Aren/NetBSD
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
341c4563e3
NetBSD/sys/dev/ic/ispvar.h

933 lines
32 KiB
C
Raw Normal View History

DMA, not dma nor Dma.
2003-05-03 22:10:37 +04:00
/* $NetBSD: ispvar.h,v 1.60 2003/05/03 18:11:21 wiz Exp $ */
Add a maintenance note. Fix QFREE and QAVAIL macros to match change in usage.
2000-08-14 11:12:15 +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. Fix QFREE and QAVAIL macros to match change in usage.
2000-08-14 11:12:15 +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.
*/
ISP 10x0 driver from Matthew Jacob of NASA Ames Research Center. (March 12, 1997 version).
1997-03-12 23:44:50 +03:00
/*
Roll minor version. Change how f/w revisions are stored. Add some defines to get max luns for a card. Make sure the RESULT_QUEUE_LEN is not less than 64- I've seen breakage with that. Move the temp port database stuff into the softc (ick). Remove most of the target mode stuff. Make xflist storage an external outer layer thing (since it will now be allocated based upon the maximum commands that this HBA can support rather than request queue size).
1999-10-14 06:21:50 +04:00
* Copyright (C) 1999 National Aeronautics & Space Administration
ISP 10x0 driver from Matthew Jacob of NASA Ames Research Center. (March 12, 1997 version).
1997-03-12 23:44:50 +03:00
* All rights reserved.
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
*
ISP 10x0 driver from Matthew Jacob of NASA Ames Research Center. (March 12, 1997 version).
1997-03-12 23:44:50 +03:00
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
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.
ISP 10x0 driver from Matthew Jacob of NASA Ames Research Center. (March 12, 1997 version).
1997-03-12 23:44:50 +03:00
* 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
Some major thrashing to handle building for other than NetBSD. There is one change of note- build a list of completing commands in ispintr and then say you're done- this avoids some re-entrancy issues that had surfaced.
1998-07-15 23:50:16 +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.
ISP 10x0 driver from Matthew Jacob of NASA Ames Research Center. (March 12, 1997 version).
1997-03-12 23:44:50 +03:00
*/
NetBSD RCS ID tweaks, a few comment block tweaks. Also, make copyright notices consistent (per Matt Jacob).
1997-03-13 00:06:41 +03:00
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
/*
* Soft Definitions for for Qlogic ISP SCSI adapters.
* Matthew Jacob <mjacob@nas.nasa.gov>
*/
ISP 10x0 driver from Matthew Jacob of NASA Ames Research Center. (March 12, 1997 version).
1997-03-12 23:44:50 +03:00
#ifndef _ISPVAR_H
#define _ISPVAR_H
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
#if defined(__NetBSD__) || defined(__OpenBSD__)
ISP 10x0 driver from Matthew Jacob of NASA Ames Research Center. (March 12, 1997 version).
1997-03-12 23:44:50 +03:00
#include <dev/ic/ispmbox.h>
Roll core version minor. Clean out some target mode stuff. Respecify some queue related functions to take u_int16_t's. Restructure the softc a bit.
1999-12-04 06:06:05 +03:00
#ifdef ISP_TARGET_MODE
#include <dev/ic/isp_target.h>
add in 12160 (Ultra3) related definitions
2000-02-12 05:29:00 +03:00
#include <dev/ic/isp_tpublic.h>
Roll core version minor. Clean out some target mode stuff. Respecify some queue related functions to take u_int16_t's. Restructure the softc a bit.
1999-12-04 06:06:05 +03:00
#endif
Some major thrashing to handle building for other than NetBSD. There is one change of note- build a list of completing commands in ispintr and then say you're done- this avoids some re-entrancy issues that had surfaced.
1998-07-15 23:50:16 +04:00
#endif
#ifdef __FreeBSD__
#include <dev/isp/ispmbox.h>
Roll core version minor. Clean out some target mode stuff. Respecify some queue related functions to take u_int16_t's. Restructure the softc a bit.
1999-12-04 06:06:05 +03:00
#ifdef ISP_TARGET_MODE
#include <dev/isp/isp_target.h>
add in 12160 (Ultra3) related definitions
2000-02-12 05:29:00 +03:00
#include <dev/isp/isp_tpublic.h>
Roll core version minor. Clean out some target mode stuff. Respecify some queue related functions to take u_int16_t's. Restructure the softc a bit.
1999-12-04 06:06:05 +03:00
#endif
Some major thrashing to handle building for other than NetBSD. There is one change of note- build a list of completing commands in ispintr and then say you're done- this avoids some re-entrancy issues that had surfaced.
1998-07-15 23:50:16 +04:00
#endif
#ifdef __linux__
clean up headers; move uninit/watch to outer layers
1998-12-28 22:10:43 +03:00
#include "ispmbox.h"
Roll core version minor. Clean out some target mode stuff. Respecify some queue related functions to take u_int16_t's. Restructure the softc a bit.
1999-12-04 06:06:05 +03:00
#ifdef ISP_TARGET_MODE
#include "isp_target.h"
add in 12160 (Ultra3) related definitions
2000-02-12 05:29:00 +03:00
#include "isp_tpublic.h"
Roll core version minor. Clean out some target mode stuff. Respecify some queue related functions to take u_int16_t's. Restructure the softc a bit.
1999-12-04 06:06:05 +03:00
#endif
Some major thrashing to handle building for other than NetBSD. There is one change of note- build a list of completing commands in ispintr and then say you're done- this avoids some re-entrancy issues that had surfaced.
1998-07-15 23:50:16 +04:00
#endif
ISP 10x0 driver from Matthew Jacob of NASA Ames Research Center. (March 12, 1997 version).
1997-03-12 23:44:50 +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 ISP_CORE_VERSION_MAJOR 2
Roll version minor. Add DEFAULT_FRAMESIZE/DEFAULT_EXEC_THROTTLE/ISPASYNC_FW_RESTARTED defines.
2002-06-15 04:18:02 +04:00
#define ISP_CORE_VERSION_MINOR 7
roll core version number and add a global visible isp_restart function
1998-07-19 01:06:20 +04:00
ISP 10x0 driver from Matthew Jacob of NASA Ames Research Center. (March 12, 1997 version).
1997-03-12 23:44:50 +03:00
/*
Implement and use Fast Posting for both parallel && fibre. Redo a bit of the startup code. Implement a call to outer framework function so that asynchronous events can be handled (e.g., speed negotiation, target mode). Roll internal release tags.
1999-01-30 10:31:50 +03:00
* Vector for bus specific code to provide specific services.
ISP 10x0 driver from Matthew Jacob of NASA Ames Research Center. (March 12, 1997 version).
1997-03-12 23:44:50 +03:00
*/
struct ispsoftc;
struct ispmdvec {
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
int (*dv_rd_isr)
(struct ispsoftc *, u_int16_t *, u_int16_t *, u_int16_t *);
Roll core minor. ANSify.
2001-03-14 08:14:36 +03:00
u_int16_t (*dv_rd_reg) (struct ispsoftc *, int);
void (*dv_wr_reg) (struct ispsoftc *, int, u_int16_t);
int (*dv_mbxdma) (struct ispsoftc *);
int (*dv_dmaset) (struct ispsoftc *,
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
XS_T *, ispreq_t *, u_int16_t *, u_int16_t);
ISP 10x0 driver from Matthew Jacob of NASA Ames Research Center. (March 12, 1997 version).
1997-03-12 23:44:50 +03:00
void (*dv_dmaclr)
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
(struct ispsoftc *, XS_T *, u_int16_t);
Roll core minor. ANSify.
2001-03-14 08:14:36 +03:00
void (*dv_reset0) (struct ispsoftc *);
void (*dv_reset1) (struct ispsoftc *);
void (*dv_dregs) (struct ispsoftc *, const char *);
Handle Port Unavailable the same as Port Logout (i.e., send a LIP). Unconst pointer to f/w in the ispdv structure. Too many compilers get unhappy over our walking the array. Make casts as appropriate so that initialization in structure is still happy. Limit length of fabric to 256. This will all go away soon. Do a cleaner case of keeping multiple CPUs/threads from reading the same response queue entries.
2002-03-22 05:34:05 +03:00
u_int16_t *dv_ispfw; /* ptr to f/w */
ISP 10x0 driver from Matthew Jacob of NASA Ames Research Center. (March 12, 1997 version).
1997-03-12 23:44:50 +03:00
u_int16_t dv_conf1;
Need a bus/platform specific clock speed variable.
1997-04-05 06:48:36 +04:00
u_int16_t dv_clock; /* clock frequency */
ISP 10x0 driver from Matthew Jacob of NASA Ames Research Center. (March 12, 1997 version).
1997-03-12 23:44:50 +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
/*
* Overall parameters
*/
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
#define MAX_TARGETS 16
#define MAX_FC_TARG 256
Roll minor version. Change how f/w revisions are stored. Add some defines to get max luns for a card. Make sure the RESULT_QUEUE_LEN is not less than 64- I've seen breakage with that. Move the temp port database stuff into the softc (ick). Remove most of the target mode stuff. Make xflist storage an external outer layer thing (since it will now be allocated based upon the maximum commands that this HBA can support rather than request queue size).
1999-10-14 06:21:50 +04:00
#define ISP_MAX_TARGETS(isp) (IS_FC(isp)? MAX_FC_TARG : MAX_TARGETS)
Roll core version. Get rid of dogactive (no longer used). Add in some volatile flags (isp_mboxbsy- keeps a bitmask of expected output mailbox registers) and storage for them (isp_mboxtmp). Get rid of macroized LUN widths. This is determined at run time.
2000-07-06 02:28:08 +04:00
#define ISP_MAX_LUNS(isp) (isp)->isp_maxluns
Roll minor version. Change how f/w revisions are stored. Add some defines to get max luns for a card. Make sure the RESULT_QUEUE_LEN is not less than 64- I've seen breakage with that. Move the temp port database stuff into the softc (ick). Remove most of the target mode stuff. Make xflist storage an external outer layer thing (since it will now be allocated based upon the maximum commands that this HBA can support rather than request queue size).
1999-10-14 06:21:50 +04:00
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
/*
* 'Types'
*/
Remove ISP_DMA_ADDR_T as a 'type'. Instead, go off of whether ISP_DAC_SUPPORTED is defined and do a real typedef for the size of DMA addresses. ISP_DAC_SUPPORTED also decides how much of any particular DMA address is significant. Lose STRNCAT over the side. Add in (sight unseen) 10160 support.
2002-10-19 03:35:25 +04:00
#ifdef ISP_DAC_SUPPORTED
typedef u_int64_t isp_dma_addr_t;
#else
typedef u_int32_t isp_dma_addr_t;
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
#endif
Roll core version minor. Clean out some target mode stuff. Respecify some queue related functions to take u_int16_t's. Restructure the softc a bit.
1999-12-04 06:06:05 +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
* Macros to access ISP registers through bus specific layers-
* mostly wrappers to vector through the mdvec structure.
Roll core version minor. Clean out some target mode stuff. Respecify some queue related functions to take u_int16_t's. Restructure the softc a bit.
1999-12-04 06:06:05 +03:00
*/
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
#define ISP_READ_ISR(isp, isrp, semap, mbox0p) \
(*(isp)->isp_mdvec->dv_rd_isr)(isp, isrp, semap, mbox0p)
Roll core version minor. Clean out some target mode stuff. Respecify some queue related functions to take u_int16_t's. Restructure the softc a bit.
1999-12-04 06:06:05 +03:00
#define ISP_READ(isp, reg) \
(*(isp)->isp_mdvec->dv_rd_reg)((isp), (reg))
#define ISP_WRITE(isp, reg, val) \
(*(isp)->isp_mdvec->dv_wr_reg)((isp), (reg), (val))
#define ISP_MBOXDMASETUP(isp) \
(*(isp)->isp_mdvec->dv_mbxdma)((isp))
#define ISP_DMASETUP(isp, xs, req, iptrp, optr) \
(*(isp)->isp_mdvec->dv_dmaset)((isp), (xs), (req), (iptrp), (optr))
#define ISP_DMAFREE(isp, xs, hndl) \
if ((isp)->isp_mdvec->dv_dmaclr) \
(*(isp)->isp_mdvec->dv_dmaclr)((isp), (xs), (hndl))
#define ISP_RESET0(isp) \
if ((isp)->isp_mdvec->dv_reset0) (*(isp)->isp_mdvec->dv_reset0)((isp))
#define ISP_RESET1(isp) \
if ((isp)->isp_mdvec->dv_reset1) (*(isp)->isp_mdvec->dv_reset1)((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 ISP_DUMPREGS(isp, m) \
if ((isp)->isp_mdvec->dv_dregs) (*(isp)->isp_mdvec->dv_dregs)((isp),(m))
Roll core version minor. Clean out some target mode stuff. Respecify some queue related functions to take u_int16_t's. Restructure the softc a bit.
1999-12-04 06:06:05 +03:00
#define ISP_SETBITS(isp, reg, val) \
(*(isp)->isp_mdvec->dv_wr_reg)((isp), (reg), ISP_READ((isp), (reg)) | (val))
#define ISP_CLRBITS(isp, reg, val) \
(*(isp)->isp_mdvec->dv_wr_reg)((isp), (reg), ISP_READ((isp), (reg)) & ~(val))
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
/*
* The MEMORYBARRIER macro is defined per platform (to provide synchronization
* on Request and Response Queues, Scratch DMA areas, and Registers)
*
* Defined Memory Barrier Synchronization Types
*/
#define SYNC_REQUEST 0 /* request queue synchronization */
#define SYNC_RESULT 1 /* result queue synchronization */
#define SYNC_SFORDEV 2 /* scratch, sync for ISP */
#define SYNC_SFORCPU 3 /* scratch, sync for CPU */
#define SYNC_REG 4 /* for registers */
/*
* Request/Response Queue defines and macros.
