups: - Move the command timeout check from a separate repeating timeout to the kthread since the kthread is already running periodically. - Move printing the hardware print buffer to the kthread. - Properly shut down the kernel thread on detach. - Detach the child array devices on detach. - Don't issue a controller halt command on detach. Doing so requires a PCI reset to wake the controller back up. The driver can now be unloaded as long as CAM support is not enabled.
2003-02-05 23:24:25 +00:00
Take the plunge and make this driver be INTR_FAST. This re-arranges the interrupt handler so that no locks are needed, and schedules the command completion routine with a taskqueue_fast. This also corrects the locking in the command thread and removes the need for operation flags. Simple load tests show that this is now considerably faster than FreeBSD 4.x in the SMP case when multiple i/o tasks are running.
2004-01-30 07:04:39 +00:00
/*
* This might miss doing the actual wakeup. However, the
- Broaden the scope of locking in aac_command_thread() again to catch some edge cases in the loop. - Try to grab a command before dequeueing the bio from the bioq. The old behaviour of requeuing deferred bios to the end of the bioq is arguably wrong. This should be fixed in the future to check the bioq head without automatically dequeueing the bio.
2004-02-07 03:26:38 +00:00
* msleep that this is waking up has a timeout, so it will
Take the plunge and make this driver be INTR_FAST. This re-arranges the interrupt handler so that no locks are needed, and schedules the command completion routine with a taskqueue_fast. This also corrects the locking in the command thread and removes the need for operation flags. Simple load tests show that this is now considerably faster than FreeBSD 4.x in the SMP case when multiple i/o tasks are running.
2004-01-30 07:04:39 +00:00
* wake up eventually. AIFs and printfs are low enough
* priority that they can handle hanging out for a few seconds
* if needed.
*/
Various cleanups: - Move the command timeout check from a separate repeating timeout to the kthread since the kthread is already running periodically. - Move printing the hardware print buffer to the kthread. - Properly shut down the kernel thread on detach. - Detach the child array devices on detach. - Don't issue a controller halt command on detach. Doing so requires a PCI reset to wake the controller back up. The driver can now be unloaded as long as CAM support is not enabled.
2003-02-05 23:24:25 +00:00
wakeup(sc->aifthread);
Take the plunge and make this driver be INTR_FAST. This re-arranges the interrupt handler so that no locks are needed, and schedules the command completion routine with a taskqueue_fast. This also corrects the locking in the command thread and removes the need for operation flags. Simple load tests show that this is now considerably faster than FreeBSD 4.x in the SMP case when multiple i/o tasks are running.
2004-01-30 07:04:39 +00:00
}
o break newbus api: add a new argument of type driver_filter_t to bus_setup_intr() o add an int return code to all fast handlers o retire INTR_FAST/IH_FAST For more info: http://docs.freebsd.org/cgi/getmsg.cgi?fetch=465712+0+current/freebsd-current Reviewed by: many Approved by: re@
2007-02-23 12:19:07 +00:00
return (FILTER_HANDLED);
Major bugfixes for large memory and fast systems. aac.c: Re-arrange the interrupt handler to optimize the common case of the adapter interrupting us because one or more commands are complete, and do a read across the pci bus to ensure that all posted status writes are flushed. This should close a race that could cause command completion interrupts to be lost. Follow the spec a bit closer when filling out command structures. Enable the Fast Response feature to eliminate the need for the card to DMA successfull command completions back into host memory. Tell the controller how much physical memory we have. Without this there was a chance that our DMA regions would collide with the memory window used by the cache on the controller. The result would be massive data corruption. This seemed to mainly affect systems with >2GB of memory. Fix a few whitespace problems. aac_debug.c: Add an extra diagnostic when printing out commands. aac_disk.c: Add extra sanity checks. aacreg.h: Prepare for making this 64-bit clean by reducing the use of enumeration types in structures. Many thanks to Justin Gibbs for helping track these down.
2003-01-11 01:59:21 +00:00
}
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
* Command Processing
*/
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
* Start as much queued I/O as possible on the controller
*/
Add a CAM interface to the aac driver. This is useful in case you should ever connect a SCSI Cdrom/Tape/Jukebox/Scanner/Printer/kitty-litter-scooper to your high-end RAID controller. The interface to the arrays is still via the block interface; this merely provides a way to circumvent the RAID functionality and access the SCSI buses directly. Note that for somewhat obvious reasons, hard drives are not exposed to the da driver through this interface, though you can still talk to them via the pass driver. Be the first on your block to low-level format unsuspecting drives that are part of an array! To enable this, add the 'aacp' device to your kernel config. MFC after: 3 days
2002-04-27 01:31:17 +00:00
void
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
aac_startio(struct aac_softc *sc)
{
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
struct aac_command *cm;
Collapse aac_map_command() into aac_startio(). Check the AAC_QUEUE_FRZN in every iteration of aac_startio(). This ensures that a command that is deferred for lack of resources doesn't immediately get retried in the aac_startio() loop. This avoids an almost certain livelock.
2004-06-01 05:32:26 +00:00
int error;
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
for (;;) {
Collapse aac_map_command() into aac_startio(). Check the AAC_QUEUE_FRZN in every iteration of aac_startio(). This ensures that a command that is deferred for lack of resources doesn't immediately get retried in the aac_startio() loop. This avoids an almost certain livelock.
2004-06-01 05:32:26 +00:00
/*
* This flag might be set if the card is out of resources.
* Checking it here prevents an infinite loop of deferrals.
*/
if (sc->flags & AAC_QUEUE_FRZN)
break;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
Remove the unused AAC_DRIVER_BUILD_DATE macro. Nuke whitespace at EOL while I'm here. Approved by: scottl (MAINTAINER)
2005-11-13 17:26:36 +00:00
* Try to get a command that's been put off for lack of
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
* resources
*/
cm = aac_dequeue_ready(sc);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
Remove the unused AAC_DRIVER_BUILD_DATE macro. Nuke whitespace at EOL while I'm here. Approved by: scottl (MAINTAINER)
2005-11-13 17:26:36 +00:00
* Try to build a command off the bio queue (ignore error
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
* return)
*/
if (cm == NULL)
aac_bio_command(sc, &cm);
/* nothing to do? */
if (cm == NULL)
break;
Collapse aac_map_command() into aac_startio(). Check the AAC_QUEUE_FRZN in every iteration of aac_startio(). This ensures that a command that is deferred for lack of resources doesn't immediately get retried in the aac_startio() loop. This avoids an almost certain livelock.
2004-06-01 05:32:26 +00:00
/* don't map more than once */
if (cm->cm_flags & AAC_CMD_MAPPED)
panic("aac: command %p already mapped", cm);
If a command has to be deferred because there are no more resources for it on the card, unmap it first. This allows it to be picked up properly when the queue gets kicked again. This was the root problem for the lost command (i.e. stuck in getblk/vinvalb) problem. While here, panic if commands don't map correctly instead of just silently ignoring the problem and dropping command. Also slow down the dynamic allocation of new commands. It should be safe to go back into the aac waters. Thanks to everyone who suffered through this and provided good feedback.
2004-02-07 10:30:22 +00:00
/*
Collapse aac_map_command() into aac_startio(). Check the AAC_QUEUE_FRZN in every iteration of aac_startio(). This ensures that a command that is deferred for lack of resources doesn't immediately get retried in the aac_startio() loop. This avoids an almost certain livelock.
2004-06-01 05:32:26 +00:00
* Set up the command to go to the controller. If there are no
* data buffers associated with the command then it can bypass
* busdma.
If a command has to be deferred because there are no more resources for it on the card, unmap it first. This allows it to be picked up properly when the queue gets kicked again. This was the root problem for the lost command (i.e. stuck in getblk/vinvalb) problem. While here, panic if commands don't map correctly instead of just silently ignoring the problem and dropping command. Also slow down the dynamic allocation of new commands. It should be safe to go back into the aac waters. Thanks to everyone who suffered through this and provided good feedback.
2004-02-07 10:30:22 +00:00
*/
Collapse aac_map_command() into aac_startio(). Check the AAC_QUEUE_FRZN in every iteration of aac_startio(). This ensures that a command that is deferred for lack of resources doesn't immediately get retried in the aac_startio() loop. This avoids an almost certain livelock.
2004-06-01 05:32:26 +00:00
if (cm->cm_datalen != 0) {
Allow unmapped I/O via aacd(4). It shouldn't be too hard to add the same support for aacp(4), I'm lacking the necessary hardware for testing, though.
2013-05-30 00:22:07 +00:00
if (cm->cm_flags & AAC_REQ_BIO)
error = bus_dmamap_load_bio(
sc->aac_buffer_dmat, cm->cm_datamap,
(struct bio *)cm->cm_private,
aac_map_command_sg, cm, 0);
else
error = bus_dmamap_load(sc->aac_buffer_dmat,
cm->cm_datamap, cm->cm_data,
cm->cm_datalen, aac_map_command_sg, cm, 0);
Collapse aac_map_command() into aac_startio(). Check the AAC_QUEUE_FRZN in every iteration of aac_startio(). This ensures that a command that is deferred for lack of resources doesn't immediately get retried in the aac_startio() loop. This avoids an almost certain livelock.
2004-06-01 05:32:26 +00:00
if (error == EINPROGRESS) {
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_COMM_B, "freezing queue\n");
Collapse aac_map_command() into aac_startio(). Check the AAC_QUEUE_FRZN in every iteration of aac_startio(). This ensures that a command that is deferred for lack of resources doesn't immediately get retried in the aac_startio() loop. This avoids an almost certain livelock.
2004-06-01 05:32:26 +00:00
sc->flags |= AAC_QUEUE_FRZN;
Commit the correct version of the patch from last night. This fixes an immediate panic when doing any i/o, and it closes a completion race.
2004-06-01 15:50:11 +00:00
} else if (error != 0)
Collapse aac_map_command() into aac_startio(). Check the AAC_QUEUE_FRZN in every iteration of aac_startio(). This ensures that a command that is deferred for lack of resources doesn't immediately get retried in the aac_startio() loop. This avoids an almost certain livelock.
2004-06-01 05:32:26 +00:00
panic("aac_startio: unexpected error %d from "
Add \n to the end of a printf string and remove it from panic strings.
2008-03-26 15:28:56 +00:00
"busdma", error);
Collapse aac_map_command() into aac_startio(). Check the AAC_QUEUE_FRZN in every iteration of aac_startio(). This ensures that a command that is deferred for lack of resources doesn't immediately get retried in the aac_startio() loop. This avoids an almost certain livelock.
2004-06-01 05:32:26 +00:00
} else
aac_map_command_sg(cm, NULL, 0, 0);
Handle the EINPROGRESS case of bus_dmamap_load() for data buffers.
2003-07-09 21:16:23 +00:00
}
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
* Handle notification of one or more FIBs coming from the controller.
*/
static void
Various cleanups: - Move the command timeout check from a separate repeating timeout to the kthread since the kthread is already running periodically. - Move printing the hardware print buffer to the kthread. - Properly shut down the kernel thread on detach. - Detach the child array devices on detach. - Don't issue a controller halt command on detach. Doing so requires a PCI reset to wake the controller back up. The driver can now be unloaded as long as CAM support is not enabled.
2003-02-05 23:24:25 +00:00
aac_command_thread(struct aac_softc *sc)
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
{
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
struct aac_fib *fib;
u_int32_t fib_size;
Take the plunge and make this driver be INTR_FAST. This re-arranges the interrupt handler so that no locks are needed, and schedules the command completion routine with a taskqueue_fast. This also corrects the locking in the command thread and removes the need for operation flags. Simple load tests show that this is now considerably faster than FreeBSD 4.x in the SMP case when multiple i/o tasks are running.
2004-01-30 07:04:39 +00:00
int size, retval;
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Remove the AAC_LOCK macros. They no longer abstract anything and only obfuscate the code. No functional differences.
2004-08-12 05:05:06 +00:00
mtx_lock(&sc->aac_io_lock);
- Broaden the scope of locking in aac_command_thread() again to catch some edge cases in the loop. - Try to grab a command before dequeueing the bio from the bioq. The old behaviour of requeuing deferred bios to the end of the bioq is arguably wrong. This should be fixed in the future to check the bioq head without automatically dequeueing the bio.
2004-02-07 03:26:38 +00:00
sc->aifflags = AAC_AIFFLAGS_RUNNING;
while ((sc->aifflags & AAC_AIFFLAGS_EXIT) == 0) {
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
- Broaden the scope of locking in aac_command_thread() again to catch some edge cases in the loop. - Try to grab a command before dequeueing the bio from the bioq. The old behaviour of requeuing deferred bios to the end of the bioq is arguably wrong. This should be fixed in the future to check the bioq head without automatically dequeueing the bio.
2004-02-07 03:26:38 +00:00
retval = 0;
if ((sc->aifflags & AAC_AIFFLAGS_PENDING) == 0)
retval = msleep(sc->aifthread, &sc->aac_io_lock, PRIBIO,
"aifthd", AAC_PERIODIC_INTERVAL * hz);
Various cleanups: - Move the command timeout check from a separate repeating timeout to the kthread since the kthread is already running periodically. - Move printing the hardware print buffer to the kthread. - Properly shut down the kernel thread on detach. - Detach the child array devices on detach. - Don't issue a controller halt command on detach. Doing so requires a PCI reset to wake the controller back up. The driver can now be unloaded as long as CAM support is not enabled.
2003-02-05 23:24:25 +00:00
Take the plunge and make this driver be INTR_FAST. This re-arranges the interrupt handler so that no locks are needed, and schedules the command completion routine with a taskqueue_fast. This also corrects the locking in the command thread and removes the need for operation flags. Simple load tests show that this is now considerably faster than FreeBSD 4.x in the SMP case when multiple i/o tasks are running.
2004-01-30 07:04:39 +00:00
/*
* First see if any FIBs need to be allocated. This needs
* to be called without the driver lock because contigmalloc
- Fix races on detach handling of AAC_IFFLAGS_* mask - Fix races on setting AAC_AIFFLAGS_ALLOCFIBS - Remove some unused AAC_IFFLAGS_* bits. Please note that the kthread still makes a difference between the total mask and AAC_AIFFLAGS_ALLOCFIBS because more flags may be added in the future to aifflags. Sponsored by: Sandvine Incorporated Reported and reviewed by: emaste MFC after: 2 weeks
2011-06-10 20:23:56 +00:00
* can sleep.
Take the plunge and make this driver be INTR_FAST. This re-arranges the interrupt handler so that no locks are needed, and schedules the command completion routine with a taskqueue_fast. This also corrects the locking in the command thread and removes the need for operation flags. Simple load tests show that this is now considerably faster than FreeBSD 4.x in the SMP case when multiple i/o tasks are running.
2004-01-30 07:04:39 +00:00
*/
Bring aac out from under Giant: - the mutex aac_io_lock protects the main codepaths which handle queues and hardware registers. Only one acquire/release is done in the top-half and the taskqueue. This mutex also applies to the userland command path and CAM data path. - Move the taskqueue to the new Giant-free version. - Register the disk device with DISKFLAG_NOGIANT so the top-half processing runs without Giant. - Move the dynamic command allocator to the worker thread to avoid locking issues with bus_dmamem_alloc(). This gives about 20% improvement in most of my benchmarks.
2003-02-26 04:46:21 +00:00
if ((sc->aifflags & AAC_AIFFLAGS_ALLOCFIBS) != 0) {
Remove the AAC_LOCK macros. They no longer abstract anything and only obfuscate the code. No functional differences.
2004-08-12 05:05:06 +00:00
mtx_unlock(&sc->aac_io_lock);
- Broaden the scope of locking in aac_command_thread() again to catch some edge cases in the loop. - Try to grab a command before dequeueing the bio from the bioq. The old behaviour of requeuing deferred bios to the end of the bioq is arguably wrong. This should be fixed in the future to check the bioq head without automatically dequeueing the bio.
2004-02-07 03:26:38 +00:00
aac_alloc_commands(sc);
Remove the AAC_LOCK macros. They no longer abstract anything and only obfuscate the code. No functional differences.
2004-08-12 05:05:06 +00:00
mtx_lock(&sc->aac_io_lock);
If a command has to be deferred because there are no more resources for it on the card, unmap it first. This allows it to be picked up properly when the queue gets kicked again. This was the root problem for the lost command (i.e. stuck in getblk/vinvalb) problem. While here, panic if commands don't map correctly instead of just silently ignoring the problem and dropping command. Also slow down the dynamic allocation of new commands. It should be safe to go back into the aac waters. Thanks to everyone who suffered through this and provided good feedback.
2004-02-07 10:30:22 +00:00
sc->aifflags &= ~AAC_AIFFLAGS_ALLOCFIBS;
- Broaden the scope of locking in aac_command_thread() again to catch some edge cases in the loop. - Try to grab a command before dequeueing the bio from the bioq. The old behaviour of requeuing deferred bios to the end of the bioq is arguably wrong. This should be fixed in the future to check the bioq head without automatically dequeueing the bio.
2004-02-07 03:26:38 +00:00
aac_startio(sc);
Bring aac out from under Giant: - the mutex aac_io_lock protects the main codepaths which handle queues and hardware registers. Only one acquire/release is done in the top-half and the taskqueue. This mutex also applies to the userland command path and CAM data path. - Move the taskqueue to the new Giant-free version. - Register the disk device with DISKFLAG_NOGIANT so the top-half processing runs without Giant. - Move the dynamic command allocator to the worker thread to avoid locking issues with bus_dmamem_alloc(). This gives about 20% improvement in most of my benchmarks.
2003-02-26 04:46:21 +00:00
}
Mega update to the aac driver. 1. Correctly handle commands initiated by the adapter. These commands are defered to a kthread responsible for their processing, then are properly returned to the controller. 2. Add/remove disk devices when notified by the card that a container was created/deleted/offline. 3. Implement crashdump functionality. 4. Support all ioctls needed for the management CLI to work. The linux version of this app can be found at the Dell or HP website. A native version will be forthcoming. MFC-after: 4.4-RELEASE
2001-08-29 23:34:05 +00:00
Take the plunge and make this driver be INTR_FAST. This re-arranges the interrupt handler so that no locks are needed, and schedules the command completion routine with a taskqueue_fast. This also corrects the locking in the command thread and removes the need for operation flags. Simple load tests show that this is now considerably faster than FreeBSD 4.x in the SMP case when multiple i/o tasks are running.
2004-01-30 07:04:39 +00:00
/*
* While we're here, check to see if any commands are stuck.
* This is pretty low-priority, so it's ok if it doesn't
* always fire.
*/
if (retval == EWOULDBLOCK)
aac_timeout(sc);
/* Check the hardware printf message buffer */
if (sc->aac_common->ac_printf[0] != 0)
aac_print_printf(sc);
/* Also check to see if the adapter has a command for us. */
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
if (sc->flags & AAC_FLAGS_NEW_COMM)
continue;
for (;;) {
if (aac_dequeue_fib(sc, AAC_HOST_NORM_CMD_QUEUE,
Remove the unused AAC_DRIVER_BUILD_DATE macro. Nuke whitespace at EOL while I'm here. Approved by: scottl (MAINTAINER)
2005-11-13 17:26:36 +00:00
&fib_size, &fib))
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
break;
Remove the unused AAC_DRIVER_BUILD_DATE macro. Nuke whitespace at EOL while I'm here. Approved by: scottl (MAINTAINER)
2005-11-13 17:26:36 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
AAC_PRINT_FIB(sc, fib);
Remove the unused AAC_DRIVER_BUILD_DATE macro. Nuke whitespace at EOL while I'm here. Approved by: scottl (MAINTAINER)
2005-11-13 17:26:36 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
switch (fib->Header.Command) {
case AifRequest:
aac_handle_aif(sc, fib);
break;
default:
device_printf(sc->aac_dev, "unknown command "
"from controller\n");
break;
}
Mega update to the aac driver. 1. Correctly handle commands initiated by the adapter. These commands are defered to a kthread responsible for their processing, then are properly returned to the controller. 2. Add/remove disk devices when notified by the card that a container was created/deleted/offline. 3. Implement crashdump functionality. 4. Support all ioctls needed for the management CLI to work. The linux version of this app can be found at the Dell or HP website. A native version will be forthcoming. MFC-after: 4.4-RELEASE
2001-08-29 23:34:05 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
if ((fib->Header.XferState == 0) ||
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
(fib->Header.StructType != AAC_FIBTYPE_TFIB)) {
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
break;
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
}
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
Various cleanups: - Move the command timeout check from a separate repeating timeout to the kthread since the kthread is already running periodically. - Move printing the hardware print buffer to the kthread. - Properly shut down the kernel thread on detach. - Detach the child array devices on detach. - Don't issue a controller halt command on detach. Doing so requires a PCI reset to wake the controller back up. The driver can now be unloaded as long as CAM support is not enabled.
2003-02-05 23:24:25 +00:00
/* Return the AIF to the controller. */
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
if (fib->Header.XferState & AAC_FIBSTATE_FROMADAP) {
fib->Header.XferState |= AAC_FIBSTATE_DONEHOST;
*(AAC_FSAStatus*)fib->data = ST_OK;
/* XXX Compute the Size field? */
size = fib->Header.Size;
if (size > sizeof(struct aac_fib)) {
Add a CAM interface to the aac driver. This is useful in case you should ever connect a SCSI Cdrom/Tape/Jukebox/Scanner/Printer/kitty-litter-scooper to your high-end RAID controller. The interface to the arrays is still via the block interface; this merely provides a way to circumvent the RAID functionality and access the SCSI buses directly. Note that for somewhat obvious reasons, hard drives are not exposed to the da driver through this interface, though you can still talk to them via the pass driver. Be the first on your block to low-level format unsuspecting drives that are part of an array! To enable this, add the 'aacp' device to your kernel config. MFC after: 3 days
2002-04-27 01:31:17 +00:00
size = sizeof(struct aac_fib);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
fib->Header.Size = size;
}
/*
* Since we did not generate this command, it
* cannot go through the normal
* enqueue->startio chain.
*/
aac_enqueue_response(sc,
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
AAC_ADAP_NORM_RESP_QUEUE,
fib);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
}
Mega update to the aac driver. 1. Correctly handle commands initiated by the adapter. These commands are defered to a kthread responsible for their processing, then are properly returned to the controller. 2. Add/remove disk devices when notified by the card that a container was created/deleted/offline. 3. Implement crashdump functionality. 4. Support all ioctls needed for the management CLI to work. The linux version of this app can be found at the Dell or HP website. A native version will be forthcoming. MFC-after: 4.4-RELEASE
2001-08-29 23:34:05 +00:00
}
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
}
sc->aifflags &= ~AAC_AIFFLAGS_RUNNING;
Remove the AAC_LOCK macros. They no longer abstract anything and only obfuscate the code. No functional differences.
2004-08-12 05:05:06 +00:00
mtx_unlock(&sc->aac_io_lock);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
wakeup(sc->aac_dev);
Mega update to the aac driver. 1. Correctly handle commands initiated by the adapter. These commands are defered to a kthread responsible for their processing, then are properly returned to the controller. 2. Add/remove disk devices when notified by the card that a container was created/deleted/offline. 3. Implement crashdump functionality. 4. Support all ioctls needed for the management CLI to work. The linux version of this app can be found at the Dell or HP website. A native version will be forthcoming. MFC-after: 4.4-RELEASE
2001-08-29 23:34:05 +00:00
Rename the kthread_xxx (e.g. kthread_create()) calls to kproc_xxx as they actually make whole processes. Thos makes way for us to add REAL kthread_create() and friends that actually make theads. it turns out that most of these calls actually end up being moved back to the thread version when it's added. but we need to make this cosmetic change first. I'd LOVE to do this rename in 7.0 so that we can eventually MFC the new kthread_xxx() calls.
2007-10-20 23:23:23 +00:00
kproc_exit(0);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
Make the aac driver be INTR_MPSAFE. Once the interrupt handler determines that a command completion happened, all further processing is deferred to a taskqueue. The taskqueue itself runs implicetely under Giant, but we already used a taskqueue for the biodone() processing, so this at least saves the contesting of Giant in the interrupt handler.
2003-02-19 21:58:34 +00:00
* Process completed commands.
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
*/
static void
Make the aac driver be INTR_MPSAFE. Once the interrupt handler determines that a command completion happened, all further processing is deferred to a taskqueue. The taskqueue itself runs implicetely under Giant, but we already used a taskqueue for the biodone() processing, so this at least saves the contesting of Giant in the interrupt handler.
2003-02-19 21:58:34 +00:00
aac_complete(void *context, int pending)
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
{
Make the aac driver be INTR_MPSAFE. Once the interrupt handler determines that a command completion happened, all further processing is deferred to a taskqueue. The taskqueue itself runs implicetely under Giant, but we already used a taskqueue for the biodone() processing, so this at least saves the contesting of Giant in the interrupt handler.
2003-02-19 21:58:34 +00:00
struct aac_softc *sc;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
struct aac_command *cm;
struct aac_fib *fib;
u_int32_t fib_size;
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Make the aac driver be INTR_MPSAFE. Once the interrupt handler determines that a command completion happened, all further processing is deferred to a taskqueue. The taskqueue itself runs implicetely under Giant, but we already used a taskqueue for the biodone() processing, so this at least saves the contesting of Giant in the interrupt handler.
2003-02-19 21:58:34 +00:00
sc = (struct aac_softc *)context;
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
Make the aac driver be INTR_MPSAFE. Once the interrupt handler determines that a command completion happened, all further processing is deferred to a taskqueue. The taskqueue itself runs implicetely under Giant, but we already used a taskqueue for the biodone() processing, so this at least saves the contesting of Giant in the interrupt handler.
2003-02-19 21:58:34 +00:00
Remove the AAC_LOCK macros. They no longer abstract anything and only obfuscate the code. No functional differences.
2004-08-12 05:05:06 +00:00
mtx_lock(&sc->aac_io_lock);
Bring aac out from under Giant: - the mutex aac_io_lock protects the main codepaths which handle queues and hardware registers. Only one acquire/release is done in the top-half and the taskqueue. This mutex also applies to the userland command path and CAM data path. - Move the taskqueue to the new Giant-free version. - Register the disk device with DISKFLAG_NOGIANT so the top-half processing runs without Giant. - Move the dynamic command allocator to the worker thread to avoid locking issues with bus_dmamem_alloc(). This gives about 20% improvement in most of my benchmarks.
2003-02-26 04:46:21 +00:00
Make the aac driver be INTR_MPSAFE. Once the interrupt handler determines that a command completion happened, all further processing is deferred to a taskqueue. The taskqueue itself runs implicetely under Giant, but we already used a taskqueue for the biodone() processing, so this at least saves the contesting of Giant in the interrupt handler.
2003-02-19 21:58:34 +00:00
/* pull completed commands off the queue */
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
for (;;) {
/* look for completed FIBs on our queue */
if (aac_dequeue_fib(sc, AAC_HOST_NORM_RESP_QUEUE, &fib_size,
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
&fib))
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
break; /* nothing to do */
Make the aac driver be INTR_MPSAFE. Once the interrupt handler determines that a command completion happened, all further processing is deferred to a taskqueue. The taskqueue itself runs implicetely under Giant, but we already used a taskqueue for the biodone() processing, so this at least saves the contesting of Giant in the interrupt handler.
2003-02-19 21:58:34 +00:00
Do some small cleanups to comments and remove AACQ_COMPLETE definitions since the completion queue is long-gone.
2004-02-07 17:40:38 +00:00
/* get the command, unmap and hand off for processing */
Fix a 64-bit bogon. The hardware command structure only has one 32 bit field for holding driver-dependant data. Instead of putting the pointer to the driver command struct in there, take advantage of these structs being a (virtually) contiguous array and just put the array index in the field.
2003-02-19 23:33:56 +00:00
cm = sc->aac_commands + fib->Header.SenderData;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
if (cm == NULL) {
AAC_PRINT_FIB(sc, fib);
Make the aac driver be INTR_MPSAFE. Once the interrupt handler determines that a command completion happened, all further processing is deferred to a taskqueue. The taskqueue itself runs implicetely under Giant, but we already used a taskqueue for the biodone() processing, so this at least saves the contesting of Giant in the interrupt handler.
2003-02-19 21:58:34 +00:00
break;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
}
Avoid repeatedly spamming the console while a timed out command is waiting to complete. Instead, print one message after the timeout period expires, and one more when (if) the command eventually completes. MFC after: 1 month
2010-09-14 01:51:04 +00:00
if ((cm->cm_flags & AAC_CMD_TIMEDOUT) != 0)
device_printf(sc->aac_dev,
"COMMAND %p COMPLETED AFTER %d SECONDS\n",
cm, (int)(time_uptime-cm->cm_timestamp));
Make the aac driver be INTR_MPSAFE. Once the interrupt handler determines that a command completion happened, all further processing is deferred to a taskqueue. The taskqueue itself runs implicetely under Giant, but we already used a taskqueue for the biodone() processing, so this at least saves the contesting of Giant in the interrupt handler.
2003-02-19 21:58:34 +00:00
aac_remove_busy(cm);
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
aac_unmap_command(cm);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
cm->cm_flags |= AAC_CMD_COMPLETED;
/* is there a completion handler? */
if (cm->cm_complete != NULL) {
cm->cm_complete(cm);
} else {
/* assume that someone is sleeping on this command */
wakeup(cm);
}
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
Major bugfix and minor update. This should resolve the current issues with the driver locking up under load. - Restructure so that we use a static pool of commands/FIBs, rather than allocating them in clusters. The cluster allocation just made things more complicated, and allowed us to waste more memory in peak load situations. - Make queueing macros more like my other drivers. This adds queue stats for free. Add some debugging to take advantage of this. - Reimplement the periodic timeout scan. Kick the interrupt handler and the start routine every scan as well, just to be safe. Track busy commands properly. - Bring resource cleanup into line with resource allocation. We should now clean up correctly after a failed probe/unload/etc. - Try to start new commands when old ones are completed. We weren't doing this before, which could lead to deadlock when the controller was full. - Don't try to build a new command if we have found a deferred command. This could cause us to lose the deferred command. - Use diskerr() to report I/O errors. - Don't bail if the AdapterInfo structure is the wrong size. Some variation seems to be normal. We need to improve our handing of 2.x firmware sets. - Improve some comments in an attempt to try to make things clearer. - Restructure to avoid some warnings.
2000-12-27 13:14:56 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/* see if we can start some more I/O */
Handle the EINPROGRESS case of bus_dmamap_load() for data buffers.
2003-07-09 21:16:23 +00:00
sc->flags &= ~AAC_QUEUE_FRZN;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
aac_startio(sc);
Bring aac out from under Giant: - the mutex aac_io_lock protects the main codepaths which handle queues and hardware registers. Only one acquire/release is done in the top-half and the taskqueue. This mutex also applies to the userland command path and CAM data path. - Move the taskqueue to the new Giant-free version. - Register the disk device with DISKFLAG_NOGIANT so the top-half processing runs without Giant. - Move the dynamic command allocator to the worker thread to avoid locking issues with bus_dmamem_alloc(). This gives about 20% improvement in most of my benchmarks.
2003-02-26 04:46:21 +00:00
Remove the AAC_LOCK macros. They no longer abstract anything and only obfuscate the code. No functional differences.
2004-08-12 05:05:06 +00:00
mtx_unlock(&sc->aac_io_lock);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
* Handle a bio submitted from a disk device.
*/
void
aac_submit_bio(struct bio *bp)
{
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
struct aac_disk *ad;
struct aac_softc *sc;
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Move to 'struct disk*' API Submitted by: phk
2003-02-26 02:50:27 +00:00
ad = (struct aac_disk *)bp->bio_disk->d_drv1;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
sc = ad->ad_controller;
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/* queue the BIO and try to get some work done */
aac_enqueue_bio(sc, bp);
aac_startio(sc);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
* Get a bio and build a command to go with it.
*/
static int
aac_bio_command(struct aac_softc *sc, struct aac_command **cmp)
{
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
struct aac_command *cm;
struct aac_fib *fib;
struct aac_disk *ad;
struct bio *bp;
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/* get the resources we will need */
cm = NULL;
- Broaden the scope of locking in aac_command_thread() again to catch some edge cases in the loop. - Try to grab a command before dequeueing the bio from the bioq. The old behaviour of requeuing deferred bios to the end of the bioq is arguably wrong. This should be fixed in the future to check the bioq head without automatically dequeueing the bio.
2004-02-07 03:26:38 +00:00
bp = NULL;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
if (aac_alloc_command(sc, &cm)) /* get a command */
goto fail;
- Broaden the scope of locking in aac_command_thread() again to catch some edge cases in the loop. - Try to grab a command before dequeueing the bio from the bioq. The old behaviour of requeuing deferred bios to the end of the bioq is arguably wrong. This should be fixed in the future to check the bioq head without automatically dequeueing the bio.
2004-02-07 03:26:38 +00:00
if ((bp = aac_dequeue_bio(sc)) == NULL)
goto fail;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/* fill out the command */
cm->cm_datalen = bp->bio_bcount;
cm->cm_complete = aac_bio_complete;
Allow unmapped I/O via aacd(4). It shouldn't be too hard to add the same support for aacp(4), I'm lacking the necessary hardware for testing, though.
2013-05-30 00:22:07 +00:00
cm->cm_flags = AAC_REQ_BIO;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
cm->cm_private = bp;
Use time_uptime for timing commands instead of time_second. This is a bit risky because the "current time" is supposed to be fed to the card during initialization, and the current time is supposed to be put into each command that is sent to the card. Hopefully either the card doesn't actually care about the timestamps, or it doesn't care about the absolute values so long and the relative values are consistent. Not an MFC candidate until more thorough testing can be done.
2005-09-14 05:46:28 +00:00
cm->cm_timestamp = time_uptime;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/* build the FIB */
fib = cm->cm_fib;
Add the ability to send 64-bit scatter/gather elements to aac cards. This is enabled when both the size of bus_addr_t > 4 and the card claims support. Don't wake up the kthread to allocate more commands if we know that we've already allocated the max number of commands.
2003-03-30 21:47:16 +00:00
fib->Header.Size = sizeof(struct aac_fib_header);
Remove the unused AAC_DRIVER_BUILD_DATE macro. Nuke whitespace at EOL while I'm here. Approved by: scottl (MAINTAINER)
2005-11-13 17:26:36 +00:00
fib->Header.XferState =
AAC_FIBSTATE_HOSTOWNED |
AAC_FIBSTATE_INITIALISED |
AAC_FIBSTATE_EMPTY |
Major bugfixes for large memory and fast systems. aac.c: Re-arrange the interrupt handler to optimize the common case of the adapter interrupting us because one or more commands are complete, and do a read across the pci bus to ensure that all posted status writes are flushed. This should close a race that could cause command completion interrupts to be lost. Follow the spec a bit closer when filling out command structures. Enable the Fast Response feature to eliminate the need for the card to DMA successfull command completions back into host memory. Tell the controller how much physical memory we have. Without this there was a chance that our DMA regions would collide with the memory window used by the cache on the controller. The result would be massive data corruption. This seemed to mainly affect systems with >2GB of memory. Fix a few whitespace problems. aac_debug.c: Add an extra diagnostic when printing out commands. aac_disk.c: Add extra sanity checks. aacreg.h: Prepare for making this 64-bit clean by reducing the use of enumeration types in structures. Many thanks to Justin Gibbs for helping track these down.
2003-01-11 01:59:21 +00:00
AAC_FIBSTATE_FROMHOST |
AAC_FIBSTATE_REXPECTED |
AAC_FIBSTATE_NORM |
AAC_FIBSTATE_ASYNC |
AAC_FIBSTATE_FAST_RESPONSE;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/* build the read/write request */
Move to 'struct disk*' API Submitted by: phk
2003-02-26 02:50:27 +00:00
ad = (struct aac_disk *)bp->bio_disk->d_drv1;
Add the ability to send 64-bit scatter/gather elements to aac cards. This is enabled when both the size of bus_addr_t > 4 and the card claims support. Don't wake up the kthread to allocate more commands if we know that we've already allocated the max number of commands.
2003-03-30 21:47:16 +00:00
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
if (sc->flags & AAC_FLAGS_RAW_IO) {
struct aac_raw_io *raw;
raw = (struct aac_raw_io *)&fib->data[0];
fib->Header.Command = RawIo;
raw->BlockNumber = (u_int64_t)bp->bio_pblkno;
raw->ByteCount = bp->bio_bcount;
raw->ContainerId = ad->ad_container->co_mntobj.ObjectId;
raw->BpTotal = 0;
raw->BpComplete = 0;
fib->Header.Size += sizeof(struct aac_raw_io);
cm->cm_sgtable = (struct aac_sg_table *)&raw->SgMapRaw;
if (bp->bio_cmd == BIO_READ) {
raw->Flags = 1;
cm->cm_flags |= AAC_CMD_DATAIN;
} else {
raw->Flags = 0;
cm->cm_flags |= AAC_CMD_DATAOUT;
}
} else if ((sc->flags & AAC_FLAGS_SG_64BIT) == 0) {
Add the ability to send 64-bit scatter/gather elements to aac cards. This is enabled when both the size of bus_addr_t > 4 and the card claims support. Don't wake up the kthread to allocate more commands if we know that we've already allocated the max number of commands.
