Aren
/
NetBSD
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
NetBSD/sys/dev/ic/isp_netbsd.h

526 lines
14 KiB
C
Raw Normal View History

Redo fabric evaluation to not use GET ALL NEXT (GA_NXT). Switches seem to be trying to wriggle out of supporting this well. Instead, use GID_FT to get a list of Port IDs and then use GPN_ID/GNN_ID to find the port and node wwn. This should make working on fabrics a bit cleaner and more stable. This also caused some cleanup of SNS subcommand canonicalization so that we can actually check for FS_ACC and FS_RJT, and if we get an FS_RJT, print out the reason and explanation codes. We'll keep the old GA_NXT method around if people want to uncomment a controlling definition in ispvar.h. This also had us clean up ISPASYNC_FABRICDEV to use a local lportdb argument and to have the caller explicitly say that a device is at the end of the fabric list.
2002-04-05 03:38:43 +04:00
/* $NetBSD: isp_netbsd.h,v 1.50 2002/04/04 23:38:48 mjacob Exp $ */
Add a maintenance note. Move the single bit tags of islocked and onintstack to be real integers. Add ISP_ILOCK/ISP_IUNLOCK macros. Fix the isp_lock/isp_unlock inlines to stop being so embarrassingly in error. Why, or why, can't I have mutex_enter/mutex_exit, pretty please?
2000-08-14 11:10:09 +04:00
/*
* This driver, which is contained in NetBSD in the files:
*
* sys/dev/ic/isp.c
Fix pathnames in comment.
2000-12-23 04:37:57 +03:00
* sys/dev/ic/isp_inline.h
* sys/dev/ic/isp_netbsd.c
* sys/dev/ic/isp_netbsd.h
* sys/dev/ic/isp_target.c
* sys/dev/ic/isp_target.h
* sys/dev/ic/isp_tpublic.h
* sys/dev/ic/ispmbox.h
* sys/dev/ic/ispreg.h
* sys/dev/ic/ispvar.h
Add a maintenance note. Move the single bit tags of islocked and onintstack to be real integers. Add ISP_ILOCK/ISP_IUNLOCK macros. Fix the isp_lock/isp_unlock inlines to stop being so embarrassingly in error. Why, or why, can't I have mutex_enter/mutex_exit, pretty please?
2000-08-14 11:10:09 +04:00
* sys/microcode/isp/asm_sbus.h
* sys/microcode/isp/asm_1040.h
* sys/microcode/isp/asm_1080.h
* sys/microcode/isp/asm_12160.h
* sys/microcode/isp/asm_2100.h
* sys/microcode/isp/asm_2200.h
* sys/pci/isp_pci.c
* sys/sbus/isp_sbus.c
*
* Is being actively maintained by Matthew Jacob (mjacob@netbsd.org).
* This driver also is shared source with FreeBSD, OpenBSD, Linux, Solaris,
* Linux versions. This tends to be an interesting maintenance problem.
*
* Please coordinate with Matthew Jacob on changes you wish to make here.
*/
NetBSD OS specific routines and definitions.
1998-07-15 23:44:04 +04:00
/*
* NetBSD Specific definitions for the Qlogic ISP Host Adapter
Change to approved NASA/Ames copyright. Add Fabric support. Fix SCCLUN support. Add code that tries to track LoopID shifting.
1999-07-06 00:31:35 +04:00
* Matthew Jacob <mjacob@nas.nasa.gov>
*/
/*
* Copyright (C) 1997, 1998, 1999 National Aeronautics & Space Administration
NetBSD OS specific routines and definitions.
1998-07-15 23:44:04 +04:00
* 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
Change to approved NASA/Ames copyright. Add Fabric support. Fix SCCLUN support. Add code that tries to track LoopID shifting.
1999-07-06 00:31:35 +04:00
* notice, this list of conditions and the following disclaimer.
NetBSD OS specific routines and definitions.
1998-07-15 23:44:04 +04:00
* 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.
* 3. The name of the author may not be used to endorse or promote products
Change to approved NASA/Ames copyright. Add Fabric support. Fix SCCLUN support. Add code that tries to track LoopID shifting.
1999-07-06 00:31:35 +04:00
* derived from this software without specific prior written permission
NetBSD OS specific routines and definitions.
1998-07-15 23:44:04 +04:00
*
Change to approved NASA/Ames copyright. Add Fabric support. Fix SCCLUN support. Add code that tries to track LoopID shifting.
1999-07-06 00:31:35 +04:00
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``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 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.
NetBSD OS specific routines and definitions.
1998-07-15 23:44:04 +04:00
*/
#ifndef _ISP_NETBSD_H
#define _ISP_NETBSD_H
#include <sys/types.h>
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/kernel.h>
#include <sys/errno.h>
#include <sys/ioctl.h>
#include <sys/device.h>
#include <sys/malloc.h>
#include <sys/buf.h>
#include <sys/proc.h>
#include <sys/user.h>
Create a kernel thread for Fibre Channel cards. This is the new way of doing business- modulo some startup spasms and peculiarities of the way kthreads are started (*after* configuration, weird) and some strangeness with the freeze/thaw code, what now happens is that any of Loop Down, LIP, Loop Reset or Port Datbase or Name Server Database Changed ASYNC events cause the queues to freeze for this channel. The arrival of a Loop UP is not relevant. What *is* relevant is that the Port Datbase or Name Server Changed async event indicate that it's okay to go and (re)evaluate the state of the FC link and (re)probe local loop and fabric membership. We have a kthread do this because it's *sooooo* much nicer to be able to sleep while doing the 130-250 mailbox commands it'll take to re-evaluate things. When the state is well known again, we can unfreeze the channel queues. Then, as commands start arriving, we simply can start them or bounce them with XS_SELTIMEOUT (if the device in question has gone away). Previously, we did lazy evaluation, which meant that if a change occurred, we would wait until the very *next* command to go rebuild stuff. The reason this is not sensible is: a) Even with sleeping, you can hang up your system because you might be making some poor stat(2) call pay the price of re-evaluating the whole fabric. b) If we ever really want to get to dynamic attachment/detachment, we should find out sooner, rather than later, where things get to. Split off ispminphys_1020 from ispminphys- a 1020 has a 24 bit limit- not anything newer. Re-enable LIPs and Loop Resets as async events- this allows the outer layer to set policy about them. Roll platform major && minor. Remove bogus waitq (no longer used). Remove callout entry in softc (no longer used). Define some shorthands for channels. Clean up a variety of cruft left over from the thorpej_scsipi changeover.
2001-05-26 01:45:55 +04:00
#include <sys/kthread.h>
NetBSD OS specific routines and definitions.
1998-07-15 23:44:04 +04:00
Major restructuring for swizzling to the request queue and unswizzling from the response queue. Instead of the ad hoc ISP_SWIZZLE_REQUEST, we now have a complete set of inline functions in isp_inline.h. Each platform is responsible for providing just one of a set of ISP_IOX_{GET,PUT}{8,16,32} macros. The reason this needs to be done is that we need to have a single set of functions that will work correctly on multiple architectures for both little and big endian machines. It also needs to work correctly in the case that we have the request or response queues in memory that has to be treated specially (e.g., have ddi_dma_sync called on it for Solaris after we update it or before we read from it). One thing that falls out of this is that we no longer build requests in the request queue itself. Instead, we build the request locally (e.g., on the stack) and then as part of the swizzling operation, copy it to the request queue entry we've allocated. I thought long and hard about whether this was too expensive a change to make as it in a lot of cases requires an extra copy. On balance, the flexbility is worth it. With any luck, the entry that we build locally stays in a processor writeback cache (after all, it's only 64 bytes) so that the cost of actually flushing it to the memory area that is the shared queue with the PCI device is not all that expensive. We may examine this again and try to get clever in the future to try and avoid copies. Another change that falls out of this is that MEMORYBARRIER should be taken a lot more seriously. The macro ISP_ADD_REQUEST does a MEMORYBARRIER on the entry being added. But there had been many other places this had been missing. It's now very important that it be done. For NetBSD, it does a ddi_dmamap_sync as appropriate. This gets us out of the explicit ddi_dmamap_sync on the whole response queue that we did for SBus cards at each interrupt. Set things up so that platforms that cannot have an SBus don't get a lot of the SBus code checks (dead coded out). Additional changes: Fix a longstanding buglet of sorts. When we get an entry via isp_getrqentry, the iptr value that gets returned is the value we intend to eventually plug into the ISP registers as the entry *one past* the last one we've written- *not* the current entry we're updating. All along we've been calling sync functions on the wrong index value. Argh. The 'fix' here is to rename all 'iptr' variables as 'nxti' to remember that this is the 'next' pointer- not the current pointer. Devote a single bit to mboxbsy- and set aside bits for output mbox registers that we need to pick up- we can have at least one command which does not have any defined output registers (MBOX_EXECUTE_FIRMWARE). Explicitly decode GetAllNext SNS Response back *as* a GetAllNext response. Otherwise, we won't unswizzle it correctly. Nuke some additional __P macros.
2001-12-14 03:13:44 +03:00
#include <machine/bus.h>
NetBSD OS specific routines and definitions.
1998-07-15 23:44:04 +04:00
#include <dev/scsipi/scsi_all.h>
#include <dev/scsipi/scsipi_all.h>
#include <dev/scsipi/scsiconf.h>
#include <dev/scsipi/scsi_message.h>
#include <dev/scsipi/scsipi_debug.h>
Change to approved NASA/Ames copyright. Add Fabric support. Fix SCCLUN support. Add code that tries to track LoopID shifting.
1999-07-06 00:31:35 +04:00
#include "opt_isp.h"
Roll platform minor. We are now always supporting ISP2100_FABRIC. Add in new MBOX_WAIT_COMPLETE/MBOX_NOTIFY_COMPLETE macros (ready for SMPizing). Define STRNCAT inline for our usage. Stealing a bit of the xs_status flags to maintain command state that is pertinent for the hBA- should really reserve them for private usage in scsipiconf.h.
2000-07-06 02:25:06 +04:00
Major restructuring for swizzling to the request queue and unswizzling from the response queue. Instead of the ad hoc ISP_SWIZZLE_REQUEST, we now have a complete set of inline functions in isp_inline.h. Each platform is responsible for providing just one of a set of ISP_IOX_{GET,PUT}{8,16,32} macros. The reason this needs to be done is that we need to have a single set of functions that will work correctly on multiple architectures for both little and big endian machines. It also needs to work correctly in the case that we have the request or response queues in memory that has to be treated specially (e.g., have ddi_dma_sync called on it for Solaris after we update it or before we read from it). One thing that falls out of this is that we no longer build requests in the request queue itself. Instead, we build the request locally (e.g., on the stack) and then as part of the swizzling operation, copy it to the request queue entry we've allocated. I thought long and hard about whether this was too expensive a change to make as it in a lot of cases requires an extra copy. On balance, the flexbility is worth it. With any luck, the entry that we build locally stays in a processor writeback cache (after all, it's only 64 bytes) so that the cost of actually flushing it to the memory area that is the shared queue with the PCI device is not all that expensive. We may examine this again and try to get clever in the future to try and avoid copies. Another change that falls out of this is that MEMORYBARRIER should be taken a lot more seriously. The macro ISP_ADD_REQUEST does a MEMORYBARRIER on the entry being added. But there had been many other places this had been missing. It's now very important that it be done. For NetBSD, it does a ddi_dmamap_sync as appropriate. This gets us out of the explicit ddi_dmamap_sync on the whole response queue that we did for SBus cards at each interrupt. Set things up so that platforms that cannot have an SBus don't get a lot of the SBus code checks (dead coded out). Additional changes: Fix a longstanding buglet of sorts. When we get an entry via isp_getrqentry, the iptr value that gets returned is the value we intend to eventually plug into the ISP registers as the entry *one past* the last one we've written- *not* the current entry we're updating. All along we've been calling sync functions on the wrong index value. Argh. The 'fix' here is to rename all 'iptr' variables as 'nxti' to remember that this is the 'next' pointer- not the current pointer. Devote a single bit to mboxbsy- and set aside bits for output mbox registers that we need to pick up- we can have at least one command which does not have any defined output registers (MBOX_EXECUTE_FIRMWARE). Explicitly decode GetAllNext SNS Response back *as* a GetAllNext response. Otherwise, we won't unswizzle it correctly. Nuke some additional __P macros.