* The maximum is defined per platform (and can be based on board type).
*/
/* This is the size of a queue entry (request and response) */
Roll major revision. Change some definitions around. Add more current device state stuff. Make DPARM flags full 16 bits (no shifting needed) and add the 2 new ones for the 7.55 f/w.
1998-09-08 11:27:04 +04:00
#define QENTRY_LEN 64
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
/* Both request and result queue length must be a power of two */
#define RQUEST_QUEUE_LEN(x) MAXISPREQUEST(x)
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
#ifdef ISP_TARGET_MODE
#define RESULT_QUEUE_LEN(x) MAXISPREQUEST(x)
#else
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 RESULT_QUEUE_LEN(x) \
(((MAXISPREQUEST(x) >> 2) < 64)? 64 : MAXISPREQUEST(x) >> 2)
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
#endif
Roll major revision. Change some definitions around. Add more current device state stuff. Make DPARM flags full 16 bits (no shifting needed) and add the 2 new ones for the 7.55 f/w.
1998-09-08 11:27:04 +04:00
#define ISP_QUEUE_ENTRY(q, idx) ((q) + ((idx) * QENTRY_LEN))
#define ISP_QUEUE_SIZE(n) ((n) * QENTRY_LEN)
#define ISP_NXT_QENTRY(idx, qlen) (((idx) + 1) & ((qlen)-1))
Add a maintenance note. Fix QFREE and QAVAIL macros to match change in usage.
2000-08-14 11:12:15 +04:00
#define ISP_QFREE(in, out, qlen) \
cleanup header to be just NetBSD, roll rev, change QDEPTH to QAVAIL macro
1998-09-18 02:52:54 +04:00
((in == out)? (qlen - 1) : ((in > out)? \
mostly stylistic
1999-10-17 05:37:34 +04:00
((qlen - 1) - (in - out)) : (out - in - 1)))
Add a maintenance note. Fix QFREE and QAVAIL macros to match change in usage.
2000-08-14 11:12:15 +04:00
#define ISP_QAVAIL(isp) \
ISP_QFREE(isp->isp_reqidx, isp->isp_reqodx, RQUEST_QUEUE_LEN(isp))
Roll core version minor. Clean out some target mode stuff. Respecify some queue related functions to take u_int16_t's. Restructure the softc a bit.
1999-12-04 06:06:05 +03:00
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_ADD_REQUEST(isp, nxti) \
MEMORYBARRIER(isp, SYNC_REQUEST, isp->isp_reqidx, QENTRY_LEN); \
WRITE_REQUEST_QUEUE_IN_POINTER(isp, nxti); \
isp->isp_reqidx = nxti
Roll core version minor. Clean out some target mode stuff. Respecify some queue related functions to take u_int16_t's. Restructure the softc a bit.
1999-12-04 06:06:05 +03:00
Defines and structures necessary to support the Qlogic 2100 (Fibre Channel PCI card); Some additional defines to try and manage various flavors of the SCSI host adapters (parameters for various SYNC modes).
1997-08-16 04:18:36 +04:00
/*
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
* SCSI Specific Host Adapter Parameters- per bus, per target
Defines and structures necessary to support the Qlogic 2100 (Fibre Channel PCI card); Some additional defines to try and manage various flavors of the SCSI host adapters (parameters for various SYNC modes).
1997-08-16 04:18:36 +04:00
*/
typedef struct {
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
u_int isp_gotdparms : 1,
mostly stylistic
1999-10-17 05:37:34 +04:00
isp_req_ack_active_neg : 1,
isp_data_line_active_neg: 1,
Defines and structures necessary to support the Qlogic 2100 (Fibre Channel PCI card); Some additional defines to try and manage various flavors of the SCSI host adapters (parameters for various SYNC modes).
1997-08-16 04:18:36 +04:00
isp_cmd_dma_burst_enable: 1,
isp_data_dma_burst_enabl: 1,
bump isp_fifo_threshold tag top 3 bits
1999-01-10 08:04:46 +03:00
isp_fifo_threshold : 3,
Implement and use Fast Posting for both parallel && fibre. Redo a bit of the startup code. Implement a call to outer framework function so that asynchronous events can be handled (e.g., speed negotiation, target mode). Roll internal release tags.
1999-01-30 10:31:50 +03:00
isp_ultramode : 1,
Defines and structures necessary to support the Qlogic 2100 (Fibre Channel PCI card); Some additional defines to try and manage various flavors of the SCSI host adapters (parameters for various SYNC modes).
1997-08-16 04:18:36 +04:00
isp_diffmode : 1,
add isp1080 support and some basic PDB change stuff
1999-03-27 01:39:44 +03:00
isp_lvdmode : 1,
Remove ispcmd_slow routine- SCBUSACCEL routine will set/clear tag/wide/sync. Make some changes about where some things sit in the softc.
2000-08-09 02:58:30 +04:00
isp_fast_mttr : 1, /* fast sram */
Defines and structures necessary to support the Qlogic 2100 (Fibre Channel PCI card); Some additional defines to try and manage various flavors of the SCSI host adapters (parameters for various SYNC modes).
1997-08-16 04:18:36 +04:00
isp_initiator_id : 4,
mostly stylistic
1999-10-17 05:37:34 +04:00
isp_async_data_setup : 4;
u_int16_t isp_selection_timeout;
u_int16_t isp_max_queue_depth;
Defines and structures necessary to support the Qlogic 2100 (Fibre Channel PCI card); Some additional defines to try and manage various flavors of the SCSI host adapters (parameters for various SYNC modes).
1997-08-16 04:18:36 +04:00
u_int8_t isp_tag_aging;
mostly stylistic
1999-10-17 05:37:34 +04:00
u_int8_t isp_bus_reset_delay;
u_int8_t isp_retry_count;
u_int8_t isp_retry_delay;
Defines and structures necessary to support the Qlogic 2100 (Fibre Channel PCI card); Some additional defines to try and manage various flavors of the SCSI host adapters (parameters for various SYNC modes).
1997-08-16 04:18:36 +04:00
struct {
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_int32_t
exc_throttle : 8,
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
: 1,
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
dev_enable : 1, /* ignored */
Implement and use Fast Posting for both parallel && fibre. Redo a bit of the startup code. Implement a call to outer framework function so that asynchronous events can be handled (e.g., speed negotiation, target mode). Roll internal release tags.
1999-01-30 10:31:50 +03:00
dev_update : 1,
dev_refresh : 1,
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
actv_offset : 4,
goal_offset : 4,
nvrm_offset : 4;
u_int8_t actv_period; /* current sync period */
u_int8_t goal_period; /* goal sync period */
u_int8_t nvrm_period; /* nvram sync period */
u_int16_t actv_flags; /* current device flags */
u_int16_t goal_flags; /* goal device flags */
u_int16_t nvrm_flags; /* nvram device flags */
Defines and structures necessary to support the Qlogic 2100 (Fibre Channel PCI card); Some additional defines to try and manage various flavors of the SCSI host adapters (parameters for various SYNC modes).
1997-08-16 04:18:36 +04:00
} isp_devparam[MAX_TARGETS];
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
} sdparam;
Defines and structures necessary to support the Qlogic 2100 (Fibre Channel PCI card); Some additional defines to try and manage various flavors of the SCSI host adapters (parameters for various SYNC modes).
1997-08-16 04:18:36 +04:00
/*
* Device Flags
*/
Roll major revision. Change some definitions around. Add more current device state stuff. Make DPARM flags full 16 bits (no shifting needed) and add the 2 new ones for the 7.55 f/w.
1998-09-08 11:27:04 +04:00
#define DPARM_DISC 0x8000
#define DPARM_PARITY 0x4000
#define DPARM_WIDE 0x2000
#define DPARM_SYNC 0x1000
#define DPARM_TQING 0x0800
#define DPARM_ARQ 0x0400
#define DPARM_QFRZ 0x0200
#define DPARM_RENEG 0x0100
Remove ispcmd_slow routine- SCBUSACCEL routine will set/clear tag/wide/sync. Make some changes about where some things sit in the softc.
2000-08-09 02:58:30 +04:00
#define DPARM_NARROW 0x0080
#define DPARM_ASYNC 0x0040
#define DPARM_PPR 0x0020
Roll internal release tag. Roll core version minor. Fix broken DPARM_DEFAULT define. Add a new config flag param (ISP_CFG_NONVRAM) whose intent it is to cause NVRAM to be ignored. Add ISPASYNC_LOOP_DOWN and ISPASYNC_LOOP_UP isp_async enums.
1999-02-09 03:46:25 +03:00
#define DPARM_DEFAULT (0xFF00 & ~DPARM_QFRZ)
Roll major revision. Change some definitions around. Add more current device state stuff. Make DPARM flags full 16 bits (no shifting needed) and add the 2 new ones for the 7.55 f/w.
1998-09-08 11:27:04 +04:00
#define DPARM_SAFE_DFLT (DPARM_DEFAULT & ~(DPARM_WIDE|DPARM_SYNC|DPARM_TQING))
Defines and structures necessary to support the Qlogic 2100 (Fibre Channel PCI card); Some additional defines to try and manage various flavors of the SCSI host adapters (parameters for various SYNC modes).
1997-08-16 04:18:36 +04:00
add isp1080 support and some basic PDB change stuff
1999-03-27 01:39:44 +03:00
/* technically, not really correct, as they need to be rated based upon clock */
Remove ispcmd_slow routine- SCBUSACCEL routine will set/clear tag/wide/sync. Make some changes about where some things sit in the softc.
2000-08-09 02:58:30 +04:00
#define ISP_80M_SYNCPARMS 0x0c09
#define ISP_40M_SYNCPARMS 0x0c0a
#define ISP_20M_SYNCPARMS 0x0c0c
#define ISP_20M_SYNCPARMS_1040 0x080c
mostly stylistic
1999-10-17 05:37:34 +04:00
#define ISP_10M_SYNCPARMS 0x0c19
#define ISP_08M_SYNCPARMS 0x0c25
#define ISP_05M_SYNCPARMS 0x0c32
#define ISP_04M_SYNCPARMS 0x0c41
Defines and structures necessary to support the Qlogic 2100 (Fibre Channel PCI card); Some additional defines to try and manage various flavors of the SCSI host adapters (parameters for various SYNC modes).
1997-08-16 04:18:36 +04:00
/*
* Fibre Channel Specifics
*/
Roll minor version. Change how f/w revisions are stored. Add some defines to get max luns for a card. Make sure the RESULT_QUEUE_LEN is not less than 64- I've seen breakage with that. Move the temp port database stuff into the softc (ick). Remove most of the target mode stuff. Make xflist storage an external outer layer thing (since it will now be allocated based upon the maximum commands that this HBA can support rather than request queue size).
1999-10-14 06:21:50 +04:00
#define FL_PORT_ID 0x7e /* FL_Port Special ID */
#define FC_PORT_ID 0x7f /* Fabric Controller Special ID */
#define FC_SNS_ID 0x80 /* SNS Server Special ID */
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 ISP_USE_GA_NXT 1 */ /* Use GA_NXT with switches */
#ifndef GA_NXT_MAX
#define GA_NXT_MAX 256
#endif
Defines and structures necessary to support the Qlogic 2100 (Fibre Channel PCI card); Some additional defines to try and manage various flavors of the SCSI host adapters (parameters for various SYNC modes).
1997-08-16 04:18:36 +04:00
typedef struct {
mostly stylistic
1999-10-17 05:37:34 +04:00
u_int32_t isp_fwoptions : 16,
Fix various 2300 GB issues related to chip errata and documentation misunderstandings. We also now can report our connection rate.
2001-10-07 00:34:49 +04:00
isp_gbspeed : 2,
Eliminate ISP2100_FABRIC- we always allow for fabric now. Add an isp_iid_set/isp_iid for fibre channel- this is because we now fake a port database entry for ourselves. Add the additional loop states between LOOP_PDB_RCVD and LOOP_READY. Change and comment on a wad of Fibre Channel isp_control functions. Change and comment on some of the ISPASYNC Fibre Channel events. Add was_fabric_dev/fabric_dev tags to our local FC database structure (so we can see rapidly whether something was a fabric device but is now gone). Add a tag which says what role this adapter should take. It can take on the value of None, Target, Initiator or Both. None is useful for warm failover purposes. Remove the ISP_CFG_NOINIT silliness since a role of "None" does this. Add a isp_lastmbxcmd tag to store the opcode for the last mailbox command used.
2001-02-13 02:33:49 +03:00
isp_iid_set : 1,
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
loop_seen_once : 1,
White space cleanup. Fix oopslet about having overlapping loop states. Handle FW crashes as an isp_async call. Reclaim a missing logging level bit.
2001-07-06 20:19:10 +04:00
isp_loopstate : 4, /* Current Loop State */
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
isp_fwstate : 3, /* ISP F/W state */
isp_gotdparms : 1,
Update from Matthew Jacob: Correctly account for F-port vs. F-port (no FLOGI_ACC) topologies. Make sure we get a port database entry for the fabric name server. Preserve fabric logins if the device didn't change across fabric or port database changes, or the device has already logged into us (e.g., for target/initiator dual role devices like Veritas SANbox). Propagate class 3 service parameter changes where devices can change roles. Fix all occurrences of setting a sendmarker so that setting it for one bus on dual bus cards doesn't wipe a pending sendmarker for other busses on the same card :-;. Comments added and clarifications made in some of the target mode code.
