d/freebsd-skq
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
420239c773
freebsd-skq/sys/dev/isp/isp_freebsd.c

3176 lines
81 KiB
C
Raw Normal View History

Start each of the license/copyright comments with /*-, minor shuffle of lines
2005-01-06 01:43:34 +00:00
/*-
Add support for the Qlogic ISP SCSI && FC/AL Adapters
1998-04-22 17:54:58 +00:00
* Platform (FreeBSD) dependent common attachment code for Qlogic adapters.
*
First of several commits as this driver is dusted off and maybe brought up to date. Principle changes for this reelase is to support 2K Port Login firmware. This allows us to support the 2322 (and 2422 4Gb) cards which only come with the 2K Port Login firmware. The 2322 should now work- but we don't have firmware sets for it in ispfw (as the change to load 2K Port Login f/w hasn't been made- that f/w is so big it has to be loaded in more than one chunk). Other changes are the beginnings of cleaning up some long standing target mode issues. The next changes here will incorporate a lot of bug fixes from others. Finally, some copyright cleanup and attempts to make the parts of the driver that are FreeBSD specific start conforming more to FreeBSD style. MFC after: 1 month
2006-01-23 06:23:37 +00:00
* Copyright (c) 1997-2006 by Matthew Jacob
* All rights reserved.
Add support for the Qlogic ISP SCSI && FC/AL Adapters
1998-04-22 17:54:58 +00:00
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice immediately at the beginning of the file, without modification,
* this list of conditions, and the following disclaimer.
some copyright cleanups
2000-09-21 20:16:04 +00:00
* 2. The name of the author may not be used to endorse or promote products
* derived from this software without specific prior written permission.
Add support for the Qlogic ISP SCSI && FC/AL Adapters
1998-04-22 17:54:58 +00:00
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE FOR
* ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*/
Use __FBSDID(). Also some minor style cleanups.
2003-08-24 17:55:58 +00:00
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
Add support for the Qlogic ISP SCSI && FC/AL Adapters
1998-04-22 17:54:58 +00:00
#include <dev/isp/isp_freebsd.h>
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
#include <sys/unistd.h>
#include <sys/kthread.h>
Add a comment as to where stdarg.h applies.
2000-08-03 03:05:50 +00:00
#include <machine/stdarg.h> /* for use by isp_prt below */
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
#include <sys/conf.h>
Put MODULE_VERSION back here so that ispfw is happy.
2002-07-25 16:00:24 +00:00
#include <sys/module.h>
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
#include <sys/ioccom.h>
#include <dev/isp/isp_ioctl.h>
Add support for the Qlogic ISP SCSI && FC/AL Adapters
1998-04-22 17:54:58 +00:00
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
Put MODULE_VERSION back here so that ispfw is happy.
2002-07-25 16:00:24 +00:00
MODULE_VERSION(isp, 1);
Add a module dependancy. Now CAM will autoload when you load this. OK'ed by: mdodd
2003-09-15 06:41:33 +00:00
MODULE_DEPEND(isp, cam, 1, 1, 1);
'Support' for ISP SBus cards. This code does not imply that SBus cards work yet. They hang for me. But I can't netboot the latest snapshot on my ultra1e, and things hang at bus_setup_intr time. Since I'm offline for a while, I thought I'd toss this in in case somebody else who has a bit better luck wants to fart around with it. Please try and wait until I get back to check things in.
2002-07-11 03:25:04 +00:00
int isp_announced = 0;
ispfwfunc *isp_get_firmware_p = NULL;
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
static d_ioctl_t ispioctl;
Add in config_hook for catching when interrupts are safe- this allows us to not the ints are ok and also to (re)ENABLE isp interrupts. Remove all splcam()/splx() invocates and replace them with ISP_LOCK/ISP_UNLOCK macros.
2000-07-04 01:05:43 +00:00
static void isp_intr_enable(void *);
a) clean up some declaration stuff (i.e., make more modern with respect to getting rid u_int for uint and so on). b) Turn back on 64 bit DAC support. Cheeze it a bit in that we have two DMA callback functions- one when we have bus_addr_t > 4 bits in width and the other which should be normal. Even Cheezier in that we turn off setting up DMA maps to be BUS_SPACE_MAXADDR if we're in ISP_TARGET_MODE. More work on this in a week or so. c) Tested under amd64 and 1MB DFLTPHYS, sparc64, i386 (PAE, but insufficient memory to really test > 4GB). LINT check under amd64. MFC after: 1 month
2006-02-15 00:31:48 +00:00
static void isp_cam_async(void *, uint32_t, struct cam_path *, void *);
Clarify and cleanup some CAM queueing breakages.
1999-08-16 20:11:45 +00:00
static void isp_poll(struct cam_sim *);
Add in the enabling of interrupts (to isp_attach). Clean up a busted comment. Check against firmware state- not loop state when enabling target mode. Other changes have to do with no longer enabling/disabling interrupts at will. Rearchitect command watchdog timeouts- First of all, set the timeout period for a command that has a timeout (in isp_action) to the period of time requested *plus* two seconds. We don't want the Qlogic firmware and the host system to race each other to report a dead command (the watchdog is there to catch dead and/or broken firmware). Next, make sure that the command being watched isn't done yet. If it's not done yet, check for INT_PENDING and call isp_intr- if that said it serviced an interrupt, check to see whether the command is now done (this is what the "IN WATCHDOG" private flag is for- if isp_intr completes the command, it won't call xpt_done on it because isp_watchdog is still looking at the command). If no interrupt was pending, or the command wasn't completed, check to see if we've set the private 'grace period' flag. If so, the command really *is* dead, so report it as dead and complete it with a CAM_CMD_TIMEOUT value. If the grace period flag wasn't set, set it and issue a SYNCHRONIZE_ALL Marker Request Queue entry and re-set the timeout for one second from now (see Revision 1.45 isp.c notes for more on this) to give the firmware a final chance to complete this command.
2000-06-27 19:31:02 +00:00
static timeout_t isp_watchdog;
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
static void isp_kthread(void *);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
static void isp_action(struct cam_sim *, union ccb *);
Clarify and cleanup some CAM queueing breakages.
1999-08-16 20:11:45 +00:00
Add in a watchdog routine to catch cases where we've dropped the command. Apparently the f/w has finished the command, but somehow an interrupt is being lost. So, we just plain wedge when booting alphas. This is a general routine we've needed for a while.
2000-05-09 01:08:21 +00:00
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
static struct cdevsw isp_cdevsw = {
Device megapatch 4/6: Introduce d_version field in struct cdevsw, this must always be initialized to D_VERSION. Flip sense of D_NOGIANT flag to D_NEEDGIANT, this involves removing four D_NOGIANT flags and adding 145 D_NEEDGIANT flags.
2004-02-21 21:10:55 +00:00
.d_version = D_VERSION,
.d_flags = D_NEEDGIANT,
Gigacommit to improve device-driver source compatibility between branches: Initialize struct cdevsw using C99 sparse initializtion and remove all initializations to default values. This patch is automatically generated and has been tested by compiling LINT with all the fields in struct cdevsw in reverse order on alpha, sparc64 and i386. Approved by: re(scottl)
2003-03-03 12:15:54 +00:00
.d_ioctl = ispioctl,
.d_name = "isp",
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
};
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
static struct ispsoftc *isplist = NULL;
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
void
clarify headers;move uninit to outer layer;remove watchdog
1998-12-28 19:22:27 +00:00
isp_attach(struct ispsoftc *isp)
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
{
A large set of changes to handle dual bus adapters.
1999-05-11 05:10:06 +00:00
int primary, secondary;
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
struct ccb_setasync csa;
struct cam_devq *devq;
A large set of changes to handle dual bus adapters.
1999-05-11 05:10:06 +00:00
struct cam_sim *sim;
struct cam_path *path;
/*
* Establish (in case of 12X0) which bus is the primary.
*/
primary = 0;
secondary = 1;
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
/*
A large set of changes to handle dual bus adapters.
1999-05-11 05:10:06 +00:00
* Create the device queue for our SIM(s).
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
*/
Add in isp_debug variable. It defaults to zero unless CAMDEBUG is defined where it defaults to one. Change simq width allocation to the max number of commands supported by the HBA after f/w fires up- not the constant MAXISPREQUEST value. Do some stylistic changes.
1999-10-17 18:50:47 +00:00
devq = cam_simq_alloc(isp->isp_maxcmds);
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
if (devq == NULL) {
return;
}
/*
A large set of changes to handle dual bus adapters.
1999-05-11 05:10:06 +00:00
* Construct our SIM entry.
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
*/
Things have become cinched down more tightly about assertions for Giant. This uncovered some missing spots where I trade off between isp's lock and Giant as I enter CAM.
2001-07-05 17:14:57 +00:00
ISPLOCK_2_CAMLOCK(isp);
A large set of changes to handle dual bus adapters.
1999-05-11 05:10:06 +00:00
sim = cam_sim_alloc(isp_action, isp_poll, "isp", isp,
Finally eliminate as many of the printf calls as possible (still leaving ones where we have a CAM path) and replacing them with calls to isp_prt., Eliminate isp_unit references- we no longer have an isp_unit- we now have an isp_dev that device_get_unit can work with.
2001-03-01 02:14:54 +00:00
device_get_unit(isp->isp_dev), 1, isp->isp_maxcmds, devq);
A large set of changes to handle dual bus adapters.
1999-05-11 05:10:06 +00:00
if (sim == NULL) {
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
cam_simq_free(devq);
Things have become cinched down more tightly about assertions for Giant. This uncovered some missing spots where I trade off between isp's lock and Giant as I enter CAM.
2001-07-05 17:14:57 +00:00
CAMLOCK_2_ISPLOCK(isp);
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
return;
}
Things have become cinched down more tightly about assertions for Giant. This uncovered some missing spots where I trade off between isp's lock and Giant as I enter CAM.
2001-07-05 17:14:57 +00:00
CAMLOCK_2_ISPLOCK(isp);
Add in config_hook for catching when interrupts are safe- this allows us to not the ints are ok and also to (re)ENABLE isp interrupts. Remove all splcam()/splx() invocates and replace them with ISP_LOCK/ISP_UNLOCK macros.
2000-07-04 01:05:43 +00:00
isp->isp_osinfo.ehook.ich_func = isp_intr_enable;
isp->isp_osinfo.ehook.ich_arg = isp;
Things have become cinched down more tightly about assertions for Giant. This uncovered some missing spots where I trade off between isp's lock and Giant as I enter CAM.
2001-07-05 17:14:57 +00:00
ISPLOCK_2_CAMLOCK(isp);
Add in config_hook for catching when interrupts are safe- this allows us to not the ints are ok and also to (re)ENABLE isp interrupts. Remove all splcam()/splx() invocates and replace them with ISP_LOCK/ISP_UNLOCK macros.
2000-07-04 01:05:43 +00:00
if (config_intrhook_establish(&isp->isp_osinfo.ehook) != 0) {
Things have become cinched down more tightly about assertions for Giant. This uncovered some missing spots where I trade off between isp's lock and Giant as I enter CAM.
2001-07-05 17:14:57 +00:00
cam_sim_free(sim, TRUE);
CAMLOCK_2_ISPLOCK(isp);
Replace some more printfs with isp_prt's. Use isp_prt/ISP_LOGDEBUG0 for rate setting/getting printouts.
2000-12-05 07:39:54 +00:00
isp_prt(isp, ISP_LOGERR,
"could not establish interrupt enable hook");
Add in config_hook for catching when interrupts are safe- this allows us to not the ints are ok and also to (re)ENABLE isp interrupts. Remove all splcam()/splx() invocates and replace them with ISP_LOCK/ISP_UNLOCK macros.
2000-07-04 01:05:43 +00:00
return;
}
A large set of changes to handle dual bus adapters.