2003-03-30 21:47:16 +00:00
fib->Header.Command = ContainerCommand;
if (bp->bio_cmd == BIO_READ) {
struct aac_blockread *br;
br = (struct aac_blockread *)&fib->data[0];
br->Command = VM_CtBlockRead;
br->ContainerId = ad->ad_container->co_mntobj.ObjectId;
br->BlockNumber = bp->bio_pblkno;
br->ByteCount = bp->bio_bcount;
fib->Header.Size += sizeof(struct aac_blockread);
cm->cm_sgtable = &br->SgMap;
cm->cm_flags |= AAC_CMD_DATAIN;
} else {
struct aac_blockwrite *bw;
bw = (struct aac_blockwrite *)&fib->data[0];
bw->Command = VM_CtBlockWrite;
bw->ContainerId = ad->ad_container->co_mntobj.ObjectId;
bw->BlockNumber = bp->bio_pblkno;
bw->ByteCount = bp->bio_bcount;
bw->Stable = CUNSTABLE;
fib->Header.Size += sizeof(struct aac_blockwrite);
cm->cm_flags |= AAC_CMD_DATAOUT;
cm->cm_sgtable = &bw->SgMap;
}
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
} else {
Add the ability to send 64-bit scatter/gather elements to aac cards. This is enabled when both the size of bus_addr_t > 4 and the card claims support. Don't wake up the kthread to allocate more commands if we know that we've already allocated the max number of commands.
2003-03-30 21:47:16 +00:00
fib->Header.Command = ContainerCommand64;
if (bp->bio_cmd == BIO_READ) {
struct aac_blockread64 *br;
br = (struct aac_blockread64 *)&fib->data[0];
br->Command = VM_CtHostRead64;
br->ContainerId = ad->ad_container->co_mntobj.ObjectId;
br->SectorCount = bp->bio_bcount / AAC_BLOCK_SIZE;
br->BlockNumber = bp->bio_pblkno;
br->Pad = 0;
br->Flags = 0;
fib->Header.Size += sizeof(struct aac_blockread64);
Correct data direction flags in aac_bio_command() in the !AAC_FLAGS_RAW_IO && AAC_FLAGS_SG_64BIT case. Submitted by: Adaptec
2008-03-25 18:34:04 +00:00
cm->cm_flags |= AAC_CMD_DATAIN;
Avoid casts as lvalues.
2004-07-28 06:21:53 +00:00
cm->cm_sgtable = (struct aac_sg_table *)&br->SgMap64;
Add the ability to send 64-bit scatter/gather elements to aac cards. This is enabled when both the size of bus_addr_t > 4 and the card claims support. Don't wake up the kthread to allocate more commands if we know that we've already allocated the max number of commands.
2003-03-30 21:47:16 +00:00
} else {
struct aac_blockwrite64 *bw;
bw = (struct aac_blockwrite64 *)&fib->data[0];
bw->Command = VM_CtHostWrite64;
bw->ContainerId = ad->ad_container->co_mntobj.ObjectId;
bw->SectorCount = bp->bio_bcount / AAC_BLOCK_SIZE;
bw->BlockNumber = bp->bio_pblkno;
bw->Pad = 0;
bw->Flags = 0;
fib->Header.Size += sizeof(struct aac_blockwrite64);
Correct data direction flags in aac_bio_command() in the !AAC_FLAGS_RAW_IO && AAC_FLAGS_SG_64BIT case. Submitted by: Adaptec
2008-03-25 18:34:04 +00:00
cm->cm_flags |= AAC_CMD_DATAOUT;
Avoid casts as lvalues.
2004-07-28 06:21:53 +00:00
cm->cm_sgtable = (struct aac_sg_table *)&bw->SgMap64;
Add the ability to send 64-bit scatter/gather elements to aac cards. This is enabled when both the size of bus_addr_t > 4 and the card claims support. Don't wake up the kthread to allocate more commands if we know that we've already allocated the max number of commands.
2003-03-30 21:47:16 +00:00
}
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
}
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
*cmp = cm;
return(0);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
fail:
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
if (bp != NULL)
aac_enqueue_bio(sc, bp);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
if (cm != NULL)
aac_release_command(cm);
return(ENOMEM);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
* Handle a bio-instigated command that has been completed.
*/
static void
aac_bio_complete(struct aac_command *cm)
{
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
struct aac_blockread_response *brr;
struct aac_blockwrite_response *bwr;
struct bio *bp;
AAC_FSAStatus status;
/* fetch relevant status and then release the command */
bp = (struct bio *)cm->cm_private;
The aac driver has evolved enough over the last few months that it no longer resembles the 4.x version very much. Garbage collect the legacy bits.
2003-03-01 05:07:19 +00:00
if (bp->bio_cmd == BIO_READ) {
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
brr = (struct aac_blockread_response *)&cm->cm_fib->data[0];
status = brr->Status;
} else {
bwr = (struct aac_blockwrite_response *)&cm->cm_fib->data[0];
status = bwr->Status;
}
aac_release_command(cm);
/* fix up the bio based on status */
if (status == ST_OK) {
bp->bio_resid = 0;
} else {
bp->bio_error = EIO;
bp->bio_flags |= BIO_ERROR;
}
aac_biodone(bp);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
* Submit a command to the controller, return when it completes.
Add functionality and fix bugs so the driver will work with soon-to-be released management apps. 1. Implement poll(). This will check for queued aif's so that a subsequent ioctl call to retrieve the next aif will not block. 2. Don't catch signals when sleeping on a fib sent from userland. This causes a race and panic due to the pthread context switcher waking up the tsleep at inopportune times. 3. Fix some whitespace nits. MFC after: 3 days
2001-12-02 03:47:33 +00:00
* XXX This is very dangerous! If the card has gone out to lunch, we could
* be stuck here forever. At the same time, signals are not caught
Remove the 'timeout' argument from aac_wait_command() as it isn't used and never will be. Update the XXX comment for this function to accurately reflect why things are the way they are.
2004-04-14 19:11:29 +00:00
* because there is a risk that a signal could wakeup the sleep before
* the card has a chance to complete the command. Since there is no way
* to cancel a command that is in progress, we can't protect against the
* card completing a command late and spamming the command and data
* memory. So, we are held hostage until the command completes.
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
*/
static int
Remove the 'timeout' argument from aac_wait_command() as it isn't used and never will be. Update the XXX comment for this function to accurately reflect why things are the way they are.
2004-04-14 19:11:29 +00:00
aac_wait_command(struct aac_command *cm)
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
{
Bring aac out from under Giant: - the mutex aac_io_lock protects the main codepaths which handle queues and hardware registers. Only one acquire/release is done in the top-half and the taskqueue. This mutex also applies to the userland command path and CAM data path. - Move the taskqueue to the new Giant-free version. - Register the disk device with DISKFLAG_NOGIANT so the top-half processing runs without Giant. - Move the dynamic command allocator to the worker thread to avoid locking issues with bus_dmamem_alloc(). This gives about 20% improvement in most of my benchmarks.
2003-02-26 04:46:21 +00:00
struct aac_softc *sc;
Remove the 'timeout' argument from aac_wait_command() as it isn't used and never will be. Update the XXX comment for this function to accurately reflect why things are the way they are.
2004-04-14 19:11:29 +00:00
int error;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
Bring aac out from under Giant: - the mutex aac_io_lock protects the main codepaths which handle queues and hardware registers. Only one acquire/release is done in the top-half and the taskqueue. This mutex also applies to the userland command path and CAM data path. - Move the taskqueue to the new Giant-free version. - Register the disk device with DISKFLAG_NOGIANT so the top-half processing runs without Giant. - Move the dynamic command allocator to the worker thread to avoid locking issues with bus_dmamem_alloc(). This gives about 20% improvement in most of my benchmarks.
2003-02-26 04:46:21 +00:00
sc = cm->cm_sc;
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
Bring aac out from under Giant: - the mutex aac_io_lock protects the main codepaths which handle queues and hardware registers. Only one acquire/release is done in the top-half and the taskqueue. This mutex also applies to the userland command path and CAM data path. - Move the taskqueue to the new Giant-free version. - Register the disk device with DISKFLAG_NOGIANT so the top-half processing runs without Giant. - Move the dynamic command allocator to the worker thread to avoid locking issues with bus_dmamem_alloc(). This gives about 20% improvement in most of my benchmarks.
2003-02-26 04:46:21 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/* Put the command on the ready queue and get things going */
aac_enqueue_ready(cm);
Bring aac out from under Giant: - the mutex aac_io_lock protects the main codepaths which handle queues and hardware registers. Only one acquire/release is done in the top-half and the taskqueue. This mutex also applies to the userland command path and CAM data path. - Move the taskqueue to the new Giant-free version. - Register the disk device with DISKFLAG_NOGIANT so the top-half processing runs without Giant. - Move the dynamic command allocator to the worker thread to avoid locking issues with bus_dmamem_alloc(). This gives about 20% improvement in most of my benchmarks.
2003-02-26 04:46:21 +00:00
aac_startio(sc);
Remove the 'timeout' argument from aac_wait_command() as it isn't used and never will be. Update the XXX comment for this function to accurately reflect why things are the way they are.
2004-04-14 19:11:29 +00:00
error = msleep(cm, &sc->aac_io_lock, PRIBIO, "aacwait", 0);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
return(error);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
*Command Buffer Management
*/
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
* Allocate a command.
*/
Add a CAM interface to the aac driver. This is useful in case you should ever connect a SCSI Cdrom/Tape/Jukebox/Scanner/Printer/kitty-litter-scooper to your high-end RAID controller. The interface to the arrays is still via the block interface; this merely provides a way to circumvent the RAID functionality and access the SCSI buses directly. Note that for somewhat obvious reasons, hard drives are not exposed to the da driver through this interface, though you can still talk to them via the pass driver. Be the first on your block to low-level format unsuspecting drives that are part of an array! To enable this, add the 'aacp' device to your kernel config. MFC after: 3 days
2002-04-27 01:31:17 +00:00
int
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
aac_alloc_command(struct aac_softc *sc, struct aac_command **cmp)
{
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
struct aac_command *cm;
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Implement a new dynamic command allocator. FIBs are allocated in 1 page blocks now, which should eliminate problems with the driver failing to attach due to insufficient contiguous RAM. Allow the FIB pool to grow from the default of 128 to the max of 512 as demand grows. Also pad the adapter init struct to work around the 2120/2200 DMA bug now that there is no longer a FIB slab.
2003-02-10 00:34:24 +00:00
if ((cm = aac_dequeue_free(sc)) == NULL) {
Add the ability to send 64-bit scatter/gather elements to aac cards. This is enabled when both the size of bus_addr_t > 4 and the card claims support. Don't wake up the kthread to allocate more commands if we know that we've already allocated the max number of commands.
2003-03-30 21:47:16 +00:00
if (sc->total_fibs < sc->aac_max_fibs) {
- Fix races on detach handling of AAC_IFFLAGS_* mask - Fix races on setting AAC_AIFFLAGS_ALLOCFIBS - Remove some unused AAC_IFFLAGS_* bits. Please note that the kthread still makes a difference between the total mask and AAC_AIFFLAGS_ALLOCFIBS because more flags may be added in the future to aifflags. Sponsored by: Sandvine Incorporated Reported and reviewed by: emaste MFC after: 2 weeks
2011-06-10 20:23:56 +00:00
mtx_lock(&sc->aac_io_lock);
Add the ability to send 64-bit scatter/gather elements to aac cards. This is enabled when both the size of bus_addr_t > 4 and the card claims support. Don't wake up the kthread to allocate more commands if we know that we've already allocated the max number of commands.
2003-03-30 21:47:16 +00:00
sc->aifflags |= AAC_AIFFLAGS_ALLOCFIBS;
- Fix races on detach handling of AAC_IFFLAGS_* mask - Fix races on setting AAC_AIFFLAGS_ALLOCFIBS - Remove some unused AAC_IFFLAGS_* bits. Please note that the kthread still makes a difference between the total mask and AAC_AIFFLAGS_ALLOCFIBS because more flags may be added in the future to aifflags. Sponsored by: Sandvine Incorporated Reported and reviewed by: emaste MFC after: 2 weeks
2011-06-10 20:23:56 +00:00
mtx_unlock(&sc->aac_io_lock);
Add the ability to send 64-bit scatter/gather elements to aac cards. This is enabled when both the size of bus_addr_t > 4 and the card claims support. Don't wake up the kthread to allocate more commands if we know that we've already allocated the max number of commands.
2003-03-30 21:47:16 +00:00
wakeup(sc->aifthread);
}
Bring aac out from under Giant: - the mutex aac_io_lock protects the main codepaths which handle queues and hardware registers. Only one acquire/release is done in the top-half and the taskqueue. This mutex also applies to the userland command path and CAM data path. - Move the taskqueue to the new Giant-free version. - Register the disk device with DISKFLAG_NOGIANT so the top-half processing runs without Giant. - Move the dynamic command allocator to the worker thread to avoid locking issues with bus_dmamem_alloc(). This gives about 20% improvement in most of my benchmarks.
2003-02-26 04:46:21 +00:00
return (EBUSY);
Implement a new dynamic command allocator. FIBs are allocated in 1 page blocks now, which should eliminate problems with the driver failing to attach due to insufficient contiguous RAM. Allow the FIB pool to grow from the default of 128 to the max of 512 as demand grows. Also pad the adapter init struct to work around the 2120/2200 DMA bug now that there is no longer a FIB slab.
2003-02-10 00:34:24 +00:00
}
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
*cmp = cm;
return(0);
Major bugfix and minor update. This should resolve the current issues with the driver locking up under load. - Restructure so that we use a static pool of commands/FIBs, rather than allocating them in clusters. The cluster allocation just made things more complicated, and allowed us to waste more memory in peak load situations. - Make queueing macros more like my other drivers. This adds queue stats for free. Add some debugging to take advantage of this. - Reimplement the periodic timeout scan. Kick the interrupt handler and the start routine every scan as well, just to be safe. Track busy commands properly. - Bring resource cleanup into line with resource allocation. We should now clean up correctly after a failed probe/unload/etc. - Try to start new commands when old ones are completed. We weren't doing this before, which could lead to deadlock when the controller was full. - Don't try to build a new command if we have found a deferred command. This could cause us to lose the deferred command. - Use diskerr() to report I/O errors. - Don't bail if the AdapterInfo structure is the wrong size. Some variation seems to be normal. We need to improve our handing of 2.x firmware sets. - Improve some comments in an attempt to try to make things clearer. - Restructure to avoid some warnings.
2000-12-27 13:14:56 +00:00
}
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
Major bugfix and minor update. This should resolve the current issues with the driver locking up under load. - Restructure so that we use a static pool of commands/FIBs, rather than allocating them in clusters. The cluster allocation just made things more complicated, and allowed us to waste more memory in peak load situations. - Make queueing macros more like my other drivers. This adds queue stats for free. Add some debugging to take advantage of this. - Reimplement the periodic timeout scan. Kick the interrupt handler and the start routine every scan as well, just to be safe. Track busy commands properly. - Bring resource cleanup into line with resource allocation. We should now clean up correctly after a failed probe/unload/etc. - Try to start new commands when old ones are completed. We weren't doing this before, which could lead to deadlock when the controller was full. - Don't try to build a new command if we have found a deferred command. This could cause us to lose the deferred command. - Use diskerr() to report I/O errors. - Don't bail if the AdapterInfo structure is the wrong size. Some variation seems to be normal. We need to improve our handing of 2.x firmware sets. - Improve some comments in an attempt to try to make things clearer. - Restructure to avoid some warnings.
2000-12-27 13:14:56 +00:00
* Release a command back to the freelist.
*/
Add a CAM interface to the aac driver. This is useful in case you should ever connect a SCSI Cdrom/Tape/Jukebox/Scanner/Printer/kitty-litter-scooper to your high-end RAID controller. The interface to the arrays is still via the block interface; this merely provides a way to circumvent the RAID functionality and access the SCSI buses directly. Note that for somewhat obvious reasons, hard drives are not exposed to the da driver through this interface, though you can still talk to them via the pass driver. Be the first on your block to low-level format unsuspecting drives that are part of an array! To enable this, add the 'aacp' device to your kernel config. MFC after: 3 days
2002-04-27 01:31:17 +00:00
void
Major bugfix and minor update. This should resolve the current issues with the driver locking up under load. - Restructure so that we use a static pool of commands/FIBs, rather than allocating them in clusters. The cluster allocation just made things more complicated, and allowed us to waste more memory in peak load situations. - Make queueing macros more like my other drivers. This adds queue stats for free. Add some debugging to take advantage of this. - Reimplement the periodic timeout scan. Kick the interrupt handler and the start routine every scan as well, just to be safe. Track busy commands properly. - Bring resource cleanup into line with resource allocation. We should now clean up correctly after a failed probe/unload/etc. - Try to start new commands when old ones are completed. We weren't doing this before, which could lead to deadlock when the controller was full. - Don't try to build a new command if we have found a deferred command. This could cause us to lose the deferred command. - Use diskerr() to report I/O errors. - Don't bail if the AdapterInfo structure is the wrong size. Some variation seems to be normal. We need to improve our handing of 2.x firmware sets. - Improve some comments in an attempt to try to make things clearer. - Restructure to avoid some warnings.
2000-12-27 13:14:56 +00:00
aac_release_command(struct aac_command *cm)
{
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
struct aac_event *event;
struct aac_softc *sc;
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
sc = cm->cm_sc;
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
Change spelling and eliminate a typo in comments to reduce diffs with Adaptec's vendor driver. I have some fixes to bring in and this makes ongoing review of the FreeBSD-Adaptec driver diffs easier.
2008-03-14 21:59:11 +00:00
/* (re)initialize the command/FIB */
Clear the data buffer length field when freeing a command structure so that it doesn't leak through when the command structure is reused for a user command without a data buffer. PR: amd64/189668 Tested by: Pete Long <pete@nrth.org> MFC after: 1 week
2014-05-17 02:45:04 +00:00
cm->cm_datalen = 0;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
cm->cm_sgtable = NULL;
cm->cm_flags = 0;
cm->cm_complete = NULL;
cm->cm_private = NULL;
Minor diff reduction with Adaptec's driver: in aac_release_command() set cm_queue to AAC_ADAP_NORM_CMD_QUEUE by default. In every place it was set, it was set to AAC_ADAP_NORM_CMD_QUEUE anyhow.
2010-02-23 21:41:13 +00:00
cm->cm_queue = AAC_ADAP_NORM_CMD_QUEUE;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
cm->cm_fib->Header.XferState = AAC_FIBSTATE_EMPTY;
cm->cm_fib->Header.StructType = AAC_FIBTYPE_TFIB;
cm->cm_fib->Header.Flags = 0;
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
cm->cm_fib->Header.SenderSize = cm->cm_sc->aac_max_fib_size;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
Remove the unused AAC_DRIVER_BUILD_DATE macro. Nuke whitespace at EOL while I'm here. Approved by: scottl (MAINTAINER)
2005-11-13 17:26:36 +00:00
/*
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
* These are duplicated in aac_start to cover the case where an
* intermediate stage may have destroyed them. They're left
Change spelling and eliminate a typo in comments to reduce diffs with Adaptec's vendor driver. I have some fixes to bring in and this makes ongoing review of the FreeBSD-Adaptec driver diffs easier.
2008-03-14 21:59:11 +00:00
* initialized here for debugging purposes only.
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
*/
Major bugfixes for large memory and fast systems. aac.c: Re-arrange the interrupt handler to optimize the common case of the adapter interrupting us because one or more commands are complete, and do a read across the pci bus to ensure that all posted status writes are flushed. This should close a race that could cause command completion interrupts to be lost. Follow the spec a bit closer when filling out command structures. Enable the Fast Response feature to eliminate the need for the card to DMA successfull command completions back into host memory. Tell the controller how much physical memory we have. Without this there was a chance that our DMA regions would collide with the memory window used by the cache on the controller. The result would be massive data corruption. This seemed to mainly affect systems with >2GB of memory. Fix a few whitespace problems. aac_debug.c: Add an extra diagnostic when printing out commands. aac_disk.c: Add extra sanity checks. aacreg.h: Prepare for making this 64-bit clean by reducing the use of enumeration types in structures. Many thanks to Justin Gibbs for helping track these down.
2003-01-11 01:59:21 +00:00
cm->cm_fib->Header.ReceiverFibAddress = (u_int32_t)cm->cm_fibphys;
cm->cm_fib->Header.SenderData = 0;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
aac_enqueue_free(cm);
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
Revert part of r173264. Both aac_ioctl_sendfib and aac_ioctl_send_raw_srb make use of the aac_ioctl_event callback, if aac_alloc_command fails. This can end up in an infinite loop in the while loop in aac_release_command. Further investigation into the issue mentioned by Scott Long [1] will be necessary. [1] http://lists.freebsd.org/pipermail/freebsd-current/2007-October/078740.html
2011-02-03 02:14:53 +00:00
if ((event = TAILQ_FIRST(&sc->aac_ev_cmfree)) != NULL) {
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
TAILQ_REMOVE(&sc->aac_ev_cmfree, event, ev_links);
event->ev_callback(sc, event, event->ev_arg);
}
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
Major bugfix and minor update. This should resolve the current issues with the driver locking up under load. - Restructure so that we use a static pool of commands/FIBs, rather than allocating them in clusters. The cluster allocation just made things more complicated, and allowed us to waste more memory in peak load situations. - Make queueing macros more like my other drivers. This adds queue stats for free. Add some debugging to take advantage of this. - Reimplement the periodic timeout scan. Kick the interrupt handler and the start routine every scan as well, just to be safe. Track busy commands properly. - Bring resource cleanup into line with resource allocation. We should now clean up correctly after a failed probe/unload/etc. - Try to start new commands when old ones are completed. We weren't doing this before, which could lead to deadlock when the controller was full. - Don't try to build a new command if we have found a deferred command. This could cause us to lose the deferred command. - Use diskerr() to report I/O errors. - Don't bail if the AdapterInfo structure is the wrong size. Some variation seems to be normal. We need to improve our handing of 2.x firmware sets. - Improve some comments in an attempt to try to make things clearer. - Restructure to avoid some warnings.
2000-12-27 13:14:56 +00:00
* Map helper for command/FIB allocation.
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
*/
static void
Major bugfix and minor update. This should resolve the current issues with the driver locking up under load. - Restructure so that we use a static pool of commands/FIBs, rather than allocating them in clusters. The cluster allocation just made things more complicated, and allowed us to waste more memory in peak load situations. - Make queueing macros more like my other drivers. This adds queue stats for free. Add some debugging to take advantage of this. - Reimplement the periodic timeout scan. Kick the interrupt handler and the start routine every scan as well, just to be safe. Track busy commands properly. - Bring resource cleanup into line with resource allocation. We should now clean up correctly after a failed probe/unload/etc. - Try to start new commands when old ones are completed. We weren't doing this before, which could lead to deadlock when the controller was full. - Don't try to build a new command if we have found a deferred command. This could cause us to lose the deferred command. - Use diskerr() to report I/O errors. - Don't bail if the AdapterInfo structure is the wrong size. Some variation seems to be normal. We need to improve our handing of 2.x firmware sets. - Improve some comments in an attempt to try to make things clearer. - Restructure to avoid some warnings.
2000-12-27 13:14:56 +00:00
aac_map_command_helper(void *arg, bus_dma_segment_t *segs, int nseg, int error)
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
{
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
uint64_t *fibphys;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
fibphys = (uint64_t *)arg;
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Implement a new dynamic command allocator. FIBs are allocated in 1 page blocks now, which should eliminate problems with the driver failing to attach due to insufficient contiguous RAM. Allow the FIB pool to grow from the default of 128 to the max of 512 as demand grows. Also pad the adapter init struct to work around the 2120/2200 DMA bug now that there is no longer a FIB slab.
2003-02-10 00:34:24 +00:00
*fibphys = segs[0].ds_addr;
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
Change spelling and eliminate a typo in comments to reduce diffs with Adaptec's vendor driver. I have some fixes to bring in and this makes ongoing review of the FreeBSD-Adaptec driver diffs easier.
2008-03-14 21:59:11 +00:00
* Allocate and initialize commands/FIBs for this adapter.
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
*/
Major bugfix and minor update. This should resolve the current issues with the driver locking up under load. - Restructure so that we use a static pool of commands/FIBs, rather than allocating them in clusters. The cluster allocation just made things more complicated, and allowed us to waste more memory in peak load situations. - Make queueing macros more like my other drivers. This adds queue stats for free. Add some debugging to take advantage of this. - Reimplement the periodic timeout scan. Kick the interrupt handler and the start routine every scan as well, just to be safe. Track busy commands properly. - Bring resource cleanup into line with resource allocation. We should now clean up correctly after a failed probe/unload/etc. - Try to start new commands when old ones are completed. We weren't doing this before, which could lead to deadlock when the controller was full. - Don't try to build a new command if we have found a deferred command. This could cause us to lose the deferred command. - Use diskerr() to report I/O errors. - Don't bail if the AdapterInfo structure is the wrong size. Some variation seems to be normal. We need to improve our handing of 2.x firmware sets. - Improve some comments in an attempt to try to make things clearer. - Restructure to avoid some warnings.
2000-12-27 13:14:56 +00:00
static int
aac_alloc_commands(struct aac_softc *sc)
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
{
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
struct aac_command *cm;
Implement a new dynamic command allocator. FIBs are allocated in 1 page blocks now, which should eliminate problems with the driver failing to attach due to insufficient contiguous RAM. Allow the FIB pool to grow from the default of 128 to the max of 512 as demand grows. Also pad the adapter init struct to work around the 2120/2200 DMA bug now that there is no longer a FIB slab.
2003-02-10 00:34:24 +00:00
struct aac_fibmap *fm;
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
uint64_t fibphys;
Implement a new dynamic command allocator. FIBs are allocated in 1 page blocks now, which should eliminate problems with the driver failing to attach due to insufficient contiguous RAM. Allow the FIB pool to grow from the default of 128 to the max of 512 as demand grows. Also pad the adapter init struct to work around the 2120/2200 DMA bug now that there is no longer a FIB slab.
2003-02-10 00:34:24 +00:00
int i, error;
Remove the unused AAC_DRIVER_BUILD_DATE macro. Nuke whitespace at EOL while I'm here. Approved by: scottl (MAINTAINER)
2005-11-13 17:26:36 +00:00
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
if (sc->total_fibs + sc->aac_max_fibs_alloc > sc->aac_max_fibs)
Implement a new dynamic command allocator. FIBs are allocated in 1 page blocks now, which should eliminate problems with the driver failing to attach due to insufficient contiguous RAM. Allow the FIB pool to grow from the default of 128 to the max of 512 as demand grows. Also pad the adapter init struct to work around the 2120/2200 DMA bug now that there is no longer a FIB slab.
2003-02-10 00:34:24 +00:00
return (ENOMEM);
o Move the cleanup of the fib maps into aac_free_commands() so as to retain symetry with aac_alloc_commans(). Since aac_alloc_commands() allocates fib maps and places them onto the fib lists, aac_free_commands() should reverse those operations. o Combine two ifs with the same body with an ||. o Switch from uintptr_t to uint32_t for fib map load operations. The target is a uint32_t so using this type for the map load call avoids an extra cast. uintptr_t should only be used when you need an "int sized the same as the machine's poitner size" which is not the case here. o Removed the commented out M_WAITOK flag in the allocation in aac_alloc_commands(). The kernel will only block in the allocator if it can grow the size of the kernel. This usually results in a page-out which could involve this aac device. Thus, sleeping here could deadlock the machine. Assuming this operation is occurring outside of attach time, we have enough fibs to operate anyway, so waiting for fibs to free up is okay if not optimal. o In aac_alloc_commands(), if we cannot dmamem_alloc additional fib space, free the fib map. o In aac_alloc_commands(), if we cannot create per-command dmamaps, don't lose track of the fib map that is mapping all of the commands that we have already released into the free pool. Instead, just cut out of the loop and modify aac_free_commands to not attempt to free maps that have not been allocated. o Don't use a magic number when pre-allocating fibs. o Use PAGE_SIZE to allocate in page sized chunks instead of an architecture specific constant. Submitted by: gibbs
2003-02-19 21:38:29 +00:00
fm = malloc(sizeof(struct aac_fibmap), M_AACBUF, M_NOWAIT|M_ZERO);
Begin support for 64-bit address support and workarounds for newer cards: - Add data structuress for doing 64-bit scatter/gather - Move busdma tag creations around so that only the parent is created in aac_pci.c. - Retrieve the capabilities word from the firmware before setting up command structures and tags. This allows the driver to decide whether to do 64-bit commands, and if work-arounds are needed for systems with >2GB of RAM. - Only enable the SCSI passthrough if it's enabled in the capabilities word in the firmware. This should fix problems with the 2120S and 2200S cards in systems with more than 2GB of RAM. Full 64-bit support is forthcoming. MFC-After: 1 week
2003-03-26 17:50:11 +00:00
if (fm == NULL)
return (ENOMEM);
Implement a new dynamic command allocator. FIBs are allocated in 1 page blocks now, which should eliminate problems with the driver failing to attach due to insufficient contiguous RAM. Allow the FIB pool to grow from the default of 128 to the max of 512 as demand grows. Also pad the adapter init struct to work around the 2120/2200 DMA bug now that there is no longer a FIB slab.
2003-02-10 00:34:24 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/* allocate the FIBs in DMAable memory and load them */
Implement a new dynamic command allocator. FIBs are allocated in 1 page blocks now, which should eliminate problems with the driver failing to attach due to insufficient contiguous RAM. Allow the FIB pool to grow from the default of 128 to the max of 512 as demand grows. Also pad the adapter init struct to work around the 2120/2200 DMA bug now that there is no longer a FIB slab.
2003-02-10 00:34:24 +00:00
if (bus_dmamem_alloc(sc->aac_fib_dmat, (void **)&fm->aac_fibs,
BUS_DMA_NOWAIT, &fm->aac_fibmap)) {
Various cleanups: - Move the command timeout check from a separate repeating timeout to the kthread since the kthread is already running periodically. - Move printing the hardware print buffer to the kthread. - Properly shut down the kernel thread on detach. - Detach the child array devices on detach. - Don't issue a controller halt command on detach. Doing so requires a PCI reset to wake the controller back up. The driver can now be unloaded as long as CAM support is not enabled.
2003-02-05 23:24:25 +00:00
device_printf(sc->aac_dev,
"Not enough contiguous memory available.\n");
o Move the cleanup of the fib maps into aac_free_commands() so as to retain symetry with aac_alloc_commans(). Since aac_alloc_commands() allocates fib maps and places them onto the fib lists, aac_free_commands() should reverse those operations. o Combine two ifs with the same body with an ||. o Switch from uintptr_t to uint32_t for fib map load operations. The target is a uint32_t so using this type for the map load call avoids an extra cast. uintptr_t should only be used when you need an "int sized the same as the machine's poitner size" which is not the case here. o Removed the commented out M_WAITOK flag in the allocation in aac_alloc_commands(). The kernel will only block in the allocator if it can grow the size of the kernel. This usually results in a page-out which could involve this aac device. Thus, sleeping here could deadlock the machine. Assuming this operation is occurring outside of attach time, we have enough fibs to operate anyway, so waiting for fibs to free up is okay if not optimal. o In aac_alloc_commands(), if we cannot dmamem_alloc additional fib space, free the fib map. o In aac_alloc_commands(), if we cannot create per-command dmamaps, don't lose track of the fib map that is mapping all of the commands that we have already released into the free pool. Instead, just cut out of the loop and modify aac_free_commands to not attempt to free maps that have not been allocated. o Don't use a magic number when pre-allocating fibs. o Use PAGE_SIZE to allocate in page sized chunks instead of an architecture specific constant. Submitted by: gibbs
2003-02-19 21:38:29 +00:00
free(fm, M_AACBUF);
The aac driver needs 256K of contiguous physical memory in order to attach. If it can't get that (highly likely if loaded as a module on a system that's been up for a while), give a more descriptive error message. Also clean up some nearby style nits. MFC after: 2 days
2002-08-30 05:02:07 +00:00
return (ENOMEM);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
}
Hack around a bug in the 2200 and 2120 controllers that connot DMA commands from below the first 8K of physical memory. A better fix is to modify the busdma api to allow either inclusion ranges or multiple exclusion ranges, but that debate is for another day. MFC After: 2 days
2003-01-23 01:01:44 +00:00
Handle the EINPROGRESS case of bus_dmamap_load() for data buffers.
2003-07-09 21:16:23 +00:00
/* Ignore errors since this doesn't bounce */
Remove the unused AAC_DRIVER_BUILD_DATE macro. Nuke whitespace at EOL while I'm here. Approved by: scottl (MAINTAINER)
2005-11-13 17:26:36 +00:00
(void)bus_dmamap_load(sc->aac_fib_dmat, fm->aac_fibmap, fm->aac_fibs,
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
sc->aac_max_fibs_alloc * sc->aac_max_fib_size,
Handle the EINPROGRESS case of bus_dmamap_load() for data buffers.
2003-07-09 21:16:23 +00:00
aac_map_command_helper, &fibphys, 0);
Hack around a bug in the 2200 and 2120 controllers that connot DMA commands from below the first 8K of physical memory. A better fix is to modify the busdma api to allow either inclusion ranges or multiple exclusion ranges, but that debate is for another day. MFC After: 2 days
2003-01-23 01:01:44 +00:00
Change spelling and eliminate a typo in comments to reduce diffs with Adaptec's vendor driver. I have some fixes to bring in and this makes ongoing review of the FreeBSD-Adaptec driver diffs easier.
2008-03-14 21:59:11 +00:00
/* initialize constant fields in the command structure */
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
bzero(fm->aac_fibs, sc->aac_max_fibs_alloc * sc->aac_max_fib_size);
for (i = 0; i < sc->aac_max_fibs_alloc; i++) {
o Move the cleanup of the fib maps into aac_free_commands() so as to retain symetry with aac_alloc_commans(). Since aac_alloc_commands() allocates fib maps and places them onto the fib lists, aac_free_commands() should reverse those operations. o Combine two ifs with the same body with an ||. o Switch from uintptr_t to uint32_t for fib map load operations. The target is a uint32_t so using this type for the map load call avoids an extra cast. uintptr_t should only be used when you need an "int sized the same as the machine's poitner size" which is not the case here. o Removed the commented out M_WAITOK flag in the allocation in aac_alloc_commands(). The kernel will only block in the allocator if it can grow the size of the kernel. This usually results in a page-out which could involve this aac device. Thus, sleeping here could deadlock the machine. Assuming this operation is occurring outside of attach time, we have enough fibs to operate anyway, so waiting for fibs to free up is okay if not optimal. o In aac_alloc_commands(), if we cannot dmamem_alloc additional fib space, free the fib map. o In aac_alloc_commands(), if we cannot create per-command dmamaps, don't lose track of the fib map that is mapping all of the commands that we have already released into the free pool. Instead, just cut out of the loop and modify aac_free_commands to not attempt to free maps that have not been allocated. o Don't use a magic number when pre-allocating fibs. o Use PAGE_SIZE to allocate in page sized chunks instead of an architecture specific constant. Submitted by: gibbs
2003-02-19 21:38:29 +00:00
cm = sc->aac_commands + sc->total_fibs;
Implement a new dynamic command allocator. FIBs are allocated in 1 page blocks now, which should eliminate problems with the driver failing to attach due to insufficient contiguous RAM. Allow the FIB pool to grow from the default of 128 to the max of 512 as demand grows. Also pad the adapter init struct to work around the 2120/2200 DMA bug now that there is no longer a FIB slab.
2003-02-10 00:34:24 +00:00
fm->aac_commands = cm;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
cm->cm_sc = sc;
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
cm->cm_fib = (struct aac_fib *)
((u_int8_t *)fm->aac_fibs + i*sc->aac_max_fib_size);
cm->cm_fibphys = fibphys + i*sc->aac_max_fib_size;
Fix a 64-bit bogon. The hardware command structure only has one 32 bit field for holding driver-dependant data. Instead of putting the pointer to the driver command struct in there, take advantage of these structs being a (virtually) contiguous array and just put the array index in the field.
2003-02-19 23:33:56 +00:00
cm->cm_index = sc->total_fibs;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
Implement a new dynamic command allocator. FIBs are allocated in 1 page blocks now, which should eliminate problems with the driver failing to attach due to insufficient contiguous RAM. Allow the FIB pool to grow from the default of 128 to the max of 512 as demand grows. Also pad the adapter init struct to work around the 2120/2200 DMA bug now that there is no longer a FIB slab.
2003-02-10 00:34:24 +00:00
if ((error = bus_dmamap_create(sc->aac_buffer_dmat, 0,
Rearrange locking in the alloc_commands and ioctl paths to avoid problems.