2000-05-13 20:53:03 +04:00
isp_topo : 3,
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
isp_onfabric : 1;
Eliminate ISP2100_FABRIC- we always allow for fabric now. Add an isp_iid_set/isp_iid for fibre channel- this is because we now fake a port database entry for ourselves. Add the additional loop states between LOOP_PDB_RCVD and LOOP_READY. Change and comment on a wad of Fibre Channel isp_control functions. Change and comment on some of the ISPASYNC Fibre Channel events. Add was_fabric_dev/fabric_dev tags to our local FC database structure (so we can see rapidly whether something was a fabric device but is now gone). Add a tag which says what role this adapter should take. It can take on the value of None, Target, Initiator or Both. None is useful for warm failover purposes. Remove the ISP_CFG_NOINIT silliness since a role of "None" does this. Add a isp_lastmbxcmd tag to store the opcode for the last mailbox command used.
2001-02-13 02:33:49 +03:00
u_int8_t isp_iid; /* 'initiator' id */
Roll major revision. Change some definitions around. Add more current device state stuff. Make DPARM flags full 16 bits (no shifting needed) and add the 2 new ones for the 7.55 f/w.
1998-09-08 11:27:04 +04:00
u_int8_t isp_loopid; /* hard loop id */
u_int8_t isp_alpa; /* ALPA */
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
u_int32_t isp_portid;
Eliminate ISP2100_FABRIC- we always allow for fabric now. Add an isp_iid_set/isp_iid for fibre channel- this is because we now fake a port database entry for ourselves. Add the additional loop states between LOOP_PDB_RCVD and LOOP_READY. Change and comment on a wad of Fibre Channel isp_control functions. Change and comment on some of the ISPASYNC Fibre Channel events. Add was_fabric_dev/fabric_dev tags to our local FC database structure (so we can see rapidly whether something was a fabric device but is now gone). Add a tag which says what role this adapter should take. It can take on the value of None, Target, Initiator or Both. None is useful for warm failover purposes. Remove the ISP_CFG_NOINIT silliness since a role of "None" does this. Add a isp_lastmbxcmd tag to store the opcode for the last mailbox command used.
2001-02-13 02:33:49 +03:00
volatile u_int16_t isp_lipseq; /* LIP sequence # */
Synchronize with FreeBSD/Linux.
2001-09-06 02:32:38 +04:00
u_int16_t isp_fwattr; /* firmware attributes */
Roll major revision. Change some definitions around. Add more current device state stuff. Make DPARM flags full 16 bits (no shifting needed) and add the 2 new ones for the 7.55 f/w.
1998-09-08 11:27:04 +04:00
u_int8_t isp_execthrottle;
mostly stylistic
1999-10-17 05:37:34 +04:00
u_int8_t isp_retry_delay;
u_int8_t isp_retry_count;
u_int8_t isp_reserved;
Roll major revision. Change some definitions around. Add more current device state stuff. Make DPARM flags full 16 bits (no shifting needed) and add the 2 new ones for the 7.55 f/w.
1998-09-08 11:27:04 +04:00
u_int16_t isp_maxalloc;
u_int16_t isp_maxfrmlen;
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
u_int64_t isp_nodewwn;
u_int64_t isp_portwwn;
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
/*
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
* Port Data Base. This is indexed by 'target', which is invariate.
* However, elements within can move around due to loop changes,
* so the actual loop ID passed to the F/W is in this structure.
* The first time the loop is seen up, loopid will match the index
* (except for fabric nodes which are above mapped above FC_SNS_ID
* and are completely virtual), but subsequent LIPs can cause things
* to move around.
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
*/
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
struct lportdb {
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
u_int32_t
port_type : 8,
Update from Matthew Jacob: Correctly account for F-port vs. F-port (no FLOGI_ACC) topologies. Make sure we get a port database entry for the fabric name server. Preserve fabric logins if the device didn't change across fabric or port database changes, or the device has already logged into us (e.g., for target/initiator dual role devices like Veritas SANbox). Propagate class 3 service parameter changes where devices can change roles. Fix all occurrences of setting a sendmarker so that setting it for one bus on dual bus cards doesn't wipe a pending sendmarker for other busses on the same card :-;. Comments added and clarifications made in some of the target mode code.
2000-05-13 20:53:03 +04:00
loopid : 8,
Roll version minor. Add DEFAULT_FRAMESIZE/DEFAULT_EXEC_THROTTLE/ISPASYNC_FW_RESTARTED defines.
2002-06-15 04:18:02 +04:00
fc4_type : 4,
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
last_fabric_dev : 1,
Roll version minor. Add DEFAULT_FRAMESIZE/DEFAULT_EXEC_THROTTLE/ISPASYNC_FW_RESTARTED defines.
2002-06-15 04:18:02 +04:00
: 2,
relogin : 1,
White space cleanup. Fix oopslet about having overlapping loop states. Handle FW crashes as an isp_async call. Reclaim a missing logging level bit.
2001-07-06 20:19:10 +04:00
force_logout : 1,
Eliminate ISP2100_FABRIC- we always allow for fabric now. Add an isp_iid_set/isp_iid for fibre channel- this is because we now fake a port database entry for ourselves. Add the additional loop states between LOOP_PDB_RCVD and LOOP_READY. Change and comment on a wad of Fibre Channel isp_control functions. Change and comment on some of the ISPASYNC Fibre Channel events. Add was_fabric_dev/fabric_dev tags to our local FC database structure (so we can see rapidly whether something was a fabric device but is now gone). Add a tag which says what role this adapter should take. It can take on the value of None, Target, Initiator or Both. None is useful for warm failover purposes. Remove the ISP_CFG_NOINIT silliness since a role of "None" does this. Add a isp_lastmbxcmd tag to store the opcode for the last mailbox command used.
2001-02-13 02:33:49 +03:00
was_fabric_dev : 1,
fabric_dev : 1,
Update from Matthew Jacob: Correctly account for F-port vs. F-port (no FLOGI_ACC) topologies. Make sure we get a port database entry for the fabric name server. Preserve fabric logins if the device didn't change across fabric or port database changes, or the device has already logged into us (e.g., for target/initiator dual role devices like Veritas SANbox). Propagate class 3 service parameter changes where devices can change roles. Fix all occurrences of setting a sendmarker so that setting it for one bus on dual bus cards doesn't wipe a pending sendmarker for other busses on the same card :-;. Comments added and clarifications made in some of the target mode code.
2000-05-13 20:53:03 +04:00
loggedin : 1,
roles : 2,
valid : 1;
mostly stylistic
1999-10-17 05:37:34 +04:00
u_int32_t portid;
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
u_int64_t node_wwn;
u_int64_t port_wwn;
Eliminate ISP2100_FABRIC- we always allow for fabric now. Add an isp_iid_set/isp_iid for fibre channel- this is because we now fake a port database entry for ourselves. Add the additional loop states between LOOP_PDB_RCVD and LOOP_READY. Change and comment on a wad of Fibre Channel isp_control functions. Change and comment on some of the ISPASYNC Fibre Channel events. Add was_fabric_dev/fabric_dev tags to our local FC database structure (so we can see rapidly whether something was a fabric device but is now gone). Add a tag which says what role this adapter should take. It can take on the value of None, Target, Initiator or Both. None is useful for warm failover purposes. Remove the ISP_CFG_NOINIT silliness since a role of "None" does this. Add a isp_lastmbxcmd tag to store the opcode for the last mailbox command used.
2001-02-13 02:33:49 +03:00
} portdb[MAX_FC_TARG], tport[FC_PORT_ID];
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
Defines and structures necessary to support the Qlogic 2100 (Fibre Channel PCI card); Some additional defines to try and manage various flavors of the SCSI host adapters (parameters for various SYNC modes).
1997-08-16 04:18:36 +04:00
/*
* Scratch DMA mapped in area to fetch Port Database stuff, etc.
*/
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
caddr_t isp_scratch;
Remove ISP_DMA_ADDR_T as a 'type'. Instead, go off of whether ISP_DAC_SUPPORTED is defined and do a real typedef for the size of DMA addresses. ISP_DAC_SUPPORTED also decides how much of any particular DMA address is significant. Lose STRNCAT over the side. Add in (sight unseen) 10160 support.
2002-10-19 03:35:25 +04:00
isp_dma_addr_t isp_scdma;
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
#ifdef ISP_FW_CRASH_DUMP
u_int16_t *isp_dump_data;
#endif
Defines and structures necessary to support the Qlogic 2100 (Fibre Channel PCI card); Some additional defines to try and manage various flavors of the SCSI host adapters (parameters for various SYNC modes).
1997-08-16 04:18:36 +04:00
} fcparam;
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
#define FW_CONFIG_WAIT 0
#define FW_WAIT_AL_PA 1
#define FW_WAIT_LOGIN 2
#define FW_READY 3
#define FW_LOSS_OF_SYNC 4
#define FW_ERROR 5
#define FW_REINIT 6
#define FW_NON_PART 7
#define LOOP_NIL 0
#define LOOP_LIP_RCVD 1
#define LOOP_PDB_RCVD 2
Eliminate ISP2100_FABRIC- we always allow for fabric now. Add an isp_iid_set/isp_iid for fibre channel- this is because we now fake a port database entry for ourselves. Add the additional loop states between LOOP_PDB_RCVD and LOOP_READY. Change and comment on a wad of Fibre Channel isp_control functions. Change and comment on some of the ISPASYNC Fibre Channel events. Add was_fabric_dev/fabric_dev tags to our local FC database structure (so we can see rapidly whether something was a fabric device but is now gone). Add a tag which says what role this adapter should take. It can take on the value of None, Target, Initiator or Both. None is useful for warm failover purposes. Remove the ISP_CFG_NOINIT silliness since a role of "None" does this. Add a isp_lastmbxcmd tag to store the opcode for the last mailbox command used.
2001-02-13 02:33:49 +03:00
#define LOOP_SCANNING_FABRIC 3
#define LOOP_FSCAN_DONE 4
#define LOOP_SCANNING_LOOP 5
White space cleanup. Fix oopslet about having overlapping loop states. Handle FW crashes as an isp_async call. Reclaim a missing logging level bit.
2001-07-06 20:19:10 +04:00
#define LOOP_LSCAN_DONE 6
#define LOOP_SYNCING_PDB 7
#define LOOP_READY 8
Defines and structures necessary to support the Qlogic 2100 (Fibre Channel PCI card); Some additional defines to try and manage various flavors of the SCSI host adapters (parameters for various SYNC modes).
1997-08-16 04:18:36 +04:00
Update from Matthew Jacob: Correctly account for F-port vs. F-port (no FLOGI_ACC) topologies. Make sure we get a port database entry for the fabric name server. Preserve fabric logins if the device didn't change across fabric or port database changes, or the device has already logged into us (e.g., for target/initiator dual role devices like Veritas SANbox). Propagate class 3 service parameter changes where devices can change roles. Fix all occurrences of setting a sendmarker so that setting it for one bus on dual bus cards doesn't wipe a pending sendmarker for other busses on the same card :-;. Comments added and clarifications made in some of the target mode code.
2000-05-13 20:53:03 +04:00
#define TOPO_NL_PORT 0
#define TOPO_FL_PORT 1
#define TOPO_N_PORT 2
#define TOPO_F_PORT 3
#define TOPO_PTP_STUB 4
ISP 10x0 driver from Matthew Jacob of NASA Ames Research Center. (March 12, 1997 version).
1997-03-12 23:44:50 +03:00
/*
* Soft Structure per host adapter
*/
Remove ispcmd_slow routine- SCBUSACCEL routine will set/clear tag/wide/sync. Make some changes about where some things sit in the softc.
2000-08-09 02:58:30 +04:00
typedef struct ispsoftc {
Some major thrashing to handle building for other than NetBSD. There is one change of note- build a list of completing commands in ispintr and then say you're done- this avoids some re-entrancy issues that had surfaced.
1998-07-15 23:50:16 +04:00
/*
* Platform (OS) specific data
*/
struct isposinfo isp_osinfo;
/*
Roll core version minor. Clean out some target mode stuff. Respecify some queue related functions to take u_int16_t's. Restructure the softc a bit.
1999-12-04 06:06:05 +03:00
* Pointer to bus specific functions and data
Some major thrashing to handle building for other than NetBSD. There is one change of note- build a list of completing commands in ispintr and then say you're done- this avoids some re-entrancy issues that had surfaced.
1998-07-15 23:50:16 +04:00
*/
ISP 10x0 driver from Matthew Jacob of NASA Ames Research Center. (March 12, 1997 version).
1997-03-12 23:44:50 +03:00
struct ispmdvec * isp_mdvec;
Some major thrashing to handle building for other than NetBSD. There is one change of note- build a list of completing commands in ispintr and then say you're done- this avoids some re-entrancy issues that had surfaced.
1998-07-15 23:50:16 +04:00
/*
Roll core version minor. Clean out some target mode stuff. Respecify some queue related functions to take u_int16_t's. Restructure the softc a bit.
1999-12-04 06:06:05 +03:00
* (Mostly) nonvolatile state. Board specific parameters
* may contain some volatile state (e.g., current loop state).
Some major thrashing to handle building for other than NetBSD. There is one change of note- build a list of completing commands in ispintr and then say you're done- this avoids some re-entrancy issues that had surfaced.