2001-12-14 03:13:44 +03:00
/*
* Efficiency- get rid of SBus code && tests unless we need them.
*/
#if defined(__sparcv9__ ) || defined(__sparc__)
#define ISP_SBUS_SUPPORTED 1
#else
#define ISP_SBUS_SUPPORTED 0
#endif
Change to approved NASA/Ames copyright. Add Fabric support. Fix SCCLUN support. Add code that tries to track LoopID shifting.
1999-07-06 00:31:35 +04:00
Create a kernel thread for Fibre Channel cards. This is the new way of doing business- modulo some startup spasms and peculiarities of the way kthreads are started (*after* configuration, weird) and some strangeness with the freeze/thaw code, what now happens is that any of Loop Down, LIP, Loop Reset or Port Datbase or Name Server Database Changed ASYNC events cause the queues to freeze for this channel. The arrival of a Loop UP is not relevant. What *is* relevant is that the Port Datbase or Name Server Changed async event indicate that it's okay to go and (re)evaluate the state of the FC link and (re)probe local loop and fabric membership. We have a kthread do this because it's *sooooo* much nicer to be able to sleep while doing the 130-250 mailbox commands it'll take to re-evaluate things. When the state is well known again, we can unfreeze the channel queues. Then, as commands start arriving, we simply can start them or bounce them with XS_SELTIMEOUT (if the device in question has gone away). Previously, we did lazy evaluation, which meant that if a change occurred, we would wait until the very *next* command to go rebuild stuff. The reason this is not sensible is: a) Even with sleeping, you can hang up your system because you might be making some poor stat(2) call pay the price of re-evaluating the whole fabric. b) If we ever really want to get to dynamic attachment/detachment, we should find out sooner, rather than later, where things get to. Split off ispminphys_1020 from ispminphys- a 1020 has a 24 bit limit- not anything newer. Re-enable LIPs and Loop Resets as async events- this allows the outer layer to set policy about them. Roll platform major && minor. Remove bogus waitq (no longer used). Remove callout entry in softc (no longer used). Define some shorthands for channels. Clean up a variety of cruft left over from the thorpej_scsipi changeover.
2001-05-26 01:45:55 +04:00
#define ISP_PLATFORM_VERSION_MAJOR 2
Major restructuring for swizzling to the request queue and unswizzling from the response queue. Instead of the ad hoc ISP_SWIZZLE_REQUEST, we now have a complete set of inline functions in isp_inline.h. Each platform is responsible for providing just one of a set of ISP_IOX_{GET,PUT}{8,16,32} macros. The reason this needs to be done is that we need to have a single set of functions that will work correctly on multiple architectures for both little and big endian machines. It also needs to work correctly in the case that we have the request or response queues in memory that has to be treated specially (e.g., have ddi_dma_sync called on it for Solaris after we update it or before we read from it). One thing that falls out of this is that we no longer build requests in the request queue itself. Instead, we build the request locally (e.g., on the stack) and then as part of the swizzling operation, copy it to the request queue entry we've allocated. I thought long and hard about whether this was too expensive a change to make as it in a lot of cases requires an extra copy. On balance, the flexbility is worth it. With any luck, the entry that we build locally stays in a processor writeback cache (after all, it's only 64 bytes) so that the cost of actually flushing it to the memory area that is the shared queue with the PCI device is not all that expensive. We may examine this again and try to get clever in the future to try and avoid copies. Another change that falls out of this is that MEMORYBARRIER should be taken a lot more seriously. The macro ISP_ADD_REQUEST does a MEMORYBARRIER on the entry being added. But there had been many other places this had been missing. It's now very important that it be done. For NetBSD, it does a ddi_dmamap_sync as appropriate. This gets us out of the explicit ddi_dmamap_sync on the whole response queue that we did for SBus cards at each interrupt. Set things up so that platforms that cannot have an SBus don't get a lot of the SBus code checks (dead coded out). Additional changes: Fix a longstanding buglet of sorts. When we get an entry via isp_getrqentry, the iptr value that gets returned is the value we intend to eventually plug into the ISP registers as the entry *one past* the last one we've written- *not* the current entry we're updating. All along we've been calling sync functions on the wrong index value. Argh. The 'fix' here is to rename all 'iptr' variables as 'nxti' to remember that this is the 'next' pointer- not the current pointer. Devote a single bit to mboxbsy- and set aside bits for output mbox registers that we need to pick up- we can have at least one command which does not have any defined output registers (MBOX_EXECUTE_FIRMWARE). Explicitly decode GetAllNext SNS Response back *as* a GetAllNext response. Otherwise, we won't unswizzle it correctly. Nuke some additional __P macros.
2001-12-14 03:13:44 +03:00
#define ISP_PLATFORM_VERSION_MINOR 1
NetBSD OS specific routines and definitions.
1998-07-15 23:44:04 +04:00
struct isposinfo {
struct device _dev;
Pull up the thorpej_scsipi branch to main branch. This is a completely rewritten scsipi_xfer execution engine, and the associated changes to HBA drivers. Overview of changes & features: - All xfers are queued in the mid-layer, rather than doing so in an ad-hoc fashion in individual adapter drivers. - Adapter/channel resource management in the mid-layer, avoids even trying to start running an xfer if the adapter/channel doesn't have the resources. - Better communication between the mid-layer and the adapters. - Asynchronous event notification mechanism from adapter to mid-layer and peripherals. - Better peripheral queue management: freeze/thaw, sorted requeueing during recovery, etc. - Clean separation of peripherals, adapters, and adapter channels (no more scsipi_link). - Kernel thread for each scsipi_channel makes error recovery much easier (no more dealing with interrupt context when recovering from an error). - Mid-layer support for tagged queueing: commands can have the tag type set explicitly, tag IDs are allocated in the mid-layer (thus eliminating the need to use buggy tag ID allocation schemes in many adapter drivers). - support for QUEUE FULL and CHECK CONDITION status in mid-layer; the command will be requeued, or a REQUEST SENSE will be sent as appropriate. Just before the merge syssrc has been tagged with thorpej_scsipi_beforemerge
2001-04-25 21:53:04 +04:00
struct scsipi_channel _chan;
struct scsipi_channel _chan_b;
Update BA for new max_lun parameter for SCSIbusses. Clearify maximum luns for FC HB based upon a SCCLUN define (15 for normal- 255 out of a possible 65535 for SCCLUN). Propagate loopid as adapter_target. Roll minor platform version. Roll core version number. Update mailbox definitions with cleaner target mode structure definitions. Clean up some ENDIAN stuff. Correct botched ISP2100_NVRAM_HARDLOOPID offset.
1998-12-05 22:48:23 +03:00
struct scsipi_adapter _adapter;
Major restructuring for swizzling to the request queue and unswizzling from the response queue. Instead of the ad hoc ISP_SWIZZLE_REQUEST, we now have a complete set of inline functions in isp_inline.h. Each platform is responsible for providing just one of a set of ISP_IOX_{GET,PUT}{8,16,32} macros. The reason this needs to be done is that we need to have a single set of functions that will work correctly on multiple architectures for both little and big endian machines. It also needs to work correctly in the case that we have the request or response queues in memory that has to be treated specially (e.g., have ddi_dma_sync called on it for Solaris after we update it or before we read from it). One thing that falls out of this is that we no longer build requests in the request queue itself. Instead, we build the request locally (e.g., on the stack) and then as part of the swizzling operation, copy it to the request queue entry we've allocated. I thought long and hard about whether this was too expensive a change to make as it in a lot of cases requires an extra copy. On balance, the flexbility is worth it. With any luck, the entry that we build locally stays in a processor writeback cache (after all, it's only 64 bytes) so that the cost of actually flushing it to the memory area that is the shared queue with the PCI device is not all that expensive. We may examine this again and try to get clever in the future to try and avoid copies. Another change that falls out of this is that MEMORYBARRIER should be taken a lot more seriously. The macro ISP_ADD_REQUEST does a MEMORYBARRIER on the entry being added. But there had been many other places this had been missing. It's now very important that it be done. For NetBSD, it does a ddi_dmamap_sync as appropriate. This gets us out of the explicit ddi_dmamap_sync on the whole response queue that we did for SBus cards at each interrupt. Set things up so that platforms that cannot have an SBus don't get a lot of the SBus code checks (dead coded out). Additional changes: Fix a longstanding buglet of sorts. When we get an entry via isp_getrqentry, the iptr value that gets returned is the value we intend to eventually plug into the ISP registers as the entry *one past* the last one we've written- *not* the current entry we're updating. All along we've been calling sync functions on the wrong index value. Argh. The 'fix' here is to rename all 'iptr' variables as 'nxti' to remember that this is the 'next' pointer- not the current pointer. Devote a single bit to mboxbsy- and set aside bits for output mbox registers that we need to pick up- we can have at least one command which does not have any defined output registers (MBOX_EXECUTE_FIRMWARE). Explicitly decode GetAllNext SNS Response back *as* a GetAllNext response. Otherwise, we won't unswizzle it correctly. Nuke some additional __P macros.
2001-12-14 03:13:44 +03:00
bus_dma_tag_t dmatag;
bus_dmamap_t rqdmap;
bus_dmamap_t rsdmap;
bus_dmamap_t scdmap; /* FC only */
int splsaved;
int mboxwaiting;
u_int32_t islocked;
u_int32_t onintstack;
Create a kernel thread for Fibre Channel cards. This is the new way of doing business- modulo some startup spasms and peculiarities of the way kthreads are started (*after* configuration, weird) and some strangeness with the freeze/thaw code, what now happens is that any of Loop Down, LIP, Loop Reset or Port Datbase or Name Server Database Changed ASYNC events cause the queues to freeze for this channel. The arrival of a Loop UP is not relevant. What *is* relevant is that the Port Datbase or Name Server Changed async event indicate that it's okay to go and (re)evaluate the state of the FC link and (re)probe local loop and fabric membership. We have a kthread do this because it's *sooooo* much nicer to be able to sleep while doing the 130-250 mailbox commands it'll take to re-evaluate things. When the state is well known again, we can unfreeze the channel queues. Then, as commands start arriving, we simply can start them or bounce them with XS_SELTIMEOUT (if the device in question has gone away). Previously, we did lazy evaluation, which meant that if a change occurred, we would wait until the very *next* command to go rebuild stuff. The reason this is not sensible is: a) Even with sleeping, you can hang up your system because you might be making some poor stat(2) call pay the price of re-evaluating the whole fabric. b) If we ever really want to get to dynamic attachment/detachment, we should find out sooner, rather than later, where things get to. Split off ispminphys_1020 from ispminphys- a 1020 has a 24 bit limit- not anything newer. Re-enable LIPs and Loop Resets as async events- this allows the outer layer to set policy about them. Roll platform major && minor. Remove bogus waitq (no longer used). Remove callout entry in softc (no longer used). Define some shorthands for channels. Clean up a variety of cruft left over from the thorpej_scsipi changeover.