1999-05-11 05:10:06 +00:00
if (xpt_bus_register(sim, primary) != CAM_SUCCESS) {
cam_sim_free(sim, TRUE);
Things have become cinched down more tightly about assertions for Giant. This uncovered some missing spots where I trade off between isp's lock and Giant as I enter CAM.
2001-07-05 17:14:57 +00:00
CAMLOCK_2_ISPLOCK(isp);
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
return;
}
A large set of changes to handle dual bus adapters.
1999-05-11 05:10:06 +00:00
if (xpt_create_path(&path, NULL, cam_sim_path(sim),
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
CAM_TARGET_WILDCARD, CAM_LUN_WILDCARD) != CAM_REQ_CMP) {
A large set of changes to handle dual bus adapters.
1999-05-11 05:10:06 +00:00
xpt_bus_deregister(cam_sim_path(sim));
cam_sim_free(sim, TRUE);
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
config_intrhook_disestablish(&isp->isp_osinfo.ehook);
Things have become cinched down more tightly about assertions for Giant. This uncovered some missing spots where I trade off between isp's lock and Giant as I enter CAM.
2001-07-05 17:14:57 +00:00
CAMLOCK_2_ISPLOCK(isp);
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
return;
}
A large set of changes to handle dual bus adapters.
1999-05-11 05:10:06 +00:00
xpt_setup_ccb(&csa.ccb_h, path, 5);
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
csa.ccb_h.func_code = XPT_SASYNC_CB;
csa.event_enable = AC_LOST_DEVICE;
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 07:29:00 +00:00
csa.callback = isp_cam_async;
A large set of changes to handle dual bus adapters.
1999-05-11 05:10:06 +00:00
csa.callback_arg = sim;
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
xpt_action((union ccb *)&csa);
Things have become cinched down more tightly about assertions for Giant. This uncovered some missing spots where I trade off between isp's lock and Giant as I enter CAM.
2001-07-05 17:14:57 +00:00
CAMLOCK_2_ISPLOCK(isp);
A large set of changes to handle dual bus adapters.
1999-05-11 05:10:06 +00:00
isp->isp_sim = sim;
isp->isp_path = path;
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
/*
* Create a kernel thread for fibre channel instances. We
* don't have dual channel FC cards.
*/
if (IS_FC(isp)) {
Things have become cinched down more tightly about assertions for Giant. This uncovered some missing spots where I trade off between isp's lock and Giant as I enter CAM.
2001-07-05 17:14:57 +00:00
ISPLOCK_2_CAMLOCK(isp);
/* XXX: LOCK VIOLATION */
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
cv_init(&isp->isp_osinfo.kthread_cv, "isp_kthread_cv");
if (kthread_create(isp_kthread, isp, &isp->isp_osinfo.kproc,
Some kernel threads try to do significant work, and the default KSTACK_PAGES doesn't give them enough stack to do much before blowing away the pcb. This adds MI and MD code to allow the allocation of an alternate kstack who's size can be speficied when calling kthread_create. Passing the value 0 prevents the alternate kstack from being created. Note that the ia64 MD code is missing for now, and PowerPC was only partially written due to the pmap.c being incomplete there. Though this patch does not modify anything to make use of the alternate kstack, acpi and usb are good candidates. Reviewed by: jake, peter, jhb
2002-10-02 07:44:29 +00:00
RFHIGHPID, 0, "%s: fc_thrd",
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
device_get_nameunit(isp->isp_dev))) {
xpt_bus_deregister(cam_sim_path(sim));
cam_sim_free(sim, TRUE);
config_intrhook_disestablish(&isp->isp_osinfo.ehook);
Things have become cinched down more tightly about assertions for Giant. This uncovered some missing spots where I trade off between isp's lock and Giant as I enter CAM.
2001-07-05 17:14:57 +00:00
CAMLOCK_2_ISPLOCK(isp);
isp_prt(isp, ISP_LOGERR, "could not create kthread");
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
return;
}
Oops- missed a CAMLOCK_2_ISP case.
2001-07-05 19:34:06 +00:00
CAMLOCK_2_ISPLOCK(isp);
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
}
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
A large set of changes to handle dual bus adapters.
1999-05-11 05:10:06 +00:00
/*
* If we have a second channel, construct SIM entry for that.
*/
Add Dual LVD bus (1280) support
1999-12-16 05:42:02 +00:00
if (IS_DUALBUS(isp)) {
Things have become cinched down more tightly about assertions for Giant. This uncovered some missing spots where I trade off between isp's lock and Giant as I enter CAM.
2001-07-05 17:14:57 +00:00
ISPLOCK_2_CAMLOCK(isp);
A large set of changes to handle dual bus adapters.
1999-05-11 05:10:06 +00:00
sim = cam_sim_alloc(isp_action, isp_poll, "isp", isp,
Finally eliminate as many of the printf calls as possible (still leaving ones where we have a CAM path) and replacing them with calls to isp_prt., Eliminate isp_unit references- we no longer have an isp_unit- we now have an isp_dev that device_get_unit can work with.
2001-03-01 02:14:54 +00:00
device_get_unit(isp->isp_dev), 1, isp->isp_maxcmds, devq);
A large set of changes to handle dual bus adapters.
1999-05-11 05:10:06 +00:00
if (sim == NULL) {
xpt_bus_deregister(cam_sim_path(isp->isp_sim));
xpt_free_path(isp->isp_path);
cam_simq_free(devq);
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
config_intrhook_disestablish(&isp->isp_osinfo.ehook);
A large set of changes to handle dual bus adapters.
1999-05-11 05:10:06 +00:00
return;
}
if (xpt_bus_register(sim, secondary) != CAM_SUCCESS) {
xpt_bus_deregister(cam_sim_path(isp->isp_sim));
xpt_free_path(isp->isp_path);
cam_sim_free(sim, TRUE);
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
config_intrhook_disestablish(&isp->isp_osinfo.ehook);
Things have become cinched down more tightly about assertions for Giant. This uncovered some missing spots where I trade off between isp's lock and Giant as I enter CAM.
2001-07-05 17:14:57 +00:00
CAMLOCK_2_ISPLOCK(isp);
A large set of changes to handle dual bus adapters.
1999-05-11 05:10:06 +00:00
return;
}
if (xpt_create_path(&path, NULL, cam_sim_path(sim),
CAM_TARGET_WILDCARD, CAM_LUN_WILDCARD) != CAM_REQ_CMP) {
xpt_bus_deregister(cam_sim_path(isp->isp_sim));
xpt_free_path(isp->isp_path);
xpt_bus_deregister(cam_sim_path(sim));
cam_sim_free(sim, TRUE);
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
config_intrhook_disestablish(&isp->isp_osinfo.ehook);
Things have become cinched down more tightly about assertions for Giant. This uncovered some missing spots where I trade off between isp's lock and Giant as I enter CAM.
2001-07-05 17:14:57 +00:00
CAMLOCK_2_ISPLOCK(isp);
A large set of changes to handle dual bus adapters.
1999-05-11 05:10:06 +00:00
return;
}
xpt_setup_ccb(&csa.ccb_h, path, 5);
csa.ccb_h.func_code = XPT_SASYNC_CB;
csa.event_enable = AC_LOST_DEVICE;
csa.callback = isp_cam_async;
csa.callback_arg = sim;
xpt_action((union ccb *)&csa);
Things have become cinched down more tightly about assertions for Giant. This uncovered some missing spots where I trade off between isp's lock and Giant as I enter CAM.
2001-07-05 17:14:57 +00:00
CAMLOCK_2_ISPLOCK(isp);
A large set of changes to handle dual bus adapters.
1999-05-11 05:10:06 +00:00
isp->isp_sim2 = sim;
isp->isp_path2 = path;
}
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
/*
* Create device nodes
*/
(void) make_dev(&isp_cdevsw, device_get_unit(isp->isp_dev), UID_ROOT,
GID_OPERATOR, 0600, "%s", device_get_nameunit(isp->isp_dev));
Do some cleanup based upon adapter role- mainly not enabling interrupts if we're ISP_ROLE_NONE. Change ISPASYNC_LOGGED_INOUT to ISPASYNC_PROMENADE. Make sure we note if something is a fabric device. Target mode: Finally fix (to a first approximation) SCSI Target Mode again- we needed to correctly check against CAM_TARGET_WILDCARD and CAM_LUN_WILDCARD so that targbh won't confuse us. Comment out the drainqueue stuff for now. Use isp_fc_runstate instead if isp_control/ISPCTL_FCLINK_TEST.
2001-02-11 03:47:39 +00:00
if (isp->isp_role != ISP_ROLE_NONE) {
isp->isp_state = ISP_RUNSTATE;
ENABLE_INTS(isp);
}
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
if (isplist == NULL) {
isplist = isp;
} else {
struct ispsoftc *tmp = isplist;
while (tmp->isp_osinfo.next) {
tmp = tmp->isp_osinfo.next;
}
tmp->isp_osinfo.next = isp;
}
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
}
Remove use of inlines and use the functions as a library. Larger code space, possibly performance hit, but more portable. Certainly less questionable use of inlining. Suggested by: des
2006-02-02 21:31:34 +00:00
static __inline void
Add get/set param ioctl support. Remove sim queue freezes for resource shortages. I've had too many strange race conditions where I freeze on a resource shortage but never get unfrozen. Consolidate the remaining sim queue freeze condition (for loopdown) into an inline with debug messages that allows us to track problems at ISP_LOGDEBUG0 level easier. Change a bunch of debug messages about loop down/up conditions to ISP_LOGDEBUG0 level. Remove dead isp_relsim code. Change some internal flag stuff for efficiency. Complain vociferously if we try and use our FC scratch area while it's busy being used already (I mean, if we don't have solaris' ability to sleep as an interrupt thread which would allow us to just use a p/v semaphore, at least *say* when you've just borked yourself). Add infrastructure to allow overrides of hard loopid && initiator id from boot variables. Fix the usual quota of silly bugs: + 'ktmature' needs to be per-instance. Argh. + When entering isp_watchdog, set intsok to zero, preserving old value to restore later. It's not nice to try and sleep from splsoftclock. + Fix tick overflow buglet in checking timeout value. MFC after: 1 week
2002-07-08 17:42:47 +00:00
isp_freeze_loopdown(struct ispsoftc *isp, char *msg)
{
if (isp->isp_osinfo.simqfrozen == 0) {
isp_prt(isp, ISP_LOGDEBUG0, "%s: freeze simq (loopdown)", msg);
isp->isp_osinfo.simqfrozen |= SIMQFRZ_LOOPDOWN;
ISPLOCK_2_CAMLOCK(isp);
xpt_freeze_simq(isp->isp_sim, 1);
CAMLOCK_2_ISPLOCK(isp);
} else {
isp_prt(isp, ISP_LOGDEBUG0, "%s: mark frozen (loopdown)", msg);
isp->isp_osinfo.simqfrozen |= SIMQFRZ_LOOPDOWN;
}
}
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
static int
Do the dreaded s/dev_t/struct cdev */ Bump __FreeBSD_version accordingly.
2004-06-16 09:47:26 +00:00
ispioctl(struct cdev *dev, u_long cmd, caddr_t addr, int flags, struct thread *td)
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
{
struct ispsoftc *isp;
Checkpoint of work in progress in cleaning up target mode. It actually seems to work well in RELENG_4. However, 5.X locking foo means that I'll have to do some quick redesign. Add ioctl handlers for ISP_GETROLE and ISP_SETROLE ioctls.