2006-04-08 06:05:29 +00:00
&cm->cm_datamap)) != 0)
o Move the cleanup of the fib maps into aac_free_commands() so as to retain symetry with aac_alloc_commans(). Since aac_alloc_commands() allocates fib maps and places them onto the fib lists, aac_free_commands() should reverse those operations. o Combine two ifs with the same body with an ||. o Switch from uintptr_t to uint32_t for fib map load operations. The target is a uint32_t so using this type for the map load call avoids an extra cast. uintptr_t should only be used when you need an "int sized the same as the machine's poitner size" which is not the case here. o Removed the commented out M_WAITOK flag in the allocation in aac_alloc_commands(). The kernel will only block in the allocator if it can grow the size of the kernel. This usually results in a page-out which could involve this aac device. Thus, sleeping here could deadlock the machine. Assuming this operation is occurring outside of attach time, we have enough fibs to operate anyway, so waiting for fibs to free up is okay if not optimal. o In aac_alloc_commands(), if we cannot dmamem_alloc additional fib space, free the fib map. o In aac_alloc_commands(), if we cannot create per-command dmamaps, don't lose track of the fib map that is mapping all of the commands that we have already released into the free pool. Instead, just cut out of the loop and modify aac_free_commands to not attempt to free maps that have not been allocated. o Don't use a magic number when pre-allocating fibs. o Use PAGE_SIZE to allocate in page sized chunks instead of an architecture specific constant. Submitted by: gibbs
2003-02-19 21:38:29 +00:00
break;
Rearrange locking in the alloc_commands and ioctl paths to avoid problems.
2006-04-08 06:05:29 +00:00
mtx_lock(&sc->aac_io_lock);
aac_release_command(cm);
o Move the cleanup of the fib maps into aac_free_commands() so as to retain symetry with aac_alloc_commans(). Since aac_alloc_commands() allocates fib maps and places them onto the fib lists, aac_free_commands() should reverse those operations. o Combine two ifs with the same body with an ||. o Switch from uintptr_t to uint32_t for fib map load operations. The target is a uint32_t so using this type for the map load call avoids an extra cast. uintptr_t should only be used when you need an "int sized the same as the machine's poitner size" which is not the case here. o Removed the commented out M_WAITOK flag in the allocation in aac_alloc_commands(). The kernel will only block in the allocator if it can grow the size of the kernel. This usually results in a page-out which could involve this aac device. Thus, sleeping here could deadlock the machine. Assuming this operation is occurring outside of attach time, we have enough fibs to operate anyway, so waiting for fibs to free up is okay if not optimal. o In aac_alloc_commands(), if we cannot dmamem_alloc additional fib space, free the fib map. o In aac_alloc_commands(), if we cannot create per-command dmamaps, don't lose track of the fib map that is mapping all of the commands that we have already released into the free pool. Instead, just cut out of the loop and modify aac_free_commands to not attempt to free maps that have not been allocated. o Don't use a magic number when pre-allocating fibs. o Use PAGE_SIZE to allocate in page sized chunks instead of an architecture specific constant. Submitted by: gibbs
2003-02-19 21:38:29 +00:00
sc->total_fibs++;
Rearrange locking in the alloc_commands and ioctl paths to avoid problems.
2006-04-08 06:05:29 +00:00
mtx_unlock(&sc->aac_io_lock);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
}
Implement a new dynamic command allocator. FIBs are allocated in 1 page blocks now, which should eliminate problems with the driver failing to attach due to insufficient contiguous RAM. Allow the FIB pool to grow from the default of 128 to the max of 512 as demand grows. Also pad the adapter init struct to work around the 2120/2200 DMA bug now that there is no longer a FIB slab.
2003-02-10 00:34:24 +00:00
o Move the cleanup of the fib maps into aac_free_commands() so as to retain symetry with aac_alloc_commans(). Since aac_alloc_commands() allocates fib maps and places them onto the fib lists, aac_free_commands() should reverse those operations. o Combine two ifs with the same body with an ||. o Switch from uintptr_t to uint32_t for fib map load operations. The target is a uint32_t so using this type for the map load call avoids an extra cast. uintptr_t should only be used when you need an "int sized the same as the machine's poitner size" which is not the case here. o Removed the commented out M_WAITOK flag in the allocation in aac_alloc_commands(). The kernel will only block in the allocator if it can grow the size of the kernel. This usually results in a page-out which could involve this aac device. Thus, sleeping here could deadlock the machine. Assuming this operation is occurring outside of attach time, we have enough fibs to operate anyway, so waiting for fibs to free up is okay if not optimal. o In aac_alloc_commands(), if we cannot dmamem_alloc additional fib space, free the fib map. o In aac_alloc_commands(), if we cannot create per-command dmamaps, don't lose track of the fib map that is mapping all of the commands that we have already released into the free pool. Instead, just cut out of the loop and modify aac_free_commands to not attempt to free maps that have not been allocated. o Don't use a magic number when pre-allocating fibs. o Use PAGE_SIZE to allocate in page sized chunks instead of an architecture specific constant. Submitted by: gibbs
2003-02-19 21:38:29 +00:00
if (i > 0) {
Rearrange locking in the alloc_commands and ioctl paths to avoid problems.
2006-04-08 06:05:29 +00:00
mtx_lock(&sc->aac_io_lock);
o Move the cleanup of the fib maps into aac_free_commands() so as to retain symetry with aac_alloc_commans(). Since aac_alloc_commands() allocates fib maps and places them onto the fib lists, aac_free_commands() should reverse those operations. o Combine two ifs with the same body with an ||. o Switch from uintptr_t to uint32_t for fib map load operations. The target is a uint32_t so using this type for the map load call avoids an extra cast. uintptr_t should only be used when you need an "int sized the same as the machine's poitner size" which is not the case here. o Removed the commented out M_WAITOK flag in the allocation in aac_alloc_commands(). The kernel will only block in the allocator if it can grow the size of the kernel. This usually results in a page-out which could involve this aac device. Thus, sleeping here could deadlock the machine. Assuming this operation is occurring outside of attach time, we have enough fibs to operate anyway, so waiting for fibs to free up is okay if not optimal. o In aac_alloc_commands(), if we cannot dmamem_alloc additional fib space, free the fib map. o In aac_alloc_commands(), if we cannot create per-command dmamaps, don't lose track of the fib map that is mapping all of the commands that we have already released into the free pool. Instead, just cut out of the loop and modify aac_free_commands to not attempt to free maps that have not been allocated. o Don't use a magic number when pre-allocating fibs. o Use PAGE_SIZE to allocate in page sized chunks instead of an architecture specific constant. Submitted by: gibbs
2003-02-19 21:38:29 +00:00
TAILQ_INSERT_TAIL(&sc->aac_fibmap_tqh, fm, fm_link);
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_COMM_B, "total_fibs= %d\n", sc->total_fibs);
Remove the AAC_LOCK macros. They no longer abstract anything and only obfuscate the code. No functional differences.
2004-08-12 05:05:06 +00:00
mtx_unlock(&sc->aac_io_lock);
o Move the cleanup of the fib maps into aac_free_commands() so as to retain symetry with aac_alloc_commans(). Since aac_alloc_commands() allocates fib maps and places them onto the fib lists, aac_free_commands() should reverse those operations. o Combine two ifs with the same body with an ||. o Switch from uintptr_t to uint32_t for fib map load operations. The target is a uint32_t so using this type for the map load call avoids an extra cast. uintptr_t should only be used when you need an "int sized the same as the machine's poitner size" which is not the case here. o Removed the commented out M_WAITOK flag in the allocation in aac_alloc_commands(). The kernel will only block in the allocator if it can grow the size of the kernel. This usually results in a page-out which could involve this aac device. Thus, sleeping here could deadlock the machine. Assuming this operation is occurring outside of attach time, we have enough fibs to operate anyway, so waiting for fibs to free up is okay if not optimal. o In aac_alloc_commands(), if we cannot dmamem_alloc additional fib space, free the fib map. o In aac_alloc_commands(), if we cannot create per-command dmamaps, don't lose track of the fib map that is mapping all of the commands that we have already released into the free pool. Instead, just cut out of the loop and modify aac_free_commands to not attempt to free maps that have not been allocated. o Don't use a magic number when pre-allocating fibs. o Use PAGE_SIZE to allocate in page sized chunks instead of an architecture specific constant. Submitted by: gibbs
2003-02-19 21:38:29 +00:00
return (0);
Remove the unused AAC_DRIVER_BUILD_DATE macro. Nuke whitespace at EOL while I'm here. Approved by: scottl (MAINTAINER)
2005-11-13 17:26:36 +00:00
}
Implement a new dynamic command allocator. FIBs are allocated in 1 page blocks now, which should eliminate problems with the driver failing to attach due to insufficient contiguous RAM. Allow the FIB pool to grow from the default of 128 to the max of 512 as demand grows. Also pad the adapter init struct to work around the 2120/2200 DMA bug now that there is no longer a FIB slab.
2003-02-10 00:34:24 +00:00
o Move the cleanup of the fib maps into aac_free_commands() so as to retain symetry with aac_alloc_commans(). Since aac_alloc_commands() allocates fib maps and places them onto the fib lists, aac_free_commands() should reverse those operations. o Combine two ifs with the same body with an ||. o Switch from uintptr_t to uint32_t for fib map load operations. The target is a uint32_t so using this type for the map load call avoids an extra cast. uintptr_t should only be used when you need an "int sized the same as the machine's poitner size" which is not the case here. o Removed the commented out M_WAITOK flag in the allocation in aac_alloc_commands(). The kernel will only block in the allocator if it can grow the size of the kernel. This usually results in a page-out which could involve this aac device. Thus, sleeping here could deadlock the machine. Assuming this operation is occurring outside of attach time, we have enough fibs to operate anyway, so waiting for fibs to free up is okay if not optimal. o In aac_alloc_commands(), if we cannot dmamem_alloc additional fib space, free the fib map. o In aac_alloc_commands(), if we cannot create per-command dmamaps, don't lose track of the fib map that is mapping all of the commands that we have already released into the free pool. Instead, just cut out of the loop and modify aac_free_commands to not attempt to free maps that have not been allocated. o Don't use a magic number when pre-allocating fibs. o Use PAGE_SIZE to allocate in page sized chunks instead of an architecture specific constant. Submitted by: gibbs
2003-02-19 21:38:29 +00:00
bus_dmamap_unload(sc->aac_fib_dmat, fm->aac_fibmap);
bus_dmamem_free(sc->aac_fib_dmat, fm->aac_fibs, fm->aac_fibmap);
free(fm, M_AACBUF);
return (ENOMEM);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
Major bugfix and minor update. This should resolve the current issues with the driver locking up under load. - Restructure so that we use a static pool of commands/FIBs, rather than allocating them in clusters. The cluster allocation just made things more complicated, and allowed us to waste more memory in peak load situations. - Make queueing macros more like my other drivers. This adds queue stats for free. Add some debugging to take advantage of this. - Reimplement the periodic timeout scan. Kick the interrupt handler and the start routine every scan as well, just to be safe. Track busy commands properly. - Bring resource cleanup into line with resource allocation. We should now clean up correctly after a failed probe/unload/etc. - Try to start new commands when old ones are completed. We weren't doing this before, which could lead to deadlock when the controller was full. - Don't try to build a new command if we have found a deferred command. This could cause us to lose the deferred command. - Use diskerr() to report I/O errors. - Don't bail if the AdapterInfo structure is the wrong size. Some variation seems to be normal. We need to improve our handing of 2.x firmware sets. - Improve some comments in an attempt to try to make things clearer. - Restructure to avoid some warnings.
2000-12-27 13:14:56 +00:00
* Free FIBs owned by this adapter.
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
*/
static void
o Move the cleanup of the fib maps into aac_free_commands() so as to retain symetry with aac_alloc_commans(). Since aac_alloc_commands() allocates fib maps and places them onto the fib lists, aac_free_commands() should reverse those operations. o Combine two ifs with the same body with an ||. o Switch from uintptr_t to uint32_t for fib map load operations. The target is a uint32_t so using this type for the map load call avoids an extra cast. uintptr_t should only be used when you need an "int sized the same as the machine's poitner size" which is not the case here. o Removed the commented out M_WAITOK flag in the allocation in aac_alloc_commands(). The kernel will only block in the allocator if it can grow the size of the kernel. This usually results in a page-out which could involve this aac device. Thus, sleeping here could deadlock the machine. Assuming this operation is occurring outside of attach time, we have enough fibs to operate anyway, so waiting for fibs to free up is okay if not optimal. o In aac_alloc_commands(), if we cannot dmamem_alloc additional fib space, free the fib map. o In aac_alloc_commands(), if we cannot create per-command dmamaps, don't lose track of the fib map that is mapping all of the commands that we have already released into the free pool. Instead, just cut out of the loop and modify aac_free_commands to not attempt to free maps that have not been allocated. o Don't use a magic number when pre-allocating fibs. o Use PAGE_SIZE to allocate in page sized chunks instead of an architecture specific constant. Submitted by: gibbs
2003-02-19 21:38:29 +00:00
aac_free_commands(struct aac_softc *sc)
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
{
o Move the cleanup of the fib maps into aac_free_commands() so as to retain symetry with aac_alloc_commans(). Since aac_alloc_commands() allocates fib maps and places them onto the fib lists, aac_free_commands() should reverse those operations. o Combine two ifs with the same body with an ||. o Switch from uintptr_t to uint32_t for fib map load operations. The target is a uint32_t so using this type for the map load call avoids an extra cast. uintptr_t should only be used when you need an "int sized the same as the machine's poitner size" which is not the case here. o Removed the commented out M_WAITOK flag in the allocation in aac_alloc_commands(). The kernel will only block in the allocator if it can grow the size of the kernel. This usually results in a page-out which could involve this aac device. Thus, sleeping here could deadlock the machine. Assuming this operation is occurring outside of attach time, we have enough fibs to operate anyway, so waiting for fibs to free up is okay if not optimal. o In aac_alloc_commands(), if we cannot dmamem_alloc additional fib space, free the fib map. o In aac_alloc_commands(), if we cannot create per-command dmamaps, don't lose track of the fib map that is mapping all of the commands that we have already released into the free pool. Instead, just cut out of the loop and modify aac_free_commands to not attempt to free maps that have not been allocated. o Don't use a magic number when pre-allocating fibs. o Use PAGE_SIZE to allocate in page sized chunks instead of an architecture specific constant. Submitted by: gibbs
2003-02-19 21:38:29 +00:00
struct aac_fibmap *fm;
Implement a new dynamic command allocator. FIBs are allocated in 1 page blocks now, which should eliminate problems with the driver failing to attach due to insufficient contiguous RAM. Allow the FIB pool to grow from the default of 128 to the max of 512 as demand grows. Also pad the adapter init struct to work around the 2120/2200 DMA bug now that there is no longer a FIB slab.
2003-02-10 00:34:24 +00:00
struct aac_command *cm;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
int i;
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
o Move the cleanup of the fib maps into aac_free_commands() so as to retain symetry with aac_alloc_commans(). Since aac_alloc_commands() allocates fib maps and places them onto the fib lists, aac_free_commands() should reverse those operations. o Combine two ifs with the same body with an ||. o Switch from uintptr_t to uint32_t for fib map load operations. The target is a uint32_t so using this type for the map load call avoids an extra cast. uintptr_t should only be used when you need an "int sized the same as the machine's poitner size" which is not the case here. o Removed the commented out M_WAITOK flag in the allocation in aac_alloc_commands(). The kernel will only block in the allocator if it can grow the size of the kernel. This usually results in a page-out which could involve this aac device. Thus, sleeping here could deadlock the machine. Assuming this operation is occurring outside of attach time, we have enough fibs to operate anyway, so waiting for fibs to free up is okay if not optimal. o In aac_alloc_commands(), if we cannot dmamem_alloc additional fib space, free the fib map. o In aac_alloc_commands(), if we cannot create per-command dmamaps, don't lose track of the fib map that is mapping all of the commands that we have already released into the free pool. Instead, just cut out of the loop and modify aac_free_commands to not attempt to free maps that have not been allocated. o Don't use a magic number when pre-allocating fibs. o Use PAGE_SIZE to allocate in page sized chunks instead of an architecture specific constant. Submitted by: gibbs
2003-02-19 21:38:29 +00:00
while ((fm = TAILQ_FIRST(&sc->aac_fibmap_tqh)) != NULL) {
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
o Move the cleanup of the fib maps into aac_free_commands() so as to retain symetry with aac_alloc_commans(). Since aac_alloc_commands() allocates fib maps and places them onto the fib lists, aac_free_commands() should reverse those operations. o Combine two ifs with the same body with an ||. o Switch from uintptr_t to uint32_t for fib map load operations. The target is a uint32_t so using this type for the map load call avoids an extra cast. uintptr_t should only be used when you need an "int sized the same as the machine's poitner size" which is not the case here. o Removed the commented out M_WAITOK flag in the allocation in aac_alloc_commands(). The kernel will only block in the allocator if it can grow the size of the kernel. This usually results in a page-out which could involve this aac device. Thus, sleeping here could deadlock the machine. Assuming this operation is occurring outside of attach time, we have enough fibs to operate anyway, so waiting for fibs to free up is okay if not optimal. o In aac_alloc_commands(), if we cannot dmamem_alloc additional fib space, free the fib map. o In aac_alloc_commands(), if we cannot create per-command dmamaps, don't lose track of the fib map that is mapping all of the commands that we have already released into the free pool. Instead, just cut out of the loop and modify aac_free_commands to not attempt to free maps that have not been allocated. o Don't use a magic number when pre-allocating fibs. o Use PAGE_SIZE to allocate in page sized chunks instead of an architecture specific constant. Submitted by: gibbs
2003-02-19 21:38:29 +00:00
TAILQ_REMOVE(&sc->aac_fibmap_tqh, fm, fm_link);
/*
* We check against total_fibs to handle partially
* allocated blocks.
*/
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
for (i = 0; i < sc->aac_max_fibs_alloc && sc->total_fibs--; i++) {
o Move the cleanup of the fib maps into aac_free_commands() so as to retain symetry with aac_alloc_commans(). Since aac_alloc_commands() allocates fib maps and places them onto the fib lists, aac_free_commands() should reverse those operations. o Combine two ifs with the same body with an ||. o Switch from uintptr_t to uint32_t for fib map load operations. The target is a uint32_t so using this type for the map load call avoids an extra cast. uintptr_t should only be used when you need an "int sized the same as the machine's poitner size" which is not the case here. o Removed the commented out M_WAITOK flag in the allocation in aac_alloc_commands(). The kernel will only block in the allocator if it can grow the size of the kernel. This usually results in a page-out which could involve this aac device. Thus, sleeping here could deadlock the machine. Assuming this operation is occurring outside of attach time, we have enough fibs to operate anyway, so waiting for fibs to free up is okay if not optimal. o In aac_alloc_commands(), if we cannot dmamem_alloc additional fib space, free the fib map. o In aac_alloc_commands(), if we cannot create per-command dmamaps, don't lose track of the fib map that is mapping all of the commands that we have already released into the free pool. Instead, just cut out of the loop and modify aac_free_commands to not attempt to free maps that have not been allocated. o Don't use a magic number when pre-allocating fibs. o Use PAGE_SIZE to allocate in page sized chunks instead of an architecture specific constant. Submitted by: gibbs
2003-02-19 21:38:29 +00:00
cm = fm->aac_commands + i;
bus_dmamap_destroy(sc->aac_buffer_dmat, cm->cm_datamap);
}
bus_dmamap_unload(sc->aac_fib_dmat, fm->aac_fibmap);
bus_dmamem_free(sc->aac_fib_dmat, fm->aac_fibs, fm->aac_fibmap);
free(fm, M_AACBUF);
}
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
* Command-mapping helper function - populate this command's s/g table.
*/
static void
aac_map_command_sg(void *arg, bus_dma_segment_t *segs, int nseg, int error)
{
Handle the EINPROGRESS case of bus_dmamap_load() for data buffers.
2003-07-09 21:16:23 +00:00
struct aac_softc *sc;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
struct aac_command *cm;
struct aac_fib *fib;
int i;
cm = (struct aac_command *)arg;
Handle the EINPROGRESS case of bus_dmamap_load() for data buffers.
2003-07-09 21:16:23 +00:00
sc = cm->cm_sc;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
fib = cm->cm_fib;
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/* copy into the FIB */
Add the ability to send 64-bit scatter/gather elements to aac cards. This is enabled when both the size of bus_addr_t > 4 and the card claims support. Don't wake up the kthread to allocate more commands if we know that we've already allocated the max number of commands.
2003-03-30 21:47:16 +00:00
if (cm->cm_sgtable != NULL) {
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
if (fib->Header.Command == RawIo) {
struct aac_sg_tableraw *sg;
sg = (struct aac_sg_tableraw *)cm->cm_sgtable;
sg->SgCount = nseg;
for (i = 0; i < nseg; i++) {
sg->SgEntryRaw[i].SgAddress = segs[i].ds_addr;
sg->SgEntryRaw[i].SgByteCount = segs[i].ds_len;
sg->SgEntryRaw[i].Next = 0;
sg->SgEntryRaw[i].Prev = 0;
sg->SgEntryRaw[i].Flags = 0;
}
/* update the FIB size for the s/g count */
fib->Header.Size += nseg*sizeof(struct aac_sg_entryraw);
} else if ((cm->cm_sc->flags & AAC_FLAGS_SG_64BIT) == 0) {
Add the ability to send 64-bit scatter/gather elements to aac cards. This is enabled when both the size of bus_addr_t > 4 and the card claims support. Don't wake up the kthread to allocate more commands if we know that we've already allocated the max number of commands.
2003-03-30 21:47:16 +00:00
struct aac_sg_table *sg;
sg = cm->cm_sgtable;
sg->SgCount = nseg;
for (i = 0; i < nseg; i++) {
sg->SgEntry[i].SgAddress = segs[i].ds_addr;
sg->SgEntry[i].SgByteCount = segs[i].ds_len;
}
/* update the FIB size for the s/g count */
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
fib->Header.Size += nseg*sizeof(struct aac_sg_entry);
Add the ability to send 64-bit scatter/gather elements to aac cards. This is enabled when both the size of bus_addr_t > 4 and the card claims support. Don't wake up the kthread to allocate more commands if we know that we've already allocated the max number of commands.
2003-03-30 21:47:16 +00:00
} else {
struct aac_sg_table64 *sg;
sg = (struct aac_sg_table64 *)cm->cm_sgtable;
sg->SgCount = nseg;
for (i = 0; i < nseg; i++) {
sg->SgEntry64[i].SgAddress = segs[i].ds_addr;
sg->SgEntry64[i].SgByteCount = segs[i].ds_len;
}
/* update the FIB size for the s/g count */
fib->Header.Size += nseg*sizeof(struct aac_sg_entry64);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
}
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
Handle the EINPROGRESS case of bus_dmamap_load() for data buffers.
2003-07-09 21:16:23 +00:00
/* Fix up the address values in the FIB. Use the command array index
* instead of a pointer since these fields are only 32 bits. Shift
Remove the unused AAC_DRIVER_BUILD_DATE macro. Nuke whitespace at EOL while I'm here. Approved by: scottl (MAINTAINER)
2005-11-13 17:26:36 +00:00
* the SenderFibAddress over to make room for the fast response bit
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
* and for the AIF bit
Handle the EINPROGRESS case of bus_dmamap_load() for data buffers.
2003-07-09 21:16:23 +00:00
*/
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
cm->cm_fib->Header.SenderFibAddress = (cm->cm_index << 2);
cm->cm_fib->Header.ReceiverFibAddress = (u_int32_t)cm->cm_fibphys;
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Handle the EINPROGRESS case of bus_dmamap_load() for data buffers.
2003-07-09 21:16:23 +00:00
/* save a pointer to the command for speedy reverse-lookup */
cm->cm_fib->Header.SenderData = cm->cm_index;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
Handle the EINPROGRESS case of bus_dmamap_load() for data buffers.
2003-07-09 21:16:23 +00:00
if (cm->cm_flags & AAC_CMD_DATAIN)
bus_dmamap_sync(sc->aac_buffer_dmat, cm->cm_datamap,
BUS_DMASYNC_PREREAD);
if (cm->cm_flags & AAC_CMD_DATAOUT)
bus_dmamap_sync(sc->aac_buffer_dmat, cm->cm_datamap,
BUS_DMASYNC_PREWRITE);
cm->cm_flags |= AAC_CMD_MAPPED;
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
if (sc->flags & AAC_FLAGS_NEW_COMM) {
int count = 10000000L;
while (AAC_SEND_COMMAND(sc, cm) != 0) {
if (--count == 0) {
aac_unmap_command(cm);
sc->flags |= AAC_QUEUE_FRZN;
aac_requeue_ready(cm);
}
DELAY(5); /* wait 5 usec. */
}
} else {
/* Put the FIB on the outbound queue */
if (aac_enqueue_fib(sc, cm->cm_queue, cm) == EBUSY) {
aac_unmap_command(cm);
sc->flags |= AAC_QUEUE_FRZN;
aac_requeue_ready(cm);
}
If a command has to be deferred because there are no more resources for it on the card, unmap it first. This allows it to be picked up properly when the queue gets kicked again. This was the root problem for the lost command (i.e. stuck in getblk/vinvalb) problem. While here, panic if commands don't map correctly instead of just silently ignoring the problem and dropping command. Also slow down the dynamic allocation of new commands. It should be safe to go back into the aac waters. Thanks to everyone who suffered through this and provided good feedback.
2004-02-07 10:30:22 +00:00
}
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
* Unmap a command from controller-visible space.
*/
static void
aac_unmap_command(struct aac_command *cm)
{
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
struct aac_softc *sc;
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
sc = cm->cm_sc;
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
if (!(cm->cm_flags & AAC_CMD_MAPPED))
return;
if (cm->cm_datalen != 0) {
if (cm->cm_flags & AAC_CMD_DATAIN)
bus_dmamap_sync(sc->aac_buffer_dmat, cm->cm_datamap,
BUS_DMASYNC_POSTREAD);
if (cm->cm_flags & AAC_CMD_DATAOUT)
bus_dmamap_sync(sc->aac_buffer_dmat, cm->cm_datamap,
BUS_DMASYNC_POSTWRITE);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
bus_dmamap_unload(sc->aac_buffer_dmat, cm->cm_datamap);
}
cm->cm_flags &= ~AAC_CMD_MAPPED;
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
* Hardware Interface
*/
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
Change spelling and eliminate a typo in comments to reduce diffs with Adaptec's vendor driver. I have some fixes to bring in and this makes ongoing review of the FreeBSD-Adaptec driver diffs easier.
2008-03-14 21:59:11 +00:00
* Initialize the adapter.
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
*/
static void
aac_common_map(void *arg, bus_dma_segment_t *segs, int nseg, int error)
{
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
struct aac_softc *sc;
sc = (struct aac_softc *)arg;
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
sc->aac_common_busaddr = segs[0].ds_addr;
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
Don't attach to Dell PERC2/QC cards that have a firmware rev of 1.x. This check is complicated by the fact that the Adaptec 5400S cards claim to use 1.x firmware also. PERC2/QC 1.x firmware is not compatible with this driver and will cause a system hang. MFC after: 3 days
2002-02-06 01:34:09 +00:00
static int
aac_check_firmware(struct aac_softc *sc)
{
Diff reduction to Adaptec driver build 15317 (refactoring and code shuffling): - Resource allocation in aac_alloc (moved from from aac_init) - Interrupt setup in aac_setup_intr (from aac_attach) - Container probing in aac_get_container_info (from aac_startup and aac_handle_aif) - Firmware status check moved to aac_check_firmware from aac_init
2008-03-24 16:38:47 +00:00
u_int32_t code, major, minor, options = 0, atu_size = 0;
- Make tables, device ID strings etc const. This includes #ifdef'ing 0 aac_command_status_table, which is actually unused since r111532. While at it, make aac_if a pointer to the now const interface tables instead of copying them over to the softc (this alone already reduces the size of aac.ko on amd64 by ~1 KiB). - Remove redundant softc members. - Use DEVMETHOD_END. - Use NULL instead of 0 for pointers. - Remove redundant bzero(9)'ing of the softc. - Use pci_enable_busmaster(9) instead of duplicating it. - Remove redundant checking for PCIM_CMD_MEMEN (resource allocation will just fail). - Canonicalize the error messages in case of resource allocation failures. - Add support for using MSI instead of INTx, controllable via the tunable hw.aac.enable_msi (defaulting to on). MFC after: 1 month
2013-03-01 19:55:10 +00:00
int rid, status;
Diff reduction to Adaptec driver build 15317 (refactoring and code shuffling): - Resource allocation in aac_alloc (moved from from aac_init) - Interrupt setup in aac_setup_intr (from aac_attach) - Container probing in aac_get_container_info (from aac_startup and aac_handle_aif) - Firmware status check moved to aac_check_firmware from aac_init
2008-03-24 16:38:47 +00:00
time_t then;
Don't attach to Dell PERC2/QC cards that have a firmware rev of 1.x. This check is complicated by the fact that the Adaptec 5400S cards claim to use 1.x firmware also. PERC2/QC 1.x firmware is not compatible with this driver and will cause a system hang. MFC after: 3 days
2002-02-06 01:34:09 +00:00
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
Diff reduction to Adaptec driver build 15317 (refactoring and code shuffling): - Resource allocation in aac_alloc (moved from from aac_init) - Interrupt setup in aac_setup_intr (from aac_attach) - Container probing in aac_get_container_info (from aac_startup and aac_handle_aif) - Firmware status check moved to aac_check_firmware from aac_init
2008-03-24 16:38:47 +00:00
/*
* Wait for the adapter to come ready.
*/
then = time_uptime;
do {
code = AAC_GET_FWSTATUS(sc);
if (code & AAC_SELF_TEST_FAILED) {
device_printf(sc->aac_dev, "FATAL: selftest failed\n");
return(ENXIO);
}
if (code & AAC_KERNEL_PANIC) {
device_printf(sc->aac_dev,
Add \n to the end of a printf string and remove it from panic strings.
2008-03-26 15:28:56 +00:00
"FATAL: controller kernel panic");
Diff reduction to Adaptec driver build 15317 (refactoring and code shuffling): - Resource allocation in aac_alloc (moved from from aac_init) - Interrupt setup in aac_setup_intr (from aac_attach) - Container probing in aac_get_container_info (from aac_startup and aac_handle_aif) - Firmware status check moved to aac_check_firmware from aac_init
2008-03-24 16:38:47 +00:00
return(ENXIO);
}
if (time_uptime > (then + AAC_BOOT_TIMEOUT)) {
device_printf(sc->aac_dev,
"FATAL: controller not coming ready, "
"status %x\n", code);
return(ENXIO);
}
} while (!(code & AAC_UP_AND_RUNNING));
Don't attach to Dell PERC2/QC cards that have a firmware rev of 1.x. This check is complicated by the fact that the Adaptec 5400S cards claim to use 1.x firmware also. PERC2/QC 1.x firmware is not compatible with this driver and will cause a system hang. MFC after: 3 days
2002-02-06 01:34:09 +00:00
Begin support for 64-bit address support and workarounds for newer cards: - Add data structuress for doing 64-bit scatter/gather - Move busdma tag creations around so that only the parent is created in aac_pci.c. - Retrieve the capabilities word from the firmware before setting up command structures and tags. This allows the driver to decide whether to do 64-bit commands, and if work-arounds are needed for systems with >2GB of RAM. - Only enable the SCSI passthrough if it's enabled in the capabilities word in the firmware. This should fix problems with the 2120S and 2200S cards in systems with more than 2GB of RAM. Full 64-bit support is forthcoming. MFC-After: 1 week
2003-03-26 17:50:11 +00:00
/*
* Retrieve the firmware version numbers. Dell PERC2/QC cards with
* firmware version 1.x are not compatible with this driver.
*/
if (sc->flags & AAC_FLAGS_PERC2QC) {
Don't attach to Dell PERC2/QC cards that have a firmware rev of 1.x. This check is complicated by the fact that the Adaptec 5400S cards claim to use 1.x firmware also. PERC2/QC 1.x firmware is not compatible with this driver and will cause a system hang. MFC after: 3 days
2002-02-06 01:34:09 +00:00
if (aac_sync_command(sc, AAC_MONKER_GETKERNVER, 0, 0, 0, 0,
NULL)) {
device_printf(sc->aac_dev,
"Error reading firmware version\n");
return (EIO);
}
/* These numbers are stored as ASCII! */
Begin support for 64-bit address support and workarounds for newer cards: - Add data structuress for doing 64-bit scatter/gather - Move busdma tag creations around so that only the parent is created in aac_pci.c. - Retrieve the capabilities word from the firmware before setting up command structures and tags. This allows the driver to decide whether to do 64-bit commands, and if work-arounds are needed for systems with >2GB of RAM. - Only enable the SCSI passthrough if it's enabled in the capabilities word in the firmware. This should fix problems with the 2120S and 2200S cards in systems with more than 2GB of RAM. Full 64-bit support is forthcoming. MFC-After: 1 week
2003-03-26 17:50:11 +00:00
major = (AAC_GET_MAILBOX(sc, 1) & 0xff) - 0x30;
minor = (AAC_GET_MAILBOX(sc, 2) & 0xff) - 0x30;
Don't attach to Dell PERC2/QC cards that have a firmware rev of 1.x. This check is complicated by the fact that the Adaptec 5400S cards claim to use 1.x firmware also. PERC2/QC 1.x firmware is not compatible with this driver and will cause a system hang. MFC after: 3 days
2002-02-06 01:34:09 +00:00
if (major == 1) {
device_printf(sc->aac_dev,
"Firmware version %d.%d is not supported.\n",
major, minor);
return (EINVAL);
}
}
Begin support for 64-bit address support and workarounds for newer cards: - Add data structuress for doing 64-bit scatter/gather - Move busdma tag creations around so that only the parent is created in aac_pci.c. - Retrieve the capabilities word from the firmware before setting up command structures and tags. This allows the driver to decide whether to do 64-bit commands, and if work-arounds are needed for systems with >2GB of RAM. - Only enable the SCSI passthrough if it's enabled in the capabilities word in the firmware. This should fix problems with the 2120S and 2200S cards in systems with more than 2GB of RAM. Full 64-bit support is forthcoming. MFC-After: 1 week
2003-03-26 17:50:11 +00:00
/*
* Retrieve the capabilities/supported options word so we know what
Fix a regression that prevented PERC3 hardware from working. Apparently they do not support the GETINFO immediate command, unlike just about every other variant of the hardware. Also document some magic values and fix some minor nearby whitespace. MFC After: 3 days
2005-10-14 16:22:45 +00:00
* work-arounds to enable. Some firmware revs don't support this
* command.