2001-05-26 01:45:55 +04:00
unsigned int : 26,
loop_checked : 1,
Long overdue- put locks on the usage of the mailbox command stuff. This keeps us from stomping on ourselves. Say we're in the middle of re-evaluating a loop (and sleeping) when a command completes, or a timer refires, and this *other* thread decides *it* wants to start re-evaluating the loop. Bad news. We have to be a bit careful- if we can't acquire the MBOX semaphore at interrupt level, we will simulate a host interface error. This is a bit of a temp workaround. There's some work underway driven by a NetBSD commercial user that will try and force most mailbox stuff into a part A/part B interrupt driven model.
2001-05-16 07:55:15 +04:00
mbox_wanted : 1,
mbox_locked : 1,
no_mbox_ints : 1,
Create a kernel thread for Fibre Channel cards. This is the new way of doing business- modulo some startup spasms and peculiarities of the way kthreads are started (*after* configuration, weird) and some strangeness with the freeze/thaw code, what now happens is that any of Loop Down, LIP, Loop Reset or Port Datbase or Name Server Database Changed ASYNC events cause the queues to freeze for this channel. The arrival of a Loop UP is not relevant. What *is* relevant is that the Port Datbase or Name Server Changed async event indicate that it's okay to go and (re)evaluate the state of the FC link and (re)probe local loop and fabric membership. We have a kthread do this because it's *sooooo* much nicer to be able to sleep while doing the 130-250 mailbox commands it'll take to re-evaluate things. When the state is well known again, we can unfreeze the channel queues. Then, as commands start arriving, we simply can start them or bounce them with XS_SELTIMEOUT (if the device in question has gone away). Previously, we did lazy evaluation, which meant that if a change occurred, we would wait until the very *next* command to go rebuild stuff. The reason this is not sensible is: a) Even with sleeping, you can hang up your system because you might be making some poor stat(2) call pay the price of re-evaluating the whole fabric. b) If we ever really want to get to dynamic attachment/detachment, we should find out sooner, rather than later, where things get to. Split off ispminphys_1020 from ispminphys- a 1020 has a 24 bit limit- not anything newer. Re-enable LIPs and Loop Resets as async events- this allows the outer layer to set policy about them. Roll platform major && minor. Remove bogus waitq (no longer used). Remove callout entry in softc (no longer used). Define some shorthands for channels. Clean up a variety of cruft left over from the thorpej_scsipi changeover.
2001-05-26 01:45:55 +04:00
paused : 1,
Long overdue- put locks on the usage of the mailbox command stuff. This keeps us from stomping on ourselves. Say we're in the middle of re-evaluating a loop (and sleeping) when a command completes, or a timer refires, and this *other* thread decides *it* wants to start re-evaluating the loop. Bad news. We have to be a bit careful- if we can't acquire the MBOX semaphore at interrupt level, we will simulate a host interface error. This is a bit of a temp workaround. There's some work underway driven by a NetBSD commercial user that will try and force most mailbox stuff into a part A/part B interrupt driven model.
2001-05-16 07:55:15 +04:00
blocked : 1;
Roll platform version. Keep a bitmap of 'discovered' devices for SCSI adapters so we'll know when it's time to switch from 'slow' command mode to normal. Change some settings for configuration printfs and debug levels. Redo the internal ispscsicmd return definitions and let isp_cmd translate them as approrpriate to NetBSD values. Remove the inline functions from here- they're now in isp_inline.h. Put in the start of the correct SWIZZLE/UNSWIZZLE functions.
1999-10-14 06:33:38 +04:00
union {
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
u_int64_t _wwn;
Roll platform version. Keep a bitmap of 'discovered' devices for SCSI adapters so we'll know when it's time to switch from 'slow' command mode to normal. Change some settings for configuration printfs and debug levels. Redo the internal ispscsicmd return definitions and let isp_cmd translate them as approrpriate to NetBSD values. Remove the inline functions from here- they're now in isp_inline.h. Put in the start of the correct SWIZZLE/UNSWIZZLE functions.
1999-10-14 06:33:38 +04:00
u_int16_t _discovered[2];
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
#define wwn_seed un._wwn
Roll platform version. Keep a bitmap of 'discovered' devices for SCSI adapters so we'll know when it's time to switch from 'slow' command mode to normal. Change some settings for configuration printfs and debug levels. Redo the internal ispscsicmd return definitions and let isp_cmd translate them as approrpriate to NetBSD values. Remove the inline functions from here- they're now in isp_inline.h. Put in the start of the correct SWIZZLE/UNSWIZZLE functions.
1999-10-14 06:33:38 +04:00
#define discovered un._discovered
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
} un;
Major restructuring for swizzling to the request queue and unswizzling from the response queue. Instead of the ad hoc ISP_SWIZZLE_REQUEST, we now have a complete set of inline functions in isp_inline.h. Each platform is responsible for providing just one of a set of ISP_IOX_{GET,PUT}{8,16,32} macros. The reason this needs to be done is that we need to have a single set of functions that will work correctly on multiple architectures for both little and big endian machines. It also needs to work correctly in the case that we have the request or response queues in memory that has to be treated specially (e.g., have ddi_dma_sync called on it for Solaris after we update it or before we read from it). One thing that falls out of this is that we no longer build requests in the request queue itself. Instead, we build the request locally (e.g., on the stack) and then as part of the swizzling operation, copy it to the request queue entry we've allocated. I thought long and hard about whether this was too expensive a change to make as it in a lot of cases requires an extra copy. On balance, the flexbility is worth it. With any luck, the entry that we build locally stays in a processor writeback cache (after all, it's only 64 bytes) so that the cost of actually flushing it to the memory area that is the shared queue with the PCI device is not all that expensive. We may examine this again and try to get clever in the future to try and avoid copies. Another change that falls out of this is that MEMORYBARRIER should be taken a lot more seriously. The macro ISP_ADD_REQUEST does a MEMORYBARRIER on the entry being added. But there had been many other places this had been missing. It's now very important that it be done. For NetBSD, it does a ddi_dmamap_sync as appropriate. This gets us out of the explicit ddi_dmamap_sync on the whole response queue that we did for SBus cards at each interrupt. Set things up so that platforms that cannot have an SBus don't get a lot of the SBus code checks (dead coded out). Additional changes: Fix a longstanding buglet of sorts. When we get an entry via isp_getrqentry, the iptr value that gets returned is the value we intend to eventually plug into the ISP registers as the entry *one past* the last one we've written- *not* the current entry we're updating. All along we've been calling sync functions on the wrong index value. Argh. The 'fix' here is to rename all 'iptr' variables as 'nxti' to remember that this is the 'next' pointer- not the current pointer. Devote a single bit to mboxbsy- and set aside bits for output mbox registers that we need to pick up- we can have at least one command which does not have any defined output registers (MBOX_EXECUTE_FIRMWARE). Explicitly decode GetAllNext SNS Response back *as* a GetAllNext response. Otherwise, we won't unswizzle it correctly. Nuke some additional __P macros.
2001-12-14 03:13:44 +03:00
u_int32_t threadwork;
struct proc * thread;
NetBSD OS specific routines and definitions.
1998-07-15 23:44:04 +04:00
};
Major restructuring for swizzling to the request queue and unswizzling from the response queue. Instead of the ad hoc ISP_SWIZZLE_REQUEST, we now have a complete set of inline functions in isp_inline.h. Each platform is responsible for providing just one of a set of ISP_IOX_{GET,PUT}{8,16,32} macros. The reason this needs to be done is that we need to have a single set of functions that will work correctly on multiple architectures for both little and big endian machines. It also needs to work correctly in the case that we have the request or response queues in memory that has to be treated specially (e.g., have ddi_dma_sync called on it for Solaris after we update it or before we read from it). One thing that falls out of this is that we no longer build requests in the request queue itself. Instead, we build the request locally (e.g., on the stack) and then as part of the swizzling operation, copy it to the request queue entry we've allocated. I thought long and hard about whether this was too expensive a change to make as it in a lot of cases requires an extra copy. On balance, the flexbility is worth it. With any luck, the entry that we build locally stays in a processor writeback cache (after all, it's only 64 bytes) so that the cost of actually flushing it to the memory area that is the shared queue with the PCI device is not all that expensive. We may examine this again and try to get clever in the future to try and avoid copies. Another change that falls out of this is that MEMORYBARRIER should be taken a lot more seriously. The macro ISP_ADD_REQUEST does a MEMORYBARRIER on the entry being added. But there had been many other places this had been missing. It's now very important that it be done. For NetBSD, it does a ddi_dmamap_sync as appropriate. This gets us out of the explicit ddi_dmamap_sync on the whole response queue that we did for SBus cards at each interrupt. Set things up so that platforms that cannot have an SBus don't get a lot of the SBus code checks (dead coded out). Additional changes: Fix a longstanding buglet of sorts. When we get an entry via isp_getrqentry, the iptr value that gets returned is the value we intend to eventually plug into the ISP registers as the entry *one past* the last one we've written- *not* the current entry we're updating. All along we've been calling sync functions on the wrong index value. Argh. The 'fix' here is to rename all 'iptr' variables as 'nxti' to remember that this is the 'next' pointer- not the current pointer. Devote a single bit to mboxbsy- and set aside bits for output mbox registers that we need to pick up- we can have at least one command which does not have any defined output registers (MBOX_EXECUTE_FIRMWARE). Explicitly decode GetAllNext SNS Response back *as* a GetAllNext response. Otherwise, we won't unswizzle it correctly. Nuke some additional __P macros.
2001-12-14 03:13:44 +03:00
#define isp_dmatag isp_osinfo.dmatag
#define isp_rqdmap isp_osinfo.rqdmap
#define isp_rsdmap isp_osinfo.rsdmap
#define isp_scdmap isp_osinfo.scdmap
Finally fix this driver to be sensible about the ENDIAN dance. It's not quite simply a question of the Qlogic being little endian and having to have stuff swapped on big endian machines- it also has to do with the fact that the SBus and PCI DMA layouts are wierd with respect to this. At any rate, now finally fixed- works on Mac G4, tested it on a SS10 for sparc, checked on alpha to see if I've broken anything, and as soon as I get another spare afternoon I'll finally install a sparc64 version which should just work (as it'll be like the Mac).
2000-12-09 11:06:31 +03:00
#define ISP_MUSTPOLL(isp) \
(isp->isp_osinfo.onintstack || isp->isp_osinfo.no_mbox_ints)
Roll platform version. Keep a bitmap of 'discovered' devices for SCSI adapters so we'll know when it's time to switch from 'slow' command mode to normal. Change some settings for configuration printfs and debug levels. Redo the internal ispscsicmd return definitions and let isp_cmd translate them as approrpriate to NetBSD values. Remove the inline functions from here- they're now in isp_inline.h. Put in the start of the correct SWIZZLE/UNSWIZZLE functions.