2004-02-07 03:47:33 +00:00
int nr, retval = ENOTTY;
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
isp = isplist;
while (isp) {
if (minor(dev) == device_get_unit(isp->isp_dev)) {
break;
}
isp = isp->isp_osinfo.next;
}
if (isp == NULL)
return (ENXIO);
switch (cmd) {
More for f/w crash dumps (bug fixing and adding ioctl entry points and hints to enable for specific units) MFC after: 1 week
2002-02-18 00:00:34 +00:00
#ifdef ISP_FW_CRASH_DUMP
case ISP_GET_FW_CRASH_DUMP:
{
a) clean up some declaration stuff (i.e., make more modern with respect to getting rid u_int for uint and so on). b) Turn back on 64 bit DAC support. Cheeze it a bit in that we have two DMA callback functions- one when we have bus_addr_t > 4 bits in width and the other which should be normal. Even Cheezier in that we turn off setting up DMA maps to be BUS_SPACE_MAXADDR if we're in ISP_TARGET_MODE. More work on this in a week or so. c) Tested under amd64 and 1MB DFLTPHYS, sparc64, i386 (PAE, but insufficient memory to really test > 4GB). LINT check under amd64. MFC after: 1 month
2006-02-15 00:31:48 +00:00
uint16_t *ptr = FCPARAM(isp)->isp_dump_data;
More for f/w crash dumps (bug fixing and adding ioctl entry points and hints to enable for specific units) MFC after: 1 week
2002-02-18 00:00:34 +00:00
size_t sz;
retval = 0;
if (IS_2200(isp))
sz = QLA2200_RISC_IMAGE_DUMP_SIZE;
else
sz = QLA2300_RISC_IMAGE_DUMP_SIZE;
ISP_LOCK(isp);
if (ptr && *ptr) {
void *uaddr = *((void **) addr);
if (copyout(ptr, uaddr, sz)) {
retval = EFAULT;
} else {
*ptr = 0;
}
} else {
retval = ENXIO;
}
ISP_UNLOCK(isp);
break;
}
case ISP_FORCE_CRASH_DUMP:
ISP_LOCK(isp);
Add get/set param ioctl support. Remove sim queue freezes for resource shortages. I've had too many strange race conditions where I freeze on a resource shortage but never get unfrozen. Consolidate the remaining sim queue freeze condition (for loopdown) into an inline with debug messages that allows us to track problems at ISP_LOGDEBUG0 level easier. Change a bunch of debug messages about loop down/up conditions to ISP_LOGDEBUG0 level. Remove dead isp_relsim code. Change some internal flag stuff for efficiency. Complain vociferously if we try and use our FC scratch area while it's busy being used already (I mean, if we don't have solaris' ability to sleep as an interrupt thread which would allow us to just use a p/v semaphore, at least *say* when you've just borked yourself). Add infrastructure to allow overrides of hard loopid && initiator id from boot variables. Fix the usual quota of silly bugs: + 'ktmature' needs to be per-instance. Argh. + When entering isp_watchdog, set intsok to zero, preserving old value to restore later. It's not nice to try and sleep from splsoftclock. + Fix tick overflow buglet in checking timeout value. MFC after: 1 week
2002-07-08 17:42:47 +00:00
isp_freeze_loopdown(isp, "ispioctl(ISP_FORCE_CRASH_DUMP)");
More for f/w crash dumps (bug fixing and adding ioctl entry points and hints to enable for specific units) MFC after: 1 week
2002-02-18 00:00:34 +00:00
isp_fw_dump(isp);
isp_reinit(isp);
ISP_UNLOCK(isp);
retval = 0;
break;
#endif
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
case ISP_SDBLEV:
{
int olddblev = isp->isp_dblev;
isp->isp_dblev = *(int *)addr;
*(int *)addr = olddblev;
retval = 0;
break;
}
Checkpoint of work in progress in cleaning up target mode. It actually seems to work well in RELENG_4. However, 5.X locking foo means that I'll have to do some quick redesign. Add ioctl handlers for ISP_GETROLE and ISP_SETROLE ioctls.
2004-02-07 03:47:33 +00:00
case ISP_GETROLE:
*(int *)addr = isp->isp_role;
retval = 0;
break;
case ISP_SETROLE:
nr = *(int *)addr;
if (nr & ~(ISP_ROLE_INITIATOR|ISP_ROLE_TARGET)) {
retval = EINVAL;
break;
}
*(int *)addr = isp->isp_role;
isp->isp_role = nr;
/* FALLTHROUGH */
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
case ISP_RESETHBA:
ISP_LOCK(isp);
isp_reinit(isp);
ISP_UNLOCK(isp);
retval = 0;
break;
Reorder some of the ioctls and add a few new ones. MFC after: 1 day
2002-02-21 23:30:05 +00:00
case ISP_RESCAN:
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
if (IS_FC(isp)) {
ISP_LOCK(isp);
if (isp_fc_runstate(isp, 5 * 1000000)) {
retval = EIO;
} else {
retval = 0;
}
ISP_UNLOCK(isp);
}
break;
case ISP_FC_LIP:
if (IS_FC(isp)) {
ISP_LOCK(isp);
if (isp_control(isp, ISPCTL_SEND_LIP, 0)) {
retval = EIO;
} else {
retval = 0;
}
ISP_UNLOCK(isp);
}
break;
case ISP_FC_GETDINFO:
{
struct isp_fc_device *ifc = (struct isp_fc_device *) addr;
struct lportdb *lp;
if (ifc->loopid < 0 || ifc->loopid >= MAX_FC_TARG) {
retval = EINVAL;
break;
}
ISP_LOCK(isp);
lp = &FCPARAM(isp)->portdb[ifc->loopid];
if (lp->valid) {
propagate role of device for ISP_GET_PDBINFO
2006-02-26 22:40:56 +00:00
ifc->role = lp->roles;
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
ifc->loopid = lp->loopid;
ifc->portid = lp->portid;
ifc->node_wwn = lp->node_wwn;
ifc->port_wwn = lp->port_wwn;
retval = 0;
} else {
retval = ENODEV;
}
ISP_UNLOCK(isp);
break;
}
Implement REDUCED INTERRUPT OPERATION usage form 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. MFC after: 2 weeks
2002-01-03 20:43:22 +00:00
case ISP_GET_STATS:
{
isp_stats_t *sp = (isp_stats_t *) addr;
MEMZERO(sp, sizeof (*sp));
sp->isp_stat_version = ISP_STATS_VERSION;
sp->isp_type = isp->isp_type;
sp->isp_revision = isp->isp_revision;
ISP_LOCK(isp);
sp->isp_stats[ISP_INTCNT] = isp->isp_intcnt;
sp->isp_stats[ISP_INTBOGUS] = isp->isp_intbogus;
sp->isp_stats[ISP_INTMBOXC] = isp->isp_intmboxc;
sp->isp_stats[ISP_INGOASYNC] = isp->isp_intoasync;
sp->isp_stats[ISP_RSLTCCMPLT] = isp->isp_rsltccmplt;
sp->isp_stats[ISP_FPHCCMCPLT] = isp->isp_fphccmplt;
sp->isp_stats[ISP_RSCCHIWAT] = isp->isp_rscchiwater;
sp->isp_stats[ISP_FPCCHIWAT] = isp->isp_fpcchiwater;
ISP_UNLOCK(isp);
retval = 0;
break;
}
case ISP_CLR_STATS:
ISP_LOCK(isp);
isp->isp_intcnt = 0;
isp->isp_intbogus = 0;
isp->isp_intmboxc = 0;
isp->isp_intoasync = 0;
isp->isp_rsltccmplt = 0;
isp->isp_fphccmplt = 0;
isp->isp_rscchiwater = 0;
isp->isp_fpcchiwater = 0;
ISP_UNLOCK(isp);
retval = 0;
break;
Add support for ISP_FC_GETHINFO, which returns current connection topology, speed, loopid, WWPN/WWNN, etc. Beef up target mode. Add isp_handle_platform_notify_scsi and isp_handle_platform_notify_fc platform handlers to handle immediate notifies (isp_handle_platform_notify_scsi is still stubbed out). In implementation of isp_handle_platform_notify_fc, for IN_ABORT_TASK, peel off a pending XPT_IMMED_NOTIFY and call xpt_done on it and hope that somebody upstream is listening. Make sure on final CTIO2s that we set residual correctly. These are absolutely crucial. Make sure we set relative offset for each CTIO2 based upon bytes we've already xferred. This is what the private adjunct datat to the original ATIO is. Note state of command so we can figure out where to find it if we get an ABORT from the firmware. Make sure we *always* set CAM_TAG_ACTION_VALID for ATIO2s. Make sure we keep track of the original lun. If se sent status (or we're otherwise done with the command), don't forget to free the adjunct structure.
2002-06-16 05:08:02 +00:00
case ISP_FC_GETHINFO:
{
struct isp_hba_device *hba = (struct isp_hba_device *) addr;
MEMZERO(hba, sizeof (*hba));
ISP_LOCK(isp);
Add firmware major, minor and micro revsions to the ISP_FC_GETHINFO ioctl structure.
2004-01-23 23:22:11 +00:00
hba->fc_fw_major = ISP_FW_MAJORX(isp->isp_fwrev);
hba->fc_fw_minor = ISP_FW_MINORX(isp->isp_fwrev);
hba->fc_fw_micro = ISP_FW_MICROX(isp->isp_fwrev);
Add support for ISP_FC_GETHINFO, which returns current connection topology, speed, loopid, WWPN/WWNN, etc. Beef up target mode. Add isp_handle_platform_notify_scsi and isp_handle_platform_notify_fc platform handlers to handle immediate notifies (isp_handle_platform_notify_scsi is still stubbed out). In implementation of isp_handle_platform_notify_fc, for IN_ABORT_TASK, peel off a pending XPT_IMMED_NOTIFY and call xpt_done on it and hope that somebody upstream is listening. Make sure on final CTIO2s that we set residual correctly. These are absolutely crucial. Make sure we set relative offset for each CTIO2 based upon bytes we've already xferred. This is what the private adjunct datat to the original ATIO is. Note state of command so we can figure out where to find it if we get an ABORT from the firmware. Make sure we *always* set CAM_TAG_ACTION_VALID for ATIO2s. Make sure we keep track of the original lun. If se sent status (or we're otherwise done with the command), don't forget to free the adjunct structure.
2002-06-16 05:08:02 +00:00
hba->fc_speed = FCPARAM(isp)->isp_gbspeed;
hba->fc_scsi_supported = 1;
hba->fc_topology = FCPARAM(isp)->isp_topo + 1;
hba->fc_loopid = FCPARAM(isp)->isp_loopid;
Report correct active vs. nvram node/port WWNs in an ioctl.
2003-09-13 01:56:24 +00:00
hba->nvram_node_wwn = FCPARAM(isp)->isp_nodewwn;
hba->nvram_port_wwn = FCPARAM(isp)->isp_portwwn;
hba->active_node_wwn = ISP_NODEWWN(isp);
hba->active_port_wwn = ISP_PORTWWN(isp);
Add support for ISP_FC_GETHINFO, which returns current connection topology, speed, loopid, WWPN/WWNN, etc. Beef up target mode. Add isp_handle_platform_notify_scsi and isp_handle_platform_notify_fc platform handlers to handle immediate notifies (isp_handle_platform_notify_scsi is still stubbed out). In implementation of isp_handle_platform_notify_fc, for IN_ABORT_TASK, peel off a pending XPT_IMMED_NOTIFY and call xpt_done on it and hope that somebody upstream is listening. Make sure on final CTIO2s that we set residual correctly. These are absolutely crucial. Make sure we set relative offset for each CTIO2 based upon bytes we've already xferred. This is what the private adjunct datat to the original ATIO is. Note state of command so we can figure out where to find it if we get an ABORT from the firmware. Make sure we *always* set CAM_TAG_ACTION_VALID for ATIO2s. Make sure we keep track of the original lun. If se sent status (or we're otherwise done with the command), don't forget to free the adjunct structure.