1998-07-15 23:50:16 +04:00
*/
Roll core version minor. Clean out some target mode stuff. Respecify some queue related functions to take u_int16_t's. Restructure the softc a bit.
1999-12-04 06:06:05 +03:00
void * isp_param; /* type specific */
u_int16_t isp_fwrev[3]; /* Loaded F/W revision */
u_int16_t isp_romfw_rev[3]; /* PROM F/W revision */
u_int16_t isp_maxcmds; /* max possible I/O cmds */
u_int8_t isp_type; /* HBA Chip Type */
u_int8_t isp_revision; /* HBA Chip H/W Revision */
Roll core version. Get rid of dogactive (no longer used). Add in some volatile flags (isp_mboxbsy- keeps a bitmask of expected output mailbox registers) and storage for them (isp_mboxtmp). Get rid of macroized LUN widths. This is determined at run time.
2000-07-06 02:28:08 +04:00
u_int32_t isp_maxluns; /* maximum luns supported */
Roll core version minor. Clean out some target mode stuff. Respecify some queue related functions to take u_int16_t's. Restructure the softc a bit.
1999-12-04 06:06:05 +03:00
Eliminate ISP2100_FABRIC- we always allow for fabric now. Add an isp_iid_set/isp_iid for fibre channel- this is because we now fake a port database entry for ourselves. Add the additional loop states between LOOP_PDB_RCVD and LOOP_READY. Change and comment on a wad of Fibre Channel isp_control functions. Change and comment on some of the ISPASYNC Fibre Channel events. Add was_fabric_dev/fabric_dev tags to our local FC database structure (so we can see rapidly whether something was a fabric device but is now gone). Add a tag which says what role this adapter should take. It can take on the value of None, Target, Initiator or Both. None is useful for warm failover purposes. Remove the ISP_CFG_NOINIT silliness since a role of "None" does this. Add a isp_lastmbxcmd tag to store the opcode for the last mailbox command used.
2001-02-13 02:33:49 +03:00
u_int32_t isp_clock : 8, /* input clock */
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
: 4,
isp_port : 1, /* 23XX only */
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
isp_failed : 1, /* board failed */
isp_open : 1, /* opened (ioctl) */
Eliminate ISP2100_FABRIC- we always allow for fabric now. Add an isp_iid_set/isp_iid for fibre channel- this is because we now fake a port database entry for ourselves. Add the additional loop states between LOOP_PDB_RCVD and LOOP_READY. Change and comment on a wad of Fibre Channel isp_control functions. Change and comment on some of the ISPASYNC Fibre Channel events. Add was_fabric_dev/fabric_dev tags to our local FC database structure (so we can see rapidly whether something was a fabric device but is now gone). Add a tag which says what role this adapter should take. It can take on the value of None, Target, Initiator or Both. None is useful for warm failover purposes. Remove the ISP_CFG_NOINIT silliness since a role of "None" does this. Add a isp_lastmbxcmd tag to store the opcode for the last mailbox command used.
2001-02-13 02:33:49 +03:00
isp_touched : 1, /* board ever seen? */
Remove ispcmd_slow routine- SCBUSACCEL routine will set/clear tag/wide/sync. Make some changes about where some things sit in the softc.
2000-08-09 02:58:30 +04:00
isp_bustype : 1, /* SBus or PCI */
isp_loaded_fw : 1, /* loaded firmware */
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
isp_role : 2, /* roles supported */
Eliminate ISP2100_FABRIC- we always allow for fabric now. Add an isp_iid_set/isp_iid for fibre channel- this is because we now fake a port database entry for ourselves. Add the additional loop states between LOOP_PDB_RCVD and LOOP_READY. Change and comment on a wad of Fibre Channel isp_control functions. Change and comment on some of the ISPASYNC Fibre Channel events. Add was_fabric_dev/fabric_dev tags to our local FC database structure (so we can see rapidly whether something was a fabric device but is now gone). Add a tag which says what role this adapter should take. It can take on the value of None, Target, Initiator or Both. None is useful for warm failover purposes. Remove the ISP_CFG_NOINIT silliness since a role of "None" does this. Add a isp_lastmbxcmd tag to store the opcode for the last mailbox command used.
2001-02-13 02:33:49 +03:00
isp_dblev : 12; /* debug log mask */
u_int32_t isp_confopts; /* config options */
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 isp_rqstinrp; /* register for REQINP */
u_int16_t isp_rqstoutrp; /* register for REQOUTP */
u_int16_t isp_respinrp; /* register for RESINP */
u_int16_t isp_respoutrp; /* register for RESOUTP */
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
/*
* Instrumentation
*/
u_int64_t isp_intcnt; /* total int count */
u_int64_t isp_intbogus; /* spurious int count */
Implement REDUCED INTERRUPT OPERATION usage for FC cards- this allows the firmware to delay completion of commands so that it can attempt to batch a bunch of completions at once- either returning 16 bit handles in mailbox registers, or in a resposne queue entry that has a whole wad of 16 bit handles. Distinguish between 2300 and 2312 chipsets- if only because the revisions on the chips have different meanings. Add more instrumentation plus ISP_GET_STATS and ISP_CLR_STATS ioctls. Run up the maximum number of response queue entities we'll look at per interrupt. If we haven't set HBA role yet, always return success from isp_fc_runstate.
2002-01-04 00:45:06 +03:00
u_int64_t isp_intmboxc; /* mbox completions */
u_int64_t isp_intoasync; /* other async */
u_int64_t isp_rsltccmplt; /* CMDs on result q */
u_int64_t isp_fphccmplt; /* CMDs via fastpost */
u_int16_t isp_rscchiwater;
u_int16_t isp_fpcchiwater;
Defines and structures necessary to support the Qlogic 2100 (Fibre Channel PCI card); Some additional defines to try and manage various flavors of the SCSI host adapters (parameters for various SYNC modes).
1997-08-16 04:18:36 +04:00
ISP 10x0 driver from Matthew Jacob of NASA Ames Research Center. (March 12, 1997 version).
1997-03-12 23:44:50 +03:00
/*
Roll major revision. Change some definitions around. Add more current device state stuff. Make DPARM flags full 16 bits (no shifting needed) and add the 2 new ones for the 7.55 f/w.
1998-09-08 11:27:04 +04:00
* Volatile state
ISP 10x0 driver from Matthew Jacob of NASA Ames Research Center. (March 12, 1997 version).
1997-03-12 23:44:50 +03:00
*/
Roll major revision. Change some definitions around. Add more current device state stuff. Make DPARM flags full 16 bits (no shifting needed) and add the 2 new ones for the 7.55 f/w.
1998-09-08 11:27:04 +04:00
Roll core version. Get rid of dogactive (no longer used). Add in some volatile flags (isp_mboxbsy- keeps a bitmask of expected output mailbox registers) and storage for them (isp_mboxtmp). Get rid of macroized LUN widths. This is determined at run time.
2000-07-06 02:28:08 +04:00
volatile u_int32_t
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
isp_obits : 8, /* mailbox command output */
isp_mboxbsy : 1, /* mailbox command active */
Roll major revision. Change some definitions around. Add more current device state stuff. Make DPARM flags full 16 bits (no shifting needed) and add the 2 new ones for the 7.55 f/w.
1998-09-08 11:27:04 +04:00
isp_state : 3,
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
isp_sendmarker : 2, /* send a marker entry */
isp_update : 2, /* update parameters */
Roll minor version. Change how f/w revisions are stored. Add some defines to get max luns for a card. Make sure the RESULT_QUEUE_LEN is not less than 64- I've seen breakage with that. Move the temp port database stuff into the softc (ick). Remove most of the target mode stuff. Make xflist storage an external outer layer thing (since it will now be allocated based upon the maximum commands that this HBA can support rather than request queue size).
1999-10-14 06:21:50 +04:00
isp_nactive : 16; /* how many commands active */
volatile u_int16_t isp_reqodx; /* index of last ISP pickup */
volatile u_int16_t isp_reqidx; /* index of next request */
volatile u_int16_t isp_residx; /* index of next result */
Handle Port Unavailable the same as Port Logout (i.e., send a LIP). Unconst pointer to f/w in the ispdv structure. Too many compilers get unhappy over our walking the array. Make casts as appropriate so that initialization in structure is still happy. Limit length of fabric to 256. This will all go away soon. Do a cleaner case of keeping multiple CPUs/threads from reading the same response queue entries.
2002-03-22 05:34:05 +03:00
volatile u_int16_t isp_resodx; /* index of next result */
volatile u_int16_t isp_rspbsy;
Roll minor version. Change how f/w revisions are stored. Add some defines to get max luns for a card. Make sure the RESULT_QUEUE_LEN is not less than 64- I've seen breakage with that. Move the temp port database stuff into the softc (ick). Remove most of the target mode stuff. Make xflist storage an external outer layer thing (since it will now be allocated based upon the maximum commands that this HBA can support rather than request queue size).
1999-10-14 06:21:50 +04:00
volatile u_int16_t isp_lasthdls; /* last handle seed */
Roll core version. Get rid of dogactive (no longer used). Add in some volatile flags (isp_mboxbsy- keeps a bitmask of expected output mailbox registers) and storage for them (isp_mboxtmp). Get rid of macroized LUN widths. This is determined at run time.
2000-07-06 02:28:08 +04:00
volatile u_int16_t isp_mboxtmp[MAX_MAILBOX];
Eliminate ISP2100_FABRIC- we always allow for fabric now. Add an isp_iid_set/isp_iid for fibre channel- this is because we now fake a port database entry for ourselves. Add the additional loop states between LOOP_PDB_RCVD and LOOP_READY. Change and comment on a wad of Fibre Channel isp_control functions. Change and comment on some of the ISPASYNC Fibre Channel events. Add was_fabric_dev/fabric_dev tags to our local FC database structure (so we can see rapidly whether something was a fabric device but is now gone). Add a tag which says what role this adapter should take. It can take on the value of None, Target, Initiator or Both. None is useful for warm failover purposes. Remove the ISP_CFG_NOINIT silliness since a role of "None" does this. Add a isp_lastmbxcmd tag to store the opcode for the last mailbox command used.
2001-02-13 02:33:49 +03:00
volatile u_int16_t isp_lastmbxcmd; /* last mbox command sent */
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
volatile u_int16_t isp_mbxwrk0;
volatile u_int16_t isp_mbxwrk1;
volatile u_int16_t isp_mbxwrk2;
void * isp_mbxworkp;
Defines and structures necessary to support the Qlogic 2100 (Fibre Channel PCI card); Some additional defines to try and manage various flavors of the SCSI host adapters (parameters for various SYNC modes).
1997-08-16 04:18:36 +04:00
ISP 10x0 driver from Matthew Jacob of NASA Ames Research Center. (March 12, 1997 version).
1997-03-12 23:44:50 +03:00
/*
Roll core version minor. Clean out some target mode stuff. Respecify some queue related functions to take u_int16_t's. Restructure the softc a bit.
1999-12-04 06:06:05 +03:00
* Active commands are stored here, indexed by handle functions.
ISP 10x0 driver from Matthew Jacob of NASA Ames Research Center. (March 12, 1997 version).
1997-03-12 23:44:50 +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
XS_T **isp_xflist;
Defines and structures necessary to support the Qlogic 2100 (Fibre Channel PCI card); Some additional defines to try and manage various flavors of the SCSI host adapters (parameters for various SYNC modes).
1997-08-16 04:18:36 +04:00
ISP 10x0 driver from Matthew Jacob of NASA Ames Research Center. (March 12, 1997 version).
1997-03-12 23:44:50 +03:00
/*
DMA, not dma nor Dma.
2003-05-03 22:10:37 +04:00
* request/result queue pointers and DMA handles for them.
ISP 10x0 driver from Matthew Jacob of NASA Ames Research Center. (March 12, 1997 version).
1997-03-12 23:44:50 +03:00
*/
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
caddr_t isp_rquest;
caddr_t isp_result;
Remove ISP_DMA_ADDR_T as a 'type'. Instead, go off of whether ISP_DAC_SUPPORTED is defined and do a real typedef for the size of DMA addresses. ISP_DAC_SUPPORTED also decides how much of any particular DMA address is significant. Lose STRNCAT over the side. Add in (sight unseen) 10160 support.
2002-10-19 03:35:25 +04:00
isp_dma_addr_t isp_rquest_dma;
isp_dma_addr_t isp_result_dma;
Remove ispcmd_slow routine- SCBUSACCEL routine will set/clear tag/wide/sync. Make some changes about where some things sit in the softc.
2000-08-09 02:58:30 +04:00
} ispsoftc_t;
ISP 10x0 driver from Matthew Jacob of NASA Ames Research Center. (March 12, 1997 version).
1997-03-12 23:44:50 +03:00
Roll minor version. Change how f/w revisions are stored. Add some defines to get max luns for a card. Make sure the RESULT_QUEUE_LEN is not less than 64- I've seen breakage with that. Move the temp port database stuff into the softc (ick). Remove most of the target mode stuff. Make xflist storage an external outer layer thing (since it will now be allocated based upon the maximum commands that this HBA can support rather than request queue size).
1999-10-14 06:21:50 +04:00
#define SDPARAM(isp) ((sdparam *) (isp)->isp_param)
#define FCPARAM(isp) ((fcparam *) (isp)->isp_param)
Defines and structures necessary to support the Qlogic 2100 (Fibre Channel PCI card); Some additional defines to try and manage various flavors of the SCSI host adapters (parameters for various SYNC modes).