Begin support for 64-bit address support and workarounds for newer cards: - Add data structuress for doing 64-bit scatter/gather - Move busdma tag creations around so that only the parent is created in aac_pci.c. - Retrieve the capabilities word from the firmware before setting up command structures and tags. This allows the driver to decide whether to do 64-bit commands, and if work-arounds are needed for systems with >2GB of RAM. - Only enable the SCSI passthrough if it's enabled in the capabilities word in the firmware. This should fix problems with the 2120S and 2200S cards in systems with more than 2GB of RAM. Full 64-bit support is forthcoming. MFC-After: 1 week
2003-03-26 17:50:11 +00:00
*/
Fix a regression that prevented PERC3 hardware from working. Apparently they do not support the GETINFO immediate command, unlike just about every other variant of the hardware. Also document some magic values and fix some minor nearby whitespace. MFC After: 3 days
2005-10-14 16:22:45 +00:00
if (aac_sync_command(sc, AAC_MONKER_GETINFO, 0, 0, 0, 0, &status)) {
if (status != AAC_SRB_STS_INVALID_REQUEST) {
device_printf(sc->aac_dev,
"RequestAdapterInfo failed\n");
return (EIO);
}
} else {
options = AAC_GET_MAILBOX(sc, 1);
atu_size = AAC_GET_MAILBOX(sc, 2);
sc->supported_options = options;
if ((options & AAC_SUPPORTED_4GB_WINDOW) != 0 &&
(sc->flags & AAC_FLAGS_NO4GB) == 0)
sc->flags |= AAC_FLAGS_4GB_WINDOW;
if (options & AAC_SUPPORTED_NONDASD)
sc->flags |= AAC_FLAGS_ENABLE_CAM;
if ((options & AAC_SUPPORTED_SGMAP_HOST64) != 0
&& (sizeof(bus_addr_t) > 4)) {
device_printf(sc->aac_dev,
"Enabling 64-bit address support\n");
sc->flags |= AAC_FLAGS_SG_64BIT;
}
if ((options & AAC_SUPPORTED_NEW_COMM)
- Make tables, device ID strings etc const. This includes #ifdef'ing 0 aac_command_status_table, which is actually unused since r111532. While at it, make aac_if a pointer to the now const interface tables instead of copying them over to the softc (this alone already reduces the size of aac.ko on amd64 by ~1 KiB). - Remove redundant softc members. - Use DEVMETHOD_END. - Use NULL instead of 0 for pointers. - Remove redundant bzero(9)'ing of the softc. - Use pci_enable_busmaster(9) instead of duplicating it. - Remove redundant checking for PCIM_CMD_MEMEN (resource allocation will just fail). - Canonicalize the error messages in case of resource allocation failures. - Add support for using MSI instead of INTx, controllable via the tunable hw.aac.enable_msi (defaulting to on). MFC after: 1 month
2013-03-01 19:55:10 +00:00
&& sc->aac_if->aif_send_command)
Fix a regression that prevented PERC3 hardware from working. Apparently they do not support the GETINFO immediate command, unlike just about every other variant of the hardware. Also document some magic values and fix some minor nearby whitespace. MFC After: 3 days
2005-10-14 16:22:45 +00:00
sc->flags |= AAC_FLAGS_NEW_COMM;
if (options & AAC_SUPPORTED_64BIT_ARRAYSIZE)
sc->flags |= AAC_FLAGS_ARRAY_64BIT;
Begin support for 64-bit address support and workarounds for newer cards: - Add data structuress for doing 64-bit scatter/gather - Move busdma tag creations around so that only the parent is created in aac_pci.c. - Retrieve the capabilities word from the firmware before setting up command structures and tags. This allows the driver to decide whether to do 64-bit commands, and if work-arounds are needed for systems with >2GB of RAM. - Only enable the SCSI passthrough if it's enabled in the capabilities word in the firmware. This should fix problems with the 2120S and 2200S cards in systems with more than 2GB of RAM. Full 64-bit support is forthcoming. MFC-After: 1 week
2003-03-26 17:50:11 +00:00
}
/* Check for broken hardware that does a lower number of commands */
Remove the unused AAC_DRIVER_BUILD_DATE macro. Nuke whitespace at EOL while I'm here. Approved by: scottl (MAINTAINER)
2005-11-13 17:26:36 +00:00
sc->aac_max_fibs = (sc->flags & AAC_FLAGS_256FIBS ? 256:512);
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
/* Remap mem. resource, if required */
Remove the unused AAC_DRIVER_BUILD_DATE macro. Nuke whitespace at EOL while I'm here. Approved by: scottl (MAINTAINER)
2005-11-13 17:26:36 +00:00
if ((sc->flags & AAC_FLAGS_NEW_COMM) &&
- Make tables, device ID strings etc const. This includes #ifdef'ing 0 aac_command_status_table, which is actually unused since r111532. While at it, make aac_if a pointer to the now const interface tables instead of copying them over to the softc (this alone already reduces the size of aac.ko on amd64 by ~1 KiB). - Remove redundant softc members. - Use DEVMETHOD_END. - Use NULL instead of 0 for pointers. - Remove redundant bzero(9)'ing of the softc. - Use pci_enable_busmaster(9) instead of duplicating it. - Remove redundant checking for PCIM_CMD_MEMEN (resource allocation will just fail). - Canonicalize the error messages in case of resource allocation failures. - Add support for using MSI instead of INTx, controllable via the tunable hw.aac.enable_msi (defaulting to on). MFC after: 1 month
2013-03-01 19:55:10 +00:00
atu_size > rman_get_size(sc->aac_regs_res1)) {
rid = rman_get_rid(sc->aac_regs_res1);
bus_release_resource(sc->aac_dev, SYS_RES_MEMORY, rid,
sc->aac_regs_res1);
Migrate many bus_alloc_resource() calls to bus_alloc_resource_anywhere(). Most calls to bus_alloc_resource() use "anywhere" as the range, with a given count. Migrate these to use the new bus_alloc_resource_anywhere() API. Reviewed by: jhb Differential Revision: https://reviews.freebsd.org/D5370
2016-02-27 03:38:01 +00:00
sc->aac_regs_res1 = bus_alloc_resource_anywhere(sc->aac_dev,
SYS_RES_MEMORY, &rid, atu_size, RF_ACTIVE);
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
if (sc->aac_regs_res1 == NULL) {
sc->aac_regs_res1 = bus_alloc_resource_any(
- Make tables, device ID strings etc const. This includes #ifdef'ing 0 aac_command_status_table, which is actually unused since r111532. While at it, make aac_if a pointer to the now const interface tables instead of copying them over to the softc (this alone already reduces the size of aac.ko on amd64 by ~1 KiB). - Remove redundant softc members. - Use DEVMETHOD_END. - Use NULL instead of 0 for pointers. - Remove redundant bzero(9)'ing of the softc. - Use pci_enable_busmaster(9) instead of duplicating it. - Remove redundant checking for PCIM_CMD_MEMEN (resource allocation will just fail). - Canonicalize the error messages in case of resource allocation failures. - Add support for using MSI instead of INTx, controllable via the tunable hw.aac.enable_msi (defaulting to on). MFC after: 1 month
2013-03-01 19:55:10 +00:00
sc->aac_dev, SYS_RES_MEMORY, &rid, RF_ACTIVE);
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
if (sc->aac_regs_res1 == NULL) {
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
device_printf(sc->aac_dev,
Ue a better msleep identifier. Fix some whitespace.
2005-10-08 22:41:57 +00:00
"couldn't allocate register window\n");
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
return (ENXIO);
}
sc->flags &= ~AAC_FLAGS_NEW_COMM;
}
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
sc->aac_btag1 = rman_get_bustag(sc->aac_regs_res1);
sc->aac_bhandle1 = rman_get_bushandle(sc->aac_regs_res1);
if (sc->aac_hwif == AAC_HWIF_NARK) {
sc->aac_regs_res0 = sc->aac_regs_res1;
sc->aac_btag0 = sc->aac_btag1;
sc->aac_bhandle0 = sc->aac_bhandle1;
}
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
}
Begin support for 64-bit address support and workarounds for newer cards: - Add data structuress for doing 64-bit scatter/gather - Move busdma tag creations around so that only the parent is created in aac_pci.c. - Retrieve the capabilities word from the firmware before setting up command structures and tags. This allows the driver to decide whether to do 64-bit commands, and if work-arounds are needed for systems with >2GB of RAM. - Only enable the SCSI passthrough if it's enabled in the capabilities word in the firmware. This should fix problems with the 2120S and 2200S cards in systems with more than 2GB of RAM. Full 64-bit support is forthcoming. MFC-After: 1 week
2003-03-26 17:50:11 +00:00
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
/* Read preferred settings */
sc->aac_max_fib_size = sizeof(struct aac_fib);
sc->aac_max_sectors = 128; /* 64KB */
if (sc->flags & AAC_FLAGS_SG_64BIT)
Fix a regression that prevented PERC3 hardware from working. Apparently they do not support the GETINFO immediate command, unlike just about every other variant of the hardware. Also document some magic values and fix some minor nearby whitespace. MFC After: 3 days
2005-10-14 16:22:45 +00:00
sc->aac_sg_tablesize = (AAC_FIB_DATASIZE
Correct calculation of aac_sg_tablesize. Obtained from: Adaptec, via driver b11669
2007-10-15 17:34:58 +00:00
- sizeof(struct aac_blockwrite64))
/ sizeof(struct aac_sg_entry64);
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
else
Fix a regression that prevented PERC3 hardware from working. Apparently they do not support the GETINFO immediate command, unlike just about every other variant of the hardware. Also document some magic values and fix some minor nearby whitespace. MFC After: 3 days
2005-10-14 16:22:45 +00:00
sc->aac_sg_tablesize = (AAC_FIB_DATASIZE
Correct calculation of aac_sg_tablesize. Obtained from: Adaptec, via driver b11669
2007-10-15 17:34:58 +00:00
- sizeof(struct aac_blockwrite))
/ sizeof(struct aac_sg_entry);
Fix a regression that prevented PERC3 hardware from working. Apparently they do not support the GETINFO immediate command, unlike just about every other variant of the hardware. Also document some magic values and fix some minor nearby whitespace. MFC After: 3 days
2005-10-14 16:22:45 +00:00
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
if (!aac_sync_command(sc, AAC_MONKER_GETCOMMPREF, 0, 0, 0, 0, NULL)) {
options = AAC_GET_MAILBOX(sc, 1);
sc->aac_max_fib_size = (options & 0xFFFF);
sc->aac_max_sectors = (options >> 16) << 1;
options = AAC_GET_MAILBOX(sc, 2);
sc->aac_sg_tablesize = (options >> 16);
options = AAC_GET_MAILBOX(sc, 3);
sc->aac_max_fibs = (options & 0xFFFF);
}
if (sc->aac_max_fib_size > PAGE_SIZE)
sc->aac_max_fib_size = PAGE_SIZE;
sc->aac_max_fibs_alloc = PAGE_SIZE / sc->aac_max_fib_size;
Remove the unused AAC_DRIVER_BUILD_DATE macro. Nuke whitespace at EOL while I'm here. Approved by: scottl (MAINTAINER)
2005-11-13 17:26:36 +00:00
Add ioctls FSACTL_SEND_LARGE_FIB, FSACTL_SEND_RAW_SRB, FSACTL_LNX_SEND_LARGE_FIB, and FSACTL_LNX_SEND_RAW_SRB, and correct size checks on FIBs passed in from userspace. Both changes were obtained from Adaptec's driver build 15317. Adaptec's commandline RAID tool arcconf uses these ioctls when creating a RAID-10 array (and probably other operations too).
2008-03-20 17:59:19 +00:00
if (sc->aac_max_fib_size > sizeof(struct aac_fib)) {
sc->flags |= AAC_FLAGS_RAW_IO;
device_printf(sc->aac_dev, "Enable Raw I/O\n");
}
Add 64-bit array support for RAIDs > 2TB. This corresponds to ~ Adaptec driver build 15317. Tested on: Adaptec 2230S, Firmware 4.2-0 (8205) ICP ICP5085BL, Firmware 5.2-0 (12814) Submitted by: Adaptec
2008-03-25 21:39:06 +00:00
if ((sc->flags & AAC_FLAGS_RAW_IO) &&
(sc->flags & AAC_FLAGS_ARRAY_64BIT)) {
sc->flags |= AAC_FLAGS_LBA_64BIT;
device_printf(sc->aac_dev, "Enable 64-bit array\n");
}
Add ioctls FSACTL_SEND_LARGE_FIB, FSACTL_SEND_RAW_SRB, FSACTL_LNX_SEND_LARGE_FIB, and FSACTL_LNX_SEND_RAW_SRB, and correct size checks on FIBs passed in from userspace. Both changes were obtained from Adaptec's driver build 15317. Adaptec's commandline RAID tool arcconf uses these ioctls when creating a RAID-10 array (and probably other operations too).
2008-03-20 17:59:19 +00:00
Don't attach to Dell PERC2/QC cards that have a firmware rev of 1.x. This check is complicated by the fact that the Adaptec 5400S cards claim to use 1.x firmware also. PERC2/QC 1.x firmware is not compatible with this driver and will cause a system hang. MFC after: 3 days
2002-02-06 01:34:09 +00:00
return (0);
}
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
static int
aac_init(struct aac_softc *sc)
{
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
struct aac_adapter_init *ip;
Diff reduction to Adaptec driver build 15317 (refactoring and code shuffling): - Resource allocation in aac_alloc (moved from from aac_init) - Interrupt setup in aac_setup_intr (from aac_attach) - Container probing in aac_get_container_info (from aac_startup and aac_handle_aif) - Firmware status check moved to aac_check_firmware from aac_init
2008-03-24 16:38:47 +00:00
u_int32_t qoffset;
Begin support for 64-bit address support and workarounds for newer cards: - Add data structuress for doing 64-bit scatter/gather - Move busdma tag creations around so that only the parent is created in aac_pci.c. - Retrieve the capabilities word from the firmware before setting up command structures and tags. This allows the driver to decide whether to do 64-bit commands, and if work-arounds are needed for systems with >2GB of RAM. - Only enable the SCSI passthrough if it's enabled in the capabilities word in the firmware. This should fix problems with the 2120S and 2200S cards in systems with more than 2GB of RAM. Full 64-bit support is forthcoming. MFC-After: 1 week
2003-03-26 17:50:11 +00:00
int error;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
Remove the unused AAC_DRIVER_BUILD_DATE macro. Nuke whitespace at EOL while I'm here. Approved by: scottl (MAINTAINER)
2005-11-13 17:26:36 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
* Fill in the init structure. This tells the adapter about the
* physical location of various important shared data structures.
*/
ip = &sc->aac_common->ac_init;
ip->InitStructRevision = AAC_INIT_STRUCT_REVISION;
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
if (sc->aac_max_fib_size > sizeof(struct aac_fib)) {
ip->InitStructRevision = AAC_INIT_STRUCT_REVISION_4;
sc->flags |= AAC_FLAGS_RAW_IO;
}
Major bugfixes for large memory and fast systems. aac.c: Re-arrange the interrupt handler to optimize the common case of the adapter interrupting us because one or more commands are complete, and do a read across the pci bus to ensure that all posted status writes are flushed. This should close a race that could cause command completion interrupts to be lost. Follow the spec a bit closer when filling out command structures. Enable the Fast Response feature to eliminate the need for the card to DMA successfull command completions back into host memory. Tell the controller how much physical memory we have. Without this there was a chance that our DMA regions would collide with the memory window used by the cache on the controller. The result would be massive data corruption. This seemed to mainly affect systems with >2GB of memory. Fix a few whitespace problems. aac_debug.c: Add an extra diagnostic when printing out commands. aac_disk.c: Add extra sanity checks. aacreg.h: Prepare for making this 64-bit clean by reducing the use of enumeration types in structures. Many thanks to Justin Gibbs for helping track these down.
2003-01-11 01:59:21 +00:00
ip->MiniPortRevision = AAC_INIT_STRUCT_MINIPORT_REVISION;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
ip->AdapterFibsPhysicalAddress = sc->aac_common_busaddr +
offsetof(struct aac_common, ac_fibs);
Don't pass pointers to kernel data structures through 32-bit fields as 64-bit architectures won't like this. Use virtual array indexes instead. This *should* allow the driver to work on 64-bit platforms, though it's still not endian clean.
2003-04-28 06:16:20 +00:00
ip->AdapterFibsVirtualAddress = 0;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
ip->AdapterFibsSize = AAC_ADAPTER_FIBS * sizeof(struct aac_fib);
ip->AdapterFibAlign = sizeof(struct aac_fib);
ip->PrintfBufferAddress = sc->aac_common_busaddr +
offsetof(struct aac_common, ac_printf);
ip->PrintfBufferSize = AAC_PRINTF_BUFSIZE;
Remove the unused AAC_DRIVER_BUILD_DATE macro. Nuke whitespace at EOL while I'm here. Approved by: scottl (MAINTAINER)
2005-11-13 17:26:36 +00:00
/*
Add a new quirk for cards that incorrectly interpret the amount of memory in the system. This might also have a small performance gain.
2003-07-09 19:19:16 +00:00
* The adapter assumes that pages are 4K in size, except on some
* broken firmware versions that do the page->byte conversion twice,
* therefore 'assuming' that this value is in 16MB units (2^24).
* Round up since the granularity is so high.
*/
Major bugfixes for large memory and fast systems. aac.c: Re-arrange the interrupt handler to optimize the common case of the adapter interrupting us because one or more commands are complete, and do a read across the pci bus to ensure that all posted status writes are flushed. This should close a race that could cause command completion interrupts to be lost. Follow the spec a bit closer when filling out command structures. Enable the Fast Response feature to eliminate the need for the card to DMA successfull command completions back into host memory. Tell the controller how much physical memory we have. Without this there was a chance that our DMA regions would collide with the memory window used by the cache on the controller. The result would be massive data corruption. This seemed to mainly affect systems with >2GB of memory. Fix a few whitespace problems. aac_debug.c: Add an extra diagnostic when printing out commands. aac_disk.c: Add extra sanity checks. aacreg.h: Prepare for making this 64-bit clean by reducing the use of enumeration types in structures. Many thanks to Justin Gibbs for helping track these down.
2003-01-11 01:59:21 +00:00
ip->HostPhysMemPages = ctob(physmem) / AAC_PAGE_SIZE;
Add a new quirk for cards that incorrectly interpret the amount of memory in the system. This might also have a small performance gain.
2003-07-09 19:19:16 +00:00
if (sc->flags & AAC_FLAGS_BROKEN_MEMMAP) {
ip->HostPhysMemPages =
(ip->HostPhysMemPages + AAC_PAGE_SIZE) / AAC_PAGE_SIZE;
Fix a missing } that got dropped from the last commit.
2003-07-09 19:30:30 +00:00
}
Use time_uptime for timing commands instead of time_second. This is a bit risky because the "current time" is supposed to be fed to the card during initialization, and the current time is supposed to be put into each command that is sent to the card. Hopefully either the card doesn't actually care about the timestamps, or it doesn't care about the absolute values so long and the relative values are consistent. Not an MFC candidate until more thorough testing can be done.
2005-09-14 05:46:28 +00:00
ip->HostElapsedSeconds = time_uptime; /* reset later if invalid */
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
ip->InitFlags = 0;
if (sc->flags & AAC_FLAGS_NEW_COMM) {
Sync some minor items with the upstream driver. Should have no functional change.
2010-04-13 01:16:15 +00:00
ip->InitFlags |= AAC_INITFLAGS_NEW_COMM_SUPPORTED;
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
device_printf(sc->aac_dev, "New comm. interface enabled\n");
}
ip->MaxIoCommands = sc->aac_max_fibs;
ip->MaxIoSize = sc->aac_max_sectors << 9;
ip->MaxFibSize = sc->aac_max_fib_size;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
Change spelling and eliminate a typo in comments to reduce diffs with Adaptec's vendor driver. I have some fixes to bring in and this makes ongoing review of the FreeBSD-Adaptec driver diffs easier.
2008-03-14 21:59:11 +00:00
* Initialize FIB queues. Note that it appears that the layout of the
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
* indexes and the segmentation of the entries may be mandated by the
* adapter, which is only told about the base of the queue index fields.
*
* The initial values of the indices are assumed to inform the adapter
Remove the unused AAC_DRIVER_BUILD_DATE macro. Nuke whitespace at EOL while I'm here. Approved by: scottl (MAINTAINER)
2005-11-13 17:26:36 +00:00
* of the sizes of the respective queues, and theoretically it could
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
* work out the entire layout of the queue structures from this. We
* take the easy route and just lay this area out like everyone else
* does.
*
Remove the unused AAC_DRIVER_BUILD_DATE macro. Nuke whitespace at EOL while I'm here. Approved by: scottl (MAINTAINER)
2005-11-13 17:26:36 +00:00
* The Linux driver uses a much more complex scheme whereby several
* header records are kept for each queue. We use a couple of generic
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
* list manipulation functions which 'know' the size of each list by
* virtue of a table.
*/
Make aac(4) compile cleanly on 64-bit machines. The code was already 64-bit safe, but some (unneeded and/or harmless) downcasts were generating warnings. The driver still is not endian-clean.
2003-08-19 21:31:45 +00:00
qoffset = offsetof(struct aac_common, ac_qbuf) + AAC_QUEUE_ALIGN;
If ~ chars were pennies, I'd be pennyless. This should fix all of the 'command not in queue' panics. Also fix a nearby style problem.
2003-09-01 00:20:29 +00:00
qoffset &= ~(AAC_QUEUE_ALIGN - 1);
sc->aac_queues =
(struct aac_queue_table *)((uintptr_t)sc->aac_common + qoffset);
Make aac(4) compile cleanly on 64-bit machines. The code was already 64-bit safe, but some (unneeded and/or harmless) downcasts were generating warnings. The driver still is not endian-clean.
2003-08-19 21:31:45 +00:00
ip->CommHeaderAddress = sc->aac_common_busaddr + qoffset;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
sc->aac_queues->qt_qindex[AAC_HOST_NORM_CMD_QUEUE][AAC_PRODUCER_INDEX] =
Reformat for 80 columns. Sorry, but I had to do it.
2001-08-03 00:50:30 +00:00
AAC_HOST_NORM_CMD_ENTRIES;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
sc->aac_queues->qt_qindex[AAC_HOST_NORM_CMD_QUEUE][AAC_CONSUMER_INDEX] =
Reformat for 80 columns. Sorry, but I had to do it.
2001-08-03 00:50:30 +00:00
AAC_HOST_NORM_CMD_ENTRIES;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
sc->aac_queues->qt_qindex[AAC_HOST_HIGH_CMD_QUEUE][AAC_PRODUCER_INDEX] =
Reformat for 80 columns. Sorry, but I had to do it.
2001-08-03 00:50:30 +00:00
AAC_HOST_HIGH_CMD_ENTRIES;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
sc->aac_queues->qt_qindex[AAC_HOST_HIGH_CMD_QUEUE][AAC_CONSUMER_INDEX] =
Reformat for 80 columns. Sorry, but I had to do it.
2001-08-03 00:50:30 +00:00
AAC_HOST_HIGH_CMD_ENTRIES;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
sc->aac_queues->qt_qindex[AAC_ADAP_NORM_CMD_QUEUE][AAC_PRODUCER_INDEX] =
Reformat for 80 columns. Sorry, but I had to do it.
2001-08-03 00:50:30 +00:00
AAC_ADAP_NORM_CMD_ENTRIES;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
sc->aac_queues->qt_qindex[AAC_ADAP_NORM_CMD_QUEUE][AAC_CONSUMER_INDEX] =
Reformat for 80 columns. Sorry, but I had to do it.
2001-08-03 00:50:30 +00:00
AAC_ADAP_NORM_CMD_ENTRIES;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
sc->aac_queues->qt_qindex[AAC_ADAP_HIGH_CMD_QUEUE][AAC_PRODUCER_INDEX] =
Reformat for 80 columns. Sorry, but I had to do it.
2001-08-03 00:50:30 +00:00
AAC_ADAP_HIGH_CMD_ENTRIES;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
sc->aac_queues->qt_qindex[AAC_ADAP_HIGH_CMD_QUEUE][AAC_CONSUMER_INDEX] =
Reformat for 80 columns. Sorry, but I had to do it.
2001-08-03 00:50:30 +00:00
AAC_ADAP_HIGH_CMD_ENTRIES;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
sc->aac_queues->qt_qindex[AAC_HOST_NORM_RESP_QUEUE][AAC_PRODUCER_INDEX]=
Reformat for 80 columns. Sorry, but I had to do it.
2001-08-03 00:50:30 +00:00
AAC_HOST_NORM_RESP_ENTRIES;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
sc->aac_queues->qt_qindex[AAC_HOST_NORM_RESP_QUEUE][AAC_CONSUMER_INDEX]=
Reformat for 80 columns. Sorry, but I had to do it.
2001-08-03 00:50:30 +00:00
AAC_HOST_NORM_RESP_ENTRIES;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
sc->aac_queues->qt_qindex[AAC_HOST_HIGH_RESP_QUEUE][AAC_PRODUCER_INDEX]=
Reformat for 80 columns. Sorry, but I had to do it.
2001-08-03 00:50:30 +00:00
AAC_HOST_HIGH_RESP_ENTRIES;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
sc->aac_queues->qt_qindex[AAC_HOST_HIGH_RESP_QUEUE][AAC_CONSUMER_INDEX]=
Reformat for 80 columns. Sorry, but I had to do it.
2001-08-03 00:50:30 +00:00
AAC_HOST_HIGH_RESP_ENTRIES;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
sc->aac_queues->qt_qindex[AAC_ADAP_NORM_RESP_QUEUE][AAC_PRODUCER_INDEX]=
Reformat for 80 columns. Sorry, but I had to do it.
2001-08-03 00:50:30 +00:00
AAC_ADAP_NORM_RESP_ENTRIES;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
sc->aac_queues->qt_qindex[AAC_ADAP_NORM_RESP_QUEUE][AAC_CONSUMER_INDEX]=
Reformat for 80 columns. Sorry, but I had to do it.
2001-08-03 00:50:30 +00:00
AAC_ADAP_NORM_RESP_ENTRIES;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
sc->aac_queues->qt_qindex[AAC_ADAP_HIGH_RESP_QUEUE][AAC_PRODUCER_INDEX]=
Reformat for 80 columns. Sorry, but I had to do it.
2001-08-03 00:50:30 +00:00
AAC_ADAP_HIGH_RESP_ENTRIES;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
sc->aac_queues->qt_qindex[AAC_ADAP_HIGH_RESP_QUEUE][AAC_CONSUMER_INDEX]=
Reformat for 80 columns. Sorry, but I had to do it.
2001-08-03 00:50:30 +00:00
AAC_ADAP_HIGH_RESP_ENTRIES;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
sc->aac_qentries[AAC_HOST_NORM_CMD_QUEUE] =
Reformat for 80 columns. Sorry, but I had to do it.
2001-08-03 00:50:30 +00:00
&sc->aac_queues->qt_HostNormCmdQueue[0];
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
sc->aac_qentries[AAC_HOST_HIGH_CMD_QUEUE] =
Reformat for 80 columns. Sorry, but I had to do it.
2001-08-03 00:50:30 +00:00
&sc->aac_queues->qt_HostHighCmdQueue[0];
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
sc->aac_qentries[AAC_ADAP_NORM_CMD_QUEUE] =
Reformat for 80 columns. Sorry, but I had to do it.
2001-08-03 00:50:30 +00:00
&sc->aac_queues->qt_AdapNormCmdQueue[0];
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
sc->aac_qentries[AAC_ADAP_HIGH_CMD_QUEUE] =
Reformat for 80 columns. Sorry, but I had to do it.
2001-08-03 00:50:30 +00:00
&sc->aac_queues->qt_AdapHighCmdQueue[0];
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
sc->aac_qentries[AAC_HOST_NORM_RESP_QUEUE] =
Reformat for 80 columns. Sorry, but I had to do it.
2001-08-03 00:50:30 +00:00
&sc->aac_queues->qt_HostNormRespQueue[0];
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
sc->aac_qentries[AAC_HOST_HIGH_RESP_QUEUE] =
Reformat for 80 columns. Sorry, but I had to do it.
2001-08-03 00:50:30 +00:00
&sc->aac_queues->qt_HostHighRespQueue[0];
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
sc->aac_qentries[AAC_ADAP_NORM_RESP_QUEUE] =
Reformat for 80 columns. Sorry, but I had to do it.
2001-08-03 00:50:30 +00:00
&sc->aac_queues->qt_AdapNormRespQueue[0];
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
sc->aac_qentries[AAC_ADAP_HIGH_RESP_QUEUE] =
Reformat for 80 columns. Sorry, but I had to do it.
2001-08-03 00:50:30 +00:00
&sc->aac_queues->qt_AdapHighRespQueue[0];
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
* Do controller-type-specific initialisation
*/
switch (sc->aac_hwif) {
case AAC_HWIF_I960RX:
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
AAC_MEM0_SETREG4(sc, AAC_RX_ODBR, ~0);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
break;
Add support for the Adaptec RAID-On-Chip architecture. This in turn provides support for the Adaptec 2130S adapter. Thanks to Adaptec for providing hardware for this.
2004-08-13 01:44:09 +00:00
case AAC_HWIF_RKT:
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
AAC_MEM0_SETREG4(sc, AAC_RKT_ODBR, ~0);
Add support for the Adaptec RAID-On-Chip architecture. This in turn provides support for the Adaptec 2130S adapter. Thanks to Adaptec for providing hardware for this.
2004-08-13 01:44:09 +00:00
break;
default:
break;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
}
/*
* Give the init structure to the controller.
*/
Remove the unused AAC_DRIVER_BUILD_DATE macro. Nuke whitespace at EOL while I'm here. Approved by: scottl (MAINTAINER)
2005-11-13 17:26:36 +00:00
if (aac_sync_command(sc, AAC_MONKER_INITSTRUCT,
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
sc->aac_common_busaddr +
offsetof(struct aac_common, ac_init), 0, 0, 0,
NULL)) {
device_printf(sc->aac_dev,
"error establishing init structure\n");
Begin support for 64-bit address support and workarounds for newer cards: - Add data structuress for doing 64-bit scatter/gather - Move busdma tag creations around so that only the parent is created in aac_pci.c. - Retrieve the capabilities word from the firmware before setting up command structures and tags. This allows the driver to decide whether to do 64-bit commands, and if work-arounds are needed for systems with >2GB of RAM. - Only enable the SCSI passthrough if it's enabled in the capabilities word in the firmware. This should fix problems with the 2120S and 2200S cards in systems with more than 2GB of RAM. Full 64-bit support is forthcoming. MFC-After: 1 week
2003-03-26 17:50:11 +00:00
error = EIO;
goto out;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
}
Begin support for 64-bit address support and workarounds for newer cards: - Add data structuress for doing 64-bit scatter/gather - Move busdma tag creations around so that only the parent is created in aac_pci.c. - Retrieve the capabilities word from the firmware before setting up command structures and tags. This allows the driver to decide whether to do 64-bit commands, and if work-arounds are needed for systems with >2GB of RAM. - Only enable the SCSI passthrough if it's enabled in the capabilities word in the firmware. This should fix problems with the 2120S and 2200S cards in systems with more than 2GB of RAM. Full 64-bit support is forthcoming. MFC-After: 1 week
2003-03-26 17:50:11 +00:00
error = 0;
out:
return(error);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
Diff reduction to Adaptec driver build 15317 (refactoring and code shuffling): - Resource allocation in aac_alloc (moved from from aac_init) - Interrupt setup in aac_setup_intr (from aac_attach) - Container probing in aac_get_container_info (from aac_startup and aac_handle_aif) - Firmware status check moved to aac_check_firmware from aac_init
2008-03-24 16:38:47 +00:00
static int
aac_setup_intr(struct aac_softc *sc)
{
- Make tables, device ID strings etc const. This includes #ifdef'ing 0 aac_command_status_table, which is actually unused since r111532. While at it, make aac_if a pointer to the now const interface tables instead of copying them over to the softc (this alone already reduces the size of aac.ko on amd64 by ~1 KiB). - Remove redundant softc members. - Use DEVMETHOD_END. - Use NULL instead of 0 for pointers. - Remove redundant bzero(9)'ing of the softc. - Use pci_enable_busmaster(9) instead of duplicating it. - Remove redundant checking for PCIM_CMD_MEMEN (resource allocation will just fail). - Canonicalize the error messages in case of resource allocation failures. - Add support for using MSI instead of INTx, controllable via the tunable hw.aac.enable_msi (defaulting to on). MFC after: 1 month
2013-03-01 19:55:10 +00:00
Diff reduction to Adaptec driver build 15317 (refactoring and code shuffling): - Resource allocation in aac_alloc (moved from from aac_init) - Interrupt setup in aac_setup_intr (from aac_attach) - Container probing in aac_get_container_info (from aac_startup and aac_handle_aif) - Firmware status check moved to aac_check_firmware from aac_init
2008-03-24 16:38:47 +00:00
if (sc->flags & AAC_FLAGS_NEW_COMM) {
if (bus_setup_intr(sc->aac_dev, sc->aac_irq,
Whitespace cleanup, in advance of next sync with Adaptec's driver. No functional change.
2010-04-13 00:33:07 +00:00
INTR_MPSAFE|INTR_TYPE_BIO, NULL,
Diff reduction to Adaptec driver build 15317 (refactoring and code shuffling): - Resource allocation in aac_alloc (moved from from aac_init) - Interrupt setup in aac_setup_intr (from aac_attach) - Container probing in aac_get_container_info (from aac_startup and aac_handle_aif) - Firmware status check moved to aac_check_firmware from aac_init
2008-03-24 16:38:47 +00:00
aac_new_intr, sc, &sc->aac_intr)) {
device_printf(sc->aac_dev, "can't set up interrupt\n");
return (EINVAL);
}
} else {
if (bus_setup_intr(sc->aac_dev, sc->aac_irq,
Rename aac_fast_intr to aac_filter to reflect its current use. Eliminate the fallback of using the filter as an interrupt handler, as it is no longer needed. Discussed with: scottl, jhb
2009-10-29 14:53:45 +00:00
INTR_TYPE_BIO, aac_filter, NULL,
Diff reduction to Adaptec driver build 15317 (refactoring and code shuffling): - Resource allocation in aac_alloc (moved from from aac_init) - Interrupt setup in aac_setup_intr (from aac_attach) - Container probing in aac_get_container_info (from aac_startup and aac_handle_aif) - Firmware status check moved to aac_check_firmware from aac_init
2008-03-24 16:38:47 +00:00
sc, &sc->aac_intr)) {
device_printf(sc->aac_dev,
Rename aac_fast_intr to aac_filter to reflect its current use. Eliminate the fallback of using the filter as an interrupt handler, as it is no longer needed. Discussed with: scottl, jhb
2009-10-29 14:53:45 +00:00
"can't set up interrupt filter\n");
return (EINVAL);
Diff reduction to Adaptec driver build 15317 (refactoring and code shuffling): - Resource allocation in aac_alloc (moved from from aac_init) - Interrupt setup in aac_setup_intr (from aac_attach) - Container probing in aac_get_container_info (from aac_startup and aac_handle_aif) - Firmware status check moved to aac_check_firmware from aac_init
2008-03-24 16:38:47 +00:00
}
}
return (0);
}
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
* Send a synchronous command to the controller and wait for a result.
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
* Indicate if the controller completed the command with an error status.
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
*/
static int
aac_sync_command(struct aac_softc *sc, u_int32_t command,
Major bugfix and minor update. This should resolve the current issues with the driver locking up under load. - Restructure so that we use a static pool of commands/FIBs, rather than allocating them in clusters. The cluster allocation just made things more complicated, and allowed us to waste more memory in peak load situations. - Make queueing macros more like my other drivers. This adds queue stats for free. Add some debugging to take advantage of this. - Reimplement the periodic timeout scan. Kick the interrupt handler and the start routine every scan as well, just to be safe. Track busy commands properly. - Bring resource cleanup into line with resource allocation. We should now clean up correctly after a failed probe/unload/etc. - Try to start new commands when old ones are completed. We weren't doing this before, which could lead to deadlock when the controller was full. - Don't try to build a new command if we have found a deferred command. This could cause us to lose the deferred command. - Use diskerr() to report I/O errors. - Don't bail if the AdapterInfo structure is the wrong size. Some variation seems to be normal. We need to improve our handing of 2.x firmware sets. - Improve some comments in an attempt to try to make things clearer. - Restructure to avoid some warnings.
2000-12-27 13:14:56 +00:00
u_int32_t arg0, u_int32_t arg1, u_int32_t arg2, u_int32_t arg3,
u_int32_t *sp)
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
{
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
time_t then;
u_int32_t status;
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/* populate the mailbox */
AAC_SET_MAILBOX(sc, command, arg0, arg1, arg2, arg3);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/* ensure the sync command doorbell flag is cleared */
AAC_CLEAR_ISTATUS(sc, AAC_DB_SYNC_COMMAND);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/* then set it to signal the adapter */
AAC_QNOTIFY(sc, AAC_DB_SYNC_COMMAND);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/* spin waiting for the command to complete */
Use time_uptime for timing commands instead of time_second. This is a bit risky because the "current time" is supposed to be fed to the card during initialization, and the current time is supposed to be put into each command that is sent to the card. Hopefully either the card doesn't actually care about the timestamps, or it doesn't care about the absolute values so long and the relative values are consistent. Not an MFC candidate until more thorough testing can be done.
2005-09-14 05:46:28 +00:00
then = time_uptime;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
do {
Use time_uptime for timing commands instead of time_second. This is a bit risky because the "current time" is supposed to be fed to the card during initialization, and the current time is supposed to be put into each command that is sent to the card. Hopefully either the card doesn't actually care about the timestamps, or it doesn't care about the absolute values so long and the relative values are consistent. Not an MFC candidate until more thorough testing can be done.