2002-06-16 05:08:02 +00:00
ISP_UNLOCK(isp);
retval = 0;
break;
}
Add get/set param ioctl support. Remove sim queue freezes for resource shortages. I've had too many strange race conditions where I freeze on a resource shortage but never get unfrozen. Consolidate the remaining sim queue freeze condition (for loopdown) into an inline with debug messages that allows us to track problems at ISP_LOGDEBUG0 level easier. Change a bunch of debug messages about loop down/up conditions to ISP_LOGDEBUG0 level. Remove dead isp_relsim code. Change some internal flag stuff for efficiency. Complain vociferously if we try and use our FC scratch area while it's busy being used already (I mean, if we don't have solaris' ability to sleep as an interrupt thread which would allow us to just use a p/v semaphore, at least *say* when you've just borked yourself). Add infrastructure to allow overrides of hard loopid && initiator id from boot variables. Fix the usual quota of silly bugs: + 'ktmature' needs to be per-instance. Argh. + When entering isp_watchdog, set intsok to zero, preserving old value to restore later. It's not nice to try and sleep from splsoftclock. + Fix tick overflow buglet in checking timeout value. MFC after: 1 week
2002-07-08 17:42:47 +00:00
case ISP_GET_FC_PARAM:
{
struct isp_fc_param *f = (struct isp_fc_param *) addr;
if (!IS_FC(isp)) {
retval = EINVAL;
break;
}
f->parameter = 0;
if (strcmp(f->param_name, "framelength") == 0) {
f->parameter = FCPARAM(isp)->isp_maxfrmlen;
retval = 0;
break;
}
if (strcmp(f->param_name, "exec_throttle") == 0) {
f->parameter = FCPARAM(isp)->isp_execthrottle;
retval = 0;
break;
}
if (strcmp(f->param_name, "fullduplex") == 0) {
if (FCPARAM(isp)->isp_fwoptions & ICBOPT_FULL_DUPLEX)
f->parameter = 1;
retval = 0;
break;
}
if (strcmp(f->param_name, "loopid") == 0) {
f->parameter = FCPARAM(isp)->isp_loopid;
retval = 0;
break;
}
retval = EINVAL;
break;
}
case ISP_SET_FC_PARAM:
{
struct isp_fc_param *f = (struct isp_fc_param *) addr;
a) clean up some declaration stuff (i.e., make more modern with respect to getting rid u_int for uint and so on). b) Turn back on 64 bit DAC support. Cheeze it a bit in that we have two DMA callback functions- one when we have bus_addr_t > 4 bits in width and the other which should be normal. Even Cheezier in that we turn off setting up DMA maps to be BUS_SPACE_MAXADDR if we're in ISP_TARGET_MODE. More work on this in a week or so. c) Tested under amd64 and 1MB DFLTPHYS, sparc64, i386 (PAE, but insufficient memory to really test > 4GB). LINT check under amd64. MFC after: 1 month
2006-02-15 00:31:48 +00:00
uint32_t param = f->parameter;
Add get/set param ioctl support. Remove sim queue freezes for resource shortages. I've had too many strange race conditions where I freeze on a resource shortage but never get unfrozen. Consolidate the remaining sim queue freeze condition (for loopdown) into an inline with debug messages that allows us to track problems at ISP_LOGDEBUG0 level easier. Change a bunch of debug messages about loop down/up conditions to ISP_LOGDEBUG0 level. Remove dead isp_relsim code. Change some internal flag stuff for efficiency. Complain vociferously if we try and use our FC scratch area while it's busy being used already (I mean, if we don't have solaris' ability to sleep as an interrupt thread which would allow us to just use a p/v semaphore, at least *say* when you've just borked yourself). Add infrastructure to allow overrides of hard loopid && initiator id from boot variables. Fix the usual quota of silly bugs: + 'ktmature' needs to be per-instance. Argh. + When entering isp_watchdog, set intsok to zero, preserving old value to restore later. It's not nice to try and sleep from splsoftclock. + Fix tick overflow buglet in checking timeout value. MFC after: 1 week
2002-07-08 17:42:47 +00:00
if (!IS_FC(isp)) {
retval = EINVAL;
break;
}
f->parameter = 0;
if (strcmp(f->param_name, "framelength") == 0) {
if (param != 512 && param != 1024 && param != 1024) {
retval = EINVAL;
break;
}
FCPARAM(isp)->isp_maxfrmlen = param;
retval = 0;
break;
}
if (strcmp(f->param_name, "exec_throttle") == 0) {
if (param < 16 || param > 255) {
retval = EINVAL;
break;
}
FCPARAM(isp)->isp_execthrottle = param;
retval = 0;
break;
}
if (strcmp(f->param_name, "fullduplex") == 0) {
if (param != 0 && param != 1) {
retval = EINVAL;
break;
}
if (param) {
FCPARAM(isp)->isp_fwoptions |=
ICBOPT_FULL_DUPLEX;
} else {
FCPARAM(isp)->isp_fwoptions &=
~ICBOPT_FULL_DUPLEX;
}
retval = 0;
break;
}
if (strcmp(f->param_name, "loopid") == 0) {
if (param < 0 || param > 125) {
retval = EINVAL;
break;
}
FCPARAM(isp)->isp_loopid = param;
retval = 0;
break;
}
retval = EINVAL;
break;
}
Add an ioctl framework for doing FC task management functions from a user space tool- useful for doing FC target mode certification.
2005-10-29 02:46:59 +00:00
case ISP_TSK_MGMT:
{
int needmarker;
struct isp_fc_tsk_mgmt *fct = (struct isp_fc_tsk_mgmt *) addr;
a) clean up some declaration stuff (i.e., make more modern with respect to getting rid u_int for uint and so on). b) Turn back on 64 bit DAC support. Cheeze it a bit in that we have two DMA callback functions- one when we have bus_addr_t > 4 bits in width and the other which should be normal. Even Cheezier in that we turn off setting up DMA maps to be BUS_SPACE_MAXADDR if we're in ISP_TARGET_MODE. More work on this in a week or so. c) Tested under amd64 and 1MB DFLTPHYS, sparc64, i386 (PAE, but insufficient memory to really test > 4GB). LINT check under amd64. MFC after: 1 month
2006-02-15 00:31:48 +00:00
uint16_t loopid;
Add an ioctl framework for doing FC task management functions from a user space tool- useful for doing FC target mode certification.
2005-10-29 02:46:59 +00:00
mbreg_t mbs;
if (IS_SCSI(isp)) {
retval = EINVAL;
break;
}
memset(&mbs, 0, sizeof (mbs));
needmarker = retval = 0;
First of several commits as this driver is dusted off and maybe brought up to date. Principle changes for this reelase is to support 2K Port Login firmware. This allows us to support the 2322 (and 2422 4Gb) cards which only come with the 2K Port Login firmware. The 2322 should now work- but we don't have firmware sets for it in ispfw (as the change to load 2K Port Login f/w hasn't been made- that f/w is so big it has to be loaded in more than one chunk). Other changes are the beginnings of cleaning up some long standing target mode issues. The next changes here will incorporate a lot of bug fixes from others. Finally, some copyright cleanup and attempts to make the parts of the driver that are FreeBSD specific start conforming more to FreeBSD style. MFC after: 1 month
2006-01-23 06:23:37 +00:00
loopid = fct->loopid;
if (IS_2KLOGIN(isp) == 0) {
loopid <<= 8;
}
Add an ioctl framework for doing FC task management functions from a user space tool- useful for doing FC target mode certification.
2005-10-29 02:46:59 +00:00
switch (fct->action) {
case CLEAR_ACA:
mbs.param[0] = MBOX_CLEAR_ACA;
First of several commits as this driver is dusted off and maybe brought up to date. Principle changes for this reelase is to support 2K Port Login firmware. This allows us to support the 2322 (and 2422 4Gb) cards which only come with the 2K Port Login firmware. The 2322 should now work- but we don't have firmware sets for it in ispfw (as the change to load 2K Port Login f/w hasn't been made- that f/w is so big it has to be loaded in more than one chunk). Other changes are the beginnings of cleaning up some long standing target mode issues. The next changes here will incorporate a lot of bug fixes from others. Finally, some copyright cleanup and attempts to make the parts of the driver that are FreeBSD specific start conforming more to FreeBSD style. MFC after: 1 month
2006-01-23 06:23:37 +00:00
mbs.param[1] = loopid;
Add an ioctl framework for doing FC task management functions from a user space tool- useful for doing FC target mode certification.
2005-10-29 02:46:59 +00:00
mbs.param[2] = fct->lun;
break;
case TARGET_RESET:
mbs.param[0] = MBOX_TARGET_RESET;
First of several commits as this driver is dusted off and maybe brought up to date. Principle changes for this reelase is to support 2K Port Login firmware. This allows us to support the 2322 (and 2422 4Gb) cards which only come with the 2K Port Login firmware. The 2322 should now work- but we don't have firmware sets for it in ispfw (as the change to load 2K Port Login f/w hasn't been made- that f/w is so big it has to be loaded in more than one chunk). Other changes are the beginnings of cleaning up some long standing target mode issues. The next changes here will incorporate a lot of bug fixes from others. Finally, some copyright cleanup and attempts to make the parts of the driver that are FreeBSD specific start conforming more to FreeBSD style. MFC after: 1 month
2006-01-23 06:23:37 +00:00
mbs.param[1] = loopid;
Add an ioctl framework for doing FC task management functions from a user space tool- useful for doing FC target mode certification.
2005-10-29 02:46:59 +00:00
needmarker = 1;
break;
case LUN_RESET:
mbs.param[0] = MBOX_LUN_RESET;
First of several commits as this driver is dusted off and maybe brought up to date. Principle changes for this reelase is to support 2K Port Login firmware. This allows us to support the 2322 (and 2422 4Gb) cards which only come with the 2K Port Login firmware. The 2322 should now work- but we don't have firmware sets for it in ispfw (as the change to load 2K Port Login f/w hasn't been made- that f/w is so big it has to be loaded in more than one chunk). Other changes are the beginnings of cleaning up some long standing target mode issues. The next changes here will incorporate a lot of bug fixes from others. Finally, some copyright cleanup and attempts to make the parts of the driver that are FreeBSD specific start conforming more to FreeBSD style. MFC after: 1 month
2006-01-23 06:23:37 +00:00
mbs.param[1] = loopid;
Add an ioctl framework for doing FC task management functions from a user space tool- useful for doing FC target mode certification.
2005-10-29 02:46:59 +00:00
mbs.param[2] = fct->lun;
needmarker = 1;
break;
case CLEAR_TASK_SET:
mbs.param[0] = MBOX_CLEAR_TASK_SET;
First of several commits as this driver is dusted off and maybe brought up to date. Principle changes for this reelase is to support 2K Port Login firmware. This allows us to support the 2322 (and 2422 4Gb) cards which only come with the 2K Port Login firmware. The 2322 should now work- but we don't have firmware sets for it in ispfw (as the change to load 2K Port Login f/w hasn't been made- that f/w is so big it has to be loaded in more than one chunk). Other changes are the beginnings of cleaning up some long standing target mode issues. The next changes here will incorporate a lot of bug fixes from others. Finally, some copyright cleanup and attempts to make the parts of the driver that are FreeBSD specific start conforming more to FreeBSD style. MFC after: 1 month
2006-01-23 06:23:37 +00:00
mbs.param[1] = loopid;
Add an ioctl framework for doing FC task management functions from a user space tool- useful for doing FC target mode certification.
2005-10-29 02:46:59 +00:00
mbs.param[2] = fct->lun;
needmarker = 1;
break;
case ABORT_TASK_SET:
mbs.param[0] = MBOX_ABORT_TASK_SET;
First of several commits as this driver is dusted off and maybe brought up to date. Principle changes for this reelase is to support 2K Port Login firmware. This allows us to support the 2322 (and 2422 4Gb) cards which only come with the 2K Port Login firmware. The 2322 should now work- but we don't have firmware sets for it in ispfw (as the change to load 2K Port Login f/w hasn't been made- that f/w is so big it has to be loaded in more than one chunk). Other changes are the beginnings of cleaning up some long standing target mode issues. The next changes here will incorporate a lot of bug fixes from others. Finally, some copyright cleanup and attempts to make the parts of the driver that are FreeBSD specific start conforming more to FreeBSD style. MFC after: 1 month
2006-01-23 06:23:37 +00:00
mbs.param[1] = loopid;
Add an ioctl framework for doing FC task management functions from a user space tool- useful for doing FC target mode certification.