1997-08-16 04:18:36 +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 Driver Run States
Some major thrashing to handle building for other than NetBSD. There is one change of note- build a list of completing commands in ispintr and then say you're done- this avoids some re-entrancy issues that had surfaced.
1998-07-15 23:50:16 +04:00
*/
#define ISP_NILSTATE 0
#define ISP_RESETSTATE 1
#define ISP_INITSTATE 2
#define ISP_RUNSTATE 3
/*
* ISP Configuration Options
*/
#define ISP_CFG_NORELOAD 0x80 /* don't download f/w */
Roll internal release tag. Roll core version minor. Fix broken DPARM_DEFAULT define. Add a new config flag param (ISP_CFG_NONVRAM) whose intent it is to cause NVRAM to be ignored. Add ISPASYNC_LOOP_DOWN and ISPASYNC_LOOP_UP isp_async enums.
1999-02-09 03:46:25 +03:00
#define ISP_CFG_NONVRAM 0x40 /* ignore NVRAM */
Fix various 2300 GB issues related to chip errata and documentation misunderstandings. We also now can report our connection rate.
2001-10-07 00:34:49 +04:00
#define ISP_CFG_TWOGB 0x20 /* force 2GB connection (23XX only) */
#define ISP_CFG_ONEGB 0x10 /* force 1GB connection (23XX only) */
Roll core version minor. Clean out some target mode stuff. Respecify some queue related functions to take u_int16_t's. Restructure the softc a bit.
1999-12-04 06:06:05 +03:00
#define ISP_CFG_FULL_DUPLEX 0x01 /* Full Duplex (Fibre Channel only) */
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_CFG_PORT_PREF 0x0C /* Mask for Port Prefs (2200 only) */
#define ISP_CFG_LPORT 0x00 /* prefer {N/F}L-Port connection */
#define ISP_CFG_NPORT 0x04 /* prefer {N/F}-Port connection */
#define ISP_CFG_NPORT_ONLY 0x08 /* insist on {N/F}-Port connection */
#define ISP_CFG_LPORT_ONLY 0x0C /* insist on {N/F}L-Port connection */
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 ISP_CFG_OWNWWPN 0x100 /* override NVRAM wwpn */
#define ISP_CFG_OWNWWNN 0x200 /* override NVRAM wwnn */
Roll version minor. Add DEFAULT_FRAMESIZE/DEFAULT_EXEC_THROTTLE/ISPASYNC_FW_RESTARTED defines.
2002-06-15 04:18:02 +04:00
#define ISP_CFG_OWNFSZ 0x400 /* override NVRAM frame size */
#define ISP_CFG_OWNLOOPID 0x800 /* override NVRAM loopid */
#define ISP_CFG_OWNEXCTHROTTLE 0x1000 /* override NVRAM execution throttle */
Some major thrashing to handle building for other than NetBSD. There is one change of note- build a list of completing commands in ispintr and then say you're done- this avoids some re-entrancy issues that had surfaced.
1998-07-15 23:50:16 +04:00
Eliminate ISP2100_FABRIC- we always allow for fabric now. Add an isp_iid_set/isp_iid for fibre channel- this is because we now fake a port database entry for ourselves. Add the additional loop states between LOOP_PDB_RCVD and LOOP_READY. Change and comment on a wad of Fibre Channel isp_control functions. Change and comment on some of the ISPASYNC Fibre Channel events. Add was_fabric_dev/fabric_dev tags to our local FC database structure (so we can see rapidly whether something was a fabric device but is now gone). Add a tag which says what role this adapter should take. It can take on the value of None, Target, Initiator or Both. None is useful for warm failover purposes. Remove the ISP_CFG_NOINIT silliness since a role of "None" does this. Add a isp_lastmbxcmd tag to store the opcode for the last mailbox command used.
2001-02-13 02:33:49 +03:00
/*
* Prior to calling isp_reset for the first time, the outer layer
* should set isp_role to one of NONE, INITIATOR, TARGET, BOTH.
*
* If you set ISP_ROLE_NONE, the cards will be reset, new firmware loaded,
* NVRAM read, and defaults set, but any further initialization (e.g.
* INITIALIZE CONTROL BLOCK commands for 2X00 cards) won't be done.
*
* If INITIATOR MODE isn't set, attempts to run commands will be stopped
* at isp_start and completed with the moral equivalent of SELECTION TIMEOUT.
*
* If TARGET MODE is set, it doesn't mean that the rest of target mode support
* needs to be enabled, or will even work. What happens with the 2X00 cards
* here is that if you have enabled it with TARGET MODE as part of the ICB
* options, but you haven't given the f/w any ram resources for ATIOs or
* Immediate Notifies, the f/w just handles what it can and you never see
* anything. Basically, it sends a single byte of data (the first byte,
* which you can set as part of the INITIALIZE CONTROL BLOCK command) for
* INQUIRY, and sends back QUEUE FULL status for any other command.
White space cleanup. Fix oopslet about having overlapping loop states. Handle FW crashes as an isp_async call. Reclaim a missing logging level bit.
2001-07-06 20:19:10 +04:00
*
Eliminate ISP2100_FABRIC- we always allow for fabric now. Add an isp_iid_set/isp_iid for fibre channel- this is because we now fake a port database entry for ourselves. Add the additional loop states between LOOP_PDB_RCVD and LOOP_READY. Change and comment on a wad of Fibre Channel isp_control functions. Change and comment on some of the ISPASYNC Fibre Channel events. Add was_fabric_dev/fabric_dev tags to our local FC database structure (so we can see rapidly whether something was a fabric device but is now gone). Add a tag which says what role this adapter should take. It can take on the value of None, Target, Initiator or Both. None is useful for warm failover purposes. Remove the ISP_CFG_NOINIT silliness since a role of "None" does this. Add a isp_lastmbxcmd tag to store the opcode for the last mailbox command used.
2001-02-13 02:33:49 +03:00
*/
#define ISP_ROLE_NONE 0x0
#define ISP_ROLE_INITIATOR 0x1
#define ISP_ROLE_TARGET 0x2
#define ISP_ROLE_BOTH (ISP_ROLE_TARGET|ISP_ROLE_INITIATOR)
#define ISP_ROLE_EITHER ISP_ROLE_BOTH
#ifndef ISP_DEFAULT_ROLES
#define ISP_DEFAULT_ROLES ISP_ROLE_INITIATOR
#endif
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
/*
* Firmware related defines
*/
Eliminate ISP2100_FABRIC- we always allow for fabric now. Add an isp_iid_set/isp_iid for fibre channel- this is because we now fake a port database entry for ourselves. Add the additional loop states between LOOP_PDB_RCVD and LOOP_READY. Change and comment on a wad of Fibre Channel isp_control functions. Change and comment on some of the ISPASYNC Fibre Channel events. Add was_fabric_dev/fabric_dev tags to our local FC database structure (so we can see rapidly whether something was a fabric device but is now gone). Add a tag which says what role this adapter should take. It can take on the value of None, Target, Initiator or Both. None is useful for warm failover purposes. Remove the ISP_CFG_NOINIT silliness since a role of "None" does this. Add a isp_lastmbxcmd tag to store the opcode for the last mailbox command used.
2001-02-13 02:33:49 +03:00
#define ISP_CODE_ORG 0x1000 /* default f/w code start */
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
#define ISP_CODE_ORG_2300 0x0800 /* ..except for 2300s */
Change f/w revision to major,minor,micro version. Add in ISP1040C definition.
1999-04-04 05:27:06 +04:00
#define ISP_FW_REV(maj, min, mic) ((maj << 24) | (min << 16) | mic)
Synchronize with FreeBSD/Linux.
2001-09-06 02:32:38 +04:00
#define ISP_FW_MAJOR(code) ((code >> 24) & 0xff)
#define ISP_FW_MINOR(code) ((code >> 16) & 0xff)
#define ISP_FW_MICRO(code) ((code >> 8) & 0xff)
Eliminate ISP2100_FABRIC- we always allow for fabric now. Add an isp_iid_set/isp_iid for fibre channel- this is because we now fake a port database entry for ourselves. Add the additional loop states between LOOP_PDB_RCVD and LOOP_READY. Change and comment on a wad of Fibre Channel isp_control functions. Change and comment on some of the ISPASYNC Fibre Channel events. Add was_fabric_dev/fabric_dev tags to our local FC database structure (so we can see rapidly whether something was a fabric device but is now gone). Add a tag which says what role this adapter should take. It can take on the value of None, Target, Initiator or Both. None is useful for warm failover purposes. Remove the ISP_CFG_NOINIT silliness since a role of "None" does this. Add a isp_lastmbxcmd tag to store the opcode for the last mailbox command used.
2001-02-13 02:33:49 +03:00
#define ISP_FW_REVX(xp) ((xp[0]<<24) | (xp[1] << 16) | xp[2])
Synchronize with FreeBSD/Linux.
2001-09-06 02:32:38 +04:00
#define ISP_FW_MAJORX(xp) (xp[0])
#define ISP_FW_MINORX(xp) (xp[1])
#define ISP_FW_MICROX(xp) (xp[2])
Further fix problems with 23XX ISR handling and correctly initialize the 23XX to use ZIO in preference to fast posting.
2002-08-17 01:43:14 +04:00
#define ISP_FW_NEWER_THAN(i, major, minor, micro) \
(ISP_FW_REVX((i)->isp_fwrev) > ISP_FW_REV(major, minor, micro))
Roll major revision. Change some definitions around. Add more current device state stuff. Make DPARM flags full 16 bits (no shifting needed) and add the 2 new ones for the 7.55 f/w.
1998-09-08 11:27:04 +04:00
Some major thrashing to handle building for other than NetBSD. There is one change of note- build a list of completing commands in ispintr and then say you're done- this avoids some re-entrancy issues that had surfaced.
1998-07-15 23:50:16 +04:00
/*
Roll major revision. Change some definitions around. Add more current device state stuff. Make DPARM flags full 16 bits (no shifting needed) and add the 2 new ones for the 7.55 f/w.
1998-09-08 11:27:04 +04:00
* Bus (implementation) types
*/
#define ISP_BT_PCI 0 /* PCI Implementations */
#define ISP_BT_SBUS 1 /* SBus Implementations */
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
/*
* If we have not otherwise defined SBus support away make sure
* it is defined here such that the code is included as default
*/
#ifndef ISP_SBUS_SUPPORTED
#define ISP_SBUS_SUPPORTED 1
#endif
Roll major revision. Change some definitions around. Add more current device state stuff. Make DPARM flags full 16 bits (no shifting needed) and add the 2 new ones for the 7.55 f/w.
1998-09-08 11:27:04 +04:00
/*
* Chip Types
Defines and structures necessary to support the Qlogic 2100 (Fibre Channel PCI card); Some additional defines to try and manage various flavors of the SCSI host adapters (parameters for various SYNC modes).
1997-08-16 04:18:36 +04:00
*/
#define ISP_HA_SCSI 0xf
Roll major revision. Change some definitions around. Add more current device state stuff. Make DPARM flags full 16 bits (no shifting needed) and add the 2 new ones for the 7.55 f/w.
1998-09-08 11:27:04 +04:00
#define ISP_HA_SCSI_UNKNOWN 0x1
#define ISP_HA_SCSI_1020 0x2
#define ISP_HA_SCSI_1020A 0x3
#define ISP_HA_SCSI_1040 0x4
#define ISP_HA_SCSI_1040A 0x5
#define ISP_HA_SCSI_1040B 0x6
Change f/w revision to major,minor,micro version. Add in ISP1040C definition.
1999-04-04 05:27:06 +04:00
#define ISP_HA_SCSI_1040C 0x7
Dual LVD (1280) support.
1999-12-16 08:35:42 +03:00
#define ISP_HA_SCSI_1240 0x8
#define ISP_HA_SCSI_1080 0x9
#define ISP_HA_SCSI_1280 0xa
Remove ISP_DMA_ADDR_T as a 'type'. Instead, go off of whether ISP_DAC_SUPPORTED is defined and do a real typedef for the size of DMA addresses. ISP_DAC_SUPPORTED also decides how much of any particular DMA address is significant. Lose STRNCAT over the side. Add in (sight unseen) 10160 support.
2002-10-19 03:35:25 +04:00
#define ISP_HA_SCSI_10160 0xb
#define ISP_HA_SCSI_12160 0xc
Defines and structures necessary to support the Qlogic 2100 (Fibre Channel PCI card); Some additional defines to try and manage various flavors of the SCSI host adapters (parameters for various SYNC modes).
1997-08-16 04:18:36 +04:00
#define ISP_HA_FC 0xf0
#define ISP_HA_FC_2100 0x10
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
#define ISP_HA_FC_2200 0x20
beginning basic 2300 infrstructure support
2001-05-16 07:58:34 +04:00
#define ISP_HA_FC_2300 0x30
Implement REDUCED INTERRUPT OPERATION usage for FC cards- this allows the firmware to delay completion of commands so that it can attempt to batch a bunch of completions at once- either returning 16 bit handles in mailbox registers, or in a resposne queue entry that has a whole wad of 16 bit handles. Distinguish between 2300 and 2312 chipsets- if only because the revisions on the chips have different meanings. Add more instrumentation plus ISP_GET_STATS and ISP_CLR_STATS ioctls. Run up the maximum number of response queue entities we'll look at per interrupt. If we haven't set HBA role yet, always return success from isp_fc_runstate.