2005-09-14 05:46:28 +00:00
if (time_uptime > (then + AAC_IMMEDIATE_TIMEOUT)) {
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_ERROR_B, "timed out");
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
return(EIO);
}
} while (!(AAC_GET_ISTATUS(sc) & AAC_DB_SYNC_COMMAND));
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/* clear the completion flag */
AAC_CLEAR_ISTATUS(sc, AAC_DB_SYNC_COMMAND);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/* get the command status */
Begin support for 64-bit address support and workarounds for newer cards: - Add data structuress for doing 64-bit scatter/gather - Move busdma tag creations around so that only the parent is created in aac_pci.c. - Retrieve the capabilities word from the firmware before setting up command structures and tags. This allows the driver to decide whether to do 64-bit commands, and if work-arounds are needed for systems with >2GB of RAM. - Only enable the SCSI passthrough if it's enabled in the capabilities word in the firmware. This should fix problems with the 2120S and 2200S cards in systems with more than 2GB of RAM. Full 64-bit support is forthcoming. MFC-After: 1 week
2003-03-26 17:50:11 +00:00
status = AAC_GET_MAILBOX(sc, 0);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
if (sp != NULL)
*sp = status;
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
Fix a regression that prevented PERC3 hardware from working. Apparently they do not support the GETINFO immediate command, unlike just about every other variant of the hardware. Also document some magic values and fix some minor nearby whitespace. MFC After: 3 days
2005-10-14 16:22:45 +00:00
if (status != AAC_SRB_STS_SUCCESS)
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
return (-1);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
return(0);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
Prepare for a major update to the aac driver: Update the aac driver with the new crashdump api. Protect sync fibs with a mutex. Align all DMA buffers on a PAGE_SIZE boundary. MFC after: 3 days
2002-04-24 05:12:50 +00:00
int
Remove the unused AAC_DRIVER_BUILD_DATE macro. Nuke whitespace at EOL while I'm here. Approved by: scottl (MAINTAINER)
2005-11-13 17:26:36 +00:00
aac_sync_fib(struct aac_softc *sc, u_int32_t command, u_int32_t xferstate,
Prepare for a major update to the aac driver: Update the aac driver with the new crashdump api. Protect sync fibs with a mutex. Align all DMA buffers on a PAGE_SIZE boundary. MFC after: 3 days
2002-04-24 05:12:50 +00:00
struct aac_fib *fib, u_int16_t datasize)
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
{
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
mtx_assert(&sc->aac_io_lock, MA_OWNED);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
if (datasize > AAC_FIB_DATASIZE)
return(EINVAL);
/*
* Set up the sync FIB
*/
fib->Header.XferState = AAC_FIBSTATE_HOSTOWNED |
AAC_FIBSTATE_INITIALISED |
AAC_FIBSTATE_EMPTY;
fib->Header.XferState |= xferstate;
fib->Header.Command = command;
fib->Header.StructType = AAC_FIBTYPE_TFIB;
Restore creation of passthrough devices with newer controller firmware by putting the correct size in the fib header. Presumably the older firmware silently ignored a bad size field. (This change tested with a 3805 controller. Passthrough devices were created when running firmware build 12814, but not 15323 or later. With this change they're created for both old and new firmware versions.) Submitted by: Adaptec
2008-03-20 20:33:48 +00:00
fib->Header.Size = sizeof(struct aac_fib_header) + datasize;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
fib->Header.SenderSize = sizeof(struct aac_fib);
Make aac(4) compile cleanly on 64-bit machines. The code was already 64-bit safe, but some (unneeded and/or harmless) downcasts were generating warnings. The driver still is not endian-clean.
2003-08-19 21:31:45 +00:00
fib->Header.SenderFibAddress = 0; /* Not needed */
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
fib->Header.ReceiverFibAddress = sc->aac_common_busaddr +
offsetof(struct aac_common,
ac_sync_fib);
/*
* Give the FIB to the controller, wait for a response.
*/
if (aac_sync_command(sc, AAC_MONKER_SYNCFIB,
fib->Header.ReceiverFibAddress, 0, 0, 0, NULL)) {
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_ERROR_B, "IO error");
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
return(EIO);
}
Prepare for a major update to the aac driver: Update the aac driver with the new crashdump api. Protect sync fibs with a mutex. Align all DMA buffers on a PAGE_SIZE boundary. MFC after: 3 days
2002-04-24 05:12:50 +00:00
return (0);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
* Adapter-space FIB queue manipulation
*
* Note that the queue implementation here is a little funky; neither the PI or
* CI will ever be zero. This behaviour is a controller feature.
*/
- Make tables, device ID strings etc const. This includes #ifdef'ing 0 aac_command_status_table, which is actually unused since r111532. While at it, make aac_if a pointer to the now const interface tables instead of copying them over to the softc (this alone already reduces the size of aac.ko on amd64 by ~1 KiB). - Remove redundant softc members. - Use DEVMETHOD_END. - Use NULL instead of 0 for pointers. - Remove redundant bzero(9)'ing of the softc. - Use pci_enable_busmaster(9) instead of duplicating it. - Remove redundant checking for PCIM_CMD_MEMEN (resource allocation will just fail). - Canonicalize the error messages in case of resource allocation failures. - Add support for using MSI instead of INTx, controllable via the tunable hw.aac.enable_msi (defaulting to on). MFC after: 1 month
2013-03-01 19:55:10 +00:00
static const struct {
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
int size;
int notify;
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
} aac_qinfo[] = {
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
{AAC_HOST_NORM_CMD_ENTRIES, AAC_DB_COMMAND_NOT_FULL},
{AAC_HOST_HIGH_CMD_ENTRIES, 0},
{AAC_ADAP_NORM_CMD_ENTRIES, AAC_DB_COMMAND_READY},
{AAC_ADAP_HIGH_CMD_ENTRIES, 0},
{AAC_HOST_NORM_RESP_ENTRIES, AAC_DB_RESPONSE_NOT_FULL},
{AAC_HOST_HIGH_RESP_ENTRIES, 0},
{AAC_ADAP_NORM_RESP_ENTRIES, AAC_DB_RESPONSE_READY},
{AAC_ADAP_HIGH_RESP_ENTRIES, 0}
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
};
/*
Reformat for 80 columns. Sorry, but I had to do it.
2001-08-03 00:50:30 +00:00
* Atomically insert an entry into the nominated queue, returns 0 on success or
* EBUSY if the queue is full.
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
*
Major bugfix and minor update. This should resolve the current issues with the driver locking up under load. - Restructure so that we use a static pool of commands/FIBs, rather than allocating them in clusters. The cluster allocation just made things more complicated, and allowed us to waste more memory in peak load situations. - Make queueing macros more like my other drivers. This adds queue stats for free. Add some debugging to take advantage of this. - Reimplement the periodic timeout scan. Kick the interrupt handler and the start routine every scan as well, just to be safe. Track busy commands properly. - Bring resource cleanup into line with resource allocation. We should now clean up correctly after a failed probe/unload/etc. - Try to start new commands when old ones are completed. We weren't doing this before, which could lead to deadlock when the controller was full. - Don't try to build a new command if we have found a deferred command. This could cause us to lose the deferred command. - Use diskerr() to report I/O errors. - Don't bail if the AdapterInfo structure is the wrong size. Some variation seems to be normal. We need to improve our handing of 2.x firmware sets. - Improve some comments in an attempt to try to make things clearer. - Restructure to avoid some warnings.
2000-12-27 13:14:56 +00:00
* Note: it would be more efficient to defer notifying the controller in
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
* the case where we may be inserting several entries in rapid succession,
* but implementing this usefully may be difficult (it would involve a
* separate queue/notify interface).
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
*/
static int
Bugfixes. Close a race and logic bug in the timeout handling, don't call the interrupt handler from the upper half, etc. This fixes some serious stability problems that we were seeing on our production server. These patches have been tested for almost 6 months and are a highly recommended MFC candidate. Reviewed by: gibbs, merry, msmith MFC after: 4 days
2001-08-05 06:20:13 +00:00
aac_enqueue_fib(struct aac_softc *sc, int queue, struct aac_command *cm)
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
{
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
u_int32_t pi, ci;
The aac driver has evolved enough over the last few months that it no longer resembles the 4.x version very much. Garbage collect the legacy bits.
2003-03-01 05:07:19 +00:00
int error;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
u_int32_t fib_size;
u_int32_t fib_addr;
Bugfixes. Close a race and logic bug in the timeout handling, don't call the interrupt handler from the upper half, etc. This fixes some serious stability problems that we were seeing on our production server. These patches have been tested for almost 6 months and are a highly recommended MFC candidate. Reviewed by: gibbs, merry, msmith MFC after: 4 days
2001-08-05 06:20:13 +00:00
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Remove the unused AAC_DRIVER_BUILD_DATE macro. Nuke whitespace at EOL while I'm here. Approved by: scottl (MAINTAINER)
2005-11-13 17:26:36 +00:00
fib_size = cm->cm_fib->Header.Size;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
fib_addr = cm->cm_fib->Header.ReceiverFibAddress;
Mega update to the aac driver. 1. Correctly handle commands initiated by the adapter. These commands are defered to a kthread responsible for their processing, then are properly returned to the controller. 2. Add/remove disk devices when notified by the card that a container was created/deleted/offline. 3. Implement crashdump functionality. 4. Support all ioctls needed for the management CLI to work. The linux version of this app can be found at the Dell or HP website. A native version will be forthcoming. MFC-after: 4.4-RELEASE
2001-08-29 23:34:05 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/* get the producer/consumer indices */
pi = sc->aac_queues->qt_qindex[queue][AAC_PRODUCER_INDEX];
ci = sc->aac_queues->qt_qindex[queue][AAC_CONSUMER_INDEX];
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/* wrap the queue? */
if (pi >= aac_qinfo[queue].size)
pi = 0;
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/* check for queue full */
if ((pi + 1) == ci) {
error = EBUSY;
goto out;
}
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Commit the correct version of the patch from last night. This fixes an immediate panic when doing any i/o, and it closes a completion race.
2004-06-01 15:50:11 +00:00
/*
* To avoid a race with its completion interrupt, place this command on
* the busy queue prior to advertising it to the controller.
*/
aac_enqueue_busy(cm);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/* populate queue entry */
(sc->aac_qentries[queue] + pi)->aq_fib_size = fib_size;
(sc->aac_qentries[queue] + pi)->aq_fib_addr = fib_addr;
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/* update producer index */
sc->aac_queues->qt_qindex[queue][AAC_PRODUCER_INDEX] = pi + 1;
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/* notify the adapter if we know how */
if (aac_qinfo[queue].notify != 0)
AAC_QNOTIFY(sc, aac_qinfo[queue].notify);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
error = 0;
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
out:
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
return(error);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
/*
Mega update to the aac driver. 1. Correctly handle commands initiated by the adapter. These commands are defered to a kthread responsible for their processing, then are properly returned to the controller. 2. Add/remove disk devices when notified by the card that a container was created/deleted/offline. 3. Implement crashdump functionality. 4. Support all ioctls needed for the management CLI to work. The linux version of this app can be found at the Dell or HP website. A native version will be forthcoming. MFC-after: 4.4-RELEASE
2001-08-29 23:34:05 +00:00
* Atomically remove one entry from the nominated queue, returns 0 on
* success or ENOENT if the queue is empty.
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
*/
static int
Reformat for 80 columns. Sorry, but I had to do it.
2001-08-03 00:50:30 +00:00
aac_dequeue_fib(struct aac_softc *sc, int queue, u_int32_t *fib_size,
struct aac_fib **fib_addr)
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
{
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
u_int32_t pi, ci;
Don't pass pointers to kernel data structures through 32-bit fields as 64-bit architectures won't like this. Use virtual array indexes instead. This *should* allow the driver to work on 64-bit platforms, though it's still not endian clean.
2003-04-28 06:16:20 +00:00
u_int32_t fib_index;
The aac driver has evolved enough over the last few months that it no longer resembles the 4.x version very much. Garbage collect the legacy bits.
2003-03-01 05:07:19 +00:00
int error;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
int notify;
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/* get the producer/consumer indices */
pi = sc->aac_queues->qt_qindex[queue][AAC_PRODUCER_INDEX];
ci = sc->aac_queues->qt_qindex[queue][AAC_CONSUMER_INDEX];
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/* check for queue empty */
if (ci == pi) {
error = ENOENT;
goto out;
}
Correctly wrap the producer queue index when dequeuing commands. This wasn't a problem for command responses since we rarely ever filled the queue. However, adapter-initiated commands have a much smaller queue and could tickle this bug. It's possible that this might fix the recently reported problems with the aac-2120s, though I haven't been able to reproduce the problem locally. MFC-After: 1 day
2003-09-16 16:07:15 +00:00
/* wrap the pi so the following test works */
if (pi >= aac_qinfo[queue].size)
pi = 0;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
notify = 0;
if (ci == pi + 1)
notify++;
Bugfixes. Close a race and logic bug in the timeout handling, don't call the interrupt handler from the upper half, etc. This fixes some serious stability problems that we were seeing on our production server. These patches have been tested for almost 6 months and are a highly recommended MFC candidate. Reviewed by: gibbs, merry, msmith MFC after: 4 days
2001-08-05 06:20:13 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/* wrap the queue? */
if (ci >= aac_qinfo[queue].size)
ci = 0;
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/* fetch the entry */
*fib_size = (sc->aac_qentries[queue] + ci)->aq_fib_size;
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Don't pass pointers to kernel data structures through 32-bit fields as 64-bit architectures won't like this. Use virtual array indexes instead. This *should* allow the driver to work on 64-bit platforms, though it's still not endian clean.
2003-04-28 06:16:20 +00:00
switch (queue) {
case AAC_HOST_NORM_CMD_QUEUE:
case AAC_HOST_HIGH_CMD_QUEUE:
/*
* The aq_fib_addr is only 32 bits wide so it can't be counted
* on to hold an address. For AIF's, the adapter assumes
* that it's giving us an address into the array of AIF fibs.
* Therefore, we have to convert it to an index.
*/
fib_index = (sc->aac_qentries[queue] + ci)->aq_fib_addr /
sizeof(struct aac_fib);
*fib_addr = &sc->aac_common->ac_fibs[fib_index];
break;
case AAC_HOST_NORM_RESP_QUEUE:
case AAC_HOST_HIGH_RESP_QUEUE:
{
struct aac_command *cm;
/*
* As above, an index is used instead of an actual address.
* Gotta shift the index to account for the fast response
* bit. No other correction is needed since this value was
* originally provided by the driver via the SenderFibAddress
* field.
*/
fib_index = (sc->aac_qentries[queue] + ci)->aq_fib_addr;
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
cm = sc->aac_commands + (fib_index >> 2);
Don't pass pointers to kernel data structures through 32-bit fields as 64-bit architectures won't like this. Use virtual array indexes instead. This *should* allow the driver to work on 64-bit platforms, though it's still not endian clean.
2003-04-28 06:16:20 +00:00
*fib_addr = cm->cm_fib;
/*
* Is this a fast response? If it is, update the fib fields in
* local memory since the whole fib isn't DMA'd back up.
*/
if (fib_index & 0x01) {
(*fib_addr)->Header.XferState |= AAC_FIBSTATE_DONEADAP;
*((u_int32_t*)((*fib_addr)->data)) = AAC_ERROR_NORMAL;
}
break;
}
default:
panic("Invalid queue in aac_dequeue_fib()");
break;
Major bugfixes for large memory and fast systems. aac.c: Re-arrange the interrupt handler to optimize the common case of the adapter interrupting us because one or more commands are complete, and do a read across the pci bus to ensure that all posted status writes are flushed. This should close a race that could cause command completion interrupts to be lost. Follow the spec a bit closer when filling out command structures. Enable the Fast Response feature to eliminate the need for the card to DMA successfull command completions back into host memory. Tell the controller how much physical memory we have. Without this there was a chance that our DMA regions would collide with the memory window used by the cache on the controller. The result would be massive data corruption. This seemed to mainly affect systems with >2GB of memory. Fix a few whitespace problems. aac_debug.c: Add an extra diagnostic when printing out commands. aac_disk.c: Add extra sanity checks. aacreg.h: Prepare for making this 64-bit clean by reducing the use of enumeration types in structures. Many thanks to Justin Gibbs for helping track these down.
2003-01-11 01:59:21 +00:00
}
Don't pass pointers to kernel data structures through 32-bit fields as 64-bit architectures won't like this. Use virtual array indexes instead. This *should* allow the driver to work on 64-bit platforms, though it's still not endian clean.
2003-04-28 06:16:20 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/* update consumer index */
sc->aac_queues->qt_qindex[queue][AAC_CONSUMER_INDEX] = ci + 1;
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/* if we have made the queue un-full, notify the adapter */
if (notify && (aac_qinfo[queue].notify != 0))
AAC_QNOTIFY(sc, aac_qinfo[queue].notify);
error = 0;
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
out:
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
return(error);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
Mega update to the aac driver. 1. Correctly handle commands initiated by the adapter. These commands are defered to a kthread responsible for their processing, then are properly returned to the controller. 2. Add/remove disk devices when notified by the card that a container was created/deleted/offline. 3. Implement crashdump functionality. 4. Support all ioctls needed for the management CLI to work. The linux version of this app can be found at the Dell or HP website. A native version will be forthcoming. MFC-after: 4.4-RELEASE
2001-08-29 23:34:05 +00:00
* Put our response to an Adapter Initialed Fib on the response queue
*/
static int
aac_enqueue_response(struct aac_softc *sc, int queue, struct aac_fib *fib)
{
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
u_int32_t pi, ci;
The aac driver has evolved enough over the last few months that it no longer resembles the 4.x version very much. Garbage collect the legacy bits.
2003-03-01 05:07:19 +00:00
int error;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
u_int32_t fib_size;
u_int32_t fib_addr;
Mega update to the aac driver. 1. Correctly handle commands initiated by the adapter. These commands are defered to a kthread responsible for their processing, then are properly returned to the controller. 2. Add/remove disk devices when notified by the card that a container was created/deleted/offline. 3. Implement crashdump functionality. 4. Support all ioctls needed for the management CLI to work. The linux version of this app can be found at the Dell or HP website. A native version will be forthcoming. MFC-after: 4.4-RELEASE
2001-08-29 23:34:05 +00:00
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
Mega update to the aac driver. 1. Correctly handle commands initiated by the adapter. These commands are defered to a kthread responsible for their processing, then are properly returned to the controller. 2. Add/remove disk devices when notified by the card that a container was created/deleted/offline. 3. Implement crashdump functionality. 4. Support all ioctls needed for the management CLI to work. The linux version of this app can be found at the Dell or HP website. A native version will be forthcoming. MFC-after: 4.4-RELEASE
2001-08-29 23:34:05 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/* Tell the adapter where the FIB is */
Remove the unused AAC_DRIVER_BUILD_DATE macro. Nuke whitespace at EOL while I'm here. Approved by: scottl (MAINTAINER)
2005-11-13 17:26:36 +00:00
fib_size = fib->Header.Size;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
fib_addr = fib->Header.SenderFibAddress;
fib->Header.ReceiverFibAddress = fib_addr;
Mega update to the aac driver. 1. Correctly handle commands initiated by the adapter. These commands are defered to a kthread responsible for their processing, then are properly returned to the controller. 2. Add/remove disk devices when notified by the card that a container was created/deleted/offline. 3. Implement crashdump functionality. 4. Support all ioctls needed for the management CLI to work. The linux version of this app can be found at the Dell or HP website. A native version will be forthcoming. MFC-after: 4.4-RELEASE
2001-08-29 23:34:05 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/* get the producer/consumer indices */
pi = sc->aac_queues->qt_qindex[queue][AAC_PRODUCER_INDEX];
ci = sc->aac_queues->qt_qindex[queue][AAC_CONSUMER_INDEX];
Mega update to the aac driver. 1. Correctly handle commands initiated by the adapter. These commands are defered to a kthread responsible for their processing, then are properly returned to the controller. 2. Add/remove disk devices when notified by the card that a container was created/deleted/offline. 3. Implement crashdump functionality. 4. Support all ioctls needed for the management CLI to work. The linux version of this app can be found at the Dell or HP website. A native version will be forthcoming. MFC-after: 4.4-RELEASE
2001-08-29 23:34:05 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/* wrap the queue? */
if (pi >= aac_qinfo[queue].size)
pi = 0;
Mega update to the aac driver. 1. Correctly handle commands initiated by the adapter. These commands are defered to a kthread responsible for their processing, then are properly returned to the controller. 2. Add/remove disk devices when notified by the card that a container was created/deleted/offline. 3. Implement crashdump functionality. 4. Support all ioctls needed for the management CLI to work. The linux version of this app can be found at the Dell or HP website. A native version will be forthcoming. MFC-after: 4.4-RELEASE
2001-08-29 23:34:05 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/* check for queue full */
if ((pi + 1) == ci) {
error = EBUSY;
goto out;
}
Mega update to the aac driver. 1. Correctly handle commands initiated by the adapter. These commands are defered to a kthread responsible for their processing, then are properly returned to the controller. 2. Add/remove disk devices when notified by the card that a container was created/deleted/offline. 3. Implement crashdump functionality. 4. Support all ioctls needed for the management CLI to work. The linux version of this app can be found at the Dell or HP website. A native version will be forthcoming. MFC-after: 4.4-RELEASE
2001-08-29 23:34:05 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/* populate queue entry */
(sc->aac_qentries[queue] + pi)->aq_fib_size = fib_size;
(sc->aac_qentries[queue] + pi)->aq_fib_addr = fib_addr;
Mega update to the aac driver. 1. Correctly handle commands initiated by the adapter. These commands are defered to a kthread responsible for their processing, then are properly returned to the controller. 2. Add/remove disk devices when notified by the card that a container was created/deleted/offline. 3. Implement crashdump functionality. 4. Support all ioctls needed for the management CLI to work. The linux version of this app can be found at the Dell or HP website. A native version will be forthcoming. MFC-after: 4.4-RELEASE
2001-08-29 23:34:05 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/* update producer index */
sc->aac_queues->qt_qindex[queue][AAC_PRODUCER_INDEX] = pi + 1;
Mega update to the aac driver. 1. Correctly handle commands initiated by the adapter. These commands are defered to a kthread responsible for their processing, then are properly returned to the controller. 2. Add/remove disk devices when notified by the card that a container was created/deleted/offline. 3. Implement crashdump functionality. 4. Support all ioctls needed for the management CLI to work. The linux version of this app can be found at the Dell or HP website. A native version will be forthcoming. MFC-after: 4.4-RELEASE
2001-08-29 23:34:05 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/* notify the adapter if we know how */
if (aac_qinfo[queue].notify != 0)
AAC_QNOTIFY(sc, aac_qinfo[queue].notify);
Mega update to the aac driver. 1. Correctly handle commands initiated by the adapter. These commands are defered to a kthread responsible for their processing, then are properly returned to the controller. 2. Add/remove disk devices when notified by the card that a container was created/deleted/offline. 3. Implement crashdump functionality. 4. Support all ioctls needed for the management CLI to work. The linux version of this app can be found at the Dell or HP website. A native version will be forthcoming. MFC-after: 4.4-RELEASE
2001-08-29 23:34:05 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
error = 0;
Mega update to the aac driver. 1. Correctly handle commands initiated by the adapter. These commands are defered to a kthread responsible for their processing, then are properly returned to the controller. 2. Add/remove disk devices when notified by the card that a container was created/deleted/offline. 3. Implement crashdump functionality. 4. Support all ioctls needed for the management CLI to work. The linux version of this app can be found at the Dell or HP website. A native version will be forthcoming. MFC-after: 4.4-RELEASE
2001-08-29 23:34:05 +00:00
out:
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
return(error);
Mega update to the aac driver. 1. Correctly handle commands initiated by the adapter. These commands are defered to a kthread responsible for their processing, then are properly returned to the controller. 2. Add/remove disk devices when notified by the card that a container was created/deleted/offline. 3. Implement crashdump functionality. 4. Support all ioctls needed for the management CLI to work. The linux version of this app can be found at the Dell or HP website. A native version will be forthcoming. MFC-after: 4.4-RELEASE
2001-08-29 23:34:05 +00:00
}
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
Major bugfix and minor update. This should resolve the current issues with the driver locking up under load. - Restructure so that we use a static pool of commands/FIBs, rather than allocating them in clusters. The cluster allocation just made things more complicated, and allowed us to waste more memory in peak load situations. - Make queueing macros more like my other drivers. This adds queue stats for free. Add some debugging to take advantage of this. - Reimplement the periodic timeout scan. Kick the interrupt handler and the start routine every scan as well, just to be safe. Track busy commands properly. - Bring resource cleanup into line with resource allocation. We should now clean up correctly after a failed probe/unload/etc. - Try to start new commands when old ones are completed. We weren't doing this before, which could lead to deadlock when the controller was full. - Don't try to build a new command if we have found a deferred command. This could cause us to lose the deferred command. - Use diskerr() to report I/O errors. - Don't bail if the AdapterInfo structure is the wrong size. Some variation seems to be normal. We need to improve our handing of 2.x firmware sets. - Improve some comments in an attempt to try to make things clearer. - Restructure to avoid some warnings.
2000-12-27 13:14:56 +00:00
* Check for commands that have been outstanding for a suspiciously long time,
* and complain about them.
*/
static void
aac_timeout(struct aac_softc *sc)
{
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
struct aac_command *cm;
time_t deadline;
If the timeout handler runs and notices that commands are timed out, check the firmware status register on the card to see if the firmware is still running. There is no way to recover from this, but at least it can give a hint as whether the car has crashed (which happens all too often). MFC after: 3 days
2004-09-16 02:37:40 +00:00
int timedout, code;
Major bugfix and minor update. This should resolve the current issues with the driver locking up under load. - Restructure so that we use a static pool of commands/FIBs, rather than allocating them in clusters. The cluster allocation just made things more complicated, and allowed us to waste more memory in peak load situations. - Make queueing macros more like my other drivers. This adds queue stats for free. Add some debugging to take advantage of this. - Reimplement the periodic timeout scan. Kick the interrupt handler and the start routine every scan as well, just to be safe. Track busy commands properly. - Bring resource cleanup into line with resource allocation. We should now clean up correctly after a failed probe/unload/etc. - Try to start new commands when old ones are completed. We weren't doing this before, which could lead to deadlock when the controller was full. - Don't try to build a new command if we have found a deferred command. This could cause us to lose the deferred command. - Use diskerr() to report I/O errors. - Don't bail if the AdapterInfo structure is the wrong size. Some variation seems to be normal. We need to improve our handing of 2.x firmware sets. - Improve some comments in an attempt to try to make things clearer. - Restructure to avoid some warnings.
2000-12-27 13:14:56 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
Various cleanups: - Move the command timeout check from a separate repeating timeout to the kthread since the kthread is already running periodically. - Move printing the hardware print buffer to the kthread. - Properly shut down the kernel thread on detach. - Detach the child array devices on detach. - Don't issue a controller halt command on detach. Doing so requires a PCI reset to wake the controller back up. The driver can now be unloaded as long as CAM support is not enabled.
2003-02-05 23:24:25 +00:00
* Traverse the busy command list, bitch about late commands once
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
* only.
*/
If the timeout handler runs and notices that commands are timed out, check the firmware status register on the card to see if the firmware is still running. There is no way to recover from this, but at least it can give a hint as whether the car has crashed (which happens all too often). MFC after: 3 days
2004-09-16 02:37:40 +00:00
timedout = 0;
Use time_uptime for timing commands instead of time_second. This is a bit risky because the "current time" is supposed to be fed to the card during initialization, and the current time is supposed to be put into each command that is sent to the card. Hopefully either the card doesn't actually care about the timestamps, or it doesn't care about the absolute values so long and the relative values are consistent. Not an MFC candidate until more thorough testing can be done.
2005-09-14 05:46:28 +00:00
deadline = time_uptime - AAC_CMD_TIMEOUT;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
TAILQ_FOREACH(cm, &sc->aac_busy, cm_link) {
if ((cm->cm_timestamp < deadline)
Avoid repeatedly spamming the console while a timed out command is waiting to complete. Instead, print one message after the timeout period expires, and one more when (if) the command eventually completes. MFC after: 1 month
2010-09-14 01:51:04 +00:00
&& !(cm->cm_flags & AAC_CMD_TIMEDOUT)) {
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
cm->cm_flags |= AAC_CMD_TIMEDOUT;
device_printf(sc->aac_dev,
Include command type in COMMAND TIMEOUT messages to aid in debugging.
2010-02-17 22:28:37 +00:00
"COMMAND %p (TYPE %d) TIMEOUT AFTER %d SECONDS\n",
cm, cm->cm_fib->Header.Command,
(int)(time_uptime-cm->cm_timestamp));
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
AAC_PRINT_FIB(sc, cm->cm_fib);
If the timeout handler runs and notices that commands are timed out, check the firmware status register on the card to see if the firmware is still running. There is no way to recover from this, but at least it can give a hint as whether the car has crashed (which happens all too often). MFC after: 3 days
2004-09-16 02:37:40 +00:00
timedout++;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
}
Major bugfix and minor update. This should resolve the current issues with the driver locking up under load. - Restructure so that we use a static pool of commands/FIBs, rather than allocating them in clusters. The cluster allocation just made things more complicated, and allowed us to waste more memory in peak load situations. - Make queueing macros more like my other drivers. This adds queue stats for free. Add some debugging to take advantage of this. - Reimplement the periodic timeout scan. Kick the interrupt handler and the start routine every scan as well, just to be safe. Track busy commands properly. - Bring resource cleanup into line with resource allocation. We should now clean up correctly after a failed probe/unload/etc. - Try to start new commands when old ones are completed. We weren't doing this before, which could lead to deadlock when the controller was full. - Don't try to build a new command if we have found a deferred command. This could cause us to lose the deferred command. - Use diskerr() to report I/O errors. - Don't bail if the AdapterInfo structure is the wrong size. Some variation seems to be normal. We need to improve our handing of 2.x firmware sets. - Improve some comments in an attempt to try to make things clearer. - Restructure to avoid some warnings.
2000-12-27 13:14:56 +00:00
}
If the timeout handler runs and notices that commands are timed out, check the firmware status register on the card to see if the firmware is still running. There is no way to recover from this, but at least it can give a hint as whether the car has crashed (which happens all too often). MFC after: 3 days
2004-09-16 02:37:40 +00:00
if (timedout) {
code = AAC_GET_FWSTATUS(sc);
if (code != AAC_UP_AND_RUNNING) {
device_printf(sc->aac_dev, "WARNING! Controller is no "
"longer running! code= 0x%x\n", code);
}
}
Major bugfix and minor update. This should resolve the current issues with the driver locking up under load. - Restructure so that we use a static pool of commands/FIBs, rather than allocating them in clusters. The cluster allocation just made things more complicated, and allowed us to waste more memory in peak load situations. - Make queueing macros more like my other drivers. This adds queue stats for free. Add some debugging to take advantage of this. - Reimplement the periodic timeout scan. Kick the interrupt handler and the start routine every scan as well, just to be safe. Track busy commands properly. - Bring resource cleanup into line with resource allocation. We should now clean up correctly after a failed probe/unload/etc. - Try to start new commands when old ones are completed. We weren't doing this before, which could lead to deadlock when the controller was full. - Don't try to build a new command if we have found a deferred command. This could cause us to lose the deferred command. - Use diskerr() to report I/O errors. - Don't bail if the AdapterInfo structure is the wrong size. Some variation seems to be normal. We need to improve our handing of 2.x firmware sets. - Improve some comments in an attempt to try to make things clearer. - Restructure to avoid some warnings.
2000-12-27 13:14:56 +00:00
}
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
* Interface Function Vectors
*/
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
* Read the current firmware status word.
*/
static int
aac_sa_get_fwstatus(struct aac_softc *sc)
{
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
return(AAC_MEM0_GETREG4(sc, AAC_SA_FWSTATUS));
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
static int
aac_rx_get_fwstatus(struct aac_softc *sc)
{
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
SVN rev 188743 modified aac_rx_get_fwstatus to use the AAC_RX_OMR0 register instead of AAC_RX_FWSTATUS, as that is the way it's done in Adaptec's vendor driver and in the Linux drivers. (The same applies to aac_rkt_get_fwstatus as well.) However, a concern has been raised about the compatibility of this change and old hardware / firmware versions. In the absense of specific information, revert to the original behaviour if the firmware does not support the "New comm." interface. Users of old cards or firmware haven't reported the problems that are potentially solved by switching to OMR0.
2009-02-23 18:22:06 +00:00
return(AAC_MEM0_GETREG4(sc, sc->flags & AAC_FLAGS_NEW_COMM ?
AAC_RX_OMR0 : AAC_RX_FWSTATUS));
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
Add support for the Adaptec RAID-On-Chip architecture. This in turn provides support for the Adaptec 2130S adapter. Thanks to Adaptec for providing hardware for this.
2004-08-13 01:44:09 +00:00
static int
aac_rkt_get_fwstatus(struct aac_softc *sc)
{
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
Add support for the Adaptec RAID-On-Chip architecture. This in turn provides support for the Adaptec 2130S adapter. Thanks to Adaptec for providing hardware for this.
2004-08-13 01:44:09 +00:00
SVN rev 188743 modified aac_rx_get_fwstatus to use the AAC_RX_OMR0 register instead of AAC_RX_FWSTATUS, as that is the way it's done in Adaptec's vendor driver and in the Linux drivers. (The same applies to aac_rkt_get_fwstatus as well.) However, a concern has been raised about the compatibility of this change and old hardware / firmware versions. In the absense of specific information, revert to the original behaviour if the firmware does not support the "New comm." interface. Users of old cards or firmware haven't reported the problems that are potentially solved by switching to OMR0.
2009-02-23 18:22:06 +00:00
return(AAC_MEM0_GETREG4(sc, sc->flags & AAC_FLAGS_NEW_COMM ?
AAC_RKT_OMR0 : AAC_RKT_FWSTATUS));
Add support for the Adaptec RAID-On-Chip architecture. This in turn provides support for the Adaptec 2130S adapter. Thanks to Adaptec for providing hardware for this.
2004-08-13 01:44:09 +00:00
}
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
* Notify the controller of a change in a given queue
*/
static void
aac_sa_qnotify(struct aac_softc *sc, int qbit)
{
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
AAC_MEM0_SETREG2(sc, AAC_SA_DOORBELL1_SET, qbit);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
static void
aac_rx_qnotify(struct aac_softc *sc, int qbit)
{
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
AAC_MEM0_SETREG4(sc, AAC_RX_IDBR, qbit);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
Add support for the Adaptec RAID-On-Chip architecture. This in turn provides support for the Adaptec 2130S adapter. Thanks to Adaptec for providing hardware for this.
2004-08-13 01:44:09 +00:00
static void
aac_rkt_qnotify(struct aac_softc *sc, int qbit)
{
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
Add support for the Adaptec RAID-On-Chip architecture. This in turn provides support for the Adaptec 2130S adapter. Thanks to Adaptec for providing hardware for this.
2004-08-13 01:44:09 +00:00
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
AAC_MEM0_SETREG4(sc, AAC_RKT_IDBR, qbit);
Add support for the Adaptec RAID-On-Chip architecture. This in turn provides support for the Adaptec 2130S adapter. Thanks to Adaptec for providing hardware for this.
2004-08-13 01:44:09 +00:00
}
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
* Get the interrupt reason bits
*/
static int
aac_sa_get_istatus(struct aac_softc *sc)
{
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
return(AAC_MEM0_GETREG2(sc, AAC_SA_DOORBELL0));
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
static int
aac_rx_get_istatus(struct aac_softc *sc)
{
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
return(AAC_MEM0_GETREG4(sc, AAC_RX_ODBR));
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
Add support for the Adaptec RAID-On-Chip architecture. This in turn provides support for the Adaptec 2130S adapter. Thanks to Adaptec for providing hardware for this.
2004-08-13 01:44:09 +00:00
static int
aac_rkt_get_istatus(struct aac_softc *sc)
{
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
Add support for the Adaptec RAID-On-Chip architecture. This in turn provides support for the Adaptec 2130S adapter. Thanks to Adaptec for providing hardware for this.
2004-08-13 01:44:09 +00:00
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
return(AAC_MEM0_GETREG4(sc, AAC_RKT_ODBR));
Add support for the Adaptec RAID-On-Chip architecture. This in turn provides support for the Adaptec 2130S adapter. Thanks to Adaptec for providing hardware for this.
2004-08-13 01:44:09 +00:00
}
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
* Clear some interrupt reason bits
*/
static void
aac_sa_clear_istatus(struct aac_softc *sc, int mask)
{
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
AAC_MEM0_SETREG2(sc, AAC_SA_DOORBELL0_CLEAR, mask);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
static void
aac_rx_clear_istatus(struct aac_softc *sc, int mask)
{
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
AAC_MEM0_SETREG4(sc, AAC_RX_ODBR, mask);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
Add support for the Adaptec RAID-On-Chip architecture. This in turn provides support for the Adaptec 2130S adapter. Thanks to Adaptec for providing hardware for this.
2004-08-13 01:44:09 +00:00
static void
aac_rkt_clear_istatus(struct aac_softc *sc, int mask)
{
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
Add support for the Adaptec RAID-On-Chip architecture. This in turn provides support for the Adaptec 2130S adapter. Thanks to Adaptec for providing hardware for this.
2004-08-13 01:44:09 +00:00
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
AAC_MEM0_SETREG4(sc, AAC_RKT_ODBR, mask);
Add support for the Adaptec RAID-On-Chip architecture. This in turn provides support for the Adaptec 2130S adapter. Thanks to Adaptec for providing hardware for this.