2005-10-29 02:46:59 +00:00
mbs.param[2] = fct->lun;
needmarker = 1;
break;
default:
retval = EINVAL;
break;
}
if (retval == 0) {
ISP_LOCK(isp);
if (needmarker) {
isp->isp_sendmarker |= 1;
}
retval = isp_control(isp, ISPCTL_RUN_MBOXCMD, &mbs);
ISP_UNLOCK(isp);
if (retval)
retval = EIO;
}
break;
}
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
default:
break;
}
return (retval);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
}
Add in config_hook for catching when interrupts are safe- this allows us to not the ints are ok and also to (re)ENABLE isp interrupts. Remove all splcam()/splx() invocates and replace them with ISP_LOCK/ISP_UNLOCK macros.
2000-07-04 01:05:43 +00:00
static void
isp_intr_enable(void *arg)
{
struct ispsoftc *isp = arg;
Do some cleanup based upon adapter role- mainly not enabling interrupts if we're ISP_ROLE_NONE. Change ISPASYNC_LOGGED_INOUT to ISPASYNC_PROMENADE. Make sure we note if something is a fabric device. Target mode: Finally fix (to a first approximation) SCSI Target Mode again- we needed to correctly check against CAM_TARGET_WILDCARD and CAM_LUN_WILDCARD so that targbh won't confuse us. Comment out the drainqueue stuff for now. Use isp_fc_runstate instead if isp_control/ISPCTL_FCLINK_TEST.
2001-02-11 03:47:39 +00:00
if (isp->isp_role != ISP_ROLE_NONE) {
ENABLE_INTS(isp);
Until I can get a clearer architecture from PHK about why he wants the geometry code to grab a mutex that prohibits any driver on the stack below it from sleeping, it's not safe to allow anything in the top half of isp to sleep (excepting the thread that Fibre Channel instances use to re-scan loops/fabrics).
2004-08-23 19:04:19 +00:00
#if 0
Do some cleanup based upon adapter role- mainly not enabling interrupts if we're ISP_ROLE_NONE. Change ISPASYNC_LOGGED_INOUT to ISPASYNC_PROMENADE. Make sure we note if something is a fabric device. Target mode: Finally fix (to a first approximation) SCSI Target Mode again- we needed to correctly check against CAM_TARGET_WILDCARD and CAM_LUN_WILDCARD so that targbh won't confuse us. Comment out the drainqueue stuff for now. Use isp_fc_runstate instead if isp_control/ISPCTL_FCLINK_TEST.
2001-02-11 03:47:39 +00:00
isp->isp_osinfo.intsok = 1;
Until I can get a clearer architecture from PHK about why he wants the geometry code to grab a mutex that prohibits any driver on the stack below it from sleeping, it's not safe to allow anything in the top half of isp to sleep (excepting the thread that Fibre Channel instances use to re-scan loops/fabrics).
2004-08-23 19:04:19 +00:00
#endif
Do some cleanup based upon adapter role- mainly not enabling interrupts if we're ISP_ROLE_NONE. Change ISPASYNC_LOGGED_INOUT to ISPASYNC_PROMENADE. Make sure we note if something is a fabric device. Target mode: Finally fix (to a first approximation) SCSI Target Mode again- we needed to correctly check against CAM_TARGET_WILDCARD and CAM_LUN_WILDCARD so that targbh won't confuse us. Comment out the drainqueue stuff for now. Use isp_fc_runstate instead if isp_control/ISPCTL_FCLINK_TEST.
2001-02-11 03:47:39 +00:00
}
Add in config_hook for catching when interrupts are safe- this allows us to not the ints are ok and also to (re)ENABLE isp interrupts. Remove all splcam()/splx() invocates and replace them with ISP_LOCK/ISP_UNLOCK macros.
2000-07-04 01:05:43 +00:00
/* Release our hook so that the boot can continue. */
config_intrhook_disestablish(&isp->isp_osinfo.ehook);
}
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
/*
* Put the target mode functions here, because some are inlines
*/
#ifdef ISP_TARGET_MODE
Remove use of inlines and use the functions as a library. Larger code space, possibly performance hit, but more portable. Certainly less questionable use of inlining. Suggested by: des
2006-02-02 21:31:34 +00:00
static __inline int is_lun_enabled(struct ispsoftc *, int, lun_id_t);
static __inline int are_any_luns_enabled(struct ispsoftc *, int);
static __inline tstate_t *get_lun_statep(struct ispsoftc *, int, lun_id_t);
static __inline void rls_lun_statep(struct ispsoftc *, tstate_t *);
static __inline atio_private_data_t *isp_get_atpd(struct ispsoftc *, int);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
static cam_status
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
create_lun_state(struct ispsoftc *, int, struct cam_path *, tstate_t **);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
static void destroy_lun_state(struct ispsoftc *, tstate_t *);
Remove condition variables and status associated with target mode enabling. Instead, go to an interrupt/polled model. Fix get_lun_statep so we don't panic if there are no wildcard luns enabled. MFC after: 6 days
2004-02-08 19:17:56 +00:00
static int isp_en_lun(struct ispsoftc *, union ccb *);
static void isp_ledone(struct ispsoftc *, lun_entry_t *);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
static cam_status isp_abort_tgt_ccb(struct ispsoftc *, union ccb *);
Add a isp_target_putback_atio- we aren't using CCINCR at this time, so we need a function that tells the Qlogic f/w that a target mode command is done, so increase the resource count for that lun. Add in a timeout function to kick the putback again if we fail to do it the first time (we may not have the request queue space for ATIO push). Split the function isp_handle_platform_ctio into two parts so that the timeout function for the ATIO push or isp_handle_platform_ctio can inform CAM that the requested CTIO(s) are now done. Clean up (cough) residual handling. What we need for Fibre Channel is to preserve the at_datalen field from the original incoming ATIO so we can calculate a 'true' residual. Unfortunately, we're not guaranteed to get that back from CAM. We'll *try* to find it hiding in the periph_priv field (layering violation)- but if an ATIO was passed in from user land- forget it. This means that we'll probably get residuals wrong for Fibre Channel commands we're completing with an error. It's too late to 4.1 release to fix this- too bad. Luckily the only device we'd really care about this occurring on is a tape device and they're still so rare as FC attached devices that this can be considered an untested combination anyway. Remove all CCINCR usage (resource autoreplenish). When we've proved to ourself that things are working properly, we can add it back in. Make sure we propage 'suggested' sense data from the incoming ATIO into the created system ATIO- and set sense_len appropriately. Correctly propagate tag values. Fall back to the model of generating (well, the functions in isp_pci.c do the work) multiple CTIOs based upon what we get from XPT. Instead of being able to pair Qlogic generated ATIOs with CAM ATIOs, and then to pair CAM CTIOs with Qlogic CTIOs, we have to take the CTIO passed to us from XPT, and if it implies that we have to generate extra Qlogic CTIOs, so be it. This means that we have to wait until the last CTIO in a sequence we generated completes before calling xpt_done. Executive summary- target mode actually now pretty much works well enough to tell folks about.
2000-07-18 06:58:28 +00:00
static timeout_t isp_refire_putback_atio;
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
static void isp_complete_ctio(union ccb *);
static void isp_target_putback_atio(union ccb *);
static cam_status isp_target_start_ctio(struct ispsoftc *, union ccb *);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
static int isp_handle_platform_atio(struct ispsoftc *, at_entry_t *);
static int isp_handle_platform_atio2(struct ispsoftc *, at2_entry_t *);
static int isp_handle_platform_ctio(struct ispsoftc *, void *);
Add support for ISP_FC_GETHINFO, which returns current connection topology, speed, loopid, WWPN/WWNN, etc. Beef up target mode. Add isp_handle_platform_notify_scsi and isp_handle_platform_notify_fc platform handlers to handle immediate notifies (isp_handle_platform_notify_scsi is still stubbed out). In implementation of isp_handle_platform_notify_fc, for IN_ABORT_TASK, peel off a pending XPT_IMMED_NOTIFY and call xpt_done on it and hope that somebody upstream is listening. Make sure on final CTIO2s that we set residual correctly. These are absolutely crucial. Make sure we set relative offset for each CTIO2 based upon bytes we've already xferred. This is what the private adjunct datat to the original ATIO is. Note state of command so we can figure out where to find it if we get an ABORT from the firmware. Make sure we *always* set CAM_TAG_ACTION_VALID for ATIO2s. Make sure we keep track of the original lun. If se sent status (or we're otherwise done with the command), don't forget to free the adjunct structure.