2002-01-04 00:45:06 +03:00
#define ISP_HA_FC_2312 0x40
Defines and structures necessary to support the Qlogic 2100 (Fibre Channel PCI card); Some additional defines to try and manage various flavors of the SCSI host adapters (parameters for various SYNC modes).
1997-08-16 04:18:36 +04:00
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
#define IS_SCSI(isp) (isp->isp_type & ISP_HA_SCSI)
Dual LVD (1280) support.
1999-12-16 08:35:42 +03:00
#define IS_1240(isp) (isp->isp_type == ISP_HA_SCSI_1240)
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
#define IS_1080(isp) (isp->isp_type == ISP_HA_SCSI_1080)
Dual LVD (1280) support.
1999-12-16 08:35:42 +03:00
#define IS_1280(isp) (isp->isp_type == ISP_HA_SCSI_1280)
Remove ISP_DMA_ADDR_T as a 'type'. Instead, go off of whether ISP_DAC_SUPPORTED is defined and do a real typedef for the size of DMA addresses. ISP_DAC_SUPPORTED also decides how much of any particular DMA address is significant. Lose STRNCAT over the side. Add in (sight unseen) 10160 support.
2002-10-19 03:35:25 +04:00
#define IS_10160(isp) (isp->isp_type == ISP_HA_SCSI_10160)
add in 12160 (Ultra3) related definitions
2000-02-12 05:29:00 +03:00
#define IS_12160(isp) (isp->isp_type == ISP_HA_SCSI_12160)
#define IS_12X0(isp) (IS_1240(isp) || IS_1280(isp))
Remove ISP_DMA_ADDR_T as a 'type'. Instead, go off of whether ISP_DAC_SUPPORTED is defined and do a real typedef for the size of DMA addresses. ISP_DAC_SUPPORTED also decides how much of any particular DMA address is significant. Lose STRNCAT over the side. Add in (sight unseen) 10160 support.
2002-10-19 03:35:25 +04:00
#define IS_1X160(isp) (IS_10160(isp) || IS_12160(isp))
add in 12160 (Ultra3) related definitions
2000-02-12 05:29:00 +03:00
#define IS_DUALBUS(isp) (IS_12X0(isp) || IS_12160(isp))
Remove ISP_DMA_ADDR_T as a 'type'. Instead, go off of whether ISP_DAC_SUPPORTED is defined and do a real typedef for the size of DMA addresses. ISP_DAC_SUPPORTED also decides how much of any particular DMA address is significant. Lose STRNCAT over the side. Add in (sight unseen) 10160 support.
2002-10-19 03:35:25 +04:00
#define IS_ULTRA2(isp) (IS_1080(isp) || IS_1280(isp) || IS_1X160(isp))
#define IS_ULTRA3(isp) (IS_1X160(isp))
add in 12160 (Ultra3) related definitions
2000-02-12 05:29:00 +03:00
beginning basic 2300 infrstructure support
2001-05-16 07:58:34 +04:00
#define IS_FC(isp) ((isp)->isp_type & ISP_HA_FC)
#define IS_2100(isp) ((isp)->isp_type == ISP_HA_FC_2100)
#define IS_2200(isp) ((isp)->isp_type == ISP_HA_FC_2200)
Implement REDUCED INTERRUPT OPERATION usage for FC cards- this allows the firmware to delay completion of commands so that it can attempt to batch a bunch of completions at once- either returning 16 bit handles in mailbox registers, or in a resposne queue entry that has a whole wad of 16 bit handles. Distinguish between 2300 and 2312 chipsets- if only because the revisions on the chips have different meanings. Add more instrumentation plus ISP_GET_STATS and ISP_CLR_STATS ioctls. Run up the maximum number of response queue entities we'll look at per interrupt. If we haven't set HBA role yet, always return success from isp_fc_runstate.
2002-01-04 00:45:06 +03:00
#define IS_23XX(isp) ((isp)->isp_type >= ISP_HA_FC_2300)
#define IS_2300(isp) ((isp)->isp_type == ISP_HA_FC_2300)
#define IS_2312(isp) ((isp)->isp_type == ISP_HA_FC_2312)
add in 12160 (Ultra3) related definitions
2000-02-12 05:29:00 +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
/*
* DMA cookie macros
*/
Remove ISP_DMA_ADDR_T as a 'type'. Instead, go off of whether ISP_DAC_SUPPORTED is defined and do a real typedef for the size of DMA addresses. ISP_DAC_SUPPORTED also decides how much of any particular DMA address is significant. Lose STRNCAT over the side. Add in (sight unseen) 10160 support.
2002-10-19 03:35:25 +04:00
#ifdef ISP_DAC_SUPPORTRED
#define DMA_WD3(x) (((x) >> 48) & 0xffff)
#define DMA_WD2(x) (((x) >> 32) & 0xffff)
#else
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
#define DMA_WD3(x) 0
#define DMA_WD2(x) 0
Remove ISP_DMA_ADDR_T as a 'type'. Instead, go off of whether ISP_DAC_SUPPORTED is defined and do a real typedef for the size of DMA addresses. ISP_DAC_SUPPORTED also decides how much of any particular DMA address is significant. Lose STRNCAT over the side. Add in (sight unseen) 10160 support.
2002-10-19 03:35:25 +04:00
#endif
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
#define DMA_WD1(x) (((x) >> 16) & 0xffff)
#define DMA_WD0(x) (((x) & 0xffff))
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
ISP 10x0 driver from Matthew Jacob of NASA Ames Research Center. (March 12, 1997 version).
1997-03-12 23:44:50 +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
* Core System Function Prototypes
ISP 10x0 driver from Matthew Jacob of NASA Ames Research Center. (March 12, 1997 version).
1997-03-12 23:44:50 +03:00
*/
Some major thrashing to handle building for other than NetBSD. There is one change of note- build a list of completing commands in ispintr and then say you're done- this avoids some re-entrancy issues that had surfaced.
1998-07-15 23:50:16 +04:00
ISP 10x0 driver from Matthew Jacob of NASA Ames Research Center. (March 12, 1997 version).
1997-03-12 23:44:50 +03:00
/*
Some major thrashing to handle building for other than NetBSD. There is one change of note- build a list of completing commands in ispintr and then say you're done- this avoids some re-entrancy issues that had surfaced.
1998-07-15 23:50:16 +04:00
* Reset Hardware. Totally. Assumes that you'll follow this with
* a call to isp_init.
ISP 10x0 driver from Matthew Jacob of NASA Ames Research Center. (March 12, 1997 version).
1997-03-12 23:44:50 +03:00
*/
Roll core minor. ANSify.
2001-03-14 08:14:36 +03:00
void isp_reset(struct ispsoftc *);
ISP 10x0 driver from Matthew Jacob of NASA Ames Research Center. (March 12, 1997 version).
1997-03-12 23:44:50 +03:00
/*
* Initialize Hardware to known state
*/
Roll core minor. ANSify.
2001-03-14 08:14:36 +03:00
void isp_init(struct ispsoftc *);
ISP 10x0 driver from Matthew Jacob of NASA Ames Research Center. (March 12, 1997 version).
1997-03-12 23:44:50 +03:00
roll core version number and add a global visible isp_restart function
1998-07-19 01:06:20 +04:00
/*
* Reset the ISP and call completion for any orphaned commands.
*/
Roll core minor. ANSify.
2001-03-14 08:14:36 +03:00
void isp_reinit(struct ispsoftc *);
roll core version number and add a global visible isp_restart function
1998-07-19 01:06:20 +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
#ifdef ISP_FW_CRASH_DUMP
/*
* Dump firmware entry point.
*/
void isp_fw_dump(struct ispsoftc *isp);
#endif
ISP 10x0 driver from Matthew Jacob of NASA Ames Research Center. (March 12, 1997 version).
1997-03-12 23:44:50 +03:00
/*
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
* Internal Interrupt Service Routine
*
* The outer layers do the spade work to get the appropriate status register,
* semaphore register and first mailbox register (if appropriate). This also
* means that most spurious/bogus interrupts not for us can be filtered first.
ISP 10x0 driver from Matthew Jacob of NASA Ames Research Center. (March 12, 1997 version).
1997-03-12 23:44:50 +03:00
*/
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
void isp_intr(struct ispsoftc *, u_int16_t, u_int16_t, u_int16_t);
ISP 10x0 driver from Matthew Jacob of NASA Ames Research Center. (March 12, 1997 version).
1997-03-12 23:44:50 +03:00
Some major thrashing to handle building for other than NetBSD. There is one change of note- build a list of completing commands in ispintr and then say you're done- this avoids some re-entrancy issues that had surfaced.
1998-07-15 23:50:16 +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
* Command Entry Point- Platform Dependent layers call into this
*/
Roll core minor. ANSify.
2001-03-14 08:14:36 +03:00
int isp_start(XS_T *);
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
/* these values are what isp_start returns */
#define CMD_COMPLETE 101 /* command completed */
#define CMD_EAGAIN 102 /* busy- maybe retry later */
#define CMD_QUEUED 103 /* command has been queued for execution */
#define CMD_RQLATER 104 /* requeue this command later */
/*
* Command Completion Point- Core layers call out from this with completed cmds
Some major thrashing to handle building for other than NetBSD. There is one change of note- build a list of completing commands in ispintr and then say you're done- this avoids some re-entrancy issues that had surfaced.
1998-07-15 23:50:16 +04:00
*/
Roll core minor. ANSify.
2001-03-14 08:14:36 +03:00
void isp_done(XS_T *);
ISP 10x0 driver from Matthew Jacob of NASA Ames Research Center. (March 12, 1997 version).
1997-03-12 23:44:50 +03:00
Some major thrashing to handle building for other than NetBSD. There is one change of note- build a list of completing commands in ispintr and then say you're done- this avoids some re-entrancy issues that had surfaced.
1998-07-15 23:50:16 +04:00
/*
Implement and use Fast Posting for both parallel && fibre. Redo a bit of the startup code. Implement a call to outer framework function so that asynchronous events can be handled (e.g., speed negotiation, target mode). Roll internal release tags.
1999-01-30 10:31:50 +03:00
* Platform Dependent to External to Internal Control Function
Some major thrashing to handle building for other than NetBSD. There is one change of note- build a list of completing commands in ispintr and then say you're done- this avoids some re-entrancy issues that had surfaced.
1998-07-15 23:50:16 +04:00
*
Eliminate ISP2100_FABRIC- we always allow for fabric now. Add an isp_iid_set/isp_iid for fibre channel- this is because we now fake a port database entry for ourselves. Add the additional loop states between LOOP_PDB_RCVD and LOOP_READY. Change and comment on a wad of Fibre Channel isp_control functions. Change and comment on some of the ISPASYNC Fibre Channel events. Add was_fabric_dev/fabric_dev tags to our local FC database structure (so we can see rapidly whether something was a fabric device but is now gone). Add a tag which says what role this adapter should take. It can take on the value of None, Target, Initiator or Both. None is useful for warm failover purposes. Remove the ISP_CFG_NOINIT silliness since a role of "None" does this. Add a isp_lastmbxcmd tag to store the opcode for the last mailbox command used.
2001-02-13 02:33:49 +03:00
* Assumes locks are held on entry. You should note that with many of
* these commands and locks may be released while this is occurring.
*
* A few notes about some of these functions:
*
* ISPCTL_FCLINK_TEST tests to make sure we have good fibre channel link.
* The argument is a pointer to an integer which is the time, in microseconds,
* we should wait to see whether we have good link. This test, if successful,
* lets us know our connection topology and our Loop ID/AL_PA and so on.
* You can't get anywhere without this.
*
* ISPCTL_SCAN_FABRIC queries the name server (if we're on a fabric) for
* all entities using the FC Generic Services subcommand GET ALL NEXT.
* For each found entity, an ISPASYNC_FABRICDEV event is generated (see
* below).
*
* ISPCTL_SCAN_LOOP does a local loop scan. This is only done if the connection
* topology is NL or FL port (private or public loop). Since the Qlogic f/w
* 'automatically' manages local loop connections, this function essentially
* notes the arrival, departure, and possible shuffling around of local loop
* entities. Thus for each arrival and departure this generates an isp_async
* event of ISPASYNC_PROMENADE (see below).
Some major thrashing to handle building for other than NetBSD. There is one change of note- build a list of completing commands in ispintr and then say you're done- this avoids some re-entrancy issues that had surfaced.
1998-07-15 23:50:16 +04:00
*
Eliminate ISP2100_FABRIC- we always allow for fabric now. Add an isp_iid_set/isp_iid for fibre channel- this is because we now fake a port database entry for ourselves. Add the additional loop states between LOOP_PDB_RCVD and LOOP_READY. Change and comment on a wad of Fibre Channel isp_control functions. Change and comment on some of the ISPASYNC Fibre Channel events. Add was_fabric_dev/fabric_dev tags to our local FC database structure (so we can see rapidly whether something was a fabric device but is now gone). Add a tag which says what role this adapter should take. It can take on the value of None, Target, Initiator or Both. None is useful for warm failover purposes. Remove the ISP_CFG_NOINIT silliness since a role of "None" does this. Add a isp_lastmbxcmd tag to store the opcode for the last mailbox command used.
2001-02-13 02:33:49 +03:00
* ISPCTL_PDB_SYNC is somewhat misnamed. It actually is the final step, in
* order, of ISPCTL_FCLINK_TEST, ISPCTL_SCAN_FABRIC, and ISPCTL_SCAN_LOOP.