1999-10-14 06:33:38 +04:00
Create a kernel thread for Fibre Channel cards. This is the new way of doing business- modulo some startup spasms and peculiarities of the way kthreads are started (*after* configuration, weird) and some strangeness with the freeze/thaw code, what now happens is that any of Loop Down, LIP, Loop Reset or Port Datbase or Name Server Database Changed ASYNC events cause the queues to freeze for this channel. The arrival of a Loop UP is not relevant. What *is* relevant is that the Port Datbase or Name Server Changed async event indicate that it's okay to go and (re)evaluate the state of the FC link and (re)probe local loop and fabric membership. We have a kthread do this because it's *sooooo* much nicer to be able to sleep while doing the 130-250 mailbox commands it'll take to re-evaluate things. When the state is well known again, we can unfreeze the channel queues. Then, as commands start arriving, we simply can start them or bounce them with XS_SELTIMEOUT (if the device in question has gone away). Previously, we did lazy evaluation, which meant that if a change occurred, we would wait until the very *next* command to go rebuild stuff. The reason this is not sensible is: a) Even with sleeping, you can hang up your system because you might be making some poor stat(2) call pay the price of re-evaluating the whole fabric. b) If we ever really want to get to dynamic attachment/detachment, we should find out sooner, rather than later, where things get to. Split off ispminphys_1020 from ispminphys- a 1020 has a 24 bit limit- not anything newer. Re-enable LIPs and Loop Resets as async events- this allows the outer layer to set policy about them. Roll platform major && minor. Remove bogus waitq (no longer used). Remove callout entry in softc (no longer used). Define some shorthands for channels. Clean up a variety of cruft left over from the thorpej_scsipi changeover.
2001-05-26 01:45:55 +04:00
#define HANDLE_LOOPSTATE_IN_OUTER_LAYERS 1
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
/*
* Required Macros/Defines
*/
NetBSD OS specific routines and definitions.
1998-07-15 23:44:04 +04:00
Per helpful suggestion from LukeM- make INLINE usage consistent (and use correct __inline format).
2001-05-17 01:41:52 +04:00
#define INLINE __inline
NetBSD OS specific routines and definitions.
1998-07-15 23:44:04 +04:00
Redo fabric evaluation to not use GET ALL NEXT (GA_NXT). Switches seem to be trying to wriggle out of supporting this well. Instead, use GID_FT to get a list of Port IDs and then use GPN_ID/GNN_ID to find the port and node wwn. This should make working on fabrics a bit cleaner and more stable. This also caused some cleanup of SNS subcommand canonicalization so that we can actually check for FS_ACC and FS_RJT, and if we get an FS_RJT, print out the reason and explanation codes. We'll keep the old GA_NXT method around if people want to uncomment a controlling definition in ispvar.h. This also had us clean up ISPASYNC_FABRICDEV to use a local lportdb argument and to have the caller explicitly say that a device is at the end of the fabric list.
2002-04-05 03:38:43 +04:00
#define ISP2100_SCRLEN 0x800
roll revision, change default tagging to simple tags
1998-09-08 11:18:56 +04:00
bzero -> memset
2001-07-07 20:13:44 +04:00
#define MEMZERO(dst, amt) memset((dst), 0, (amt))
bcopy -> memcpy
2001-07-07 19:53:13 +04:00
#define MEMCPY(dst, src, amt) memcpy((dst), (src), (amt))
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
#define SNPRINTF snprintf
#define STRNCAT strncat
#define USEC_DELAY DELAY
Finally fix this driver to be sensible about the ENDIAN dance. It's not quite simply a question of the Qlogic being little endian and having to have stuff swapped on big endian machines- it also has to do with the fact that the SBus and PCI DMA layouts are wierd with respect to this. At any rate, now finally fixed- works on Mac G4, tested it on a SS10 for sparc, checked on alpha to see if I've broken anything, and as soon as I get another spare afternoon I'll finally install a sparc64 version which should just work (as it'll be like the Mac).
2000-12-09 11:06:31 +03:00
#define USEC_SLEEP(isp, x) \
if (!ISP_MUSTPOLL(isp)) \
ISP_UNLOCK(isp); \
DELAY(x); \
if (!ISP_MUSTPOLL(isp)) \
ISP_LOCK(isp)
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
#define NANOTIME_T struct timeval
#define GET_NANOTIME microtime
#define GET_NANOSEC(x) (((x)->tv_sec * 1000000 + (x)->tv_usec) * 1000)
#define NANOTIME_SUB isp_microtime_sub
#define MAXISPREQUEST(isp) 256
Major restructuring for swizzling to the request queue and unswizzling from the response queue. Instead of the ad hoc ISP_SWIZZLE_REQUEST, we now have a complete set of inline functions in isp_inline.h. Each platform is responsible for providing just one of a set of ISP_IOX_{GET,PUT}{8,16,32} macros. The reason this needs to be done is that we need to have a single set of functions that will work correctly on multiple architectures for both little and big endian machines. It also needs to work correctly in the case that we have the request or response queues in memory that has to be treated specially (e.g., have ddi_dma_sync called on it for Solaris after we update it or before we read from it). One thing that falls out of this is that we no longer build requests in the request queue itself. Instead, we build the request locally (e.g., on the stack) and then as part of the swizzling operation, copy it to the request queue entry we've allocated. I thought long and hard about whether this was too expensive a change to make as it in a lot of cases requires an extra copy. On balance, the flexbility is worth it. With any luck, the entry that we build locally stays in a processor writeback cache (after all, it's only 64 bytes) so that the cost of actually flushing it to the memory area that is the shared queue with the PCI device is not all that expensive. We may examine this again and try to get clever in the future to try and avoid copies. Another change that falls out of this is that MEMORYBARRIER should be taken a lot more seriously. The macro ISP_ADD_REQUEST does a MEMORYBARRIER on the entry being added. But there had been many other places this had been missing. It's now very important that it be done. For NetBSD, it does a ddi_dmamap_sync as appropriate. This gets us out of the explicit ddi_dmamap_sync on the whole response queue that we did for SBus cards at each interrupt. Set things up so that platforms that cannot have an SBus don't get a lot of the SBus code checks (dead coded out). Additional changes: Fix a longstanding buglet of sorts. When we get an entry via isp_getrqentry, the iptr value that gets returned is the value we intend to eventually plug into the ISP registers as the entry *one past* the last one we've written- *not* the current entry we're updating. All along we've been calling sync functions on the wrong index value. Argh. The 'fix' here is to rename all 'iptr' variables as 'nxti' to remember that this is the 'next' pointer- not the current pointer. Devote a single bit to mboxbsy- and set aside bits for output mbox registers that we need to pick up- we can have at least one command which does not have any defined output registers (MBOX_EXECUTE_FIRMWARE). Explicitly decode GetAllNext SNS Response back *as* a GetAllNext response. Otherwise, we won't unswizzle it correctly. Nuke some additional __P macros.
2001-12-14 03:13:44 +03:00
#define MEMORYBARRIER(isp, type, offset, size) \
switch (type) { \
case SYNC_REQUEST: \
{ \
off_t off = (off_t) offset * QENTRY_LEN; \
bus_dmamap_sync(isp->isp_dmatag, isp->isp_rqdmap, \
off, size, BUS_DMASYNC_PREWRITE); \
break; \
} \
case SYNC_RESULT: \
{ \
off_t off = (off_t) offset * QENTRY_LEN; \
bus_dmamap_sync(isp->isp_dmatag, isp->isp_rsdmap, \
off, size, BUS_DMASYNC_POSTREAD); \
break; \
} \
case SYNC_SFORDEV: \
{ \
Thanks to Jason Wright of OpenBSD- spotted that the offsets were all wrong for bus_dmamap_sync calls. They'd been blindly ported from Solaris which had *one* dma map for the entire control space, so offset was incremented for the Request, Response and FC Scratch spaces. Tsk. There are three maps in NetBSD. I should probably make them one anyway.
2002-03-31 21:56:54 +04:00
off_t off = (off_t) offset; \
Major restructuring for swizzling to the request queue and unswizzling from the response queue. Instead of the ad hoc ISP_SWIZZLE_REQUEST, we now have a complete set of inline functions in isp_inline.h. Each platform is responsible for providing just one of a set of ISP_IOX_{GET,PUT}{8,16,32} macros. The reason this needs to be done is that we need to have a single set of functions that will work correctly on multiple architectures for both little and big endian machines. It also needs to work correctly in the case that we have the request or response queues in memory that has to be treated specially (e.g., have ddi_dma_sync called on it for Solaris after we update it or before we read from it). One thing that falls out of this is that we no longer build requests in the request queue itself. Instead, we build the request locally (e.g., on the stack) and then as part of the swizzling operation, copy it to the request queue entry we've allocated. I thought long and hard about whether this was too expensive a change to make as it in a lot of cases requires an extra copy. On balance, the flexbility is worth it. With any luck, the entry that we build locally stays in a processor writeback cache (after all, it's only 64 bytes) so that the cost of actually flushing it to the memory area that is the shared queue with the PCI device is not all that expensive. We may examine this again and try to get clever in the future to try and avoid copies. Another change that falls out of this is that MEMORYBARRIER should be taken a lot more seriously. The macro ISP_ADD_REQUEST does a MEMORYBARRIER on the entry being added. But there had been many other places this had been missing. It's now very important that it be done. For NetBSD, it does a ddi_dmamap_sync as appropriate. This gets us out of the explicit ddi_dmamap_sync on the whole response queue that we did for SBus cards at each interrupt. Set things up so that platforms that cannot have an SBus don't get a lot of the SBus code checks (dead coded out). Additional changes: Fix a longstanding buglet of sorts. When we get an entry via isp_getrqentry, the iptr value that gets returned is the value we intend to eventually plug into the ISP registers as the entry *one past* the last one we've written- *not* the current entry we're updating. All along we've been calling sync functions on the wrong index value. Argh. The 'fix' here is to rename all 'iptr' variables as 'nxti' to remember that this is the 'next' pointer- not the current pointer. Devote a single bit to mboxbsy- and set aside bits for output mbox registers that we need to pick up- we can have at least one command which does not have any defined output registers (MBOX_EXECUTE_FIRMWARE). Explicitly decode GetAllNext SNS Response back *as* a GetAllNext response. Otherwise, we won't unswizzle it correctly. Nuke some additional __P macros.
2001-12-14 03:13:44 +03:00
bus_dmamap_sync(isp->isp_dmatag, isp->isp_scdmap, \
off, size, BUS_DMASYNC_PREWRITE); \
break; \
} \
case SYNC_SFORCPU: \
{ \
Thanks to Jason Wright of OpenBSD- spotted that the offsets were all wrong for bus_dmamap_sync calls. They'd been blindly ported from Solaris which had *one* dma map for the entire control space, so offset was incremented for the Request, Response and FC Scratch spaces. Tsk. There are three maps in NetBSD. I should probably make them one anyway.