2002-06-16 05:08:02 +00:00
static int isp_handle_platform_notify_scsi(struct ispsoftc *, in_entry_t *);
static int isp_handle_platform_notify_fc(struct ispsoftc *, in_fcentry_t *);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
Remove use of inlines and use the functions as a library. Larger code space, possibly performance hit, but more portable. Certainly less questionable use of inlining. Suggested by: des
2006-02-02 21:31:34 +00:00
static __inline int
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
is_lun_enabled(struct ispsoftc *isp, int bus, lun_id_t lun)
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
{
tstate_t *tptr;
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
tptr = isp->isp_osinfo.lun_hash[LUN_HASH_FUNC(isp, bus, lun)];
if (tptr == NULL) {
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
return (0);
}
do {
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
if (tptr->lun == (lun_id_t) lun && tptr->bus == bus) {
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
return (1);
}
} while ((tptr = tptr->next) != NULL);
return (0);
}
Remove use of inlines and use the functions as a library. Larger code space, possibly performance hit, but more portable. Certainly less questionable use of inlining. Suggested by: des
2006-02-02 21:31:34 +00:00
static __inline int
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
are_any_luns_enabled(struct ispsoftc *isp, int port)
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
{
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
int lo, hi;
if (IS_DUALBUS(isp)) {
lo = (port * (LUN_HASH_SIZE >> 1));
hi = lo + (LUN_HASH_SIZE >> 1);
} else {
lo = 0;
hi = LUN_HASH_SIZE;
}
for (lo = 0; lo < hi; lo++) {
if (isp->isp_osinfo.lun_hash[lo]) {
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
return (1);
}
}
return (0);
}
Remove use of inlines and use the functions as a library. Larger code space, possibly performance hit, but more portable. Certainly less questionable use of inlining. Suggested by: des
2006-02-02 21:31:34 +00:00
static __inline tstate_t *
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
get_lun_statep(struct ispsoftc *isp, int bus, lun_id_t lun)
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
{
I don't know what I was thinking- if I have two separate busses on on SIM (as is true for the 1280 and the 12160), then I have to have separate flags && status for *both* busses. *Whap*. Implement condition variables for coordination with some target mode events. It's nice to use these and not panic in obscure little places in the kernel like 'propagate_priority' just because we went to sleep holding a mutex, or some other absurd thing. Remove some bogus ISP_UNLOCK calls. *Whap*. No longer require that somebody do a lun enable on the wildcard device to enable target mode. They are, in fact, orthogonal. A wildcard open is a statement that somebody upstream is willing to accept commands which are otherwise unrouteable. Now, for QLogic regular SCSI target mode, this won't matter for a damn because we'll never see ATIOs for luns we haven't enabled (are listening for, if you will). But for SCCLUN fibre channel SCSI, we get all kinds of ATIOs. We can either reflect them back here with minimal info (which is isp_target.c:isp_endcmd() is for), or the wildcard device (nominally targbh) can handle them. Do further checking against firmware attributes to see whether we can, in fact, support target mode in Fibre Channel. For now, require SCCLUN f/w to supoprt FC target mode. This is an awful lot of change, but target mode *still* isn't quite right. MFC after: 4 weeks
2001-09-04 21:53:12 +00:00
tstate_t *tptr = NULL;
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
if (lun == CAM_LUN_WILDCARD) {
I don't know what I was thinking- if I have two separate busses on on SIM (as is true for the 1280 and the 12160), then I have to have separate flags && status for *both* busses. *Whap*. Implement condition variables for coordination with some target mode events. It's nice to use these and not panic in obscure little places in the kernel like 'propagate_priority' just because we went to sleep holding a mutex, or some other absurd thing. Remove some bogus ISP_UNLOCK calls. *Whap*. No longer require that somebody do a lun enable on the wildcard device to enable target mode. They are, in fact, orthogonal. A wildcard open is a statement that somebody upstream is willing to accept commands which are otherwise unrouteable. Now, for QLogic regular SCSI target mode, this won't matter for a damn because we'll never see ATIOs for luns we haven't enabled (are listening for, if you will). But for SCCLUN fibre channel SCSI, we get all kinds of ATIOs. We can either reflect them back here with minimal info (which is isp_target.c:isp_endcmd() is for), or the wildcard device (nominally targbh) can handle them. Do further checking against firmware attributes to see whether we can, in fact, support target mode in Fibre Channel. For now, require SCCLUN f/w to supoprt FC target mode. This is an awful lot of change, but target mode *still* isn't quite right. MFC after: 4 weeks
2001-09-04 21:53:12 +00:00
if (isp->isp_osinfo.tmflags[bus] & TM_WILDCARD_ENABLED) {
Add 2 Gigabit Fibre Channel support (2300 && 2312 cards). This required some reworking (and consequent cleanup) of the interrupt service code. Also begin to start a cleanup of target mode support that will (eventually) not require more inforamtion routed with the ATIO to come back with the CTIO other than tag. MFC after: 4 weeks
2001-08-31 21:39:04 +00:00
tptr = &isp->isp_osinfo.tsdflt[bus];
tptr->hold++;
return (tptr);
}
Remove condition variables and status associated with target mode enabling. Instead, go to an interrupt/polled model. Fix get_lun_statep so we don't panic if there are no wildcard luns enabled. MFC after: 6 days
2004-02-08 19:17:56 +00:00
return (NULL);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
} else {
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
tptr = isp->isp_osinfo.lun_hash[LUN_HASH_FUNC(isp, bus, lun)];
I don't know what I was thinking- if I have two separate busses on on SIM (as is true for the 1280 and the 12160), then I have to have separate flags && status for *both* busses. *Whap*. Implement condition variables for coordination with some target mode events. It's nice to use these and not panic in obscure little places in the kernel like 'propagate_priority' just because we went to sleep holding a mutex, or some other absurd thing. Remove some bogus ISP_UNLOCK calls. *Whap*. No longer require that somebody do a lun enable on the wildcard device to enable target mode. They are, in fact, orthogonal. A wildcard open is a statement that somebody upstream is willing to accept commands which are otherwise unrouteable. Now, for QLogic regular SCSI target mode, this won't matter for a damn because we'll never see ATIOs for luns we haven't enabled (are listening for, if you will). But for SCCLUN fibre channel SCSI, we get all kinds of ATIOs. We can either reflect them back here with minimal info (which is isp_target.c:isp_endcmd() is for), or the wildcard device (nominally targbh) can handle them. Do further checking against firmware attributes to see whether we can, in fact, support target mode in Fibre Channel. For now, require SCCLUN f/w to supoprt FC target mode. This is an awful lot of change, but target mode *still* isn't quite right. MFC after: 4 weeks
2001-09-04 21:53:12 +00:00
if (tptr == NULL) {
return (NULL);
}
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
}
do {
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
if (tptr->lun == lun && tptr->bus == bus) {
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
tptr->hold++;
return (tptr);
}
} while ((tptr = tptr->next) != NULL);
return (tptr);
}
Remove use of inlines and use the functions as a library. Larger code space, possibly performance hit, but more portable. Certainly less questionable use of inlining. Suggested by: des
2006-02-02 21:31:34 +00:00
static __inline void
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
rls_lun_statep(struct ispsoftc *isp, tstate_t *tptr)
{
if (tptr->hold)
tptr->hold--;
}
Remove use of inlines and use the functions as a library. Larger code space, possibly performance hit, but more portable. Certainly less questionable use of inlining. Suggested by: des
2006-02-02 21:31:34 +00:00
static __inline atio_private_data_t *
Begin to implement target mode that for Fibre Channel has a private per-command component that we *don't* try and pass thru CAM. CAM just is too risky and too much of a pain- structures get copied, but not all info of interest can be considered safely transported thru all consumers (including user space) from the incoming ATIO to the outgoing CTIO- it's just much safer to have a buddy structure, identified by the command's tag which *does* make it thru safely. Pay attention to link speed and report 200MB/s xfer speed for a 23XX card in 2GPs mode. MFC after: 1 week
2001-10-01 03:48:42 +00:00
isp_get_atpd(struct ispsoftc *isp, int tag)
{
atio_private_data_t *atp;
for (atp = isp->isp_osinfo.atpdp;
atp < &isp->isp_osinfo.atpdp[ATPDPSIZE]; atp++) {
if (atp->tag == tag)
return (atp);
}
return (NULL);
}
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
static cam_status
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
create_lun_state(struct ispsoftc *isp, int bus,
struct cam_path *path, tstate_t **rslt)
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
{
cam_status status;
lun_id_t lun;
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
int hfx;
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
tstate_t *tptr, *new;
lun = xpt_path_lun_id(path);
if (lun < 0) {
return (CAM_LUN_INVALID);
}
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
if (is_lun_enabled(isp, bus, lun)) {
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
return (CAM_LUN_ALRDY_ENA);
}
More M_ZERO patches. Submitted by: josh@zipperup.org Submitted by: Robert Drehmel <robd@gmx.net> Approved by: mjacob
2000-12-03 20:46:54 +00:00
new = (tstate_t *) malloc(sizeof (tstate_t), M_DEVBUF, M_NOWAIT|M_ZERO);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
if (new == NULL) {
return (CAM_RESRC_UNAVAIL);
}
status = xpt_create_path(&new->owner, NULL, xpt_path_path_id(path),
xpt_path_target_id(path), xpt_path_lun_id(path));
if (status != CAM_REQ_CMP) {
free(new, M_DEVBUF);
return (status);
}
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
new->bus = bus;
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
new->lun = lun;
SLIST_INIT(&new->atios);
SLIST_INIT(&new->inots);
new->hold = 1;
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
hfx = LUN_HASH_FUNC(isp, new->bus, new->lun);
tptr = isp->isp_osinfo.lun_hash[hfx];
if (tptr == NULL) {
isp->isp_osinfo.lun_hash[hfx] = new;
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
} else {
while (tptr->next)
tptr = tptr->next;
tptr->next = new;
}
*rslt = new;
return (CAM_REQ_CMP);
}
Remove use of inlines and use the functions as a library. Larger code space, possibly performance hit, but more portable. Certainly less questionable use of inlining. Suggested by: des
2006-02-02 21:31:34 +00:00
static __inline void
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
destroy_lun_state(struct ispsoftc *isp, tstate_t *tptr)
{
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
int hfx;
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
tstate_t *lw, *pw;
if (tptr->hold) {
return;
}
Remove condition variables and status associated with target mode enabling. Instead, go to an interrupt/polled model. Fix get_lun_statep so we don't panic if there are no wildcard luns enabled. MFC after: 6 days
2004-02-08 19:17:56 +00:00
hfx = LUN_HASH_FUNC(isp, tptr->bus, tptr->lun);
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
pw = isp->isp_osinfo.lun_hash[hfx];
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
if (pw == NULL) {
return;
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
} else if (pw->lun == tptr->lun && pw->bus == tptr->bus) {
isp->isp_osinfo.lun_hash[hfx] = pw->next;
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
} else {
lw = pw;
pw = lw->next;
while (pw) {
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
if (pw->lun == tptr->lun && pw->bus == tptr->bus) {
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
lw->next = pw->next;
break;
}
lw = pw;
pw = pw->next;
}
if (pw == NULL) {
return;
}
}
free(tptr, M_DEVBUF);
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
}
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
/*
Remove condition variables and status associated with target mode enabling. Instead, go to an interrupt/polled model. Fix get_lun_statep so we don't panic if there are no wildcard luns enabled. MFC after: 6 days
2004-02-08 19:17:56 +00:00
* Enable luns.
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
*/
Remove condition variables and status associated with target mode enabling. Instead, go to an interrupt/polled model. Fix get_lun_statep so we don't panic if there are no wildcard luns enabled. MFC after: 6 days
2004-02-08 19:17:56 +00:00
static int
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
isp_en_lun(struct ispsoftc *isp, union ccb *ccb)
{
struct ccb_en_lun *cel = &ccb->cel;
tstate_t *tptr;
a) clean up some declaration stuff (i.e., make more modern with respect to getting rid u_int for uint and so on). b) Turn back on 64 bit DAC support. Cheeze it a bit in that we have two DMA callback functions- one when we have bus_addr_t > 4 bits in width and the other which should be normal. Even Cheezier in that we turn off setting up DMA maps to be BUS_SPACE_MAXADDR if we're in ISP_TARGET_MODE. More work on this in a week or so. c) Tested under amd64 and 1MB DFLTPHYS, sparc64, i386 (PAE, but insufficient memory to really test > 4GB). LINT check under amd64. MFC after: 1 month
2006-02-15 00:31:48 +00:00
uint32_t seq;
Checkpoint of work in progress in cleaning up target mode. It actually seems to work well in RELENG_4. However, 5.X locking foo means that I'll have to do some quick redesign. Add ioctl handlers for ISP_GETROLE and ISP_SETROLE ioctls.
2004-02-07 03:47:33 +00:00
int bus, cmd, av, wildcard, tm_on;
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
lun_id_t lun;
target_id_t tgt;
Remove condition variables and status associated with target mode enabling. Instead, go to an interrupt/polled model. Fix get_lun_statep so we don't panic if there are no wildcard luns enabled. MFC after: 6 days
2004-02-08 19:17:56 +00:00
bus = XS_CHANNEL(ccb);
if (bus > 1) {
xpt_print_path(ccb->ccb_h.path);
printf("illegal bus %d\n", bus);
ccb->ccb_h.status = CAM_PATH_INVALID;
return (-1);
}
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
tgt = ccb->ccb_h.target_id;
lun = ccb->ccb_h.target_lun;
Remove condition variables and status associated with target mode enabling. Instead, go to an interrupt/polled model. Fix get_lun_statep so we don't panic if there are no wildcard luns enabled. MFC after: 6 days
2004-02-08 19:17:56 +00:00
isp_prt(isp, ISP_LOGTDEBUG0,
"isp_en_lun: %sabling lun 0x%x on channel %d",
cel->enable? "en" : "dis", lun, bus);
various fixes
2000-08-27 23:38:44 +00:00
Do some cleanup based upon adapter role- mainly not enabling interrupts if we're ISP_ROLE_NONE. Change ISPASYNC_LOGGED_INOUT to ISPASYNC_PROMENADE. Make sure we note if something is a fabric device. Target mode: Finally fix (to a first approximation) SCSI Target Mode again- we needed to correctly check against CAM_TARGET_WILDCARD and CAM_LUN_WILDCARD so that targbh won't confuse us. Comment out the drainqueue stuff for now. Use isp_fc_runstate instead if isp_control/ISPCTL_FCLINK_TEST.