* The main purpose of ISPCTL_PDB_SYNC is to complete management of logging
* and logging out of fabric devices (if one is on a fabric) and then marking
* the 'loop state' as being ready to now be used for sending commands to
* devices. Originally fabric name server and local loop scanning were
seperate -> separate
2001-07-22 17:33:58 +04:00
* part of this function. It's now been separated to allow for finer control.
Some major thrashing to handle building for other than NetBSD. There is one change of note- build a list of completing commands in ispintr and then say you're done- this avoids some re-entrancy issues that had surfaced.
1998-07-15 23:50:16 +04:00
*/
typedef enum {
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
ISPCTL_RESET_BUS, /* Reset Bus */
ISPCTL_RESET_DEV, /* Reset Device */
ISPCTL_ABORT_CMD, /* Abort Command */
Eliminate ISP2100_FABRIC- we always allow for fabric now. Add an isp_iid_set/isp_iid for fibre channel- this is because we now fake a port database entry for ourselves. Add the additional loop states between LOOP_PDB_RCVD and LOOP_READY. Change and comment on a wad of Fibre Channel isp_control functions. Change and comment on some of the ISPASYNC Fibre Channel events. Add was_fabric_dev/fabric_dev tags to our local FC database structure (so we can see rapidly whether something was a fabric device but is now gone). Add a tag which says what role this adapter should take. It can take on the value of None, Target, Initiator or Both. None is useful for warm failover purposes. Remove the ISP_CFG_NOINIT silliness since a role of "None" does this. Add a isp_lastmbxcmd tag to store the opcode for the last mailbox command used.
2001-02-13 02:33:49 +03:00
ISPCTL_UPDATE_PARAMS, /* Update Operating Parameters (SCSI) */
roll revision level and add some taret mode changes
2000-01-06 06:00:11 +03:00
ISPCTL_FCLINK_TEST, /* Test FC Link Status */
Eliminate ISP2100_FABRIC- we always allow for fabric now. Add an isp_iid_set/isp_iid for fibre channel- this is because we now fake a port database entry for ourselves. Add the additional loop states between LOOP_PDB_RCVD and LOOP_READY. Change and comment on a wad of Fibre Channel isp_control functions. Change and comment on some of the ISPASYNC Fibre Channel events. Add was_fabric_dev/fabric_dev tags to our local FC database structure (so we can see rapidly whether something was a fabric device but is now gone). Add a tag which says what role this adapter should take. It can take on the value of None, Target, Initiator or Both. None is useful for warm failover purposes. Remove the ISP_CFG_NOINIT silliness since a role of "None" does this. Add a isp_lastmbxcmd tag to store the opcode for the last mailbox command used.
2001-02-13 02:33:49 +03:00
ISPCTL_SCAN_FABRIC, /* (Re)scan Fabric Name Server */
ISPCTL_SCAN_LOOP, /* (Re)scan Local Loop */
roll revision level and add some taret mode changes
2000-01-06 06:00:11 +03:00
ISPCTL_PDB_SYNC, /* Synchronize Port Database */
Eliminate ISP2100_FABRIC- we always allow for fabric now. Add an isp_iid_set/isp_iid for fibre channel- this is because we now fake a port database entry for ourselves. Add the additional loop states between LOOP_PDB_RCVD and LOOP_READY. Change and comment on a wad of Fibre Channel isp_control functions. Change and comment on some of the ISPASYNC Fibre Channel events. Add was_fabric_dev/fabric_dev tags to our local FC database structure (so we can see rapidly whether something was a fabric device but is now gone). Add a tag which says what role this adapter should take. It can take on the value of None, Target, Initiator or Both. None is useful for warm failover purposes. Remove the ISP_CFG_NOINIT silliness since a role of "None" does this. Add a isp_lastmbxcmd tag to store the opcode for the last mailbox command used.
2001-02-13 02:33:49 +03:00
ISPCTL_SEND_LIP, /* Send a LIP */
ISPCTL_GET_POSMAP, /* Get FC-AL position map */
Add ISPCTL_RUN_MBOXCMD control op and ISPASYNC_UNHANDLED_RESPONSE async event.
2001-02-20 04:11:35 +03:00
ISPCTL_RUN_MBOXCMD, /* run a mailbox command */
roll revision level and add some taret mode changes
2000-01-06 06:00:11 +03:00
ISPCTL_TOGGLE_TMODE /* toggle target mode */
Some major thrashing to handle building for other than NetBSD. There is one change of note- build a list of completing commands in ispintr and then say you're done- this avoids some re-entrancy issues that had surfaced.
1998-07-15 23:50:16 +04:00
} ispctl_t;
Roll core minor. ANSify.
2001-03-14 08:14:36 +03:00
int isp_control(struct ispsoftc *, ispctl_t, void *);
ISP 10x0 driver from Matthew Jacob of NASA Ames Research Center. (March 12, 1997 version).
1997-03-12 23:44:50 +03:00
Implement and use Fast Posting for both parallel && fibre. Redo a bit of the startup code. Implement a call to outer framework function so that asynchronous events can be handled (e.g., speed negotiation, target mode). Roll internal release tags.
1999-01-30 10:31:50 +03:00
/*
* Platform Dependent to Internal to External Control Function
* (each platform must provide such a function)
*
Eliminate ISP2100_FABRIC- we always allow for fabric now. Add an isp_iid_set/isp_iid for fibre channel- this is because we now fake a port database entry for ourselves. Add the additional loop states between LOOP_PDB_RCVD and LOOP_READY. Change and comment on a wad of Fibre Channel isp_control functions. Change and comment on some of the ISPASYNC Fibre Channel events. Add was_fabric_dev/fabric_dev tags to our local FC database structure (so we can see rapidly whether something was a fabric device but is now gone). Add a tag which says what role this adapter should take. It can take on the value of None, Target, Initiator or Both. None is useful for warm failover purposes. Remove the ISP_CFG_NOINIT silliness since a role of "None" does this. Add a isp_lastmbxcmd tag to store the opcode for the last mailbox command used.
2001-02-13 02:33:49 +03:00
* Assumes locks are held.
*
* A few notes about some of these functions:
*
* ISPASYNC_CHANGE_NOTIFY notifies the outer layer that a change has
* occurred that invalidates the list of fabric devices known and/or
* the list of known loop devices. The argument passed is a pointer
* whose values are defined below (local loop change, name server
* change, other). 'Other' may simply be a LIP, or a change in
* connection topology.
*
* ISPASYNC_FABRIC_DEV announces the next element in a list of
* fabric device names we're getting out of the name server. The
* argument points to a GET ALL NEXT response structure. The list
* is known to terminate with an entry that refers to ourselves.
* One of the main purposes of this function is to allow outer
* layers, which are OS dependent, to set policy as to which fabric
* devices might actually be logged into (and made visible) later
* at ISPCTL_PDB_SYNC time. Since there's a finite number of fabric
* devices that we can log into (256 less 3 'reserved' for F-port
* topologies), and fabrics can grow up to 8 million or so entries
* (24 bits of Port Address, less a wad of reserved spaces), clearly
* we had better let the OS determine login policy.
*
* ISPASYNC_PROMENADE has an argument that is a pointer to an integer which
Remove ISP_DMA_ADDR_T as a 'type'. Instead, go off of whether ISP_DAC_SUPPORTED is defined and do a real typedef for the size of DMA addresses. ISP_DAC_SUPPORTED also decides how much of any particular DMA address is significant. Lose STRNCAT over the side. Add in (sight unseen) 10160 support.
2002-10-19 03:35:25 +04:00
* is an index into the portdb in the softc ('target'). Whether that entry's
Eliminate ISP2100_FABRIC- we always allow for fabric now. Add an isp_iid_set/isp_iid for fibre channel- this is because we now fake a port database entry for ourselves. Add the additional loop states between LOOP_PDB_RCVD and LOOP_READY. Change and comment on a wad of Fibre Channel isp_control functions. Change and comment on some of the ISPASYNC Fibre Channel events. Add was_fabric_dev/fabric_dev tags to our local FC database structure (so we can see rapidly whether something was a fabric device but is now gone). Add a tag which says what role this adapter should take. It can take on the value of None, Target, Initiator or Both. None is useful for warm failover purposes. Remove the ISP_CFG_NOINIT silliness since a role of "None" does this. Add a isp_lastmbxcmd tag to store the opcode for the last mailbox command used.
2001-02-13 02:33:49 +03:00
* valid tag is set or not says whether something has arrived or departed.
* The name refers to a favorite pastime of many city dwellers- watching
* people come and go, talking of Michaelangelo, and so on..
Add ISPCTL_RUN_MBOXCMD control op and ISPASYNC_UNHANDLED_RESPONSE async event.
2001-02-20 04:11:35 +03:00
*
* ISPASYNC_UNHANDLED_RESPONSE gives outer layers a chance to parse a
* response queue entry not otherwise handled. The outer layer should
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
* return non-zero if it handled it. The 'arg' points to an unmassaged
* response queue entry.
Implement and use Fast Posting for both parallel && fibre. Redo a bit of the startup code. Implement a call to outer framework function so that asynchronous events can be handled (e.g., speed negotiation, target mode). Roll internal release tags.
1999-01-30 10:31:50 +03:00
*/
typedef enum {
Eliminate ISP2100_FABRIC- we always allow for fabric now. Add an isp_iid_set/isp_iid for fibre channel- this is because we now fake a port database entry for ourselves. Add the additional loop states between LOOP_PDB_RCVD and LOOP_READY. Change and comment on a wad of Fibre Channel isp_control functions. Change and comment on some of the ISPASYNC Fibre Channel events. Add was_fabric_dev/fabric_dev tags to our local FC database structure (so we can see rapidly whether something was a fabric device but is now gone). Add a tag which says what role this adapter should take. It can take on the value of None, Target, Initiator or Both. None is useful for warm failover purposes. Remove the ISP_CFG_NOINIT silliness since a role of "None" does this. Add a isp_lastmbxcmd tag to store the opcode for the last mailbox command used.
2001-02-13 02:33:49 +03:00
ISPASYNC_NEW_TGT_PARAMS, /* New Target Parameters Negotiated */
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
ISPASYNC_BUS_RESET, /* Bus Was Reset */
ISPASYNC_LOOP_DOWN, /* FC Loop Down */
ISPASYNC_LOOP_UP, /* FC Loop Up */
Add back in (again) ISPASYNC_LIP and ISPASYNC_LOOP_RESET async notifications. Let each platform set policy as to what to do with these.
2001-05-22 23:32:41 +04:00
ISPASYNC_LIP, /* LIP Received */
ISPASYNC_LOOP_RESET, /* Loop Reset Received */
Eliminate ISP2100_FABRIC- we always allow for fabric now. Add an isp_iid_set/isp_iid for fibre channel- this is because we now fake a port database entry for ourselves. Add the additional loop states between LOOP_PDB_RCVD and LOOP_READY. Change and comment on a wad of Fibre Channel isp_control functions. Change and comment on some of the ISPASYNC Fibre Channel events. Add was_fabric_dev/fabric_dev tags to our local FC database structure (so we can see rapidly whether something was a fabric device but is now gone). Add a tag which says what role this adapter should take. It can take on the value of None, Target, Initiator or Both. None is useful for warm failover purposes. Remove the ISP_CFG_NOINIT silliness since a role of "None" does this. Add a isp_lastmbxcmd tag to store the opcode for the last mailbox command used.
2001-02-13 02:33:49 +03:00
ISPASYNC_CHANGE_NOTIFY, /* FC Change Notification */
ISPASYNC_FABRIC_DEV, /* FC Fabric Device Arrival */
ISPASYNC_PROMENADE, /* FC Objects coming && going */
roll revision level and add some taret mode changes
2000-01-06 06:00:11 +03:00
ISPASYNC_TARGET_MESSAGE, /* target message */
ISPASYNC_TARGET_EVENT, /* target asynchronous event */
Eliminate ISP2100_FABRIC- we always allow for fabric now. Add an isp_iid_set/isp_iid for fibre channel- this is because we now fake a port database entry for ourselves. Add the additional loop states between LOOP_PDB_RCVD and LOOP_READY. Change and comment on a wad of Fibre Channel isp_control functions. Change and comment on some of the ISPASYNC Fibre Channel events. Add was_fabric_dev/fabric_dev tags to our local FC database structure (so we can see rapidly whether something was a fabric device but is now gone). Add a tag which says what role this adapter should take. It can take on the value of None, Target, Initiator or Both. None is useful for warm failover purposes. Remove the ISP_CFG_NOINIT silliness since a role of "None" does this. Add a isp_lastmbxcmd tag to store the opcode for the last mailbox command used.
2001-02-13 02:33:49 +03:00
ISPASYNC_TARGET_ACTION, /* other target command action */
Add ISPCTL_RUN_MBOXCMD control op and ISPASYNC_UNHANDLED_RESPONSE async event.
2001-02-20 04:11:35 +03:00
ISPASYNC_CONF_CHANGE, /* Platform Configuration Change */
White space cleanup. Fix oopslet about having overlapping loop states. Handle FW crashes as an isp_async call. Reclaim a missing logging level bit.
2001-07-06 20:19:10 +04:00
ISPASYNC_UNHANDLED_RESPONSE, /* Unhandled Response Entry */
Roll version minor. Add DEFAULT_FRAMESIZE/DEFAULT_EXEC_THROTTLE/ISPASYNC_FW_RESTARTED defines.
2002-06-15 04:18:02 +04:00
ISPASYNC_FW_CRASH, /* Firmware has crashed */
Various changes to allow for overriding just loopid and/or iid. Fix a random 23XX ISR reading bug. Be chattier about HBA_BOTCH errors. More f/w cashdump layering.