2002-03-31 21:56:54 +04:00
off_t off = (off_t) offset; \
Major restructuring for swizzling to the request queue and unswizzling from the response queue. Instead of the ad hoc ISP_SWIZZLE_REQUEST, we now have a complete set of inline functions in isp_inline.h. Each platform is responsible for providing just one of a set of ISP_IOX_{GET,PUT}{8,16,32} macros. The reason this needs to be done is that we need to have a single set of functions that will work correctly on multiple architectures for both little and big endian machines. It also needs to work correctly in the case that we have the request or response queues in memory that has to be treated specially (e.g., have ddi_dma_sync called on it for Solaris after we update it or before we read from it). One thing that falls out of this is that we no longer build requests in the request queue itself. Instead, we build the request locally (e.g., on the stack) and then as part of the swizzling operation, copy it to the request queue entry we've allocated. I thought long and hard about whether this was too expensive a change to make as it in a lot of cases requires an extra copy. On balance, the flexbility is worth it. With any luck, the entry that we build locally stays in a processor writeback cache (after all, it's only 64 bytes) so that the cost of actually flushing it to the memory area that is the shared queue with the PCI device is not all that expensive. We may examine this again and try to get clever in the future to try and avoid copies. Another change that falls out of this is that MEMORYBARRIER should be taken a lot more seriously. The macro ISP_ADD_REQUEST does a MEMORYBARRIER on the entry being added. But there had been many other places this had been missing. It's now very important that it be done. For NetBSD, it does a ddi_dmamap_sync as appropriate. This gets us out of the explicit ddi_dmamap_sync on the whole response queue that we did for SBus cards at each interrupt. Set things up so that platforms that cannot have an SBus don't get a lot of the SBus code checks (dead coded out). Additional changes: Fix a longstanding buglet of sorts. When we get an entry via isp_getrqentry, the iptr value that gets returned is the value we intend to eventually plug into the ISP registers as the entry *one past* the last one we've written- *not* the current entry we're updating. All along we've been calling sync functions on the wrong index value. Argh. The 'fix' here is to rename all 'iptr' variables as 'nxti' to remember that this is the 'next' pointer- not the current pointer. Devote a single bit to mboxbsy- and set aside bits for output mbox registers that we need to pick up- we can have at least one command which does not have any defined output registers (MBOX_EXECUTE_FIRMWARE). Explicitly decode GetAllNext SNS Response back *as* a GetAllNext response. Otherwise, we won't unswizzle it correctly. Nuke some additional __P macros.
2001-12-14 03:13:44 +03:00
bus_dmamap_sync(isp->isp_dmatag, isp->isp_scdmap, \
off, size, BUS_DMASYNC_POSTREAD); \
break; \
} \
case SYNC_REG: \
default: \
break; \
}
cross platform define MEMZERO added
1999-01-10 08:04:18 +03:00
Long overdue- put locks on the usage of the mailbox command stuff. This keeps us from stomping on ourselves. Say we're in the middle of re-evaluating a loop (and sleeping) when a command completes, or a timer refires, and this *other* thread decides *it* wants to start re-evaluating the loop. Bad news. We have to be a bit careful- if we can't acquire the MBOX semaphore at interrupt level, we will simulate a host interface error. This is a bit of a temp workaround. There's some work underway driven by a NetBSD commercial user that will try and force most mailbox stuff into a part A/part B interrupt driven model.
2001-05-16 07:55:15 +04:00
#define MBOX_ACQUIRE(isp) \
if (isp->isp_osinfo.onintstack) { \
if (isp->isp_osinfo.mbox_locked) { \
mbp->param[0] = MBOX_HOST_INTERFACE_ERROR; \
isp_prt(isp, ISP_LOGERR, "cannot acquire MBOX lock"); \
return; \
} \
} else { \
while (isp->isp_osinfo.mbox_locked) { \
isp->isp_osinfo.mbox_wanted = 1; \
(void) tsleep(&isp->isp_osinfo.mboxwaiting+1, \
PRIBIO, "mbox_acquire", 0); \
} \
} \
isp->isp_osinfo.mbox_locked = 1
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
#define MBOX_WAIT_COMPLETE isp_wait_complete
NetBSD OS specific routines and definitions.
1998-07-15 23:44:04 +04:00
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
#define MBOX_NOTIFY_COMPLETE(isp) \
if (isp->isp_osinfo.mboxwaiting) { \
isp->isp_osinfo.mboxwaiting = 0; \
wakeup(&isp->isp_osinfo.mboxwaiting); \
} \
isp->isp_mboxbsy = 0
Roll platform minor. We are now always supporting ISP2100_FABRIC. Add in new MBOX_WAIT_COMPLETE/MBOX_NOTIFY_COMPLETE macros (ready for SMPizing). Define STRNCAT inline for our usage. Stealing a bit of the xs_status flags to maintain command state that is pertinent for the hBA- should really reserve them for private usage in scsipiconf.h.
2000-07-06 02:25:06 +04:00
Long overdue- put locks on the usage of the mailbox command stuff. This keeps us from stomping on ourselves. Say we're in the middle of re-evaluating a loop (and sleeping) when a command completes, or a timer refires, and this *other* thread decides *it* wants to start re-evaluating the loop. Bad news. We have to be a bit careful- if we can't acquire the MBOX semaphore at interrupt level, we will simulate a host interface error. This is a bit of a temp workaround. There's some work underway driven by a NetBSD commercial user that will try and force most mailbox stuff into a part A/part B interrupt driven model.
2001-05-16 07:55:15 +04:00
#define MBOX_RELEASE(isp) \
if (isp->isp_osinfo.mbox_wanted) { \
isp->isp_osinfo.mbox_wanted = 0; \
wakeup(&isp->isp_osinfo.mboxwaiting+1); \
} \
isp->isp_osinfo.mbox_locked = 0
Roll platform minor. We are now always supporting ISP2100_FABRIC. Add in new MBOX_WAIT_COMPLETE/MBOX_NOTIFY_COMPLETE macros (ready for SMPizing). Define STRNCAT inline for our usage. Stealing a bit of the xs_status flags to maintain command state that is pertinent for the hBA- should really reserve them for private usage in scsipiconf.h.
2000-07-06 02:25:06 +04:00
Implement F/W crashdumps for 2200 && 2300 cards (kernel compile option). Distinguish between 2312 and 2300 cards (they *are* different). Enable RIO (Reduced Interrupt Operation) for the LVD cards (hey- I've seen batched completions of the 30 commands at a time with this,....)... If we get a Port Logout on local loop topologies, we have to force the f/w to log back in. The easiest way (for us) to do this is to force a LIP. This also will wake up the disk that probably just had a f/w crash. Implement mailbox 'continuations'- this allows interrupts to re-drive a mailbox command if it's one that just essentially repeats the previous mailbox command (e.g., f/w download). This saves a boatload of sleep/wakeup twitches. If we're not a 2300 and we're about to return with a 'bogus interrupt'- check the semaphore register to be non-zero at all and outgoing mailbox 0- this seems to be where some of the lost ISP1080 commands came from.
2002-02-22 01:32:40 +03:00
#define FC_SCRATCH_ACQUIRE(isp)
#define FC_SCRATCH_RELEASE(isp)
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
#ifndef SCSI_GOOD
#define SCSI_GOOD 0x0
#endif
#ifndef SCSI_CHECK
#define SCSI_CHECK 0x2
#endif
#ifndef SCSI_BUSY
#define SCSI_BUSY 0x8
#endif
#ifndef SCSI_QFULL
#define SCSI_QFULL 0x28
#endif
NetBSD OS specific routines and definitions.
1998-07-15 23:44:04 +04:00
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
#define XS_T struct scsipi_xfer
Pull up the thorpej_scsipi branch to main branch. This is a completely rewritten scsipi_xfer execution engine, and the associated changes to HBA drivers. Overview of changes & features: - All xfers are queued in the mid-layer, rather than doing so in an ad-hoc fashion in individual adapter drivers. - Adapter/channel resource management in the mid-layer, avoids even trying to start running an xfer if the adapter/channel doesn't have the resources. - Better communication between the mid-layer and the adapters. - Asynchronous event notification mechanism from adapter to mid-layer and peripherals. - Better peripheral queue management: freeze/thaw, sorted requeueing during recovery, etc. - Clean separation of peripherals, adapters, and adapter channels (no more scsipi_link). - Kernel thread for each scsipi_channel makes error recovery much easier (no more dealing with interrupt context when recovering from an error). - Mid-layer support for tagged queueing: commands can have the tag type set explicitly, tag IDs are allocated in the mid-layer (thus eliminating the need to use buggy tag ID allocation schemes in many adapter drivers). - support for QUEUE FULL and CHECK CONDITION status in mid-layer; the command will be requeued, or a REQUEST SENSE will be sent as appropriate. Just before the merge syssrc has been tagged with thorpej_scsipi_beforemerge
2001-04-25 21:53:04 +04:00
#define XS_CHANNEL(xs) ((int) (xs)->xs_periph->periph_channel->chan_channel)
#define XS_ISP(xs) \
((void *)(xs)->xs_periph->periph_channel->chan_adapter->adapt_dev)
#define XS_LUN(xs) ((int) (xs)->xs_periph->periph_lun)
#define XS_TGT(xs) ((int) (xs)->xs_periph->periph_target)
Do a fairly large internal restructuring to accomodate dual-bus host adapters (e.g., the 1240). Include the new 1080/1240 NVRAM layout reading code. Some moderately significant mailbox changes were necessary also to accomodate a second channel.
1999-05-12 22:59:23 +04:00
#define XS_CDBP(xs) ((caddr_t) (xs)->cmd)
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
#define XS_CDBLEN(xs) (xs)->cmdlen
#define XS_XFRLEN(xs) (xs)->datalen
NetBSD OS specific routines and definitions.
1998-07-15 23:44:04 +04:00
#define XS_TIME(xs) (xs)->timeout
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
#define XS_RESID(xs) (xs)->resid
#define XS_STSP(xs) (&(xs)->status)
NetBSD OS specific routines and definitions.
1998-07-15 23:44:04 +04:00
#define XS_SNSP(xs) (&(xs)->sense.scsi_sense)
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
#define XS_SNSLEN(xs) (sizeof (xs)->sense)
NetBSD OS specific routines and definitions.
1998-07-15 23:44:04 +04:00
#define XS_SNSKEY(xs) ((xs)->sense.scsi_sense.flags)
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
#define XS_TAG_P(ccb) (((xs)->xs_control & XS_CTL_POLL) != 0)
#define XS_TAG_TYPE(xs) \
Make our policy WRT tagged queueing consistent and sane: ordered tags for sync writes, simple tags for all else. Should make ahc and adv a bit more reliable (metadata writes won't get reordered incorrectly...) and isp a bit more performant (it was using ordered tags all the time).
2000-08-12 01:31:19 +04:00
(((xs)->xs_control & XS_CTL_URGENT) ? REQFLAG_HTAG : \
((xs)->bp && xs->bp->b_flags & B_ASYNC) ? REQFLAG_STAG : REQFLAG_OTAG)
NetBSD OS specific routines and definitions.
1998-07-15 23:44:04 +04:00
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
#define XS_SETERR(xs, v) (xs)->error = v
NetBSD OS specific routines and definitions.
1998-07-15 23:44:04 +04:00
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
# define HBA_NOERROR XS_NOERROR
# define HBA_BOTCH XS_DRIVER_STUFFUP
# define HBA_CMDTIMEOUT XS_TIMEOUT
# define HBA_SELTIMEOUT XS_SELTIMEOUT
# define HBA_TGTBSY XS_BUSY
# define HBA_BUSRESET XS_RESET
# define HBA_ABORTED XS_DRIVER_STUFFUP
# define HBA_DATAOVR XS_DRIVER_STUFFUP
# define HBA_ARQFAIL XS_DRIVER_STUFFUP
Roll platform minor. We are now always supporting ISP2100_FABRIC. Add in new MBOX_WAIT_COMPLETE/MBOX_NOTIFY_COMPLETE macros (ready for SMPizing). Define STRNCAT inline for our usage. Stealing a bit of the xs_status flags to maintain command state that is pertinent for the hBA- should really reserve them for private usage in scsipiconf.h.