2004-08-13 01:44:09 +00:00
}
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
* Populate the mailbox and set the command word
*/
static void
aac_sa_set_mailbox(struct aac_softc *sc, u_int32_t command,
u_int32_t arg0, u_int32_t arg1, u_int32_t arg2, u_int32_t arg3)
{
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
AAC_MEM1_SETREG4(sc, AAC_SA_MAILBOX, command);
AAC_MEM1_SETREG4(sc, AAC_SA_MAILBOX + 4, arg0);
AAC_MEM1_SETREG4(sc, AAC_SA_MAILBOX + 8, arg1);
AAC_MEM1_SETREG4(sc, AAC_SA_MAILBOX + 12, arg2);
AAC_MEM1_SETREG4(sc, AAC_SA_MAILBOX + 16, arg3);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
static void
aac_rx_set_mailbox(struct aac_softc *sc, u_int32_t command,
u_int32_t arg0, u_int32_t arg1, u_int32_t arg2, u_int32_t arg3)
{
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
AAC_MEM1_SETREG4(sc, AAC_RX_MAILBOX, command);
AAC_MEM1_SETREG4(sc, AAC_RX_MAILBOX + 4, arg0);
AAC_MEM1_SETREG4(sc, AAC_RX_MAILBOX + 8, arg1);
AAC_MEM1_SETREG4(sc, AAC_RX_MAILBOX + 12, arg2);
AAC_MEM1_SETREG4(sc, AAC_RX_MAILBOX + 16, arg3);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
Add support for the Adaptec RAID-On-Chip architecture. This in turn provides support for the Adaptec 2130S adapter. Thanks to Adaptec for providing hardware for this.
2004-08-13 01:44:09 +00:00
static void
aac_rkt_set_mailbox(struct aac_softc *sc, u_int32_t command, u_int32_t arg0,
u_int32_t arg1, u_int32_t arg2, u_int32_t arg3)
{
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
Add support for the Adaptec RAID-On-Chip architecture. This in turn provides support for the Adaptec 2130S adapter. Thanks to Adaptec for providing hardware for this.
2004-08-13 01:44:09 +00:00
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
AAC_MEM1_SETREG4(sc, AAC_RKT_MAILBOX, command);
AAC_MEM1_SETREG4(sc, AAC_RKT_MAILBOX + 4, arg0);
AAC_MEM1_SETREG4(sc, AAC_RKT_MAILBOX + 8, arg1);
AAC_MEM1_SETREG4(sc, AAC_RKT_MAILBOX + 12, arg2);
AAC_MEM1_SETREG4(sc, AAC_RKT_MAILBOX + 16, arg3);
Add support for the Adaptec RAID-On-Chip architecture. This in turn provides support for the Adaptec 2130S adapter. Thanks to Adaptec for providing hardware for this.
2004-08-13 01:44:09 +00:00
}
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
* Fetch the immediate command status word
*/
static int
Begin support for 64-bit address support and workarounds for newer cards: - Add data structuress for doing 64-bit scatter/gather - Move busdma tag creations around so that only the parent is created in aac_pci.c. - Retrieve the capabilities word from the firmware before setting up command structures and tags. This allows the driver to decide whether to do 64-bit commands, and if work-arounds are needed for systems with >2GB of RAM. - Only enable the SCSI passthrough if it's enabled in the capabilities word in the firmware. This should fix problems with the 2120S and 2200S cards in systems with more than 2GB of RAM. Full 64-bit support is forthcoming. MFC-After: 1 week
2003-03-26 17:50:11 +00:00
aac_sa_get_mailbox(struct aac_softc *sc, int mb)
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
{
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
return(AAC_MEM1_GETREG4(sc, AAC_SA_MAILBOX + (mb * 4)));
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
static int
Begin support for 64-bit address support and workarounds for newer cards: - Add data structuress for doing 64-bit scatter/gather - Move busdma tag creations around so that only the parent is created in aac_pci.c. - Retrieve the capabilities word from the firmware before setting up command structures and tags. This allows the driver to decide whether to do 64-bit commands, and if work-arounds are needed for systems with >2GB of RAM. - Only enable the SCSI passthrough if it's enabled in the capabilities word in the firmware. This should fix problems with the 2120S and 2200S cards in systems with more than 2GB of RAM. Full 64-bit support is forthcoming. MFC-After: 1 week
2003-03-26 17:50:11 +00:00
aac_rx_get_mailbox(struct aac_softc *sc, int mb)
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
{
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
return(AAC_MEM1_GETREG4(sc, AAC_RX_MAILBOX + (mb * 4)));
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
Add support for the Adaptec RAID-On-Chip architecture. This in turn provides support for the Adaptec 2130S adapter. Thanks to Adaptec for providing hardware for this.
2004-08-13 01:44:09 +00:00
static int
aac_rkt_get_mailbox(struct aac_softc *sc, int mb)
{
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
Add support for the Adaptec RAID-On-Chip architecture. This in turn provides support for the Adaptec 2130S adapter. Thanks to Adaptec for providing hardware for this.
2004-08-13 01:44:09 +00:00
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
return(AAC_MEM1_GETREG4(sc, AAC_RKT_MAILBOX + (mb * 4)));
Add support for the Adaptec RAID-On-Chip architecture. This in turn provides support for the Adaptec 2130S adapter. Thanks to Adaptec for providing hardware for this.
2004-08-13 01:44:09 +00:00
}
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
* Set/clear interrupt masks
*/
static void
aac_sa_set_interrupts(struct aac_softc *sc, int enable)
{
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "%sable interrupts", enable ? "en" : "dis");
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
if (enable) {
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
AAC_MEM0_SETREG2((sc), AAC_SA_MASK0_CLEAR, AAC_DB_INTERRUPTS);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
} else {
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
AAC_MEM0_SETREG2((sc), AAC_SA_MASK0_SET, ~0);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
}
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
static void
aac_rx_set_interrupts(struct aac_softc *sc, int enable)
{
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "%sable interrupts", enable ? "en" : "dis");
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
if (enable) {
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
if (sc->flags & AAC_FLAGS_NEW_COMM)
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
AAC_MEM0_SETREG4(sc, AAC_RX_OIMR, ~AAC_DB_INT_NEW_COMM);
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
else
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
AAC_MEM0_SETREG4(sc, AAC_RX_OIMR, ~AAC_DB_INTERRUPTS);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
} else {
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
AAC_MEM0_SETREG4(sc, AAC_RX_OIMR, ~0);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
}
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
Add support for the Adaptec RAID-On-Chip architecture. This in turn provides support for the Adaptec 2130S adapter. Thanks to Adaptec for providing hardware for this.
2004-08-13 01:44:09 +00:00
static void
aac_rkt_set_interrupts(struct aac_softc *sc, int enable)
{
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "%sable interrupts", enable ? "en" : "dis");
Add support for the Adaptec RAID-On-Chip architecture. This in turn provides support for the Adaptec 2130S adapter. Thanks to Adaptec for providing hardware for this.
2004-08-13 01:44:09 +00:00
if (enable) {
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
if (sc->flags & AAC_FLAGS_NEW_COMM)
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
AAC_MEM0_SETREG4(sc, AAC_RKT_OIMR, ~AAC_DB_INT_NEW_COMM);
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
else
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
AAC_MEM0_SETREG4(sc, AAC_RKT_OIMR, ~AAC_DB_INTERRUPTS);
Add support for the Adaptec RAID-On-Chip architecture. This in turn provides support for the Adaptec 2130S adapter. Thanks to Adaptec for providing hardware for this.
2004-08-13 01:44:09 +00:00
} else {
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
AAC_MEM0_SETREG4(sc, AAC_RKT_OIMR, ~0);
Add support for the Adaptec RAID-On-Chip architecture. This in turn provides support for the Adaptec 2130S adapter. Thanks to Adaptec for providing hardware for this.
2004-08-13 01:44:09 +00:00
}
}
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
/*
* New comm. interface: Send command functions
*/
static int
aac_rx_send_command(struct aac_softc *sc, struct aac_command *cm)
{
u_int32_t index, device;
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "send command (new comm.)");
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
index = AAC_MEM0_GETREG4(sc, AAC_RX_IQUE);
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
if (index == 0xffffffffL)
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
index = AAC_MEM0_GETREG4(sc, AAC_RX_IQUE);
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
if (index == 0xffffffffL)
return index;
aac_enqueue_busy(cm);
device = index;
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
AAC_MEM1_SETREG4(sc, device, (u_int32_t)(cm->cm_fibphys & 0xffffffffUL));
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
device += 4;
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
AAC_MEM1_SETREG4(sc, device, (u_int32_t)(cm->cm_fibphys >> 32));
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
device += 4;
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
AAC_MEM1_SETREG4(sc, device, cm->cm_fib->Header.Size);
AAC_MEM0_SETREG4(sc, AAC_RX_IQUE, index);
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
return 0;
}
static int
aac_rkt_send_command(struct aac_softc *sc, struct aac_command *cm)
{
u_int32_t index, device;
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "send command (new comm.)");
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
index = AAC_MEM0_GETREG4(sc, AAC_RKT_IQUE);
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
if (index == 0xffffffffL)
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
index = AAC_MEM0_GETREG4(sc, AAC_RKT_IQUE);
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
if (index == 0xffffffffL)
return index;
aac_enqueue_busy(cm);
device = index;
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
AAC_MEM1_SETREG4(sc, device, (u_int32_t)(cm->cm_fibphys & 0xffffffffUL));
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
device += 4;
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
AAC_MEM1_SETREG4(sc, device, (u_int32_t)(cm->cm_fibphys >> 32));
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
device += 4;
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
AAC_MEM1_SETREG4(sc, device, cm->cm_fib->Header.Size);
AAC_MEM0_SETREG4(sc, AAC_RKT_IQUE, index);
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
return 0;
}
Remove the unused AAC_DRIVER_BUILD_DATE macro. Nuke whitespace at EOL while I'm here. Approved by: scottl (MAINTAINER)
2005-11-13 17:26:36 +00:00
/*
* New comm. interface: get, set outbound queue index
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
*/
static int
aac_rx_get_outb_queue(struct aac_softc *sc)
{
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
return(AAC_MEM0_GETREG4(sc, AAC_RX_OQUE));
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
}
static int
aac_rkt_get_outb_queue(struct aac_softc *sc)
{
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
return(AAC_MEM0_GETREG4(sc, AAC_RKT_OQUE));
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
}
static void
aac_rx_set_outb_queue(struct aac_softc *sc, int index)
{
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
AAC_MEM0_SETREG4(sc, AAC_RX_OQUE, index);
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
}
static void
aac_rkt_set_outb_queue(struct aac_softc *sc, int index)
{
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
Sync with the official Adaptec vendor driver: [1] Add the support for the NARK controller which seems a variant of the i960Rx. [2] Split up memory regions and other resources in 2 different parts as long as NARK uses them separately (it is not clear to me why though as long as there are no more informations available on this controller). Please note that in all the other cases, the regions overlaps leaving the default behaviour for all the other controllers. [3] Implement a clock daemon responsible for maintain updated the wall clock time of the controller (run any 30 minutes)*. Submitted by: Adaptec (driver build 15317 [1, 2] and 15727 [3]) Reviewed by: emaste Tested by: emaste Sponsored by: Sandvine Incorporated * Please note that originally, in the Adaptec driver, the clock daemon is not implemented with callouts as in our in-tree driver.
2009-02-21 15:40:03 +00:00
AAC_MEM0_SETREG4(sc, AAC_RKT_OQUE, index);
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
}
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
* Debugging and Diagnostics
*/
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
* Print some information about the controller.
*/
static void
aac_describe_controller(struct aac_softc *sc)
{
Prepare for a major update to the aac driver: Update the aac driver with the new crashdump api. Protect sync fibs with a mutex. Align all DMA buffers on a PAGE_SIZE boundary. MFC after: 3 days
2002-04-24 05:12:50 +00:00
struct aac_fib *fib;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
struct aac_adapter_info *info;
Add family support to allow the driver to attach to new devices that share the same interface. Submitted by: Achim Leubner at Adaptec
2007-12-07 18:05:41 +00:00
char *adapter_type = "Adaptec RAID controller";
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
Fix a missing set of lock operations. Submitted by: green PR: 87191
2005-10-10 20:13:31 +00:00
mtx_lock(&sc->aac_io_lock);
Collapse sync fib locking into normal i/o locking. The former didn't protect the registers so it was trivially possible for a sync command and i/o command to fight each other and confuse the controller. Make the sync fib alloc/release functions inline and remove the somewhat worthless AAC_SYNC_LOCK_FORCE flag. Thanks to Adil Katchi for helping me to track this down in RELENG_4.
2004-06-02 18:15:48 +00:00
aac_alloc_sync_fib(sc, &fib);
Prepare for a major update to the aac driver: Update the aac driver with the new crashdump api. Protect sync fibs with a mutex. Align all DMA buffers on a PAGE_SIZE boundary. MFC after: 3 days
2002-04-24 05:12:50 +00:00
fib->data[0] = 0;
if (aac_sync_fib(sc, RequestAdapterInfo, 0, fib, 1)) {
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
device_printf(sc->aac_dev, "RequestAdapterInfo failed\n");
Add a CAM interface to the aac driver. This is useful in case you should ever connect a SCSI Cdrom/Tape/Jukebox/Scanner/Printer/kitty-litter-scooper to your high-end RAID controller. The interface to the arrays is still via the block interface; this merely provides a way to circumvent the RAID functionality and access the SCSI buses directly. Note that for somewhat obvious reasons, hard drives are not exposed to the da driver through this interface, though you can still talk to them via the pass driver. Be the first on your block to low-level format unsuspecting drives that are part of an array! To enable this, add the 'aacp' device to your kernel config. MFC after: 3 days
2002-04-27 01:31:17 +00:00
aac_release_sync_fib(sc);
Fix a missing set of lock operations. Submitted by: green PR: 87191
2005-10-10 20:13:31 +00:00
mtx_unlock(&sc->aac_io_lock);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
return;
}
/* save the kernel revision structure for later use */
Remove the unused AAC_DRIVER_BUILD_DATE macro. Nuke whitespace at EOL while I'm here. Approved by: scottl (MAINTAINER)
2005-11-13 17:26:36 +00:00
info = (struct aac_adapter_info *)&fib->data[0];
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
sc->aac_revision = info->KernelRevision;
Add a CAM interface to the aac driver. This is useful in case you should ever connect a SCSI Cdrom/Tape/Jukebox/Scanner/Printer/kitty-litter-scooper to your high-end RAID controller. The interface to the arrays is still via the block interface; this merely provides a way to circumvent the RAID functionality and access the SCSI buses directly. Note that for somewhat obvious reasons, hard drives are not exposed to the da driver through this interface, though you can still talk to them via the pass driver. Be the first on your block to low-level format unsuspecting drives that are part of an array! To enable this, add the 'aacp' device to your kernel config. MFC after: 3 days
2002-04-27 01:31:17 +00:00
Put some of the probe messages under bootverbose so to lessen the noise.
2004-09-12 03:19:32 +00:00
if (bootverbose) {
Add a text description for the Intel IOP302/303 processors. Be slightly more verbose about the allocation of RAM on the controller. Sbumitted by: Jeremy Chadwick PR: kern/81259 MFC-After: 3 days
2005-06-01 07:11:17 +00:00
device_printf(sc->aac_dev, "%s %dMHz, %dMB memory "
Remove the unused AAC_DRIVER_BUILD_DATE macro. Nuke whitespace at EOL while I'm here. Approved by: scottl (MAINTAINER)
2005-11-13 17:26:36 +00:00
"(%dMB cache, %dMB execution), %s\n",
Put some of the probe messages under bootverbose so to lessen the noise.
2004-09-12 03:19:32 +00:00
aac_describe_code(aac_cpu_variant, info->CpuVariant),
Remove the unused AAC_DRIVER_BUILD_DATE macro. Nuke whitespace at EOL while I'm here. Approved by: scottl (MAINTAINER)
2005-11-13 17:26:36 +00:00
info->ClockSpeed, info->TotalMem / (1024 * 1024),
Add a text description for the Intel IOP302/303 processors. Be slightly more verbose about the allocation of RAM on the controller. Sbumitted by: Jeremy Chadwick PR: kern/81259 MFC-After: 3 days
2005-06-01 07:11:17 +00:00
info->BufferMem / (1024 * 1024),
info->ExecutionMem / (1024 * 1024),
Put some of the probe messages under bootverbose so to lessen the noise.
2004-09-12 03:19:32 +00:00
aac_describe_code(aac_battery_platform,
info->batteryPlatform));
device_printf(sc->aac_dev,
"Kernel %d.%d-%d, Build %d, S/N %6X\n",
info->KernelRevision.external.comp.major,
info->KernelRevision.external.comp.minor,
info->KernelRevision.external.comp.dash,
info->KernelRevision.buildNumber,
(u_int32_t)(info->SerialNumber & 0xffffff));
Begin support for 64-bit address support and workarounds for newer cards: - Add data structuress for doing 64-bit scatter/gather - Move busdma tag creations around so that only the parent is created in aac_pci.c. - Retrieve the capabilities word from the firmware before setting up command structures and tags. This allows the driver to decide whether to do 64-bit commands, and if work-arounds are needed for systems with >2GB of RAM. - Only enable the SCSI passthrough if it's enabled in the capabilities word in the firmware. This should fix problems with the 2120S and 2200S cards in systems with more than 2GB of RAM. Full 64-bit support is forthcoming. MFC-After: 1 week
2003-03-26 17:50:11 +00:00
device_printf(sc->aac_dev, "Supported Options=%b\n",
sc->supported_options,
"\20"
"\1SNAPSHOT"
"\2CLUSTERS"
"\3WCACHE"
"\4DATA64"
"\5HOSTTIME"
"\6RAID50"
"\7WINDOW4GB"
"\10SCSIUPGD"
"\11SOFTERR"
"\12NORECOND"
"\13SGMAP64"
"\14ALARM"
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
"\15NONDASD"
"\16SCSIMGT"
"\17RAIDSCSI"
"\21ADPTINFO"
"\22NEWCOMM"
"\23ARRAY64BIT"
"\24HEATSENSOR");
Begin support for 64-bit address support and workarounds for newer cards: - Add data structuress for doing 64-bit scatter/gather - Move busdma tag creations around so that only the parent is created in aac_pci.c. - Retrieve the capabilities word from the firmware before setting up command structures and tags. This allows the driver to decide whether to do 64-bit commands, and if work-arounds are needed for systems with >2GB of RAM. - Only enable the SCSI passthrough if it's enabled in the capabilities word in the firmware. This should fix problems with the 2120S and 2200S cards in systems with more than 2GB of RAM. Full 64-bit support is forthcoming. MFC-After: 1 week
2003-03-26 17:50:11 +00:00
}
Calling RequestSupplementAdapterInfo before RequestAdapterInfo appears to trip a bug causing the latter to return a zeroed struct aac_adapter_info. This causes two issues. One is cosmetic only -- a verbose boot prints information about the controller, and shows all zero: aac0: Unknown processor 0MHz, 0MB memory (0MB cache, 0MB execution), unknown battery platform The second problem is that the firmware version information is stored away for aac_rev_check, for userland tools (like aaccli) to query via the FSACTL_MINIPORT_REV_CHECK and FSACTL_LNX_MINIPORT_REV_CHECK ioctls. When aaccli encounters this issue it prints Command Error: <The current AFAAPI.DLL is too old to work with the current controller software.> Move the RequestSupplementAdapterInfo call after RequestAdapterInfo, which seems to fix both problems.
2008-04-01 20:53:32 +00:00
if (sc->supported_options & AAC_SUPPORTED_SUPPLEMENT_ADAPTER_INFO) {
fib->data[0] = 0;
if (aac_sync_fib(sc, RequestSupplementAdapterInfo, 0, fib, 1))
device_printf(sc->aac_dev,
"RequestSupplementAdapterInfo failed\n");
else
adapter_type = ((struct aac_supplement_adapter_info *)
&fib->data[0])->AdapterTypeText;
}
device_printf(sc->aac_dev, "%s, aac driver %d.%d.%d-%d\n",
adapter_type,
Diff reduction with Adaptec's vendor driver. Driver version 2.1.9 chosen as that Adaptec version roughly corresponds with the current feature set merged to the in-tree driver.
2010-02-14 17:14:11 +00:00
AAC_DRIVER_MAJOR_VERSION, AAC_DRIVER_MINOR_VERSION,
AAC_DRIVER_BUGFIX_LEVEL, AAC_DRIVER_BUILD);
Calling RequestSupplementAdapterInfo before RequestAdapterInfo appears to trip a bug causing the latter to return a zeroed struct aac_adapter_info. This causes two issues. One is cosmetic only -- a verbose boot prints information about the controller, and shows all zero: aac0: Unknown processor 0MHz, 0MB memory (0MB cache, 0MB execution), unknown battery platform The second problem is that the firmware version information is stored away for aac_rev_check, for userland tools (like aaccli) to query via the FSACTL_MINIPORT_REV_CHECK and FSACTL_LNX_MINIPORT_REV_CHECK ioctls. When aaccli encounters this issue it prints Command Error: <The current AFAAPI.DLL is too old to work with the current controller software.> Move the RequestSupplementAdapterInfo call after RequestAdapterInfo, which seems to fix both problems.
2008-04-01 20:53:32 +00:00
Put some of the probe messages under bootverbose so to lessen the noise.
2004-09-12 03:19:32 +00:00
aac_release_sync_fib(sc);
Fix a missing set of lock operations. Submitted by: green PR: 87191
2005-10-10 20:13:31 +00:00
mtx_unlock(&sc->aac_io_lock);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
* Look up a text description of a numeric error code and return a pointer to
* same.
*/
- Make tables, device ID strings etc const. This includes #ifdef'ing 0 aac_command_status_table, which is actually unused since r111532. While at it, make aac_if a pointer to the now const interface tables instead of copying them over to the softc (this alone already reduces the size of aac.ko on amd64 by ~1 KiB). - Remove redundant softc members. - Use DEVMETHOD_END. - Use NULL instead of 0 for pointers. - Remove redundant bzero(9)'ing of the softc. - Use pci_enable_busmaster(9) instead of duplicating it. - Remove redundant checking for PCIM_CMD_MEMEN (resource allocation will just fail). - Canonicalize the error messages in case of resource allocation failures. - Add support for using MSI instead of INTx, controllable via the tunable hw.aac.enable_msi (defaulting to on). MFC after: 1 month
2013-03-01 19:55:10 +00:00
static const char *
aac_describe_code(const struct aac_code_lookup *table, u_int32_t code)
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
{
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
int i;
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
for (i = 0; table[i].string != NULL; i++)
if (table[i].code == code)
return(table[i].string);
return(table[i + 1].string);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
* Management Interface
*/
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
static int
We no longer need to use d_thread_t, migrate to struct thread *.
2009-05-20 17:29:21 +00:00
aac_open(struct cdev *dev, int flags, int fmt, struct thread *td)
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
{
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
struct aac_softc *sc;
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
sc = dev->si_drv1;
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
Fix bogus busying mechanism from cdevsw callbacks: - D_TRACKCLOSE may be used there as d_close() are expected to match up d_open() calls - Replace the hand-crafted counter and flag with the device_busy()/device_unbusy() proper usage. Sponsored by: Sandvine Incorporated Reported by: Mark Johnston <mjohnston at sandvine dot com> Tested by: Mark Johnston Reviewed by: emaste MFC after: 10 days
2010-09-15 14:24:21 +00:00
device_busy(sc->aac_dev);
Implement device unbusying via a cdevpriv destructor. Suggested by: jhb Tested by: Mark Johnston <mjohnston at sandvine dot com> Reviewed by: emaste, jhb MFC after: 10 days X-MFC: r212661
2010-09-16 17:49:10 +00:00
devfs_set_cdevpriv(sc, aac_cdevpriv_dtor);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
return 0;
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
static int
We no longer need to use d_thread_t, migrate to struct thread *.
2009-05-20 17:29:21 +00:00
aac_ioctl(struct cdev *dev, u_long cmd, caddr_t arg, int flag, struct thread *td)
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
{
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
union aac_statrequest *as;
struct aac_softc *sc;
int error = 0;
as = (union aac_statrequest *)arg;
sc = dev->si_drv1;
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
switch (cmd) {
case AACIO_STATS:
switch (as->as_item) {
case AACQ_FREE:
case AACQ_BIO:
case AACQ_READY:
case AACQ_BUSY:
bcopy(&sc->aac_qstat[as->as_item], &as->as_qstat,
sizeof(struct aac_qstat));
break;
default:
error = ENOENT;
break;
}
Major bugfix and minor update. This should resolve the current issues with the driver locking up under load. - Restructure so that we use a static pool of commands/FIBs, rather than allocating them in clusters. The cluster allocation just made things more complicated, and allowed us to waste more memory in peak load situations. - Make queueing macros more like my other drivers. This adds queue stats for free. Add some debugging to take advantage of this. - Reimplement the periodic timeout scan. Kick the interrupt handler and the start routine every scan as well, just to be safe. Track busy commands properly. - Bring resource cleanup into line with resource allocation. We should now clean up correctly after a failed probe/unload/etc. - Try to start new commands when old ones are completed. We weren't doing this before, which could lead to deadlock when the controller was full. - Don't try to build a new command if we have found a deferred command. This could cause us to lose the deferred command. - Use diskerr() to report I/O errors. - Don't bail if the AdapterInfo structure is the wrong size. Some variation seems to be normal. We need to improve our handing of 2.x firmware sets. - Improve some comments in an attempt to try to make things clearer. - Restructure to avoid some warnings.
2000-12-27 13:14:56 +00:00
break;
Remove the unused AAC_DRIVER_BUILD_DATE macro. Nuke whitespace at EOL while I'm here. Approved by: scottl (MAINTAINER)
2005-11-13 17:26:36 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
case FSACTL_SENDFIB:
Add ioctls FSACTL_SEND_LARGE_FIB, FSACTL_SEND_RAW_SRB, FSACTL_LNX_SEND_LARGE_FIB, and FSACTL_LNX_SEND_RAW_SRB, and correct size checks on FIBs passed in from userspace. Both changes were obtained from Adaptec's driver build 15317. Adaptec's commandline RAID tool arcconf uses these ioctls when creating a RAID-10 array (and probably other operations too).
2008-03-20 17:59:19 +00:00
case FSACTL_SEND_LARGE_FIB:
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
arg = *(caddr_t*)arg;
case FSACTL_LNX_SENDFIB:
Add ioctls FSACTL_SEND_LARGE_FIB, FSACTL_SEND_RAW_SRB, FSACTL_LNX_SEND_LARGE_FIB, and FSACTL_LNX_SEND_RAW_SRB, and correct size checks on FIBs passed in from userspace. Both changes were obtained from Adaptec's driver build 15317. Adaptec's commandline RAID tool arcconf uses these ioctls when creating a RAID-10 array (and probably other operations too).
2008-03-20 17:59:19 +00:00
case FSACTL_LNX_SEND_LARGE_FIB:
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_IOCTL_COMMANDS_B, "FSACTL_SENDFIB");
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
error = aac_ioctl_sendfib(sc, arg);
break;
Add ioctls FSACTL_SEND_LARGE_FIB, FSACTL_SEND_RAW_SRB, FSACTL_LNX_SEND_LARGE_FIB, and FSACTL_LNX_SEND_RAW_SRB, and correct size checks on FIBs passed in from userspace. Both changes were obtained from Adaptec's driver build 15317. Adaptec's commandline RAID tool arcconf uses these ioctls when creating a RAID-10 array (and probably other operations too).
2008-03-20 17:59:19 +00:00
case FSACTL_SEND_RAW_SRB:
arg = *(caddr_t*)arg;
case FSACTL_LNX_SEND_RAW_SRB:
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_IOCTL_COMMANDS_B, "FSACTL_SEND_RAW_SRB");
Add ioctls FSACTL_SEND_LARGE_FIB, FSACTL_SEND_RAW_SRB, FSACTL_LNX_SEND_LARGE_FIB, and FSACTL_LNX_SEND_RAW_SRB, and correct size checks on FIBs passed in from userspace. Both changes were obtained from Adaptec's driver build 15317. Adaptec's commandline RAID tool arcconf uses these ioctls when creating a RAID-10 array (and probably other operations too).
2008-03-20 17:59:19 +00:00
error = aac_ioctl_send_raw_srb(sc, arg);
break;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
case FSACTL_AIF_THREAD:
case FSACTL_LNX_AIF_THREAD:
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_IOCTL_COMMANDS_B, "FSACTL_AIF_THREAD");
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
error = EINVAL;
break;
case FSACTL_OPEN_GET_ADAPTER_FIB:
arg = *(caddr_t*)arg;
Add functionality and fix bugs so the driver will work with soon-to-be released management apps. 1. Implement poll(). This will check for queued aif's so that a subsequent ioctl call to retrieve the next aif will not block. 2. Don't catch signals when sleeping on a fib sent from userland. This causes a race and panic due to the pthread context switcher waking up the tsleep at inopportune times. 3. Fix some whitespace nits. MFC after: 3 days
2001-12-02 03:47:33 +00:00
case FSACTL_LNX_OPEN_GET_ADAPTER_FIB:
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_IOCTL_COMMANDS_B, "FSACTL_OPEN_GET_ADAPTER_FIB");
Allow simultaneous opens of the device for issuing commands to the controller. This is merged from Adaptec driver build 11669.
2007-12-07 00:22:23 +00:00
error = aac_open_aif(sc, arg);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
break;
case FSACTL_GET_NEXT_ADAPTER_FIB:
arg = *(caddr_t*)arg;
case FSACTL_LNX_GET_NEXT_ADAPTER_FIB:
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_IOCTL_COMMANDS_B, "FSACTL_GET_NEXT_ADAPTER_FIB");
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
error = aac_getnext_aif(sc, arg);
break;
case FSACTL_CLOSE_GET_ADAPTER_FIB:
Allow simultaneous opens of the device for issuing commands to the controller. This is merged from Adaptec driver build 11669.
2007-12-07 00:22:23 +00:00
arg = *(caddr_t*)arg;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
case FSACTL_LNX_CLOSE_GET_ADAPTER_FIB:
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_IOCTL_COMMANDS_B, "FSACTL_CLOSE_GET_ADAPTER_FIB");
Allow simultaneous opens of the device for issuing commands to the controller. This is merged from Adaptec driver build 11669.
2007-12-07 00:22:23 +00:00
error = aac_close_aif(sc, arg);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
break;
case FSACTL_MINIPORT_REV_CHECK:
arg = *(caddr_t*)arg;
case FSACTL_LNX_MINIPORT_REV_CHECK:
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_IOCTL_COMMANDS_B, "FSACTL_MINIPORT_REV_CHECK");
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
error = aac_rev_check(sc, arg);
break;
case FSACTL_QUERY_DISK:
arg = *(caddr_t*)arg;
case FSACTL_LNX_QUERY_DISK:
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_IOCTL_COMMANDS_B, "FSACTL_QUERY_DISK");
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
error = aac_query_disk(sc, arg);
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
break;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
case FSACTL_DELETE_DISK:
case FSACTL_LNX_DELETE_DISK:
/*
* We don't trust the underland to tell us when to delete a
Remove the unused AAC_DRIVER_BUILD_DATE macro. Nuke whitespace at EOL while I'm here. Approved by: scottl (MAINTAINER)
2005-11-13 17:26:36 +00:00
* container, rather we rely on an AIF coming from the
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
* controller
*/
error = 0;
break;
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
case FSACTL_GET_PCI_INFO:
arg = *(caddr_t*)arg;
case FSACTL_LNX_GET_PCI_INFO:
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_IOCTL_COMMANDS_B, "FSACTL_GET_PCI_INFO");
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
error = aac_get_pci_info(sc, arg);
break;
Implement FSACTL_LNX_GET_FEATURES and FSACTL_GET_FEATURES ioctls. RAID tools (e.g. arcconf) need this to be able to create arrays larger than 2TB. Submitted by: Adaptec, via driver build 15317
2008-03-28 19:07:25 +00:00
case FSACTL_GET_FEATURES:
arg = *(caddr_t*)arg;
case FSACTL_LNX_GET_FEATURES:
fwprintf(sc, HBA_FLAGS_DBG_IOCTL_COMMANDS_B, "FSACTL_GET_FEATURES");
error = aac_supported_features(sc, arg);
break;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
default:
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_IOCTL_COMMANDS_B, "unsupported cmd 0x%lx\n", cmd);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
error = EINVAL;
break;
}
return(error);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
Add functionality and fix bugs so the driver will work with soon-to-be released management apps. 1. Implement poll(). This will check for queued aif's so that a subsequent ioctl call to retrieve the next aif will not block. 2. Don't catch signals when sleeping on a fib sent from userland. This causes a race and panic due to the pthread context switcher waking up the tsleep at inopportune times. 3. Fix some whitespace nits. MFC after: 3 days
2001-12-02 03:47:33 +00:00
static int
We no longer need to use d_thread_t, migrate to struct thread *.
2009-05-20 17:29:21 +00:00
aac_poll(struct cdev *dev, int poll_events, struct thread *td)
Add functionality and fix bugs so the driver will work with soon-to-be released management apps. 1. Implement poll(). This will check for queued aif's so that a subsequent ioctl call to retrieve the next aif will not block. 2. Don't catch signals when sleeping on a fib sent from userland. This causes a race and panic due to the pthread context switcher waking up the tsleep at inopportune times. 3. Fix some whitespace nits. MFC after: 3 days
2001-12-02 03:47:33 +00:00
{
struct aac_softc *sc;
Fix test for waiting AIFs in aac_poll(). This seems to solve the problem where Adaptec's arcconf monitoring tool hangs after producing its expected output. Submitted by: Adaptec, via driver ver 15317 MFC after: 1 week
2008-06-24 03:26:41 +00:00
struct aac_fib_context *ctx;
Add functionality and fix bugs so the driver will work with soon-to-be released management apps. 1. Implement poll(). This will check for queued aif's so that a subsequent ioctl call to retrieve the next aif will not block. 2. Don't catch signals when sleeping on a fib sent from userland. This causes a race and panic due to the pthread context switcher waking up the tsleep at inopportune times. 3. Fix some whitespace nits. MFC after: 3 days
2001-12-02 03:47:33 +00:00
int revents;
sc = dev->si_drv1;
revents = 0;
Remove the AAC_LOCK macros. They no longer abstract anything and only obfuscate the code. No functional differences.
2004-08-12 05:05:06 +00:00
mtx_lock(&sc->aac_aifq_lock);
Add functionality and fix bugs so the driver will work with soon-to-be released management apps. 1. Implement poll(). This will check for queued aif's so that a subsequent ioctl call to retrieve the next aif will not block. 2. Don't catch signals when sleeping on a fib sent from userland. This causes a race and panic due to the pthread context switcher waking up the tsleep at inopportune times. 3. Fix some whitespace nits. MFC after: 3 days
2001-12-02 03:47:33 +00:00
if ((poll_events & (POLLRDNORM | POLLIN)) != 0) {
Fix test for waiting AIFs in aac_poll(). This seems to solve the problem where Adaptec's arcconf monitoring tool hangs after producing its expected output. Submitted by: Adaptec, via driver ver 15317 MFC after: 1 week
2008-06-24 03:26:41 +00:00
for (ctx = sc->fibctx; ctx; ctx = ctx->next) {
if (ctx->ctx_idx != sc->aifq_idx || ctx->ctx_wrap) {
revents |= poll_events & (POLLIN | POLLRDNORM);
break;
}
}
Add functionality and fix bugs so the driver will work with soon-to-be released management apps. 1. Implement poll(). This will check for queued aif's so that a subsequent ioctl call to retrieve the next aif will not block. 2. Don't catch signals when sleeping on a fib sent from userland. This causes a race and panic due to the pthread context switcher waking up the tsleep at inopportune times. 3. Fix some whitespace nits. MFC after: 3 days
2001-12-02 03:47:33 +00:00
}
Remove the AAC_LOCK macros. They no longer abstract anything and only obfuscate the code. No functional differences.
2004-08-12 05:05:06 +00:00
mtx_unlock(&sc->aac_aifq_lock);
Add functionality and fix bugs so the driver will work with soon-to-be released management apps. 1. Implement poll(). This will check for queued aif's so that a subsequent ioctl call to retrieve the next aif will not block. 2. Don't catch signals when sleeping on a fib sent from userland. This causes a race and panic due to the pthread context switcher waking up the tsleep at inopportune times. 3. Fix some whitespace nits. MFC after: 3 days
2001-12-02 03:47:33 +00:00
if (revents == 0) {
if (poll_events & (POLLIN | POLLRDNORM))
selrecord(td, &sc->rcv_select);
}
return (revents);
}
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
static void
aac_ioctl_event(struct aac_softc *sc, struct aac_event *event, void *arg)
{
switch (event->ev_type) {
case AAC_EVENT_CMFREE:
Do not recursively acquire aac_io_lock in aac_ioctl_event. MFC After: 3 days
2007-12-19 14:33:40 +00:00
mtx_assert(&sc->aac_io_lock, MA_OWNED);
The return value of aac_alloc_command() was misinterpreted in aac_ioctl_event(). Once triggered this would leak away all available commands and starve the rest of the driver. Reviewed by: scottl
2007-02-14 09:10:37 +00:00
if (aac_alloc_command(sc, (struct aac_command **)arg)) {
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
aac_add_event(sc, event);
return;
}
free(event, M_AACBUF);
Ue a better msleep identifier. Fix some whitespace.