2002-08-13 01:33:39 +04:00
ISPASYNC_FW_DUMPED, /* Firmware crashdump taken */
Roll version minor. Add DEFAULT_FRAMESIZE/DEFAULT_EXEC_THROTTLE/ISPASYNC_FW_RESTARTED defines.
2002-06-15 04:18:02 +04:00
ISPASYNC_FW_RESTARTED /* Firmware has been restarted */
Implement and use Fast Posting for both parallel && fibre. Redo a bit of the startup code. Implement a call to outer framework function so that asynchronous events can be handled (e.g., speed negotiation, target mode). Roll internal release tags.
1999-01-30 10:31:50 +03:00
} ispasync_t;
Roll core minor. ANSify.
2001-03-14 08:14:36 +03:00
int isp_async(struct ispsoftc *, ispasync_t, void *);
Implement and use Fast Posting for both parallel && fibre. Redo a bit of the startup code. Implement a call to outer framework function so that asynchronous events can be handled (e.g., speed negotiation, target mode). Roll internal release tags.
1999-01-30 10:31:50 +03:00
Eliminate ISP2100_FABRIC- we always allow for fabric now. Add an isp_iid_set/isp_iid for fibre channel- this is because we now fake a port database entry for ourselves. Add the additional loop states between LOOP_PDB_RCVD and LOOP_READY. Change and comment on a wad of Fibre Channel isp_control functions. Change and comment on some of the ISPASYNC Fibre Channel events. Add was_fabric_dev/fabric_dev tags to our local FC database structure (so we can see rapidly whether something was a fabric device but is now gone). Add a tag which says what role this adapter should take. It can take on the value of None, Target, Initiator or Both. None is useful for warm failover purposes. Remove the ISP_CFG_NOINIT silliness since a role of "None" does this. Add a isp_lastmbxcmd tag to store the opcode for the last mailbox command used.
2001-02-13 02:33:49 +03:00
#define ISPASYNC_CHANGE_PDB ((void *) 0)
#define ISPASYNC_CHANGE_SNS ((void *) 1)
#define ISPASYNC_CHANGE_OTHER ((void *) 2)
Some major thrashing to handle building for other than NetBSD. There is one change of note- build a list of completing commands in ispintr and then say you're done- this avoids some re-entrancy issues that had surfaced.
1998-07-15 23:50:16 +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 Dependent Error and Debug Printout
Some major thrashing to handle building for other than NetBSD. There is one change of note- build a list of completing commands in ispintr and then say you're done- this avoids some re-entrancy issues that had surfaced.
1998-07-15 23:50:16 +04:00
*/
Apply nearly all of Bill Sommerfeld's -Wformat patches. The one change I didn't take is the %llu format- I can't have a common across multiple platform module assume a %ll argument capability- which really pointed out that I shouldn't be trying to *print* something which could long long.
2000-12-29 01:27:46 +03:00
#ifdef __GNUC__
Roll core minor. ANSify.
2001-03-14 08:14:36 +03:00
void isp_prt(struct ispsoftc *, int level, const char *, ...)
Apply nearly all of Bill Sommerfeld's -Wformat patches. The one change I didn't take is the %llu format- I can't have a common across multiple platform module assume a %ll argument capability- which really pointed out that I shouldn't be trying to *print* something which could long long.
2000-12-29 01:27:46 +03:00
__attribute__((__format__(__printf__,3,4)));
#else
Roll core minor. ANSify.
2001-03-14 08:14:36 +03:00
void isp_prt(struct ispsoftc *, int level, const char *, ...);
Apply nearly all of Bill Sommerfeld's -Wformat patches. The one change I didn't take is the %llu format- I can't have a common across multiple platform module assume a %ll argument capability- which really pointed out that I shouldn't be trying to *print* something which could long long.
2000-12-29 01:27:46 +03:00
#endif
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_LOGALL 0x0 /* log always */
#define ISP_LOGCONFIG 0x1 /* log configuration messages */
#define ISP_LOGINFO 0x2 /* log informational messages */
#define ISP_LOGWARN 0x4 /* log warning messages */
#define ISP_LOGERR 0x8 /* log error messages */
#define ISP_LOGDEBUG0 0x10 /* log simple debug messages */
#define ISP_LOGDEBUG1 0x20 /* log intermediate debug messages */
#define ISP_LOGDEBUG2 0x40 /* log most debug messages */
White space cleanup. Fix oopslet about having overlapping loop states. Handle FW crashes as an isp_async call. Reclaim a missing logging level bit.
2001-07-06 20:19:10 +04:00
#define ISP_LOGDEBUG3 0x80 /* log high frequency debug messages */
#define ISP_LOGDEBUG4 0x100 /* log high frequency debug messages */
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_LOGTDEBUG0 0x200 /* log simple debug messages (target mode) */
#define ISP_LOGTDEBUG1 0x400 /* log intermediate debug messages (target) */
#define ISP_LOGTDEBUG2 0x800 /* log all debug messages (target) */
Roll major revision. Change some definitions around. Add more current device state stuff. Make DPARM flags full 16 bits (no shifting needed) and add the 2 new ones for the 7.55 f/w.
1998-09-08 11:27:04 +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
/*
* Each Platform provides it's own isposinfo substructure of the ispsoftc
* defined above.
*
* Each platform must also provide the following macros/defines:
*
*
* INLINE - platform specific define for 'inline' functions
*
Remove ISP_DMA_ADDR_T as a 'type'. Instead, go off of whether ISP_DAC_SUPPORTED is defined and do a real typedef for the size of DMA addresses. ISP_DAC_SUPPORTED also decides how much of any particular DMA address is significant. Lose STRNCAT over the side. Add in (sight unseen) 10160 support.
2002-10-19 03:35:25 +04:00
* ISP_DAC_SUPPORTED - Is DAC (Dual Address Cycle) is supported?
* Basically means whether or not DMA for PCI
* PCI cards (Ultra2 or better or FC) works
* above 4GB.
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
*
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
* ISP2100_SCRLEN - length for the Fibre Channel scratch DMA area
*
* MEMZERO(dst, src) platform zeroing function
* MEMCPY(dst, src, count) platform copying function
* SNPRINTF(buf, bufsize, fmt, ...) snprintf
* USEC_DELAY(usecs) microsecond spindelay function
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
* USEC_SLEEP(isp, usecs) microsecond sleep function
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
*
* NANOTIME_T nanosecond time type
*
* GET_NANOTIME(NANOTIME_T *) get current nanotime.
*
* GET_NANOSEC(NANOTIME_T *) get u_int64_t from NANOTIME_T
*
* NANOTIME_SUB(NANOTIME_T *, NANOTIME_T *)
* subtract two NANOTIME_T values
*
*
* MAXISPREQUEST(struct ispsoftc *) maximum request queue size
* for this particular board type
*
* MEMORYBARRIER(struct ispsoftc *, barrier_type, offset, size)
*
* Function/Macro the provides memory synchronization on
* various objects so that the ISP's and the system's view
* of the same object is consistent.
*
* MBOX_ACQUIRE(struct ispsoftc *) acquire lock on mailbox regs
* MBOX_WAIT_COMPLETE(struct ispsoftc *) wait for mailbox cmd to be done
* MBOX_NOTIFY_COMPLETE(struct ispsoftc *) notification of mbox cmd donee
* MBOX_RELEASE(struct ispsoftc *) release lock on mailbox regs
Eliminate ISP2100_FABRIC- we always allow for fabric now. Add an isp_iid_set/isp_iid for fibre channel- this is because we now fake a port database entry for ourselves. Add the additional loop states between LOOP_PDB_RCVD and LOOP_READY. Change and comment on a wad of Fibre Channel isp_control functions. Change and comment on some of the ISPASYNC Fibre Channel events. Add was_fabric_dev/fabric_dev tags to our local FC database structure (so we can see rapidly whether something was a fabric device but is now gone). Add a tag which says what role this adapter should take. It can take on the value of None, Target, Initiator or Both. None is useful for warm failover purposes. Remove the ISP_CFG_NOINIT silliness since a role of "None" does this. Add a isp_lastmbxcmd tag to store the opcode for the last mailbox command used.
2001-02-13 02:33:49 +03:00
*
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
* FC_SCRATCH_ACQUIRE(struct ispsoftc *) acquire lock on FC scratch area
* FC_SCRATCH_RELEASE(struct ispsoftc *) acquire lock on FC scratch area
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
*
* SCSI_GOOD SCSI 'Good' Status
* SCSI_CHECK SCSI 'Check Condition' Status
* SCSI_BUSY SCSI 'Busy' Status
* SCSI_QFULL SCSI 'Queue Full' Status
*
* XS_T Platform SCSI transaction type (i.e., command for HBA)
* XS_ISP(xs) gets an instance out of an XS_T
* XS_CHANNEL(xs) gets the channel (bus # for DUALBUS cards) ""
* XS_TGT(xs) gets the target ""
* XS_LUN(xs) gets the lun ""
* XS_CDBP(xs) gets a pointer to the scsi CDB ""
* XS_CDBLEN(xs) gets the CDB's length ""
* XS_XFRLEN(xs) gets the associated data transfer length ""
* XS_TIME(xs) gets the time (in milliseconds) for this command
* XS_RESID(xs) gets the current residual count
* XS_STSP(xs) gets a pointer to the SCSI status byte ""
* XS_SNSP(xs) gets a pointer to the associate sense data
* XS_SNSLEN(xs) gets the length of sense data storage
* XS_SNSKEY(xs) dereferences XS_SNSP to get the current stored Sense Key
* XS_TAG_P(xs) predicate of whether this command should be tagged
* XS_TAG_TYPE(xs) which type of tag to use
* XS_SETERR(xs) set error state
*
* HBA_NOERROR command has no erros
* HBA_BOTCH hba botched something
* HBA_CMDTIMEOUT command timed out
* HBA_SELTIMEOUT selection timed out (also port logouts for FC)
* HBA_TGTBSY target returned a BUSY status
* HBA_BUSRESET bus reset destroyed command
* HBA_ABORTED command was aborted (by request)
* HBA_DATAOVR a data overrun was detected
* HBA_ARQFAIL Automatic Request Sense failed
*
* XS_ERR(xs) return current error state
* XS_NOERR(xs) there is no error currently set
* XS_INITERR(xs) initialize error state
*
* XS_SAVE_SENSE(xs, sp) save sense data
*
* XS_SET_STATE_STAT(isp, sp, xs) platform dependent interpreter of
* response queue entry status bits
*
*
* DEFAULT_IID(struct ispsoftc *) Default SCSI initiator ID
* DEFAULT_LOOPID(struct ispsoftc *) Default FC Loop ID
Change some of the WWN macros- eliminate PORT_FROM_NODE_WWN and make sure we have a DEFAULT_{PORT,NODE}WWN and ISP_{NODE,PORT}WWN macro as a requirement for all platforms.
2000-10-16 09:16:56 +04:00
* DEFAULT_NODEWWN(struct ispsoftc *) Default Node WWN
* DEFAULT_PORTWWN(struct ispsoftc *) Default Port WWN
Roll version minor. Add DEFAULT_FRAMESIZE/DEFAULT_EXEC_THROTTLE/ISPASYNC_FW_RESTARTED defines.
2002-06-15 04:18:02 +04:00
* DEFAULT_FRAMESIZE(struct ispsoftc *) Default Frame Size
* DEFAULT_EXEC_THROTTLE(struct ispsoftc *) Default Execution Throttle
Change some of the WWN macros- eliminate PORT_FROM_NODE_WWN and make sure we have a DEFAULT_{PORT,NODE}WWN and ISP_{NODE,PORT}WWN macro as a requirement for all platforms.
2000-10-16 09:16:56 +04:00
* These establish reasonable defaults for each platform.
* These must be available independent of card NVRAM and are
* to be used should NVRAM not be readable.
*
* ISP_NODEWWN(struct ispsoftc *) FC Node WWN to use
* ISP_PORTWWN(struct ispsoftc *) FC Port WWN to use
*
* These are to be used after NVRAM is read. The tags
* in fcparam.isp_{node,port}wwn reflect the values
* read from NVRAM (possibly corrected for card botches).
* Each platform can take that information and override
* it or ignore and return the Node and Port WWNs to be
* used when sending the Qlogic f/w the Initialization Control
* Block.
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
*
* (XXX these do endian specific transformations- in transition XXX)
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
*
* ISP_IOXPUT_8(struct ispsoftc *, u_int8_t srcval, u_int8_t *dstptr)
* ISP_IOXPUT_16(struct ispsoftc *, u_int16_t srcval, u_int16_t *dstptr)
* ISP_IOXPUT_32(struct ispsoftc *, u_int32_t srcval, u_int32_t *dstptr)
*
* ISP_IOXGET_8(struct ispsoftc *, u_int8_t *srcptr, u_int8_t dstrval)
* ISP_IOXGET_16(struct ispsoftc *, u_int16_t *srcptr, u_int16_t dstrval)
* ISP_IOXGET_32(struct ispsoftc *, u_int32_t *srcptr, u_int32_t dstrval)
*
* ISP_SWIZZLE_NVRAM_WORD(struct ispsoftc *, u_int16_t *)
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
*/
beginning basic 2300 infrstructure support
2001-05-16 07:58:34 +04:00
ISP 10x0 driver from Matthew Jacob of NASA Ames Research Center. (March 12, 1997 version).
1997-03-12 23:44:50 +03:00
#endif /* _ISPVAR_H */
Reference in New Issue Copy Permalink