2000-07-06 02:25:06 +04:00
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
#define XS_ERR(xs) (xs)->error
Roll platform minor. We are now always supporting ISP2100_FABRIC. Add in new MBOX_WAIT_COMPLETE/MBOX_NOTIFY_COMPLETE macros (ready for SMPizing). Define STRNCAT inline for our usage. Stealing a bit of the xs_status flags to maintain command state that is pertinent for the hBA- should really reserve them for private usage in scsipiconf.h.
2000-07-06 02:25:06 +04:00
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
#define XS_NOERR(xs) (xs)->error == XS_NOERROR
Roll platform minor. We are now always supporting ISP2100_FABRIC. Add in new MBOX_WAIT_COMPLETE/MBOX_NOTIFY_COMPLETE macros (ready for SMPizing). Define STRNCAT inline for our usage. Stealing a bit of the xs_status flags to maintain command state that is pertinent for the hBA- should really reserve them for private usage in scsipiconf.h.
2000-07-06 02:25:06 +04:00
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
#define XS_INITERR(xs) (xs)->error = 0, XS_CMD_S_CLEAR(xs)
Roll platform minor. We are now always supporting ISP2100_FABRIC. Add in new MBOX_WAIT_COMPLETE/MBOX_NOTIFY_COMPLETE macros (ready for SMPizing). Define STRNCAT inline for our usage. Stealing a bit of the xs_status flags to maintain command state that is pertinent for the hBA- should really reserve them for private usage in scsipiconf.h.
2000-07-06 02:25:06 +04:00
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
#define XS_SAVE_SENSE(xs, sp) \
Ian McDonnell <apriori@world.std.com> made me see the error of my ways. *Cough*. Ahem. I hadn't been setting error to XS_SENSE when there was a check condition. Major blunder.
2001-01-05 10:02:00 +03:00
if (xs->error == XS_NOERROR) { \
xs->error = XS_SENSE; \
} \
bcopy -> memcpy
2001-07-07 19:53:13 +04:00
memcpy(&(xs)->sense, sp->req_sense_data, \
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
imin(XS_SNSLEN(xs), sp->req_sense_len))
Roll platform minor. We are now always supporting ISP2100_FABRIC. Add in new MBOX_WAIT_COMPLETE/MBOX_NOTIFY_COMPLETE macros (ready for SMPizing). Define STRNCAT inline for our usage. Stealing a bit of the xs_status flags to maintain command state that is pertinent for the hBA- should really reserve them for private usage in scsipiconf.h.
2000-07-06 02:25:06 +04:00
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
#define XS_SET_STATE_STAT(a, b, c)
NetBSD OS specific routines and definitions.
1998-07-15 23:44:04 +04:00
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
#define DEFAULT_IID(x) 7
#define DEFAULT_LOOPID(x) 108
#define DEFAULT_NODEWWN(isp) (isp)->isp_osinfo.un._wwn
Make changes relevant to changes in WWN defaults. Also fix egregious bug where we would never decrement the islocked recursion counter. I guess this means that we don't recurse on this platform! All of this should go away when we have real lock primitives to use in drivers.
2000-10-16 09:18:15 +04:00
#define DEFAULT_PORTWWN(isp) (isp)->isp_osinfo.un._wwn
#define ISP_NODEWWN(isp) FCPARAM(isp)->isp_nodewwn
#define ISP_PORTWWN(isp) FCPARAM(isp)->isp_portwwn
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
Finally fix this driver to be sensible about the ENDIAN dance. It's not quite simply a question of the Qlogic being little endian and having to have stuff swapped on big endian machines- it also has to do with the fact that the SBus and PCI DMA layouts are wierd with respect to this. At any rate, now finally fixed- works on Mac G4, tested it on a SS10 for sparc, checked on alpha to see if I've broken anything, and as soon as I get another spare afternoon I'll finally install a sparc64 version which should just work (as it'll be like the Mac).
2000-12-09 11:06:31 +03:00
#if _BYTE_ORDER == _BIG_ENDIAN
Major restructuring for swizzling to the request queue and unswizzling from the response queue. Instead of the ad hoc ISP_SWIZZLE_REQUEST, we now have a complete set of inline functions in isp_inline.h. Each platform is responsible for providing just one of a set of ISP_IOX_{GET,PUT}{8,16,32} macros. The reason this needs to be done is that we need to have a single set of functions that will work correctly on multiple architectures for both little and big endian machines. It also needs to work correctly in the case that we have the request or response queues in memory that has to be treated specially (e.g., have ddi_dma_sync called on it for Solaris after we update it or before we read from it). One thing that falls out of this is that we no longer build requests in the request queue itself. Instead, we build the request locally (e.g., on the stack) and then as part of the swizzling operation, copy it to the request queue entry we've allocated. I thought long and hard about whether this was too expensive a change to make as it in a lot of cases requires an extra copy. On balance, the flexbility is worth it. With any luck, the entry that we build locally stays in a processor writeback cache (after all, it's only 64 bytes) so that the cost of actually flushing it to the memory area that is the shared queue with the PCI device is not all that expensive. We may examine this again and try to get clever in the future to try and avoid copies. Another change that falls out of this is that MEMORYBARRIER should be taken a lot more seriously. The macro ISP_ADD_REQUEST does a MEMORYBARRIER on the entry being added. But there had been many other places this had been missing. It's now very important that it be done. For NetBSD, it does a ddi_dmamap_sync as appropriate. This gets us out of the explicit ddi_dmamap_sync on the whole response queue that we did for SBus cards at each interrupt. Set things up so that platforms that cannot have an SBus don't get a lot of the SBus code checks (dead coded out). Additional changes: Fix a longstanding buglet of sorts. When we get an entry via isp_getrqentry, the iptr value that gets returned is the value we intend to eventually plug into the ISP registers as the entry *one past* the last one we've written- *not* the current entry we're updating. All along we've been calling sync functions on the wrong index value. Argh. The 'fix' here is to rename all 'iptr' variables as 'nxti' to remember that this is the 'next' pointer- not the current pointer. Devote a single bit to mboxbsy- and set aside bits for output mbox registers that we need to pick up- we can have at least one command which does not have any defined output registers (MBOX_EXECUTE_FIRMWARE). Explicitly decode GetAllNext SNS Response back *as* a GetAllNext response. Otherwise, we won't unswizzle it correctly. Nuke some additional __P macros.
2001-12-14 03:13:44 +03:00
#ifdef ISP_SBUS_SUPPORTED
#define ISP_IOXPUT_8(isp, s, d) *(d) = s
#define ISP_IOXPUT_16(isp, s, d) \
*(d) = (isp->isp_bustype == ISP_BT_SBUS)? s : bswap16(s)
#define ISP_IOXPUT_32(isp, s, d) \
*(d) = (isp->isp_bustype == ISP_BT_SBUS)? s : bswap32(s)
#define ISP_IOXGET_8(isp, s, d) d = (*((u_int8_t *)s))
#define ISP_IOXGET_16(isp, s, d) \
d = (isp->isp_bustype == ISP_BT_SBUS)? \
*((u_int16_t *)s) : bswap16(*((u_int16_t *)s))
#define ISP_IOXGET_32(isp, s, d) \
d = (isp->isp_bustype == ISP_BT_SBUS)? \
*((u_int32_t *)s) : bswap32(*((u_int32_t *)s))
#else
#define ISP_IOXPUT_8(isp, s, d) *(d) = s
#define ISP_IOXPUT_16(isp, s, d) *(d) = bswap16(s)
#define ISP_IOXPUT_32(isp, s, d) *(d) = bswap32(s)
#define ISP_IOXGET_8(isp, s, d) d = (*((u_int8_t *)s))
#define ISP_IOXGET_16(isp, s, d) d = bswap16(*((u_int16_t *)s))
#define ISP_IOXGET_32(isp, s, d) d = bswap32(*((u_int32_t *)s))
#endif
Finally fix this driver to be sensible about the ENDIAN dance. It's not quite simply a question of the Qlogic being little endian and having to have stuff swapped on big endian machines- it also has to do with the fact that the SBus and PCI DMA layouts are wierd with respect to this. At any rate, now finally fixed- works on Mac G4, tested it on a SS10 for sparc, checked on alpha to see if I've broken anything, and as soon as I get another spare afternoon I'll finally install a sparc64 version which should just work (as it'll be like the Mac).
2000-12-09 11:06:31 +03:00
#define ISP_SWIZZLE_NVRAM_WORD(isp, rp) *rp = bswap16(*rp)
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
#else
Major restructuring for swizzling to the request queue and unswizzling from the response queue. Instead of the ad hoc ISP_SWIZZLE_REQUEST, we now have a complete set of inline functions in isp_inline.h. Each platform is responsible for providing just one of a set of ISP_IOX_{GET,PUT}{8,16,32} macros. The reason this needs to be done is that we need to have a single set of functions that will work correctly on multiple architectures for both little and big endian machines. It also needs to work correctly in the case that we have the request or response queues in memory that has to be treated specially (e.g., have ddi_dma_sync called on it for Solaris after we update it or before we read from it). One thing that falls out of this is that we no longer build requests in the request queue itself. Instead, we build the request locally (e.g., on the stack) and then as part of the swizzling operation, copy it to the request queue entry we've allocated. I thought long and hard about whether this was too expensive a change to make as it in a lot of cases requires an extra copy. On balance, the flexbility is worth it. With any luck, the entry that we build locally stays in a processor writeback cache (after all, it's only 64 bytes) so that the cost of actually flushing it to the memory area that is the shared queue with the PCI device is not all that expensive. We may examine this again and try to get clever in the future to try and avoid copies. Another change that falls out of this is that MEMORYBARRIER should be taken a lot more seriously. The macro ISP_ADD_REQUEST does a MEMORYBARRIER on the entry being added. But there had been many other places this had been missing. It's now very important that it be done. For NetBSD, it does a ddi_dmamap_sync as appropriate. This gets us out of the explicit ddi_dmamap_sync on the whole response queue that we did for SBus cards at each interrupt. Set things up so that platforms that cannot have an SBus don't get a lot of the SBus code checks (dead coded out). Additional changes: Fix a longstanding buglet of sorts. When we get an entry via isp_getrqentry, the iptr value that gets returned is the value we intend to eventually plug into the ISP registers as the entry *one past* the last one we've written- *not* the current entry we're updating. All along we've been calling sync functions on the wrong index value. Argh. The 'fix' here is to rename all 'iptr' variables as 'nxti' to remember that this is the 'next' pointer- not the current pointer. Devote a single bit to mboxbsy- and set aside bits for output mbox registers that we need to pick up- we can have at least one command which does not have any defined output registers (MBOX_EXECUTE_FIRMWARE). Explicitly decode GetAllNext SNS Response back *as* a GetAllNext response. Otherwise, we won't unswizzle it correctly. Nuke some additional __P macros.