2005-10-08 22:41:57 +00:00
wakeup(arg);
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
break;
default:
break;
}
}
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
* Send a FIB supplied from userspace
*/
static int
aac_ioctl_sendfib(struct aac_softc *sc, caddr_t ufib)
{
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
struct aac_command *cm;
int size, error;
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
cm = NULL;
/*
* Get a command
*/
Remove the AAC_LOCK macros. They no longer abstract anything and only obfuscate the code. No functional differences.
2004-08-12 05:05:06 +00:00
mtx_lock(&sc->aac_io_lock);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
if (aac_alloc_command(sc, &cm)) {
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
struct aac_event *event;
event = malloc(sizeof(struct aac_event), M_AACBUF,
M_NOWAIT | M_ZERO);
if (event == NULL) {
error = EBUSY;
Avoid holding the aac_io_lock over copyout. Submitted by: Achim Leubner @ Adaptec.
2007-12-20 21:13:58 +00:00
mtx_unlock(&sc->aac_io_lock);
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
goto out;
}
event->ev_type = AAC_EVENT_CMFREE;
event->ev_callback = aac_ioctl_event;
event->ev_arg = &cm;
aac_add_event(sc, event);
Ue a better msleep identifier. Fix some whitespace.
2005-10-08 22:41:57 +00:00
msleep(&cm, &sc->aac_io_lock, 0, "sendfib", 0);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
}
Rearrange locking in the alloc_commands and ioctl paths to avoid problems.
2006-04-08 06:05:29 +00:00
mtx_unlock(&sc->aac_io_lock);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
* Fetch the FIB header, then re-copy to get data as well.
*/
if ((error = copyin(ufib, cm->cm_fib,
sizeof(struct aac_fib_header))) != 0)
goto out;
size = cm->cm_fib->Header.Size + sizeof(struct aac_fib_header);
Add ioctls FSACTL_SEND_LARGE_FIB, FSACTL_SEND_RAW_SRB, FSACTL_LNX_SEND_LARGE_FIB, and FSACTL_LNX_SEND_RAW_SRB, and correct size checks on FIBs passed in from userspace. Both changes were obtained from Adaptec's driver build 15317. Adaptec's commandline RAID tool arcconf uses these ioctls when creating a RAID-10 array (and probably other operations too).
2008-03-20 17:59:19 +00:00
if (size > sc->aac_max_fib_size) {
device_printf(sc->aac_dev, "incoming FIB oversized (%d > %d)\n",
size, sc->aac_max_fib_size);
size = sc->aac_max_fib_size;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
}
if ((error = copyin(ufib, cm->cm_fib, size)) != 0)
goto out;
cm->cm_fib->Header.Size = size;
Use time_uptime for timing commands instead of time_second. This is a bit risky because the "current time" is supposed to be fed to the card during initialization, and the current time is supposed to be put into each command that is sent to the card. Hopefully either the card doesn't actually care about the timestamps, or it doesn't care about the absolute values so long and the relative values are consistent. Not an MFC candidate until more thorough testing can be done.
2005-09-14 05:46:28 +00:00
cm->cm_timestamp = time_uptime;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
* Pass the FIB to the controller, wait for it to complete.
*/
Rearrange locking in the alloc_commands and ioctl paths to avoid problems.
2006-04-08 06:05:29 +00:00
mtx_lock(&sc->aac_io_lock);
Avoid holding the aac_io_lock over copyout. Submitted by: Achim Leubner @ Adaptec.
2007-12-20 21:13:58 +00:00
error = aac_wait_command(cm);
mtx_unlock(&sc->aac_io_lock);
if (error != 0) {
Various cleanups: - Move the command timeout check from a separate repeating timeout to the kthread since the kthread is already running periodically. - Move printing the hardware print buffer to the kthread. - Properly shut down the kernel thread on detach. - Detach the child array devices on detach. - Don't issue a controller halt command on detach. Doing so requires a PCI reset to wake the controller back up. The driver can now be unloaded as long as CAM support is not enabled.
2003-02-05 23:24:25 +00:00
device_printf(sc->aac_dev,
"aac_wait_command return %d\n", error);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
goto out;
Add functionality and fix bugs so the driver will work with soon-to-be released management apps. 1. Implement poll(). This will check for queued aif's so that a subsequent ioctl call to retrieve the next aif will not block. 2. Don't catch signals when sleeping on a fib sent from userland. This causes a race and panic due to the pthread context switcher waking up the tsleep at inopportune times. 3. Fix some whitespace nits. MFC after: 3 days
2001-12-02 03:47:33 +00:00
}
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
* Copy the FIB and data back out to the caller.
*/
size = cm->cm_fib->Header.Size;
Add ioctls FSACTL_SEND_LARGE_FIB, FSACTL_SEND_RAW_SRB, FSACTL_LNX_SEND_LARGE_FIB, and FSACTL_LNX_SEND_RAW_SRB, and correct size checks on FIBs passed in from userspace. Both changes were obtained from Adaptec's driver build 15317. Adaptec's commandline RAID tool arcconf uses these ioctls when creating a RAID-10 array (and probably other operations too).
2008-03-20 17:59:19 +00:00
if (size > sc->aac_max_fib_size) {
device_printf(sc->aac_dev, "outbound FIB oversized (%d > %d)\n",
size, sc->aac_max_fib_size);
size = sc->aac_max_fib_size;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
}
error = copyout(cm->cm_fib, ufib, size);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
out:
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
if (cm != NULL) {
Avoid holding the aac_io_lock over copyout. Submitted by: Achim Leubner @ Adaptec.
2007-12-20 21:13:58 +00:00
mtx_lock(&sc->aac_io_lock);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
aac_release_command(cm);
Avoid holding the aac_io_lock over copyout. Submitted by: Achim Leubner @ Adaptec.
2007-12-20 21:13:58 +00:00
mtx_unlock(&sc->aac_io_lock);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
}
return(error);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
Add ioctls FSACTL_SEND_LARGE_FIB, FSACTL_SEND_RAW_SRB, FSACTL_LNX_SEND_LARGE_FIB, and FSACTL_LNX_SEND_RAW_SRB, and correct size checks on FIBs passed in from userspace. Both changes were obtained from Adaptec's driver build 15317. Adaptec's commandline RAID tool arcconf uses these ioctls when creating a RAID-10 array (and probably other operations too).
2008-03-20 17:59:19 +00:00
/*
* Send a passthrough FIB supplied from userspace
*/
static int
aac_ioctl_send_raw_srb(struct aac_softc *sc, caddr_t arg)
{
Checkin a facility for specifying a passthrough FIB from userland. arcconf tool by Adaptec already seems to use for identifying the Serial Number of the devices. Some simple things (like FIB setup and bound checks) are retrieved from the Adaptec's driver, but this implementation is quite different because it does use the normal buffer dmat area for loading segments and not a special one (like the Adaptec's one does). Sponsored by: Sandvine Incorporated Discussed with: emaste, scottl Reviewed by: emaste, scottl MFC: 2 weeks
2010-03-14 22:38:18 +00:00
struct aac_command *cm;
struct aac_event *event;
struct aac_fib *fib;
struct aac_srb *srbcmd, *user_srb;
struct aac_sg_entry *sge;
struct aac_sg_entry64 *sge64;
void *srb_sg_address, *ureply;
uint32_t fibsize, srb_sg_bytecount;
int error, transfer_data;
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
cm = NULL;
transfer_data = 0;
fibsize = 0;
user_srb = (struct aac_srb *)arg;
mtx_lock(&sc->aac_io_lock);
if (aac_alloc_command(sc, &cm)) {
event = malloc(sizeof(struct aac_event), M_AACBUF,
M_NOWAIT | M_ZERO);
if (event == NULL) {
error = EBUSY;
mtx_unlock(&sc->aac_io_lock);
goto out;
}
event->ev_type = AAC_EVENT_CMFREE;
event->ev_callback = aac_ioctl_event;
event->ev_arg = &cm;
aac_add_event(sc, event);
msleep(cm, &sc->aac_io_lock, 0, "aacraw", 0);
}
mtx_unlock(&sc->aac_io_lock);
cm->cm_data = NULL;
fib = cm->cm_fib;
srbcmd = (struct aac_srb *)fib->data;
error = copyin(&user_srb->data_len, &fibsize, sizeof(uint32_t));
if (error != 0)
goto out;
if (fibsize > (sc->aac_max_fib_size - sizeof(struct aac_fib_header))) {
error = EINVAL;
goto out;
}
error = copyin(user_srb, srbcmd, fibsize);
if (error != 0)
goto out;
srbcmd->function = 0;
srbcmd->retry_limit = 0;
if (srbcmd->sg_map.SgCount > 1) {
error = EINVAL;
goto out;
}
/* Retrieve correct SG entries. */
if (fibsize == (sizeof(struct aac_srb) +
srbcmd->sg_map.SgCount * sizeof(struct aac_sg_entry))) {
aacraid(4): Sanely copyin userland pointers and ensure that we don't get anything janky from a user. (cturt) aac(4): landergriffith+freebsdbugzilla@gmail.com pointed out that aacraid(4) had the same issue and handling of pointers, so let's change that too. PR: 206573 Submitted by: cturt@hardenedbsd.org Obtained from: HardenedBSD MFC after: 1 week
2016-04-19 18:27:28 +00:00
struct aac_sg_entry sg;
Checkin a facility for specifying a passthrough FIB from userland. arcconf tool by Adaptec already seems to use for identifying the Serial Number of the devices. Some simple things (like FIB setup and bound checks) are retrieved from the Adaptec's driver, but this implementation is quite different because it does use the normal buffer dmat area for loading segments and not a special one (like the Adaptec's one does). Sponsored by: Sandvine Incorporated Discussed with: emaste, scottl Reviewed by: emaste, scottl MFC: 2 weeks
2010-03-14 22:38:18 +00:00
sge = srbcmd->sg_map.SgEntry;
sge64 = NULL;
aacraid(4): Sanely copyin userland pointers and ensure that we don't get anything janky from a user. (cturt) aac(4): landergriffith+freebsdbugzilla@gmail.com pointed out that aacraid(4) had the same issue and handling of pointers, so let's change that too. PR: 206573 Submitted by: cturt@hardenedbsd.org Obtained from: HardenedBSD MFC after: 1 week
2016-04-19 18:27:28 +00:00
if ((error = copyin(sge, &sg, sizeof(sg))) != 0)
goto out;
srb_sg_bytecount = sg.SgByteCount;
srb_sg_address = (void *)(uintptr_t)sg.SgAddress;
Checkin a facility for specifying a passthrough FIB from userland. arcconf tool by Adaptec already seems to use for identifying the Serial Number of the devices. Some simple things (like FIB setup and bound checks) are retrieved from the Adaptec's driver, but this implementation is quite different because it does use the normal buffer dmat area for loading segments and not a special one (like the Adaptec's one does). Sponsored by: Sandvine Incorporated Discussed with: emaste, scottl Reviewed by: emaste, scottl MFC: 2 weeks
2010-03-14 22:38:18 +00:00
}
#ifdef __amd64__
else if (fibsize == (sizeof(struct aac_srb) +
srbcmd->sg_map.SgCount * sizeof(struct aac_sg_entry64))) {
aacraid(4): Sanely copyin userland pointers and ensure that we don't get anything janky from a user. (cturt) aac(4): landergriffith+freebsdbugzilla@gmail.com pointed out that aacraid(4) had the same issue and handling of pointers, so let's change that too. PR: 206573 Submitted by: cturt@hardenedbsd.org Obtained from: HardenedBSD MFC after: 1 week
2016-04-19 18:27:28 +00:00
struct aac_sg_entry64 sg;
Checkin a facility for specifying a passthrough FIB from userland. arcconf tool by Adaptec already seems to use for identifying the Serial Number of the devices. Some simple things (like FIB setup and bound checks) are retrieved from the Adaptec's driver, but this implementation is quite different because it does use the normal buffer dmat area for loading segments and not a special one (like the Adaptec's one does). Sponsored by: Sandvine Incorporated Discussed with: emaste, scottl Reviewed by: emaste, scottl MFC: 2 weeks
2010-03-14 22:38:18 +00:00
sge = NULL;
sge64 = (struct aac_sg_entry64 *)srbcmd->sg_map.SgEntry;
aacraid(4): Sanely copyin userland pointers and ensure that we don't get anything janky from a user. (cturt) aac(4): landergriffith+freebsdbugzilla@gmail.com pointed out that aacraid(4) had the same issue and handling of pointers, so let's change that too. PR: 206573 Submitted by: cturt@hardenedbsd.org Obtained from: HardenedBSD MFC after: 1 week
2016-04-19 18:27:28 +00:00
if ((error = copyin(sge64, &sg, sizeof(sg))) != 0)
goto out;
srb_sg_bytecount = sg.SgByteCount;
srb_sg_address = (void *)sg.SgAddress;
Checkin a facility for specifying a passthrough FIB from userland. arcconf tool by Adaptec already seems to use for identifying the Serial Number of the devices. Some simple things (like FIB setup and bound checks) are retrieved from the Adaptec's driver, but this implementation is quite different because it does use the normal buffer dmat area for loading segments and not a special one (like the Adaptec's one does). Sponsored by: Sandvine Incorporated Discussed with: emaste, scottl Reviewed by: emaste, scottl MFC: 2 weeks
2010-03-14 22:38:18 +00:00
if (sge64->SgAddress > 0xffffffffull &&
(sc->flags & AAC_FLAGS_SG_64BIT) == 0) {
error = EINVAL;
goto out;
}
}
#endif
else {
error = EINVAL;
goto out;
}
ureply = (char *)arg + fibsize;
srbcmd->data_len = srb_sg_bytecount;
if (srbcmd->sg_map.SgCount == 1)
transfer_data = 1;
cm->cm_sgtable = (struct aac_sg_table *)&srbcmd->sg_map;
if (transfer_data) {
cm->cm_datalen = srb_sg_bytecount;
cm->cm_data = malloc(cm->cm_datalen, M_AACBUF, M_NOWAIT);
if (cm->cm_data == NULL) {
error = ENOMEM;
goto out;
}
if (srbcmd->flags & AAC_SRB_FLAGS_DATA_IN)
cm->cm_flags |= AAC_CMD_DATAIN;
if (srbcmd->flags & AAC_SRB_FLAGS_DATA_OUT) {
cm->cm_flags |= AAC_CMD_DATAOUT;
error = copyin(srb_sg_address, cm->cm_data,
cm->cm_datalen);
if (error != 0)
goto out;
}
}
fib->Header.Size = sizeof(struct aac_fib_header) +
sizeof(struct aac_srb);
fib->Header.XferState =
AAC_FIBSTATE_HOSTOWNED |
AAC_FIBSTATE_INITIALISED |
AAC_FIBSTATE_EMPTY |
AAC_FIBSTATE_FROMHOST |
AAC_FIBSTATE_REXPECTED |
AAC_FIBSTATE_NORM |
AAC_FIBSTATE_ASYNC |
AAC_FIBSTATE_FAST_RESPONSE;
fib->Header.Command = (sc->flags & AAC_FLAGS_SG_64BIT) != 0 ?
ScsiPortCommandU64 : ScsiPortCommand;
mtx_lock(&sc->aac_io_lock);
aac_wait_command(cm);
mtx_unlock(&sc->aac_io_lock);
if (transfer_data && (srbcmd->flags & AAC_SRB_FLAGS_DATA_IN) != 0) {
error = copyout(cm->cm_data, srb_sg_address, cm->cm_datalen);
if (error != 0)
goto out;
}
error = copyout(fib->data, ureply, sizeof(struct aac_srb_response));
out:
if (cm != NULL) {
if (cm->cm_data != NULL)
free(cm->cm_data, M_AACBUF);
mtx_lock(&sc->aac_io_lock);
aac_release_command(cm);
mtx_unlock(&sc->aac_io_lock);
}
return(error);
Add ioctls FSACTL_SEND_LARGE_FIB, FSACTL_SEND_RAW_SRB, FSACTL_LNX_SEND_LARGE_FIB, and FSACTL_LNX_SEND_RAW_SRB, and correct size checks on FIBs passed in from userspace. Both changes were obtained from Adaptec's driver build 15317. Adaptec's commandline RAID tool arcconf uses these ioctls when creating a RAID-10 array (and probably other operations too).
2008-03-20 17:59:19 +00:00
}
Implement device unbusying via a cdevpriv destructor. Suggested by: jhb Tested by: Mark Johnston <mjohnston at sandvine dot com> Reviewed by: emaste, jhb MFC after: 10 days X-MFC: r212661
2010-09-16 17:49:10 +00:00
/*
* cdevpriv interface private destructor.
*/
static void
aac_cdevpriv_dtor(void *arg)
{
struct aac_softc *sc;
sc = arg;
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
mtx_lock(&Giant);
device_unbusy(sc->aac_dev);
mtx_unlock(&Giant);
}
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
* Handle an AIF sent to us by the controller; queue it for later reference.
Mega update to the aac driver. 1. Correctly handle commands initiated by the adapter. These commands are defered to a kthread responsible for their processing, then are properly returned to the controller. 2. Add/remove disk devices when notified by the card that a container was created/deleted/offline. 3. Implement crashdump functionality. 4. Support all ioctls needed for the management CLI to work. The linux version of this app can be found at the Dell or HP website. A native version will be forthcoming. MFC-after: 4.4-RELEASE
2001-08-29 23:34:05 +00:00
* If the queue fills up, then drop the older entries.
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
*/
static void
Mega update to the aac driver. 1. Correctly handle commands initiated by the adapter. These commands are defered to a kthread responsible for their processing, then are properly returned to the controller. 2. Add/remove disk devices when notified by the card that a container was created/deleted/offline. 3. Implement crashdump functionality. 4. Support all ioctls needed for the management CLI to work. The linux version of this app can be found at the Dell or HP website. A native version will be forthcoming. MFC-after: 4.4-RELEASE
2001-08-29 23:34:05 +00:00
aac_handle_aif(struct aac_softc *sc, struct aac_fib *fib)
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
{
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
struct aac_aif_command *aif;
struct aac_container *co, *co_next;
Allow simultaneous opens of the device for issuing commands to the controller. This is merged from Adaptec driver build 11669.
2007-12-07 00:22:23 +00:00
struct aac_fib_context *ctx;
Diff reduction to Adaptec driver build 15317 (refactoring and code shuffling): - Resource allocation in aac_alloc (moved from from aac_init) - Interrupt setup in aac_setup_intr (from aac_attach) - Container probing in aac_get_container_info (from aac_startup and aac_handle_aif) - Firmware status check moved to aac_check_firmware from aac_init
2008-03-24 16:38:47 +00:00
struct aac_mntinforesp *mir;
Allow simultaneous opens of the device for issuing commands to the controller. This is merged from Adaptec driver build 11669.
2007-12-07 00:22:23 +00:00
int next, current, found;
When scanning for changed containers, don't assume that the controller will respond in a sane manner. Thanks to Petri Helenius <pete@he.iki.fi> for spotting this and pestering me to fix it.
2003-06-03 02:10:55 +00:00
int count = 0, added = 0, i = 0;
Previously, the aac driver did not handle enclosure management AIFs, which were raised during hot-swap events. Now such events trigger cam rescans, as is done in the mps driver. Submitted by: Mark Johnston <mjohnston at sandvine dot com>
2010-09-29 14:22:00 +00:00
uint32_t channel;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
aif = (struct aac_aif_command*)&fib->data[0];
aac_print_aif(sc, aif);
/* Is it an event that we should care about? */
switch (aif->command) {
case AifCmdEventNotify:
switch (aif->data.EN.type) {
case AifEnAddContainer:
case AifEnDeleteContainer:
/*
Remove the unused AAC_DRIVER_BUILD_DATE macro. Nuke whitespace at EOL while I'm here. Approved by: scottl (MAINTAINER)
2005-11-13 17:26:36 +00:00
* A container was added or deleted, but the message
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
* doesn't tell us anything else! Re-enumerate the
* containers and sort things out.
*/
Collapse sync fib locking into normal i/o locking. The former didn't protect the registers so it was trivially possible for a sync command and i/o command to fight each other and confuse the controller. Make the sync fib alloc/release functions inline and remove the somewhat worthless AAC_SYNC_LOCK_FORCE flag. Thanks to Adil Katchi for helping me to track this down in RELENG_4.
2004-06-02 18:15:48 +00:00
aac_alloc_sync_fib(sc, &fib);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
do {
/*
* Ask the controller for its containers one at
* a time.
* XXX What if the controller's list changes
* midway through this enumaration?
* XXX This should be done async.
*/
Diff reduction to Adaptec driver build 15317 (refactoring and code shuffling): - Resource allocation in aac_alloc (moved from from aac_init) - Interrupt setup in aac_setup_intr (from aac_attach) - Container probing in aac_get_container_info (from aac_startup and aac_handle_aif) - Firmware status check moved to aac_check_firmware from aac_init
2008-03-24 16:38:47 +00:00
if ((mir = aac_get_container_info(sc, fib, i)) == NULL)
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
continue;
Diff reduction to Adaptec driver build 15317 (refactoring and code shuffling): - Resource allocation in aac_alloc (moved from from aac_init) - Interrupt setup in aac_setup_intr (from aac_attach) - Container probing in aac_get_container_info (from aac_startup and aac_handle_aif) - Firmware status check moved to aac_check_firmware from aac_init
2008-03-24 16:38:47 +00:00
if (i == 0)
count = mir->MntRespCount;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
* Check the container against our list.
* co->co_found was already set to 0 in a
* previous run.
*/
Prepare for a major update to the aac driver: Update the aac driver with the new crashdump api. Protect sync fibs with a mutex. Align all DMA buffers on a PAGE_SIZE boundary. MFC after: 3 days
2002-04-24 05:12:50 +00:00
if ((mir->Status == ST_OK) &&
(mir->MntTable[0].VolType != CT_NONE)) {
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
found = 0;
TAILQ_FOREACH(co,
Remove the unused AAC_DRIVER_BUILD_DATE macro. Nuke whitespace at EOL while I'm here. Approved by: scottl (MAINTAINER)
2005-11-13 17:26:36 +00:00
&sc->aac_container_tqh,
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
co_link) {
if (co->co_mntobj.ObjectId ==
Prepare for a major update to the aac driver: Update the aac driver with the new crashdump api. Protect sync fibs with a mutex. Align all DMA buffers on a PAGE_SIZE boundary. MFC after: 3 days
2002-04-24 05:12:50 +00:00
mir->MntTable[0].ObjectId) {
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
co->co_found = 1;
found = 1;
break;
}
}
/*
* If the container matched, continue
* in the list.
*/
if (found) {
i++;
continue;
}
/*
* This is a new container. Do all the
Various cleanups: - Move the command timeout check from a separate repeating timeout to the kthread since the kthread is already running periodically. - Move printing the hardware print buffer to the kthread. - Properly shut down the kernel thread on detach. - Detach the child array devices on detach. - Don't issue a controller halt command on detach. Doing so requires a PCI reset to wake the controller back up. The driver can now be unloaded as long as CAM support is not enabled.
2003-02-05 23:24:25 +00:00
* appropriate things to set it up.
*/
Prepare for a major update to the aac driver: Update the aac driver with the new crashdump api. Protect sync fibs with a mutex. Align all DMA buffers on a PAGE_SIZE boundary. MFC after: 3 days
2002-04-24 05:12:50 +00:00
aac_add_container(sc, mir, 1);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
added = 1;
}
i++;
When scanning for changed containers, don't assume that the controller will respond in a sane manner. Thanks to Petri Helenius <pete@he.iki.fi> for spotting this and pestering me to fix it.
2003-06-03 02:10:55 +00:00
} while ((i < count) && (i < AAC_MAX_CONTAINERS));
Prepare for a major update to the aac driver: Update the aac driver with the new crashdump api. Protect sync fibs with a mutex. Align all DMA buffers on a PAGE_SIZE boundary. MFC after: 3 days
2002-04-24 05:12:50 +00:00
aac_release_sync_fib(sc);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
* Go through our list of containers and see which ones
* were not marked 'found'. Since the controller didn't
* list them they must have been deleted. Do the
* appropriate steps to destroy the device. Also reset
* the co->co_found field.
*/
co = TAILQ_FIRST(&sc->aac_container_tqh);
while (co != NULL) {
if (co->co_found == 0) {
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
mtx_unlock(&sc->aac_io_lock);
Temporarily revert the new-bus locking for 8.0 release. It will be reintroduced after HEAD is reopened for commits by re@. Approved by: re (kib), attilio
2009-08-20 19:17:53 +00:00
mtx_lock(&Giant);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
device_delete_child(sc->aac_dev,
co->co_disk);
Temporarily revert the new-bus locking for 8.0 release. It will be reintroduced after HEAD is reopened for commits by re@. Approved by: re (kib), attilio
2009-08-20 19:17:53 +00:00
mtx_unlock(&Giant);
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
mtx_lock(&sc->aac_io_lock);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
co_next = TAILQ_NEXT(co, co_link);
Remove the AAC_LOCK macros. They no longer abstract anything and only obfuscate the code. No functional differences.
2004-08-12 05:05:06 +00:00
mtx_lock(&sc->aac_container_lock);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
TAILQ_REMOVE(&sc->aac_container_tqh, co,
co_link);
Remove the AAC_LOCK macros. They no longer abstract anything and only obfuscate the code. No functional differences.
2004-08-12 05:05:06 +00:00
mtx_unlock(&sc->aac_container_lock);
Change FREE() to free()
2004-08-12 05:09:17 +00:00
free(co, M_AACBUF);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
co = co_next;
} else {
co->co_found = 0;
co = TAILQ_NEXT(co, co_link);
}
Mega update to the aac driver. 1. Correctly handle commands initiated by the adapter. These commands are defered to a kthread responsible for their processing, then are properly returned to the controller. 2. Add/remove disk devices when notified by the card that a container was created/deleted/offline. 3. Implement crashdump functionality. 4. Support all ioctls needed for the management CLI to work. The linux version of this app can be found at the Dell or HP website. A native version will be forthcoming. MFC-after: 4.4-RELEASE
2001-08-29 23:34:05 +00:00
}
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/* Attach the newly created containers */
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
if (added) {
mtx_unlock(&sc->aac_io_lock);
Temporarily revert the new-bus locking for 8.0 release. It will be reintroduced after HEAD is reopened for commits by re@. Approved by: re (kib), attilio
2009-08-20 19:17:53 +00:00
mtx_lock(&Giant);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
bus_generic_attach(sc->aac_dev);
Temporarily revert the new-bus locking for 8.0 release. It will be reintroduced after HEAD is reopened for commits by re@. Approved by: re (kib), attilio
2009-08-20 19:17:53 +00:00
mtx_unlock(&Giant);
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
mtx_lock(&sc->aac_io_lock);
}
Remove the unused AAC_DRIVER_BUILD_DATE macro. Nuke whitespace at EOL while I'm here. Approved by: scottl (MAINTAINER)
2005-11-13 17:26:36 +00:00
Fix mis-indentation. Approved by: scottl
2002-10-20 16:00:43 +00:00
break;
Mega update to the aac driver. 1. Correctly handle commands initiated by the adapter. These commands are defered to a kthread responsible for their processing, then are properly returned to the controller. 2. Add/remove disk devices when notified by the card that a container was created/deleted/offline. 3. Implement crashdump functionality. 4. Support all ioctls needed for the management CLI to work. The linux version of this app can be found at the Dell or HP website. A native version will be forthcoming. MFC-after: 4.4-RELEASE
2001-08-29 23:34:05 +00:00
Previously, the aac driver did not handle enclosure management AIFs, which were raised during hot-swap events. Now such events trigger cam rescans, as is done in the mps driver. Submitted by: Mark Johnston <mjohnston at sandvine dot com>
2010-09-29 14:22:00 +00:00
case AifEnEnclosureManagement:
switch (aif->data.EN.data.EEE.eventType) {
case AIF_EM_DRIVE_INSERTION:
case AIF_EM_DRIVE_REMOVAL:
channel = aif->data.EN.data.EEE.unitID;
if (sc->cam_rescan_cb != NULL)
sc->cam_rescan_cb(sc,
(channel >> 24) & 0xF,
(channel & 0xFFFF));
break;
}
break;
case AifEnAddJBOD:
case AifEnDeleteJBOD:
channel = aif->data.EN.data.ECE.container;
if (sc->cam_rescan_cb != NULL)
sc->cam_rescan_cb(sc, (channel >> 24) & 0xF,
AAC_CAM_TARGET_WILDCARD);
break;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
default:
break;
}
Mega update to the aac driver. 1. Correctly handle commands initiated by the adapter. These commands are defered to a kthread responsible for their processing, then are properly returned to the controller. 2. Add/remove disk devices when notified by the card that a container was created/deleted/offline. 3. Implement crashdump functionality. 4. Support all ioctls needed for the management CLI to work. The linux version of this app can be found at the Dell or HP website. A native version will be forthcoming. MFC-after: 4.4-RELEASE
2001-08-29 23:34:05 +00:00
default:
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
break;
Mega update to the aac driver. 1. Correctly handle commands initiated by the adapter. These commands are defered to a kthread responsible for their processing, then are properly returned to the controller. 2. Add/remove disk devices when notified by the card that a container was created/deleted/offline. 3. Implement crashdump functionality. 4. Support all ioctls needed for the management CLI to work. The linux version of this app can be found at the Dell or HP website. A native version will be forthcoming. MFC-after: 4.4-RELEASE
2001-08-29 23:34:05 +00:00
}
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/* Copy the AIF data to the AIF queue for ioctl retrieval */
Remove the AAC_LOCK macros. They no longer abstract anything and only obfuscate the code. No functional differences.
2004-08-12 05:05:06 +00:00
mtx_lock(&sc->aac_aifq_lock);
Allow simultaneous opens of the device for issuing commands to the controller. This is merged from Adaptec driver build 11669.
2007-12-07 00:22:23 +00:00
current = sc->aifq_idx;
next = (current + 1) % AAC_AIFQ_LENGTH;
if (next == 0)
sc->aifq_filled = 1;
bcopy(fib, &sc->aac_aifq[current], sizeof(struct aac_fib));
/* modify AIF contexts */
if (sc->aifq_filled) {
for (ctx = sc->fibctx; ctx; ctx = ctx->next) {
if (next == ctx->ctx_idx)
ctx->ctx_wrap = 1;
else if (current == ctx->ctx_idx && ctx->ctx_wrap)
ctx->ctx_idx = next;
}
}
sc->aifq_idx = next;
/* On the off chance that someone is sleeping for an aif... */
if (sc->aac_state & AAC_STATE_AIF_SLEEPER)
wakeup(sc->aac_aifq);
/* Wakeup any poll()ers */
selwakeuppri(&sc->rcv_select, PRIBIO);
Remove the AAC_LOCK macros. They no longer abstract anything and only obfuscate the code. No functional differences.
2004-08-12 05:05:06 +00:00
mtx_unlock(&sc->aac_aifq_lock);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
Major bugfix and minor update. This should resolve the current issues with the driver locking up under load. - Restructure so that we use a static pool of commands/FIBs, rather than allocating them in clusters. The cluster allocation just made things more complicated, and allowed us to waste more memory in peak load situations. - Make queueing macros more like my other drivers. This adds queue stats for free. Add some debugging to take advantage of this. - Reimplement the periodic timeout scan. Kick the interrupt handler and the start routine every scan as well, just to be safe. Track busy commands properly. - Bring resource cleanup into line with resource allocation. We should now clean up correctly after a failed probe/unload/etc. - Try to start new commands when old ones are completed. We weren't doing this before, which could lead to deadlock when the controller was full. - Don't try to build a new command if we have found a deferred command. This could cause us to lose the deferred command. - Use diskerr() to report I/O errors. - Don't bail if the AdapterInfo structure is the wrong size. Some variation seems to be normal. We need to improve our handing of 2.x firmware sets. - Improve some comments in an attempt to try to make things clearer. - Restructure to avoid some warnings.
2000-12-27 13:14:56 +00:00
* Return the Revision of the driver to userspace and check to see if the
Mega update to the aac driver. 1. Correctly handle commands initiated by the adapter. These commands are defered to a kthread responsible for their processing, then are properly returned to the controller. 2. Add/remove disk devices when notified by the card that a container was created/deleted/offline. 3. Implement crashdump functionality. 4. Support all ioctls needed for the management CLI to work. The linux version of this app can be found at the Dell or HP website. A native version will be forthcoming. MFC-after: 4.4-RELEASE
2001-08-29 23:34:05 +00:00
* userspace app is possibly compatible. This is extremely bogus since
* our driver doesn't follow Adaptec's versioning system. Cheat by just
* returning what the card reported.
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
*/
static int
Put the linux compat ioctl interface on death row. Add FreeBSD definitions of all of the ioctls, plus round out all ioctl definitions to match what exists for linux. Allow ioctls to be called through either the native or linux interface, though no apps exist (yet) that can take advantage of native calling.
2001-08-06 04:34:47 +00:00
aac_rev_check(struct aac_softc *sc, caddr_t udata)
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
{
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
struct aac_rev_check rev_check;
struct aac_rev_check_resp rev_check_resp;
int error = 0;
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
* Copyin the revision struct from userspace
*/
if ((error = copyin(udata, (caddr_t)&rev_check,
Reformat for 80 columns. Sorry, but I had to do it.
2001-08-03 00:50:30 +00:00
sizeof(struct aac_rev_check))) != 0) {
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
return error;
}
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_IOCTL_COMMANDS_B, "Userland revision= %d\n",
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
rev_check.callingRevision.buildNumber);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
* Doctor up the response struct.
*/
rev_check_resp.possiblyCompatible = 1;
Diff reduction with Adaptec's vendor driver. Driver version 2.1.9 chosen as that Adaptec version roughly corresponds with the current feature set merged to the in-tree driver.
2010-02-14 17:14:11 +00:00
rev_check_resp.adapterSWRevision.external.comp.major =
AAC_DRIVER_MAJOR_VERSION;
rev_check_resp.adapterSWRevision.external.comp.minor =
AAC_DRIVER_MINOR_VERSION;
rev_check_resp.adapterSWRevision.external.comp.type =
AAC_DRIVER_TYPE;
rev_check_resp.adapterSWRevision.external.comp.dash =
AAC_DRIVER_BUGFIX_LEVEL;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
rev_check_resp.adapterSWRevision.buildNumber =
Diff reduction with Adaptec's vendor driver. Driver version 2.1.9 chosen as that Adaptec version roughly corresponds with the current feature set merged to the in-tree driver.
2010-02-14 17:14:11 +00:00
AAC_DRIVER_BUILD;
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
return(copyout((caddr_t)&rev_check_resp, udata,
sizeof(struct aac_rev_check_resp)));
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
Allow simultaneous opens of the device for issuing commands to the controller. This is merged from Adaptec driver build 11669.
2007-12-07 00:22:23 +00:00
/*
* Pass the fib context to the caller
*/
static int
aac_open_aif(struct aac_softc *sc, caddr_t arg)
{
struct aac_fib_context *fibctx, *ctx;
int error = 0;
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
Allow simultaneous opens of the device for issuing commands to the controller. This is merged from Adaptec driver build 11669.
2007-12-07 00:22:23 +00:00
fibctx = malloc(sizeof(struct aac_fib_context), M_AACBUF, M_NOWAIT|M_ZERO);
if (fibctx == NULL)
return (ENOMEM);
mtx_lock(&sc->aac_aifq_lock);
/* all elements are already 0, add to queue */
if (sc->fibctx == NULL)
sc->fibctx = fibctx;
else {
for (ctx = sc->fibctx; ctx->next; ctx = ctx->next)
;
ctx->next = fibctx;
fibctx->prev = ctx;
}
/* evaluate unique value */
fibctx->unique = (*(u_int32_t *)&fibctx & 0xffffffff);
ctx = sc->fibctx;
while (ctx != fibctx) {
if (ctx->unique == fibctx->unique) {
fibctx->unique++;
ctx = sc->fibctx;
} else {
ctx = ctx->next;
}
}
mtx_unlock(&sc->aac_aifq_lock);
error = copyout(&fibctx->unique, (void *)arg, sizeof(u_int32_t));
if (error)
aac_close_aif(sc, (caddr_t)ctx);
return error;
}
/*
* Close the caller's fib context
*/
static int
aac_close_aif(struct aac_softc *sc, caddr_t arg)
{
struct aac_fib_context *ctx;
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
Allow simultaneous opens of the device for issuing commands to the controller. This is merged from Adaptec driver build 11669.
2007-12-07 00:22:23 +00:00
mtx_lock(&sc->aac_aifq_lock);
for (ctx = sc->fibctx; ctx; ctx = ctx->next) {
if (ctx->unique == *(uint32_t *)&arg) {
if (ctx == sc->fibctx)
sc->fibctx = NULL;
else {
ctx->prev->next = ctx->next;
if (ctx->next)
ctx->next->prev = ctx->prev;
}
break;
}
}
mtx_unlock(&sc->aac_aifq_lock);
if (ctx)
free(ctx, M_AACBUF);
return 0;
}
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
* Pass the caller the next AIF in their queue
*/
static int
Put the linux compat ioctl interface on death row. Add FreeBSD definitions of all of the ioctls, plus round out all ioctl definitions to match what exists for linux. Allow ioctls to be called through either the native or linux interface, though no apps exist (yet) that can take advantage of native calling.