2001-02-11 03:47:39 +00:00
if ((lun != CAM_LUN_WILDCARD) &&
(lun < 0 || lun >= (lun_id_t) isp->isp_maxluns)) {
various fixes
2000-08-27 23:38:44 +00:00
ccb->ccb_h.status = CAM_LUN_INVALID;
Remove condition variables and status associated with target mode enabling. Instead, go to an interrupt/polled model. Fix get_lun_statep so we don't panic if there are no wildcard luns enabled. MFC after: 6 days
2004-02-08 19:17:56 +00:00
return (-1);
various fixes
2000-08-27 23:38:44 +00:00
}
I don't know what I was thinking- if I have two separate busses on on SIM (as is true for the 1280 and the 12160), then I have to have separate flags && status for *both* busses. *Whap*. Implement condition variables for coordination with some target mode events. It's nice to use these and not panic in obscure little places in the kernel like 'propagate_priority' just because we went to sleep holding a mutex, or some other absurd thing. Remove some bogus ISP_UNLOCK calls. *Whap*. No longer require that somebody do a lun enable on the wildcard device to enable target mode. They are, in fact, orthogonal. A wildcard open is a statement that somebody upstream is willing to accept commands which are otherwise unrouteable. Now, for QLogic regular SCSI target mode, this won't matter for a damn because we'll never see ATIOs for luns we haven't enabled (are listening for, if you will). But for SCCLUN fibre channel SCSI, we get all kinds of ATIOs. We can either reflect them back here with minimal info (which is isp_target.c:isp_endcmd() is for), or the wildcard device (nominally targbh) can handle them. Do further checking against firmware attributes to see whether we can, in fact, support target mode in Fibre Channel. For now, require SCCLUN f/w to supoprt FC target mode. This is an awful lot of change, but target mode *still* isn't quite right. MFC after: 4 weeks
2001-09-04 21:53:12 +00:00
various fixes
2000-08-27 23:38:44 +00:00
if (IS_SCSI(isp)) {
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
sdparam *sdp = isp->isp_param;
sdp += bus;
various fixes
2000-08-27 23:38:44 +00:00
if (tgt != CAM_TARGET_WILDCARD &&
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
tgt != sdp->isp_initiator_id) {
various fixes
2000-08-27 23:38:44 +00:00
ccb->ccb_h.status = CAM_TID_INVALID;
Remove condition variables and status associated with target mode enabling. Instead, go to an interrupt/polled model. Fix get_lun_statep so we don't panic if there are no wildcard luns enabled. MFC after: 6 days
2004-02-08 19:17:56 +00:00
return (-1);
various fixes
2000-08-27 23:38:44 +00:00
}
} else {
Checkpoint of work in progress in cleaning up target mode. It actually seems to work well in RELENG_4. However, 5.X locking foo means that I'll have to do some quick redesign. Add ioctl handlers for ISP_GETROLE and ISP_SETROLE ioctls.
2004-02-07 03:47:33 +00:00
/*
* There's really no point in doing this yet w/o multi-tid
* capability. Even then, it's problematic.
*/
#if 0
various fixes
2000-08-27 23:38:44 +00:00
if (tgt != CAM_TARGET_WILDCARD &&
Do some cleanup based upon adapter role- mainly not enabling interrupts if we're ISP_ROLE_NONE. Change ISPASYNC_LOGGED_INOUT to ISPASYNC_PROMENADE. Make sure we note if something is a fabric device. Target mode: Finally fix (to a first approximation) SCSI Target Mode again- we needed to correctly check against CAM_TARGET_WILDCARD and CAM_LUN_WILDCARD so that targbh won't confuse us. Comment out the drainqueue stuff for now. Use isp_fc_runstate instead if isp_control/ISPCTL_FCLINK_TEST.
2001-02-11 03:47:39 +00:00
tgt != FCPARAM(isp)->isp_iid) {
various fixes
2000-08-27 23:38:44 +00:00
ccb->ccb_h.status = CAM_TID_INVALID;
Remove condition variables and status associated with target mode enabling. Instead, go to an interrupt/polled model. Fix get_lun_statep so we don't panic if there are no wildcard luns enabled. MFC after: 6 days
2004-02-08 19:17:56 +00:00
return (-1);
various fixes
2000-08-27 23:38:44 +00:00
}
Checkpoint of work in progress in cleaning up target mode. It actually seems to work well in RELENG_4. However, 5.X locking foo means that I'll have to do some quick redesign. Add ioctl handlers for ISP_GETROLE and ISP_SETROLE ioctls.
2004-02-07 03:47:33 +00:00
#endif
I don't know what I was thinking- if I have two separate busses on on SIM (as is true for the 1280 and the 12160), then I have to have separate flags && status for *both* busses. *Whap*. Implement condition variables for coordination with some target mode events. It's nice to use these and not panic in obscure little places in the kernel like 'propagate_priority' just because we went to sleep holding a mutex, or some other absurd thing. Remove some bogus ISP_UNLOCK calls. *Whap*. No longer require that somebody do a lun enable on the wildcard device to enable target mode. They are, in fact, orthogonal. A wildcard open is a statement that somebody upstream is willing to accept commands which are otherwise unrouteable. Now, for QLogic regular SCSI target mode, this won't matter for a damn because we'll never see ATIOs for luns we haven't enabled (are listening for, if you will). But for SCCLUN fibre channel SCSI, we get all kinds of ATIOs. We can either reflect them back here with minimal info (which is isp_target.c:isp_endcmd() is for), or the wildcard device (nominally targbh) can handle them. Do further checking against firmware attributes to see whether we can, in fact, support target mode in Fibre Channel. For now, require SCCLUN f/w to supoprt FC target mode. This is an awful lot of change, but target mode *still* isn't quite right. MFC after: 4 weeks
2001-09-04 21:53:12 +00:00
/*
* This is as a good a place as any to check f/w capabilities.
*/
if ((FCPARAM(isp)->isp_fwattr & ISP_FW_ATTR_TMODE) == 0) {
isp_prt(isp, ISP_LOGERR,
"firmware does not support target mode");
ccb->ccb_h.status = CAM_FUNC_NOTAVAIL;
Remove condition variables and status associated with target mode enabling. Instead, go to an interrupt/polled model. Fix get_lun_statep so we don't panic if there are no wildcard luns enabled. MFC after: 6 days
2004-02-08 19:17:56 +00:00
return (-1);
Do some cleanup based upon adapter role- mainly not enabling interrupts if we're ISP_ROLE_NONE. Change ISPASYNC_LOGGED_INOUT to ISPASYNC_PROMENADE. Make sure we note if something is a fabric device. Target mode: Finally fix (to a first approximation) SCSI Target Mode again- we needed to correctly check against CAM_TARGET_WILDCARD and CAM_LUN_WILDCARD so that targbh won't confuse us. Comment out the drainqueue stuff for now. Use isp_fc_runstate instead if isp_control/ISPCTL_FCLINK_TEST.
2001-02-11 03:47:39 +00:00
}
I don't know what I was thinking- if I have two separate busses on on SIM (as is true for the 1280 and the 12160), then I have to have separate flags && status for *both* busses. *Whap*. Implement condition variables for coordination with some target mode events. It's nice to use these and not panic in obscure little places in the kernel like 'propagate_priority' just because we went to sleep holding a mutex, or some other absurd thing. Remove some bogus ISP_UNLOCK calls. *Whap*. No longer require that somebody do a lun enable on the wildcard device to enable target mode. They are, in fact, orthogonal. A wildcard open is a statement that somebody upstream is willing to accept commands which are otherwise unrouteable. Now, for QLogic regular SCSI target mode, this won't matter for a damn because we'll never see ATIOs for luns we haven't enabled (are listening for, if you will). But for SCCLUN fibre channel SCSI, we get all kinds of ATIOs. We can either reflect them back here with minimal info (which is isp_target.c:isp_endcmd() is for), or the wildcard device (nominally targbh) can handle them. Do further checking against firmware attributes to see whether we can, in fact, support target mode in Fibre Channel. For now, require SCCLUN f/w to supoprt FC target mode. This is an awful lot of change, but target mode *still* isn't quite right. MFC after: 4 weeks
2001-09-04 21:53:12 +00:00
/*
* XXX: We *could* handle non-SCCLUN f/w, but we'd have to
* XXX: dorks with our already fragile enable/disable code.
*/
if ((FCPARAM(isp)->isp_fwattr & ISP_FW_ATTR_SCCLUN) == 0) {
isp_prt(isp, ISP_LOGERR,
"firmware not SCCLUN capable");
Checkpoint of work in progress in cleaning up target mode. It actually seems to work well in RELENG_4. However, 5.X locking foo means that I'll have to do some quick redesign. Add ioctl handlers for ISP_GETROLE and ISP_SETROLE ioctls.
2004-02-07 03:47:33 +00:00
ccb->ccb_h.status = CAM_FUNC_NOTAVAIL;
Remove condition variables and status associated with target mode enabling. Instead, go to an interrupt/polled model. Fix get_lun_statep so we don't panic if there are no wildcard luns enabled. MFC after: 6 days
2004-02-08 19:17:56 +00:00
return (-1);
various fixes
2000-08-27 23:38:44 +00:00
}
}
I don't know what I was thinking- if I have two separate busses on on SIM (as is true for the 1280 and the 12160), then I have to have separate flags && status for *both* busses. *Whap*. Implement condition variables for coordination with some target mode events. It's nice to use these and not panic in obscure little places in the kernel like 'propagate_priority' just because we went to sleep holding a mutex, or some other absurd thing. Remove some bogus ISP_UNLOCK calls. *Whap*. No longer require that somebody do a lun enable on the wildcard device to enable target mode. They are, in fact, orthogonal. A wildcard open is a statement that somebody upstream is willing to accept commands which are otherwise unrouteable. Now, for QLogic regular SCSI target mode, this won't matter for a damn because we'll never see ATIOs for luns we haven't enabled (are listening for, if you will). But for SCCLUN fibre channel SCSI, we get all kinds of ATIOs. We can either reflect them back here with minimal info (which is isp_target.c:isp_endcmd() is for), or the wildcard device (nominally targbh) can handle them. Do further checking against firmware attributes to see whether we can, in fact, support target mode in Fibre Channel. For now, require SCCLUN f/w to supoprt FC target mode. This is an awful lot of change, but target mode *still* isn't quite right. MFC after: 4 weeks
2001-09-04 21:53:12 +00:00
if (tgt == CAM_TARGET_WILDCARD) {
if (lun == CAM_LUN_WILDCARD) {
wildcard = 1;
} else {
ccb->ccb_h.status = CAM_LUN_INVALID;
Remove condition variables and status associated with target mode enabling. Instead, go to an interrupt/polled model. Fix get_lun_statep so we don't panic if there are no wildcard luns enabled. MFC after: 6 days
2004-02-08 19:17:56 +00:00
return (-1);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
}
Add 2 Gigabit Fibre Channel support (2300 && 2312 cards). This required some reworking (and consequent cleanup) of the interrupt service code. Also begin to start a cleanup of target mode support that will (eventually) not require more inforamtion routed with the ATIO to come back with the CTIO other than tag. MFC after: 4 weeks
2001-08-31 21:39:04 +00:00
} else {
wildcard = 0;
}
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
Checkpoint of work in progress in cleaning up target mode. It actually seems to work well in RELENG_4. However, 5.X locking foo means that I'll have to do some quick redesign. Add ioctl handlers for ISP_GETROLE and ISP_SETROLE ioctls.
2004-02-07 03:47:33 +00:00
tm_on = (isp->isp_osinfo.tmflags[bus] & TM_TMODE_ENABLED) != 0;
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
/*
* Next check to see whether this is a target/lun wildcard action.
I don't know what I was thinking- if I have two separate busses on on SIM (as is true for the 1280 and the 12160), then I have to have separate flags && status for *both* busses. *Whap*. Implement condition variables for coordination with some target mode events. It's nice to use these and not panic in obscure little places in the kernel like 'propagate_priority' just because we went to sleep holding a mutex, or some other absurd thing. Remove some bogus ISP_UNLOCK calls. *Whap*. No longer require that somebody do a lun enable on the wildcard device to enable target mode. They are, in fact, orthogonal. A wildcard open is a statement that somebody upstream is willing to accept commands which are otherwise unrouteable. Now, for QLogic regular SCSI target mode, this won't matter for a damn because we'll never see ATIOs for luns we haven't enabled (are listening for, if you will). But for SCCLUN fibre channel SCSI, we get all kinds of ATIOs. We can either reflect them back here with minimal info (which is isp_target.c:isp_endcmd() is for), or the wildcard device (nominally targbh) can handle them. Do further checking against firmware attributes to see whether we can, in fact, support target mode in Fibre Channel. For now, require SCCLUN f/w to supoprt FC target mode. This is an awful lot of change, but target mode *still* isn't quite right. MFC after: 4 weeks
2001-09-04 21:53:12 +00:00
*
* If so, we know that we can accept commands for luns that haven't
* been enabled yet and send them upstream. Otherwise, we have to
* handle them locally (if we see them at all).