2001-12-14 03:13:44 +03:00
#define ISP_IOXPUT_8(isp, s, d) *(d) = s
#define ISP_IOXPUT_16(isp, s, d) *(d) = s
#define ISP_IOXPUT_32(isp, s, d) *(d) = s
#define ISP_IOXGET_8(isp, s, d) d = *(s)
#define ISP_IOXGET_16(isp, s, d) d = *(s)
#define ISP_IOXGET_32(isp, s, d) d = *(s)
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
#define ISP_SWIZZLE_NVRAM_WORD(isp, rp)
#endif
NetBSD OS specific routines and definitions.
1998-07-15 23:44:04 +04:00
/*
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
* Includes of common header files
NetBSD OS specific routines and definitions.
1998-07-15 23:44:04 +04:00
*/
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
#include <dev/ic/ispreg.h>
#include <dev/ic/ispvar.h>
#include <dev/ic/ispmbox.h>
Include isp_ioctl.h.
2001-04-11 01:52:15 +04:00
#include <dev/ic/isp_ioctl.h>
NetBSD OS specific routines and definitions.
1998-07-15 23:44:04 +04:00
/*
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
* isp_osinfo definitions, extensions and shorthand.
NetBSD OS specific routines and definitions.
1998-07-15 23:44:04 +04:00
*/
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
#define isp_name isp_osinfo._dev.dv_xname
#define isp_unit isp_osinfo._dev.dv_unit
Create a kernel thread for Fibre Channel cards. This is the new way of doing business- modulo some startup spasms and peculiarities of the way kthreads are started (*after* configuration, weird) and some strangeness with the freeze/thaw code, what now happens is that any of Loop Down, LIP, Loop Reset or Port Datbase or Name Server Database Changed ASYNC events cause the queues to freeze for this channel. The arrival of a Loop UP is not relevant. What *is* relevant is that the Port Datbase or Name Server Changed async event indicate that it's okay to go and (re)evaluate the state of the FC link and (re)probe local loop and fabric membership. We have a kthread do this because it's *sooooo* much nicer to be able to sleep while doing the 130-250 mailbox commands it'll take to re-evaluate things. When the state is well known again, we can unfreeze the channel queues. Then, as commands start arriving, we simply can start them or bounce them with XS_SELTIMEOUT (if the device in question has gone away). Previously, we did lazy evaluation, which meant that if a change occurred, we would wait until the very *next* command to go rebuild stuff. The reason this is not sensible is: a) Even with sleeping, you can hang up your system because you might be making some poor stat(2) call pay the price of re-evaluating the whole fabric. b) If we ever really want to get to dynamic attachment/detachment, we should find out sooner, rather than later, where things get to. Split off ispminphys_1020 from ispminphys- a 1020 has a 24 bit limit- not anything newer. Re-enable LIPs and Loop Resets as async events- this allows the outer layer to set policy about them. Roll platform major && minor. Remove bogus waitq (no longer used). Remove callout entry in softc (no longer used). Define some shorthands for channels. Clean up a variety of cruft left over from the thorpej_scsipi changeover.
2001-05-26 01:45:55 +04:00
#define isp_chanA isp_osinfo._chan
#define isp_chanB isp_osinfo._chan_b
NetBSD OS specific routines and definitions.
1998-07-15 23:44:04 +04:00
Roll platform version. Keep a bitmap of 'discovered' devices for SCSI adapters so we'll know when it's time to switch from 'slow' command mode to normal. Change some settings for configuration printfs and debug levels. Redo the internal ispscsicmd return definitions and let isp_cmd translate them as approrpriate to NetBSD values. Remove the inline functions from here- they're now in isp_inline.h. Put in the start of the correct SWIZZLE/UNSWIZZLE functions.
1999-10-14 06:33:38 +04:00
/*
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
* Driver prototypes..
Roll platform version. Keep a bitmap of 'discovered' devices for SCSI adapters so we'll know when it's time to switch from 'slow' command mode to normal. Change some settings for configuration printfs and debug levels. Redo the internal ispscsicmd return definitions and let isp_cmd translate them as approrpriate to NetBSD values. Remove the inline functions from here- they're now in isp_inline.h. Put in the start of the correct SWIZZLE/UNSWIZZLE functions.
1999-10-14 06:33:38 +04:00
*/
ANSIfy source.
2001-03-14 08:44:46 +03:00
void isp_attach(struct ispsoftc *);
void isp_uninit(struct ispsoftc *);
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
Per helpful suggestion from LukeM- make INLINE usage consistent (and use correct __inline format).
2001-05-17 01:41:52 +04:00
static INLINE void isp_lock(struct ispsoftc *);
static INLINE void isp_unlock(struct ispsoftc *);
static INLINE char *strncat(char *, const char *, size_t);
static INLINE u_int64_t
ANSIfy source.
2001-03-14 08:44:46 +03:00
isp_microtime_sub(struct timeval *, struct timeval *);
static void isp_wait_complete(struct ispsoftc *);
NetBSD OS specific routines and definitions.
1998-07-15 23:44:04 +04:00
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
/*
* Driver wide data...
*/
NetBSD OS specific routines and definitions.
1998-07-15 23:44:04 +04:00
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
/*
* Locking macros...
*/
#define ISP_LOCK isp_lock
#define ISP_UNLOCK isp_unlock
Add a maintenance note. Move the single bit tags of islocked and onintstack to be real integers. Add ISP_ILOCK/ISP_IUNLOCK macros. Fix the isp_lock/isp_unlock inlines to stop being so embarrassingly in error. Why, or why, can't I have mutex_enter/mutex_exit, pretty please?
2000-08-14 11:10:09 +04:00
#define ISP_ILOCK(x) isp_lock(x); isp->isp_osinfo.onintstack++
#define ISP_IUNLOCK(x) isp->isp_osinfo.onintstack--; isp_unlock(x)
NetBSD OS specific routines and definitions.
1998-07-15 23:44:04 +04:00
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
/*
* Platform private flags
*/
Do a fairly large internal restructuring to accomodate dual-bus host adapters (e.g., the 1240). Include the new 1080/1240 NVRAM layout reading code. Some moderately significant mailbox changes were necessary also to accomodate a second channel.
1999-05-12 22:59:23 +04:00
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
#define XS_PSTS_INWDOG 0x10000000
#define XS_PSTS_GRACE 0x20000000
#define XS_PSTS_ALL 0x30000000
NetBSD OS specific routines and definitions.
1998-07-15 23:44:04 +04:00
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
#define XS_CMD_S_WDOG(xs) (xs)->xs_status |= XS_PSTS_INWDOG
#define XS_CMD_C_WDOG(xs) (xs)->xs_status &= ~XS_PSTS_INWDOG
#define XS_CMD_WDOG_P(xs) (((xs)->xs_status & XS_PSTS_INWDOG) != 0)
NetBSD OS specific routines and definitions.
1998-07-15 23:44:04 +04:00
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
#define XS_CMD_S_GRACE(xs) (xs)->xs_status |= XS_PSTS_GRACE
#define XS_CMD_C_GRACE(xs) (xs)->xs_status &= ~XS_PSTS_GRACE
#define XS_CMD_GRACE_P(xs) (((xs)->xs_status & XS_PSTS_GRACE) != 0)
#define XS_CMD_S_DONE(xs) (xs)->xs_status |= XS_STS_DONE
#define XS_CMD_C_DONE(xs) (xs)->xs_status &= ~XS_STS_DONE
#define XS_CMD_DONE_P(xs) (((xs)->xs_status & XS_STS_DONE) != 0)
Update the driver with some infrastructure for the 1080. Fix an embarrassing clock botch bug. Additional infrastructure for PDB change stuff.
1999-03-17 09:15:47 +03:00
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
#define XS_CMD_S_CLEAR(xs) (xs)->xs_status &= ~XS_PSTS_ALL
Update the driver with some infrastructure for the 1080. Fix an embarrassing clock botch bug. Additional infrastructure for PDB change stuff.
1999-03-17 09:15:47 +03:00
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
/*
* Platform specific 'inline' or support functions
*/
Per helpful suggestion from LukeM- make INLINE usage consistent (and use correct __inline format).
2001-05-17 01:41:52 +04:00
static INLINE void
ANSIfy source.
2001-03-14 08:44:46 +03:00
isp_lock(struct ispsoftc *isp)
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
{
int s = splbio();
Add a maintenance note. Move the single bit tags of islocked and onintstack to be real integers. Add ISP_ILOCK/ISP_IUNLOCK macros. Fix the isp_lock/isp_unlock inlines to stop being so embarrassingly in error. Why, or why, can't I have mutex_enter/mutex_exit, pretty please?
2000-08-14 11:10:09 +04:00
if (isp->isp_osinfo.islocked++ == 0) {
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
isp->isp_osinfo.splsaved = s;
} else {
splx(s);
}
}
Per helpful suggestion from LukeM- make INLINE usage consistent (and use correct __inline format).
2001-05-17 01:41:52 +04:00
static INLINE void
ANSIfy source.
2001-03-14 08:44:46 +03:00
isp_unlock(struct ispsoftc *isp)
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
{
Make changes relevant to changes in WWN defaults. Also fix egregious bug where we would never decrement the islocked recursion counter. I guess this means that we don't recurse on this platform! All of this should go away when we have real lock primitives to use in drivers.
2000-10-16 09:18:15 +04:00
if (isp->isp_osinfo.islocked-- <= 1) {
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
isp->isp_osinfo.islocked = 0;
splx(isp->isp_osinfo.splsaved);
}
}
Roll platform version. Keep a bitmap of 'discovered' devices for SCSI adapters so we'll know when it's time to switch from 'slow' command mode to normal. Change some settings for configuration printfs and debug levels. Redo the internal ispscsicmd return definitions and let isp_cmd translate them as approrpriate to NetBSD values. Remove the inline functions from here- they're now in isp_inline.h. Put in the start of the correct SWIZZLE/UNSWIZZLE functions.
1999-10-14 06:33:38 +04:00
Per helpful suggestion from LukeM- make INLINE usage consistent (and use correct __inline format).
2001-05-17 01:41:52 +04:00
static INLINE char *
ANSIfy source.
2001-03-14 08:44:46 +03:00
strncat(char *d, const char *s, size_t c)
Roll platform minor. We are now always supporting ISP2100_FABRIC. Add in new MBOX_WAIT_COMPLETE/MBOX_NOTIFY_COMPLETE macros (ready for SMPizing). Define STRNCAT inline for our usage. Stealing a bit of the xs_status flags to maintain command state that is pertinent for the hBA- should really reserve them for private usage in scsipiconf.h.
2000-07-06 02:25:06 +04:00
{
char *t = d;
if (c) {
while (*d)
d++;
while ((*d++ = *s++)) {
if (--c == 0) {
*d = '\0';
break;
}
}
}
return (t);
}
Per helpful suggestion from LukeM- make INLINE usage consistent (and use correct __inline format).
2001-05-17 01:41:52 +04:00
static INLINE u_int64_t
ANSIfy source.
2001-03-14 08:44:46 +03:00
isp_microtime_sub(struct timeval *b, struct timeval *a)
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
{
struct timeval x;
u_int64_t elapsed;
timersub(b, a, &x);
elapsed = GET_NANOSEC(&x);
if (elapsed == 0)
elapsed++;
return (elapsed);
}
Roll platform minor. We are now always supporting ISP2100_FABRIC. Add in new MBOX_WAIT_COMPLETE/MBOX_NOTIFY_COMPLETE macros (ready for SMPizing). Define STRNCAT inline for our usage. Stealing a bit of the xs_status flags to maintain command state that is pertinent for the hBA- should really reserve them for private usage in scsipiconf.h.