2001-08-06 04:34:47 +00:00
aac_getnext_aif(struct aac_softc *sc, caddr_t arg)
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
{
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
struct get_adapter_fib_ioctl agf;
Allow simultaneous opens of the device for issuing commands to the controller. This is merged from Adaptec driver build 11669.
2007-12-07 00:22:23 +00:00
struct aac_fib_context *ctx;
The aac driver has evolved enough over the last few months that it no longer resembles the 4.x version very much. Garbage collect the legacy bits.
2003-03-01 05:07:19 +00:00
int error;
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
if ((error = copyin(arg, &agf, sizeof(agf))) == 0) {
Allow simultaneous opens of the device for issuing commands to the controller. This is merged from Adaptec driver build 11669.
2007-12-07 00:22:23 +00:00
for (ctx = sc->fibctx; ctx; ctx = ctx->next) {
if (agf.AdapterFibContext == ctx->unique)
break;
}
if (!ctx)
return (EFAULT);
error = aac_return_aif(sc, ctx, agf.AifFib);
if (error == EAGAIN && agf.Wait) {
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_AIF_B, "aac_getnext_aif(): waiting for AIF");
Allow simultaneous opens of the device for issuing commands to the controller. This is merged from Adaptec driver build 11669.
2007-12-07 00:22:23 +00:00
sc->aac_state |= AAC_STATE_AIF_SLEEPER;
while (error == EAGAIN) {
error = tsleep(sc->aac_aifq, PRIBIO |
PCATCH, "aacaif", 0);
if (error == 0)
error = aac_return_aif(sc, ctx, agf.AifFib);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
}
Allow simultaneous opens of the device for issuing commands to the controller. This is merged from Adaptec driver build 11669.
2007-12-07 00:22:23 +00:00
sc->aac_state &= ~AAC_STATE_AIF_SLEEPER;
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
}
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
return(error);
A new driver for PCI:SCSI RAID controllers based on the Adaptec FSA design. This includes integrated Dell RAID controllers, the Dell PERC 2/QC and the HP NetRAID-4M.
2000-09-13 03:20:35 +00:00
}
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
Major bugfix and minor update. This should resolve the current issues with the driver locking up under load. - Restructure so that we use a static pool of commands/FIBs, rather than allocating them in clusters. The cluster allocation just made things more complicated, and allowed us to waste more memory in peak load situations. - Make queueing macros more like my other drivers. This adds queue stats for free. Add some debugging to take advantage of this. - Reimplement the periodic timeout scan. Kick the interrupt handler and the start routine every scan as well, just to be safe. Track busy commands properly. - Bring resource cleanup into line with resource allocation. We should now clean up correctly after a failed probe/unload/etc. - Try to start new commands when old ones are completed. We weren't doing this before, which could lead to deadlock when the controller was full. - Don't try to build a new command if we have found a deferred command. This could cause us to lose the deferred command. - Use diskerr() to report I/O errors. - Don't bail if the AdapterInfo structure is the wrong size. Some variation seems to be normal. We need to improve our handing of 2.x firmware sets. - Improve some comments in an attempt to try to make things clearer. - Restructure to avoid some warnings.
2000-12-27 13:14:56 +00:00
* Hand the next AIF off the top of the queue out to userspace.
*/
static int
Allow simultaneous opens of the device for issuing commands to the controller. This is merged from Adaptec driver build 11669.
2007-12-07 00:22:23 +00:00
aac_return_aif(struct aac_softc *sc, struct aac_fib_context *ctx, caddr_t uptr)
Major bugfix and minor update. This should resolve the current issues with the driver locking up under load. - Restructure so that we use a static pool of commands/FIBs, rather than allocating them in clusters. The cluster allocation just made things more complicated, and allowed us to waste more memory in peak load situations. - Make queueing macros more like my other drivers. This adds queue stats for free. Add some debugging to take advantage of this. - Reimplement the periodic timeout scan. Kick the interrupt handler and the start routine every scan as well, just to be safe. Track busy commands properly. - Bring resource cleanup into line with resource allocation. We should now clean up correctly after a failed probe/unload/etc. - Try to start new commands when old ones are completed. We weren't doing this before, which could lead to deadlock when the controller was full. - Don't try to build a new command if we have found a deferred command. This could cause us to lose the deferred command. - Use diskerr() to report I/O errors. - Don't bail if the AdapterInfo structure is the wrong size. Some variation seems to be normal. We need to improve our handing of 2.x firmware sets. - Improve some comments in an attempt to try to make things clearer. - Restructure to avoid some warnings.
2000-12-27 13:14:56 +00:00
{
Allow simultaneous opens of the device for issuing commands to the controller. This is merged from Adaptec driver build 11669.
2007-12-07 00:22:23 +00:00
int current, error;
Major bugfix and minor update. This should resolve the current issues with the driver locking up under load. - Restructure so that we use a static pool of commands/FIBs, rather than allocating them in clusters. The cluster allocation just made things more complicated, and allowed us to waste more memory in peak load situations. - Make queueing macros more like my other drivers. This adds queue stats for free. Add some debugging to take advantage of this. - Reimplement the periodic timeout scan. Kick the interrupt handler and the start routine every scan as well, just to be safe. Track busy commands properly. - Bring resource cleanup into line with resource allocation. We should now clean up correctly after a failed probe/unload/etc. - Try to start new commands when old ones are completed. We weren't doing this before, which could lead to deadlock when the controller was full. - Don't try to build a new command if we have found a deferred command. This could cause us to lose the deferred command. - Use diskerr() to report I/O errors. - Don't bail if the AdapterInfo structure is the wrong size. Some variation seems to be normal. We need to improve our handing of 2.x firmware sets. - Improve some comments in an attempt to try to make things clearer. - Restructure to avoid some warnings.
2000-12-27 13:14:56 +00:00
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
Major bugfix and minor update. This should resolve the current issues with the driver locking up under load. - Restructure so that we use a static pool of commands/FIBs, rather than allocating them in clusters. The cluster allocation just made things more complicated, and allowed us to waste more memory in peak load situations. - Make queueing macros more like my other drivers. This adds queue stats for free. Add some debugging to take advantage of this. - Reimplement the periodic timeout scan. Kick the interrupt handler and the start routine every scan as well, just to be safe. Track busy commands properly. - Bring resource cleanup into line with resource allocation. We should now clean up correctly after a failed probe/unload/etc. - Try to start new commands when old ones are completed. We weren't doing this before, which could lead to deadlock when the controller was full. - Don't try to build a new command if we have found a deferred command. This could cause us to lose the deferred command. - Use diskerr() to report I/O errors. - Don't bail if the AdapterInfo structure is the wrong size. Some variation seems to be normal. We need to improve our handing of 2.x firmware sets. - Improve some comments in an attempt to try to make things clearer. - Restructure to avoid some warnings.
2000-12-27 13:14:56 +00:00
Remove the AAC_LOCK macros. They no longer abstract anything and only obfuscate the code. No functional differences.
2004-08-12 05:05:06 +00:00
mtx_lock(&sc->aac_aifq_lock);
Allow simultaneous opens of the device for issuing commands to the controller. This is merged from Adaptec driver build 11669.
2007-12-07 00:22:23 +00:00
current = ctx->ctx_idx;
if (current == sc->aifq_idx && !ctx->ctx_wrap) {
/* empty */
Remove the AAC_LOCK macros. They no longer abstract anything and only obfuscate the code. No functional differences.
2004-08-12 05:05:06 +00:00
mtx_unlock(&sc->aac_aifq_lock);
Fix a couple of bugs with AIF handling: - Correct the logic for the AIF array index pointers so that correct slot is always looked at. - Copy the full FIB payload size when copying AIF's, not just the first 64 bytes. Thanks to Mirapoint, Inc, for pointing these problems out and offering a solution.
2003-10-17 21:44:06 +00:00
return (EAGAIN);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
}
Allow simultaneous opens of the device for issuing commands to the controller. This is merged from Adaptec driver build 11669.
2007-12-07 00:22:23 +00:00
error =
copyout(&sc->aac_aifq[current], (void *)uptr, sizeof(struct aac_fib));
Fix a couple of bugs with AIF handling: - Correct the logic for the AIF array index pointers so that correct slot is always looked at. - Copy the full FIB payload size when copying AIF's, not just the first 64 bytes. Thanks to Mirapoint, Inc, for pointing these problems out and offering a solution.
2003-10-17 21:44:06 +00:00
if (error)
device_printf(sc->aac_dev,
"aac_return_aif: copyout returned %d\n", error);
Allow simultaneous opens of the device for issuing commands to the controller. This is merged from Adaptec driver build 11669.
2007-12-07 00:22:23 +00:00
else {
ctx->ctx_wrap = 0;
ctx->ctx_idx = (current + 1) % AAC_AIFQ_LENGTH;
}
Remove the AAC_LOCK macros. They no longer abstract anything and only obfuscate the code. No functional differences.
2004-08-12 05:05:06 +00:00
mtx_unlock(&sc->aac_aifq_lock);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
return(error);
Major bugfix and minor update. This should resolve the current issues with the driver locking up under load. - Restructure so that we use a static pool of commands/FIBs, rather than allocating them in clusters. The cluster allocation just made things more complicated, and allowed us to waste more memory in peak load situations. - Make queueing macros more like my other drivers. This adds queue stats for free. Add some debugging to take advantage of this. - Reimplement the periodic timeout scan. Kick the interrupt handler and the start routine every scan as well, just to be safe. Track busy commands properly. - Bring resource cleanup into line with resource allocation. We should now clean up correctly after a failed probe/unload/etc. - Try to start new commands when old ones are completed. We weren't doing this before, which could lead to deadlock when the controller was full. - Don't try to build a new command if we have found a deferred command. This could cause us to lose the deferred command. - Use diskerr() to report I/O errors. - Don't bail if the AdapterInfo structure is the wrong size. Some variation seems to be normal. We need to improve our handing of 2.x firmware sets. - Improve some comments in an attempt to try to make things clearer. - Restructure to avoid some warnings.
2000-12-27 13:14:56 +00:00
}
Mega update to the aac driver. 1. Correctly handle commands initiated by the adapter. These commands are defered to a kthread responsible for their processing, then are properly returned to the controller. 2. Add/remove disk devices when notified by the card that a container was created/deleted/offline. 3. Implement crashdump functionality. 4. Support all ioctls needed for the management CLI to work. The linux version of this app can be found at the Dell or HP website. A native version will be forthcoming. MFC-after: 4.4-RELEASE
2001-08-29 23:34:05 +00:00
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
static int
aac_get_pci_info(struct aac_softc *sc, caddr_t uptr)
{
struct aac_pci_info {
u_int32_t bus;
u_int32_t slot;
} pciinf;
int error;
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
Mega Update to the aac driver to support a whole new family of cards and the modified interface that they use. Changes include: - Register a different interrupt handler for the new interface. This one is INTR_MPSAFE, not INTR_FAST, and directly processes completions and AIFs. - Add an event registration and callback mechanism for the ioctl and CAM modules can know when a resource shortage clears. This condition was previously fatal in CAM due to programming oversights. - Fix locking to play better with newbus. - Provide access methods for talking to cards with the NEWCOMM interface. - Fix up the CAM module to be better suited for dealing with newer firmware on the PERC Si/Di series that requires talking to plain SCSI via aac. - Add a whole slew of new PCI Id's. Thanks to Adaptec for providing an initial version of this work and for answering countless questions about it. There are still some rough edges in this, but it works well enough to commit and test for now. Obtained from: Adaptec, Inc.
2005-10-08 15:55:09 +00:00
pciinf.bus = pci_get_bus(sc->aac_dev);
pciinf.slot = pci_get_slot(sc->aac_dev);
error = copyout((caddr_t)&pciinf, uptr,
sizeof(struct aac_pci_info));
return (error);
}
Implement FSACTL_LNX_GET_FEATURES and FSACTL_GET_FEATURES ioctls. RAID tools (e.g. arcconf) need this to be able to create arrays larger than 2TB. Submitted by: Adaptec, via driver build 15317
2008-03-28 19:07:25 +00:00
static int
aac_supported_features(struct aac_softc *sc, caddr_t uptr)
{
struct aac_features f;
int error;
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
if ((error = copyin(uptr, &f, sizeof (f))) != 0)
return (error);
/*
* When the management driver receives FSACTL_GET_FEATURES ioctl with
* ALL zero in the featuresState, the driver will return the current
* state of all the supported features, the data field will not be
* valid.
* When the management driver receives FSACTL_GET_FEATURES ioctl with
* a specific bit set in the featuresState, the driver will return the
* current state of this specific feature and whatever data that are
* associated with the feature in the data field or perform whatever
* action needed indicates in the data field.
*/
Whitespace tweak. Submitted by: phk, via accidental commit in r196969.
2009-09-09 04:04:09 +00:00
if (f.feat.fValue == 0) {
Implement FSACTL_LNX_GET_FEATURES and FSACTL_GET_FEATURES ioctls. RAID tools (e.g. arcconf) need this to be able to create arrays larger than 2TB. Submitted by: Adaptec, via driver build 15317
2008-03-28 19:07:25 +00:00
f.feat.fBits.largeLBA =
(sc->flags & AAC_FLAGS_LBA_64BIT) ? 1 : 0;
/* TODO: In the future, add other features state here as well */
} else {
if (f.feat.fBits.largeLBA)
f.feat.fBits.largeLBA =
(sc->flags & AAC_FLAGS_LBA_64BIT) ? 1 : 0;
/* TODO: Add other features state and data in the future */
}
error = copyout(&f, uptr, sizeof (f));
return (error);
}
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
/*
Mega update to the aac driver. 1. Correctly handle commands initiated by the adapter. These commands are defered to a kthread responsible for their processing, then are properly returned to the controller. 2. Add/remove disk devices when notified by the card that a container was created/deleted/offline. 3. Implement crashdump functionality. 4. Support all ioctls needed for the management CLI to work. The linux version of this app can be found at the Dell or HP website. A native version will be forthcoming. MFC-after: 4.4-RELEASE
2001-08-29 23:34:05 +00:00
* Give the userland some information about the container. The AAC arch
* expects the driver to be a SCSI passthrough type driver, so it expects
* the containers to have b:t:l numbers. Fake it.
*/
static int
aac_query_disk(struct aac_softc *sc, caddr_t uptr)
{
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
struct aac_query_disk query_disk;
struct aac_container *co;
struct aac_disk *disk;
int error, id;
Mega update to the aac driver. 1. Correctly handle commands initiated by the adapter. These commands are defered to a kthread responsible for their processing, then are properly returned to the controller. 2. Add/remove disk devices when notified by the card that a container was created/deleted/offline. 3. Implement crashdump functionality. 4. Support all ioctls needed for the management CLI to work. The linux version of this app can be found at the Dell or HP website. A native version will be forthcoming. MFC-after: 4.4-RELEASE
2001-08-29 23:34:05 +00:00
Diff reduction to Adaptec's driver (around build 15317): catch up with a change in debugging routines. The fwprintf macro in the AAC_DEBUG case (mapping to printf) isn't from the Adaptec driver.
2008-03-24 19:23:33 +00:00
fwprintf(sc, HBA_FLAGS_DBG_FUNCTION_ENTRY_B, "");
Mega update to the aac driver. 1. Correctly handle commands initiated by the adapter. These commands are defered to a kthread responsible for their processing, then are properly returned to the controller. 2. Add/remove disk devices when notified by the card that a container was created/deleted/offline. 3. Implement crashdump functionality. 4. Support all ioctls needed for the management CLI to work. The linux version of this app can be found at the Dell or HP website. A native version will be forthcoming. MFC-after: 4.4-RELEASE
2001-08-29 23:34:05 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
disk = NULL;
Mega update to the aac driver. 1. Correctly handle commands initiated by the adapter. These commands are defered to a kthread responsible for their processing, then are properly returned to the controller. 2. Add/remove disk devices when notified by the card that a container was created/deleted/offline. 3. Implement crashdump functionality. 4. Support all ioctls needed for the management CLI to work. The linux version of this app can be found at the Dell or HP website. A native version will be forthcoming. MFC-after: 4.4-RELEASE
2001-08-29 23:34:05 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
error = copyin(uptr, (caddr_t)&query_disk,
sizeof(struct aac_query_disk));
if (error)
return (error);
Mega update to the aac driver. 1. Correctly handle commands initiated by the adapter. These commands are defered to a kthread responsible for their processing, then are properly returned to the controller. 2. Add/remove disk devices when notified by the card that a container was created/deleted/offline. 3. Implement crashdump functionality. 4. Support all ioctls needed for the management CLI to work. The linux version of this app can be found at the Dell or HP website. A native version will be forthcoming. MFC-after: 4.4-RELEASE
2001-08-29 23:34:05 +00:00
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
id = query_disk.ContainerNumber;
if (id == -1)
return (EINVAL);
Remove the AAC_LOCK macros. They no longer abstract anything and only obfuscate the code. No functional differences.
2004-08-12 05:05:06 +00:00
mtx_lock(&sc->aac_container_lock);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
TAILQ_FOREACH(co, &sc->aac_container_tqh, co_link) {
if (co->co_mntobj.ObjectId == id)
break;
}
Fix mis-indentation. Approved by: scottl
2002-10-20 16:00:43 +00:00
if (co == NULL) {
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
query_disk.Valid = 0;
query_disk.Locked = 0;
query_disk.Deleted = 1; /* XXX is this right? */
Fix mis-indentation. Approved by: scottl
2002-10-20 16:00:43 +00:00
} else {
disk = device_get_softc(co->co_disk);
query_disk.Valid = 1;
query_disk.Locked =
(disk->ad_flags & AAC_DISK_OPEN) ? 1 : 0;
query_disk.Deleted = 0;
query_disk.Bus = device_get_unit(sc->aac_dev);
query_disk.Target = disk->unit;
query_disk.Lun = 0;
query_disk.UnMapped = 0;
Move to 'struct disk*' API Submitted by: phk
2003-02-26 02:50:27 +00:00
sprintf(&query_disk.diskDeviceName[0], "%s%d",
Whitespace fixup: 8 spaces -> tab
2009-10-27 19:37:37 +00:00
disk->ad_disk->d_name, disk->ad_disk->d_unit);
Fix mis-indentation. Approved by: scottl
2002-10-20 16:00:43 +00:00
}
Remove the AAC_LOCK macros. They no longer abstract anything and only obfuscate the code. No functional differences.
2004-08-12 05:05:06 +00:00
mtx_unlock(&sc->aac_container_lock);
Bring the aac driver *much* closer to style(9). Reviewed by: ken
2001-09-05 20:43:02 +00:00
error = copyout((caddr_t)&query_disk, uptr,
sizeof(struct aac_query_disk));
return (error);
Mega update to the aac driver. 1. Correctly handle commands initiated by the adapter. These commands are defered to a kthread responsible for their processing, then are properly returned to the controller. 2. Add/remove disk devices when notified by the card that a container was created/deleted/offline. 3. Implement crashdump functionality. 4. Support all ioctls needed for the management CLI to work. The linux version of this app can be found at the Dell or HP website. A native version will be forthcoming. MFC-after: 4.4-RELEASE
2001-08-29 23:34:05 +00:00
}
Add a CAM interface to the aac driver. This is useful in case you should ever connect a SCSI Cdrom/Tape/Jukebox/Scanner/Printer/kitty-litter-scooper to your high-end RAID controller. The interface to the arrays is still via the block interface; this merely provides a way to circumvent the RAID functionality and access the SCSI buses directly. Note that for somewhat obvious reasons, hard drives are not exposed to the da driver through this interface, though you can still talk to them via the pass driver. Be the first on your block to low-level format unsuspecting drives that are part of an array! To enable this, add the 'aacp' device to your kernel config. MFC after: 3 days
2002-04-27 01:31:17 +00:00
static void
aac_get_bus_info(struct aac_softc *sc)
{
struct aac_fib *fib;
struct aac_ctcfg *c_cmd;
struct aac_ctcfg_resp *c_resp;
struct aac_vmioctl *vmi;
struct aac_vmi_businf_resp *vmi_resp;
struct aac_getbusinf businfo;
Various cleanups: - Move the command timeout check from a separate repeating timeout to the kthread since the kthread is already running periodically. - Move printing the hardware print buffer to the kthread. - Properly shut down the kernel thread on detach. - Detach the child array devices on detach. - Don't issue a controller halt command on detach. Doing so requires a PCI reset to wake the controller back up. The driver can now be unloaded as long as CAM support is not enabled.
2003-02-05 23:24:25 +00:00
struct aac_sim *caminf;
Add a CAM interface to the aac driver. This is useful in case you should ever connect a SCSI Cdrom/Tape/Jukebox/Scanner/Printer/kitty-litter-scooper to your high-end RAID controller. The interface to the arrays is still via the block interface; this merely provides a way to circumvent the RAID functionality and access the SCSI buses directly. Note that for somewhat obvious reasons, hard drives are not exposed to the da driver through this interface, though you can still talk to them via the pass driver. Be the first on your block to low-level format unsuspecting drives that are part of an array! To enable this, add the 'aacp' device to your kernel config. MFC after: 3 days
2002-04-27 01:31:17 +00:00
device_t child;
int i, found, error;
Pickup I/O lock in aac_get_bus_info, as this code will call aac_alloc_sync_fib(). aac_alloc_sync_fib() will assert that the I/O locks are held. This fixes a panic on system boot up when the aac(4) device's bus_generic_attach() routine is called. Reviewed by: scottl
2005-10-31 00:12:19 +00:00
mtx_lock(&sc->aac_io_lock);
Collapse sync fib locking into normal i/o locking. The former didn't protect the registers so it was trivially possible for a sync command and i/o command to fight each other and confuse the controller. Make the sync fib alloc/release functions inline and remove the somewhat worthless AAC_SYNC_LOCK_FORCE flag. Thanks to Adil Katchi for helping me to track this down in RELENG_4.
2004-06-02 18:15:48 +00:00
aac_alloc_sync_fib(sc, &fib);
Add a CAM interface to the aac driver. This is useful in case you should ever connect a SCSI Cdrom/Tape/Jukebox/Scanner/Printer/kitty-litter-scooper to your high-end RAID controller. The interface to the arrays is still via the block interface; this merely provides a way to circumvent the RAID functionality and access the SCSI buses directly. Note that for somewhat obvious reasons, hard drives are not exposed to the da driver through this interface, though you can still talk to them via the pass driver. Be the first on your block to low-level format unsuspecting drives that are part of an array! To enable this, add the 'aacp' device to your kernel config. MFC after: 3 days
2002-04-27 01:31:17 +00:00
c_cmd = (struct aac_ctcfg *)&fib->data[0];
bzero out the sync command buffer when sending commands. This was causing problems when enumerating multiple arrays. This is an MFC candidate.
2002-05-03 00:07:50 +00:00
bzero(c_cmd, sizeof(struct aac_ctcfg));
Add a CAM interface to the aac driver. This is useful in case you should ever connect a SCSI Cdrom/Tape/Jukebox/Scanner/Printer/kitty-litter-scooper to your high-end RAID controller. The interface to the arrays is still via the block interface; this merely provides a way to circumvent the RAID functionality and access the SCSI buses directly. Note that for somewhat obvious reasons, hard drives are not exposed to the da driver through this interface, though you can still talk to them via the pass driver. Be the first on your block to low-level format unsuspecting drives that are part of an array! To enable this, add the 'aacp' device to your kernel config. MFC after: 3 days
2002-04-27 01:31:17 +00:00
c_cmd->Command = VM_ContainerConfig;
c_cmd->cmd = CT_GET_SCSI_METHOD;
c_cmd->param = 0;
error = aac_sync_fib(sc, ContainerCommand, 0, fib,
sizeof(struct aac_ctcfg));
if (error) {
device_printf(sc->aac_dev, "Error %d sending "
"VM_ContainerConfig command\n", error);
aac_release_sync_fib(sc);
Pickup I/O lock in aac_get_bus_info, as this code will call aac_alloc_sync_fib(). aac_alloc_sync_fib() will assert that the I/O locks are held. This fixes a panic on system boot up when the aac(4) device's bus_generic_attach() routine is called. Reviewed by: scottl
2005-10-31 00:12:19 +00:00
mtx_unlock(&sc->aac_io_lock);
Add a CAM interface to the aac driver. This is useful in case you should ever connect a SCSI Cdrom/Tape/Jukebox/Scanner/Printer/kitty-litter-scooper to your high-end RAID controller. The interface to the arrays is still via the block interface; this merely provides a way to circumvent the RAID functionality and access the SCSI buses directly. Note that for somewhat obvious reasons, hard drives are not exposed to the da driver through this interface, though you can still talk to them via the pass driver. Be the first on your block to low-level format unsuspecting drives that are part of an array! To enable this, add the 'aacp' device to your kernel config. MFC after: 3 days
2002-04-27 01:31:17 +00:00
return;
}
c_resp = (struct aac_ctcfg_resp *)&fib->data[0];
if (c_resp->Status != ST_OK) {
device_printf(sc->aac_dev, "VM_ContainerConfig returned 0x%x\n",
c_resp->Status);
aac_release_sync_fib(sc);
Pickup I/O lock in aac_get_bus_info, as this code will call aac_alloc_sync_fib(). aac_alloc_sync_fib() will assert that the I/O locks are held. This fixes a panic on system boot up when the aac(4) device's bus_generic_attach() routine is called. Reviewed by: scottl
2005-10-31 00:12:19 +00:00
mtx_unlock(&sc->aac_io_lock);
Add a CAM interface to the aac driver. This is useful in case you should ever connect a SCSI Cdrom/Tape/Jukebox/Scanner/Printer/kitty-litter-scooper to your high-end RAID controller. The interface to the arrays is still via the block interface; this merely provides a way to circumvent the RAID functionality and access the SCSI buses directly. Note that for somewhat obvious reasons, hard drives are not exposed to the da driver through this interface, though you can still talk to them via the pass driver. Be the first on your block to low-level format unsuspecting drives that are part of an array! To enable this, add the 'aacp' device to your kernel config. MFC after: 3 days
2002-04-27 01:31:17 +00:00
return;
}
sc->scsi_method_id = c_resp->param;
vmi = (struct aac_vmioctl *)&fib->data[0];
bzero out the sync command buffer when sending commands. This was causing problems when enumerating multiple arrays. This is an MFC candidate.
2002-05-03 00:07:50 +00:00
bzero(vmi, sizeof(struct aac_vmioctl));
Add a CAM interface to the aac driver. This is useful in case you should ever connect a SCSI Cdrom/Tape/Jukebox/Scanner/Printer/kitty-litter-scooper to your high-end RAID controller. The interface to the arrays is still via the block interface; this merely provides a way to circumvent the RAID functionality and access the SCSI buses directly. Note that for somewhat obvious reasons, hard drives are not exposed to the da driver through this interface, though you can still talk to them via the pass driver. Be the first on your block to low-level format unsuspecting drives that are part of an array! To enable this, add the 'aacp' device to your kernel config. MFC after: 3 days
2002-04-27 01:31:17 +00:00
vmi->Command = VM_Ioctl;
vmi->ObjType = FT_DRIVE;
vmi->MethId = sc->scsi_method_id;
vmi->ObjId = 0;
vmi->IoctlCmd = GetBusInfo;
error = aac_sync_fib(sc, ContainerCommand, 0, fib,
Restore creation of passthrough devices with newer controller firmware by putting the correct size in the fib header. Presumably the older firmware silently ignored a bad size field. (This change tested with a 3805 controller. Passthrough devices were created when running firmware build 12814, but not 15323 or later. With this change they're created for both old and new firmware versions.) Submitted by: Adaptec
2008-03-20 20:33:48 +00:00
sizeof(struct aac_vmi_businf_resp));
Add a CAM interface to the aac driver. This is useful in case you should ever connect a SCSI Cdrom/Tape/Jukebox/Scanner/Printer/kitty-litter-scooper to your high-end RAID controller. The interface to the arrays is still via the block interface; this merely provides a way to circumvent the RAID functionality and access the SCSI buses directly. Note that for somewhat obvious reasons, hard drives are not exposed to the da driver through this interface, though you can still talk to them via the pass driver. Be the first on your block to low-level format unsuspecting drives that are part of an array! To enable this, add the 'aacp' device to your kernel config. MFC after: 3 days
2002-04-27 01:31:17 +00:00
if (error) {
device_printf(sc->aac_dev, "Error %d sending VMIoctl command\n",
error);
aac_release_sync_fib(sc);
Pickup I/O lock in aac_get_bus_info, as this code will call aac_alloc_sync_fib(). aac_alloc_sync_fib() will assert that the I/O locks are held. This fixes a panic on system boot up when the aac(4) device's bus_generic_attach() routine is called. Reviewed by: scottl
2005-10-31 00:12:19 +00:00
mtx_unlock(&sc->aac_io_lock);
Add a CAM interface to the aac driver. This is useful in case you should ever connect a SCSI Cdrom/Tape/Jukebox/Scanner/Printer/kitty-litter-scooper to your high-end RAID controller. The interface to the arrays is still via the block interface; this merely provides a way to circumvent the RAID functionality and access the SCSI buses directly. Note that for somewhat obvious reasons, hard drives are not exposed to the da driver through this interface, though you can still talk to them via the pass driver. Be the first on your block to low-level format unsuspecting drives that are part of an array! To enable this, add the 'aacp' device to your kernel config. MFC after: 3 days
2002-04-27 01:31:17 +00:00
return;
}
vmi_resp = (struct aac_vmi_businf_resp *)&fib->data[0];
if (vmi_resp->Status != ST_OK) {
device_printf(sc->aac_dev, "VM_Ioctl returned %d\n",
vmi_resp->Status);
aac_release_sync_fib(sc);
Pickup I/O lock in aac_get_bus_info, as this code will call aac_alloc_sync_fib(). aac_alloc_sync_fib() will assert that the I/O locks are held. This fixes a panic on system boot up when the aac(4) device's bus_generic_attach() routine is called. Reviewed by: scottl
2005-10-31 00:12:19 +00:00
mtx_unlock(&sc->aac_io_lock);
Add a CAM interface to the aac driver. This is useful in case you should ever connect a SCSI Cdrom/Tape/Jukebox/Scanner/Printer/kitty-litter-scooper to your high-end RAID controller. The interface to the arrays is still via the block interface; this merely provides a way to circumvent the RAID functionality and access the SCSI buses directly. Note that for somewhat obvious reasons, hard drives are not exposed to the da driver through this interface, though you can still talk to them via the pass driver. Be the first on your block to low-level format unsuspecting drives that are part of an array! To enable this, add the 'aacp' device to your kernel config. MFC after: 3 days
2002-04-27 01:31:17 +00:00
return;
}
bcopy(&vmi_resp->BusInf, &businfo, sizeof(struct aac_getbusinf));
aac_release_sync_fib(sc);
Pickup I/O lock in aac_get_bus_info, as this code will call aac_alloc_sync_fib(). aac_alloc_sync_fib() will assert that the I/O locks are held. This fixes a panic on system boot up when the aac(4) device's bus_generic_attach() routine is called. Reviewed by: scottl
2005-10-31 00:12:19 +00:00
mtx_unlock(&sc->aac_io_lock);
Add a CAM interface to the aac driver. This is useful in case you should ever connect a SCSI Cdrom/Tape/Jukebox/Scanner/Printer/kitty-litter-scooper to your high-end RAID controller. The interface to the arrays is still via the block interface; this merely provides a way to circumvent the RAID functionality and access the SCSI buses directly. Note that for somewhat obvious reasons, hard drives are not exposed to the da driver through this interface, though you can still talk to them via the pass driver. Be the first on your block to low-level format unsuspecting drives that are part of an array! To enable this, add the 'aacp' device to your kernel config. MFC after: 3 days
2002-04-27 01:31:17 +00:00
found = 0;
for (i = 0; i < businfo.BusCount; i++) {
if (businfo.BusValid[i] != AAC_BUS_VALID)
continue;
Free the container and sim objects on detach. Convert several uses of MALLOC to malloc.
2003-02-06 00:37:37 +00:00
caminf = (struct aac_sim *)malloc( sizeof(struct aac_sim),
M_AACBUF, M_NOWAIT | M_ZERO);
Handle failures better in the passthrough bus creation code. Submitted by: Coverity Prevent analysis tool
2005-03-19 06:29:32 +00:00
if (caminf == NULL) {
device_printf(sc->aac_dev,
"No memory to add passthrough bus %d\n", i);
break;
Cleanup unnecessary semicolons from the kernel. Found with devel/coccinelle.
2016-04-10 23:07:00 +00:00
}
Add a CAM interface to the aac driver. This is useful in case you should ever connect a SCSI Cdrom/Tape/Jukebox/Scanner/Printer/kitty-litter-scooper to your high-end RAID controller. The interface to the arrays is still via the block interface; this merely provides a way to circumvent the RAID functionality and access the SCSI buses directly. Note that for somewhat obvious reasons, hard drives are not exposed to the da driver through this interface, though you can still talk to them via the pass driver. Be the first on your block to low-level format unsuspecting drives that are part of an array! To enable this, add the 'aacp' device to your kernel config. MFC after: 3 days
2002-04-27 01:31:17 +00:00
child = device_add_child(sc->aac_dev, "aacp", -1);
if (child == NULL) {
Handle failures better in the passthrough bus creation code. Submitted by: Coverity Prevent analysis tool
2005-03-19 06:29:32 +00:00
device_printf(sc->aac_dev,
"device_add_child failed for passthrough bus %d\n",
i);
free(caminf, M_AACBUF);
Remove the unused AAC_DRIVER_BUILD_DATE macro. Nuke whitespace at EOL while I'm here. Approved by: scottl (MAINTAINER)
2005-11-13 17:26:36 +00:00
break;
Add a CAM interface to the aac driver. This is useful in case you should ever connect a SCSI Cdrom/Tape/Jukebox/Scanner/Printer/kitty-litter-scooper to your high-end RAID controller. The interface to the arrays is still via the block interface; this merely provides a way to circumvent the RAID functionality and access the SCSI buses directly. Note that for somewhat obvious reasons, hard drives are not exposed to the da driver through this interface, though you can still talk to them via the pass driver. Be the first on your block to low-level format unsuspecting drives that are part of an array! To enable this, add the 'aacp' device to your kernel config. MFC after: 3 days
2002-04-27 01:31:17 +00:00
}
caminf->TargetsPerBus = businfo.TargetsPerBus;
caminf->BusNumber = i;
caminf->InitiatorBusId = businfo.InitiatorBusId[i];
caminf->aac_sc = sc;
Teach the CAM module how to deregister itself so it can be unloaded.
2003-02-06 03:31:02 +00:00
caminf->sim_dev = child;
Add a CAM interface to the aac driver. This is useful in case you should ever connect a SCSI Cdrom/Tape/Jukebox/Scanner/Printer/kitty-litter-scooper to your high-end RAID controller. The interface to the arrays is still via the block interface; this merely provides a way to circumvent the RAID functionality and access the SCSI buses directly. Note that for somewhat obvious reasons, hard drives are not exposed to the da driver through this interface, though you can still talk to them via the pass driver. Be the first on your block to low-level format unsuspecting drives that are part of an array! To enable this, add the 'aacp' device to your kernel config. MFC after: 3 days
2002-04-27 01:31:17 +00:00
device_set_ivars(child, caminf);
device_set_desc(child, "SCSI Passthrough Bus");
Various cleanups: - Move the command timeout check from a separate repeating timeout to the kthread since the kthread is already running periodically. - Move printing the hardware print buffer to the kthread. - Properly shut down the kernel thread on detach. - Detach the child array devices on detach. - Don't issue a controller halt command on detach. Doing so requires a PCI reset to wake the controller back up. The driver can now be unloaded as long as CAM support is not enabled.
2003-02-05 23:24:25 +00:00
TAILQ_INSERT_TAIL(&sc->aac_sim_tqh, caminf, sim_link);
Add a CAM interface to the aac driver. This is useful in case you should ever connect a SCSI Cdrom/Tape/Jukebox/Scanner/Printer/kitty-litter-scooper to your high-end RAID controller. The interface to the arrays is still via the block interface; this merely provides a way to circumvent the RAID functionality and access the SCSI buses directly. Note that for somewhat obvious reasons, hard drives are not exposed to the da driver through this interface, though you can still talk to them via the pass driver. Be the first on your block to low-level format unsuspecting drives that are part of an array! To enable this, add the 'aacp' device to your kernel config. MFC after: 3 days
2002-04-27 01:31:17 +00:00
found = 1;
}
if (found)
bus_generic_attach(sc->aac_dev);
}
Reference in New Issue Copy Permalink