2000-07-06 02:25:06 +04:00
Per helpful suggestion from LukeM- make INLINE usage consistent (and use correct __inline format).
2001-05-17 01:41:52 +04:00
static INLINE void
ANSIfy source.
2001-03-14 08:44:46 +03:00
isp_wait_complete(struct ispsoftc *isp)
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
{
Implement F/W crashdumps for 2200 && 2300 cards (kernel compile option). Distinguish between 2312 and 2300 cards (they *are* different). Enable RIO (Reduced Interrupt Operation) for the LVD cards (hey- I've seen batched completions of the 30 commands at a time with this,....)... If we get a Port Logout on local loop topologies, we have to force the f/w to log back in. The easiest way (for us) to do this is to force a LIP. This also will wake up the disk that probably just had a f/w crash. Implement mailbox 'continuations'- this allows interrupts to re-drive a mailbox command if it's one that just essentially repeats the previous mailbox command (e.g., f/w download). This saves a boatload of sleep/wakeup twitches. If we're not a 2300 and we're about to return with a 'bogus interrupt'- check the semaphore register to be non-zero at all and outgoing mailbox 0- this seems to be where some of the lost ISP1080 commands came from.
2002-02-22 01:32:40 +03:00
int lim;
if (isp->isp_mbxwrk0)
lim = 60;
else
lim = 5;
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
if (isp->isp_osinfo.onintstack || isp->isp_osinfo.no_mbox_ints) {
Implement F/W crashdumps for 2200 && 2300 cards (kernel compile option). Distinguish between 2312 and 2300 cards (they *are* different). Enable RIO (Reduced Interrupt Operation) for the LVD cards (hey- I've seen batched completions of the 30 commands at a time with this,....)... If we get a Port Logout on local loop topologies, we have to force the f/w to log back in. The easiest way (for us) to do this is to force a LIP. This also will wake up the disk that probably just had a f/w crash. Implement mailbox 'continuations'- this allows interrupts to re-drive a mailbox command if it's one that just essentially repeats the previous mailbox command (e.g., f/w download). This saves a boatload of sleep/wakeup twitches. If we're not a 2300 and we're about to return with a 'bogus interrupt'- check the semaphore register to be non-zero at all and outgoing mailbox 0- this seems to be where some of the lost ISP1080 commands came from.
2002-02-22 01:32:40 +03:00
int useclim = 1000000 * lim, usecs = 0;
Ian McDonnell <apriori@world.std.com> made me see the error of my ways. *Cough*. Ahem. I hadn't been setting error to XS_SENSE when there was a check condition. Major blunder.
2001-01-05 10:02:00 +03:00
/*
* For sanity's sake, we don't delay longer
* than 5 seconds for polled commands.
*/
Implement F/W crashdumps for 2200 && 2300 cards (kernel compile option). Distinguish between 2312 and 2300 cards (they *are* different). Enable RIO (Reduced Interrupt Operation) for the LVD cards (hey- I've seen batched completions of the 30 commands at a time with this,....)... If we get a Port Logout on local loop topologies, we have to force the f/w to log back in. The easiest way (for us) to do this is to force a LIP. This also will wake up the disk that probably just had a f/w crash. Implement mailbox 'continuations'- this allows interrupts to re-drive a mailbox command if it's one that just essentially repeats the previous mailbox command (e.g., f/w download). This saves a boatload of sleep/wakeup twitches. If we're not a 2300 and we're about to return with a 'bogus interrupt'- check the semaphore register to be non-zero at all and outgoing mailbox 0- this seems to be where some of the lost ISP1080 commands came from.
2002-02-22 01:32:40 +03:00
while (usecs < useclim) {
Add support for 2 Gigabit cards (2300/2312). This necessitated a change in how interrupts are down- the 23XX has not only a different place to check for an interrupt, but unlike all other QLogic cards, you have to read the status as a 32 bit word- not 16 bit words. Rather than have device specific functions as called from the core module (in isp_intr), it makes more sense to have the platform/bus modules do the gruntwork of splitting out the isr, semaphore register and the first outgoing mailbox register (if needed) *prior* to calling isp_intr (if calling isp_intr is necessary at all).
2001-09-01 11:12:23 +04:00
u_int16_t isr, sema, mbox;
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
if (isp->isp_mboxbsy == 0) {
break;
}
Add support for 2 Gigabit cards (2300/2312). This necessitated a change in how interrupts are down- the 23XX has not only a different place to check for an interrupt, but unlike all other QLogic cards, you have to read the status as a 32 bit word- not 16 bit words. Rather than have device specific functions as called from the core module (in isp_intr), it makes more sense to have the platform/bus modules do the gruntwork of splitting out the isr, semaphore register and the first outgoing mailbox register (if needed) *prior* to calling isp_intr (if calling isp_intr is necessary at all).
2001-09-01 11:12:23 +04:00
if (ISP_READ_ISR(isp, &isr, &sema, &mbox)) {
isp_intr(isp, isr, sema, mbox);
if (isp->isp_mboxbsy == 0) {
break;
}
}
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
USEC_DELAY(500);
usecs += 500;
}
if (isp->isp_mboxbsy != 0) {
Ian McDonnell <apriori@world.std.com> made me see the error of my ways. *Cough*. Ahem. I hadn't been setting error to XS_SENSE when there was a check condition. Major blunder.
2001-01-05 10:02:00 +03:00
isp_prt(isp, ISP_LOGWARN,
"Polled Mailbox Command (0x%x) Timeout",
Remove ISP2100_FABRIC (we're always fabric now). Fix usage of isp_lastmbxcmd to report the mailbox command that times out. Fix isp_unswizzle_sns_rsp which for reasons *I* find obscurer just doesn't work correctly on sparc64 with words past 128. I have no idea why this *does* work on SparcLinux.
2001-02-13 02:32:11 +03:00
isp->isp_lastmbxcmd);
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
}
} else {
int rv = 0;
isp->isp_osinfo.mboxwaiting = 1;
while (isp->isp_osinfo.mboxwaiting && rv == 0) {
Implement F/W crashdumps for 2200 && 2300 cards (kernel compile option). Distinguish between 2312 and 2300 cards (they *are* different). Enable RIO (Reduced Interrupt Operation) for the LVD cards (hey- I've seen batched completions of the 30 commands at a time with this,....)... If we get a Port Logout on local loop topologies, we have to force the f/w to log back in. The easiest way (for us) to do this is to force a LIP. This also will wake up the disk that probably just had a f/w crash. Implement mailbox 'continuations'- this allows interrupts to re-drive a mailbox command if it's one that just essentially repeats the previous mailbox command (e.g., f/w download). This saves a boatload of sleep/wakeup twitches. If we're not a 2300 and we're about to return with a 'bogus interrupt'- check the semaphore register to be non-zero at all and outgoing mailbox 0- this seems to be where some of the lost ISP1080 commands came from.
2002-02-22 01:32:40 +03:00
struct timeval dtime;
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
int timo;
struct timeval tv;
Implement F/W crashdumps for 2200 && 2300 cards (kernel compile option). Distinguish between 2312 and 2300 cards (they *are* different). Enable RIO (Reduced Interrupt Operation) for the LVD cards (hey- I've seen batched completions of the 30 commands at a time with this,....)... If we get a Port Logout on local loop topologies, we have to force the f/w to log back in. The easiest way (for us) to do this is to force a LIP. This also will wake up the disk that probably just had a f/w crash. Implement mailbox 'continuations'- this allows interrupts to re-drive a mailbox command if it's one that just essentially repeats the previous mailbox command (e.g., f/w download). This saves a boatload of sleep/wakeup twitches. If we're not a 2300 and we're about to return with a 'bogus interrupt'- check the semaphore register to be non-zero at all and outgoing mailbox 0- this seems to be where some of the lost ISP1080 commands came from.
2002-02-22 01:32:40 +03:00
dtime.tv_sec = lim;
dtime.tv_usec = 0;
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
microtime(&tv);
Ian McDonnell <apriori@world.std.com> made me see the error of my ways. *Cough*. Ahem. I hadn't been setting error to XS_SENSE when there was a check condition. Major blunder.
2001-01-05 10:02:00 +03:00
timeradd(&tv, &dtime, &tv);
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
if ((timo = hzto(&tv)) == 0) {
timo = 1;
}
rv = tsleep(&isp->isp_osinfo.mboxwaiting,
PRIBIO, "isp_mboxcmd", timo);
}
if (rv == EWOULDBLOCK) {
isp->isp_mboxbsy = 0;
isp->isp_osinfo.mboxwaiting = 0;
Ian McDonnell <apriori@world.std.com> made me see the error of my ways. *Cough*. Ahem. I hadn't been setting error to XS_SENSE when there was a check condition. Major blunder.
2001-01-05 10:02:00 +03:00
isp_prt(isp, ISP_LOGWARN,
"Interrupting Mailbox Command (0x%x) Timeout",
Remove ISP2100_FABRIC (we're always fabric now). Fix usage of isp_lastmbxcmd to report the mailbox command that times out. Fix isp_unswizzle_sns_rsp which for reasons *I* find obscurer just doesn't work correctly on sparc64 with words past 128. I have no idea why this *does* work on SparcLinux.
2001-02-13 02:32:11 +03:00
isp->isp_lastmbxcmd);
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
}
}
}
Finally fix this driver to be sensible about the ENDIAN dance. It's not quite simply a question of the Qlogic being little endian and having to have stuff swapped on big endian machines- it also has to do with the fact that the SBus and PCI DMA layouts are wierd with respect to this. At any rate, now finally fixed- works on Mac G4, tested it on a SS10 for sparc, checked on alpha to see if I've broken anything, and as soon as I get another spare afternoon I'll finally install a sparc64 version which should just work (as it'll be like the Mac).
2000-12-09 11:06:31 +03:00
Core version 2.0 (platform version 1.0) rewrite of ISP driver. Some interace cleanups, some new common functions. The major impact that will be noticeable right away is that if you boot with not Fibre connected to the FC cards, you no longer hang indefinitely.
2000-08-02 03:55:09 +04:00
/*
* Common inline functions
*/
Roll platform version. Keep a bitmap of 'discovered' devices for SCSI adapters so we'll know when it's time to switch from 'slow' command mode to normal. Change some settings for configuration printfs and debug levels. Redo the internal ispscsicmd return definitions and let isp_cmd translate them as approrpriate to NetBSD values. Remove the inline functions from here- they're now in isp_inline.h. Put in the start of the correct SWIZZLE/UNSWIZZLE functions.
1999-10-14 06:33:38 +04:00
#include <dev/ic/isp_inline.h>
Rearrange defines to make the file diff cleaner with OpenBSD. Change CFGPRINTF and default verbose levels. Add in a forced compile in f/w define unless either IS_DISABLE_FW or ISP_COMPILE_FW defined.
1999-12-04 05:59:29 +03:00
#if !defined(ISP_DISABLE_FW) && !defined(ISP_COMPILE_FW)
#define ISP_COMPILE_FW 1
#endif
NetBSD OS specific routines and definitions.
1998-07-15 23:44:04 +04:00
#endif /* _ISP_NETBSD_H */