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
*/
Add 2 Gigabit Fibre Channel support (2300 && 2312 cards). This required some reworking (and consequent cleanup) of the interrupt service code. Also begin to start a cleanup of target mode support that will (eventually) not require more inforamtion routed with the ATIO to come back with the CTIO other than tag. MFC after: 4 weeks
2001-08-31 21:39:04 +00:00
if (wildcard) {
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
tptr = &isp->isp_osinfo.tsdflt[bus];
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
if (cel->enable) {
Remove condition variables and status associated with target mode enabling. Instead, go to an interrupt/polled model. Fix get_lun_statep so we don't panic if there are no wildcard luns enabled. MFC after: 6 days
2004-02-08 19:17:56 +00:00
if (tm_on) {
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
ccb->ccb_h.status = CAM_LUN_ALRDY_ENA;
Remove condition variables and status associated with target mode enabling. Instead, go to an interrupt/polled model. Fix get_lun_statep so we don't panic if there are no wildcard luns enabled. MFC after: 6 days
2004-02-08 19:17:56 +00:00
return (-1);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
}
ccb->ccb_h.status =
xpt_create_path(&tptr->owner, NULL,
xpt_path_path_id(ccb->ccb_h.path),
xpt_path_target_id(ccb->ccb_h.path),
xpt_path_lun_id(ccb->ccb_h.path));
if (ccb->ccb_h.status != CAM_REQ_CMP) {
Remove condition variables and status associated with target mode enabling. Instead, go to an interrupt/polled model. Fix get_lun_statep so we don't panic if there are no wildcard luns enabled. MFC after: 6 days
2004-02-08 19:17:56 +00:00
return (-1);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
}
SLIST_INIT(&tptr->atios);
SLIST_INIT(&tptr->inots);
I don't know what I was thinking- if I have two separate busses on on SIM (as is true for the 1280 and the 12160), then I have to have separate flags && status for *both* busses. *Whap*. Implement condition variables for coordination with some target mode events. It's nice to use these and not panic in obscure little places in the kernel like 'propagate_priority' just because we went to sleep holding a mutex, or some other absurd thing. Remove some bogus ISP_UNLOCK calls. *Whap*. No longer require that somebody do a lun enable on the wildcard device to enable target mode. They are, in fact, orthogonal. A wildcard open is a statement that somebody upstream is willing to accept commands which are otherwise unrouteable. Now, for QLogic regular SCSI target mode, this won't matter for a damn because we'll never see ATIOs for luns we haven't enabled (are listening for, if you will). But for SCCLUN fibre channel SCSI, we get all kinds of ATIOs. We can either reflect them back here with minimal info (which is isp_target.c:isp_endcmd() is for), or the wildcard device (nominally targbh) can handle them. Do further checking against firmware attributes to see whether we can, in fact, support target mode in Fibre Channel. For now, require SCCLUN f/w to supoprt FC target mode. This is an awful lot of change, but target mode *still* isn't quite right. MFC after: 4 weeks
2001-09-04 21:53:12 +00:00
isp->isp_osinfo.tmflags[bus] |= TM_WILDCARD_ENABLED;
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
} else {
Remove condition variables and status associated with target mode enabling. Instead, go to an interrupt/polled model. Fix get_lun_statep so we don't panic if there are no wildcard luns enabled. MFC after: 6 days
2004-02-08 19:17:56 +00:00
if (tm_on == 0) {
Add 2 Gigabit Fibre Channel support (2300 && 2312 cards). This required some reworking (and consequent cleanup) of the interrupt service code. Also begin to start a cleanup of target mode support that will (eventually) not require more inforamtion routed with the ATIO to come back with the CTIO other than tag. MFC after: 4 weeks
2001-08-31 21:39:04 +00:00
ccb->ccb_h.status = CAM_REQ_CMP;
Remove condition variables and status associated with target mode enabling. Instead, go to an interrupt/polled model. Fix get_lun_statep so we don't panic if there are no wildcard luns enabled. MFC after: 6 days
2004-02-08 19:17:56 +00:00
return (-1);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
}
Add 2 Gigabit Fibre Channel support (2300 && 2312 cards). This required some reworking (and consequent cleanup) of the interrupt service code. Also begin to start a cleanup of target mode support that will (eventually) not require more inforamtion routed with the ATIO to come back with the CTIO other than tag. MFC after: 4 weeks
2001-08-31 21:39:04 +00:00
if (tptr->hold) {
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
ccb->ccb_h.status = CAM_SCSI_BUSY;
Remove condition variables and status associated with target mode enabling. Instead, go to an interrupt/polled model. Fix get_lun_statep so we don't panic if there are no wildcard luns enabled. MFC after: 6 days
2004-02-08 19:17:56 +00:00
return (-1);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
}
Add 2 Gigabit Fibre Channel support (2300 && 2312 cards). This required some reworking (and consequent cleanup) of the interrupt service code. Also begin to start a cleanup of target mode support that will (eventually) not require more inforamtion routed with the ATIO to come back with the CTIO other than tag. MFC after: 4 weeks
2001-08-31 21:39:04 +00:00
xpt_free_path(tptr->owner);
I don't know what I was thinking- if I have two separate busses on on SIM (as is true for the 1280 and the 12160), then I have to have separate flags && status for *both* busses. *Whap*. Implement condition variables for coordination with some target mode events. It's nice to use these and not panic in obscure little places in the kernel like 'propagate_priority' just because we went to sleep holding a mutex, or some other absurd thing. Remove some bogus ISP_UNLOCK calls. *Whap*. No longer require that somebody do a lun enable on the wildcard device to enable target mode. They are, in fact, orthogonal. A wildcard open is a statement that somebody upstream is willing to accept commands which are otherwise unrouteable. Now, for QLogic regular SCSI target mode, this won't matter for a damn because we'll never see ATIOs for luns we haven't enabled (are listening for, if you will). But for SCCLUN fibre channel SCSI, we get all kinds of ATIOs. We can either reflect them back here with minimal info (which is isp_target.c:isp_endcmd() is for), or the wildcard device (nominally targbh) can handle them. Do further checking against firmware attributes to see whether we can, in fact, support target mode in Fibre Channel. For now, require SCCLUN f/w to supoprt FC target mode. This is an awful lot of change, but target mode *still* isn't quite right. MFC after: 4 weeks
2001-09-04 21:53:12 +00:00
isp->isp_osinfo.tmflags[bus] &= ~TM_WILDCARD_ENABLED;
Add 2 Gigabit Fibre Channel support (2300 && 2312 cards). This required some reworking (and consequent cleanup) of the interrupt service code. Also begin to start a cleanup of target mode support that will (eventually) not require more inforamtion routed with the ATIO to come back with the CTIO other than tag. MFC after: 4 weeks
2001-08-31 21:39:04 +00:00
}
}
/*
* Now check to see whether this bus needs to be
* enabled/disabled with respect to target mode.
*/
av = bus << 31;
Checkpoint of work in progress in cleaning up target mode. It actually seems to work well in RELENG_4. However, 5.X locking foo means that I'll have to do some quick redesign. Add ioctl handlers for ISP_GETROLE and ISP_SETROLE ioctls.
2004-02-07 03:47:33 +00:00
if (cel->enable && tm_on == 0) {
Add 2 Gigabit Fibre Channel support (2300 && 2312 cards). This required some reworking (and consequent cleanup) of the interrupt service code. Also begin to start a cleanup of target mode support that will (eventually) not require more inforamtion routed with the ATIO to come back with the CTIO other than tag. MFC after: 4 weeks
2001-08-31 21:39:04 +00:00
av |= ENABLE_TARGET_FLAG;
av = isp_control(isp, ISPCTL_TOGGLE_TMODE, &av);
if (av) {
ccb->ccb_h.status = CAM_FUNC_NOTAVAIL;
if (wildcard) {
I don't know what I was thinking- if I have two separate busses on on SIM (as is true for the 1280 and the 12160), then I have to have separate flags && status for *both* busses. *Whap*. Implement condition variables for coordination with some target mode events. It's nice to use these and not panic in obscure little places in the kernel like 'propagate_priority' just because we went to sleep holding a mutex, or some other absurd thing. Remove some bogus ISP_UNLOCK calls. *Whap*. No longer require that somebody do a lun enable on the wildcard device to enable target mode. They are, in fact, orthogonal. A wildcard open is a statement that somebody upstream is willing to accept commands which are otherwise unrouteable. Now, for QLogic regular SCSI target mode, this won't matter for a damn because we'll never see ATIOs for luns we haven't enabled (are listening for, if you will). But for SCCLUN fibre channel SCSI, we get all kinds of ATIOs. We can either reflect them back here with minimal info (which is isp_target.c:isp_endcmd() is for), or the wildcard device (nominally targbh) can handle them. Do further checking against firmware attributes to see whether we can, in fact, support target mode in Fibre Channel. For now, require SCCLUN f/w to supoprt FC target mode. This is an awful lot of change, but target mode *still* isn't quite right. MFC after: 4 weeks
2001-09-04 21:53:12 +00:00
isp->isp_osinfo.tmflags[bus] &=
~TM_WILDCARD_ENABLED;
Add 2 Gigabit Fibre Channel support (2300 && 2312 cards). This required some reworking (and consequent cleanup) of the interrupt service code. Also begin to start a cleanup of target mode support that will (eventually) not require more inforamtion routed with the ATIO to come back with the CTIO other than tag. MFC after: 4 weeks
2001-08-31 21:39:04 +00:00
xpt_free_path(tptr->owner);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
}
Remove condition variables and status associated with target mode enabling. Instead, go to an interrupt/polled model. Fix get_lun_statep so we don't panic if there are no wildcard luns enabled. MFC after: 6 days
2004-02-08 19:17:56 +00:00
return (-1);
Add 2 Gigabit Fibre Channel support (2300 && 2312 cards). This required some reworking (and consequent cleanup) of the interrupt service code. Also begin to start a cleanup of target mode support that will (eventually) not require more inforamtion routed with the ATIO to come back with the CTIO other than tag. MFC after: 4 weeks
2001-08-31 21:39:04 +00:00
}
I don't know what I was thinking- if I have two separate busses on on SIM (as is true for the 1280 and the 12160), then I have to have separate flags && status for *both* busses. *Whap*. Implement condition variables for coordination with some target mode events. It's nice to use these and not panic in obscure little places in the kernel like 'propagate_priority' just because we went to sleep holding a mutex, or some other absurd thing. Remove some bogus ISP_UNLOCK calls. *Whap*. No longer require that somebody do a lun enable on the wildcard device to enable target mode. They are, in fact, orthogonal. A wildcard open is a statement that somebody upstream is willing to accept commands which are otherwise unrouteable. Now, for QLogic regular SCSI target mode, this won't matter for a damn because we'll never see ATIOs for luns we haven't enabled (are listening for, if you will). But for SCCLUN fibre channel SCSI, we get all kinds of ATIOs. We can either reflect them back here with minimal info (which is isp_target.c:isp_endcmd() is for), or the wildcard device (nominally targbh) can handle them. Do further checking against firmware attributes to see whether we can, in fact, support target mode in Fibre Channel. For now, require SCCLUN f/w to supoprt FC target mode. This is an awful lot of change, but target mode *still* isn't quite right. MFC after: 4 weeks
2001-09-04 21:53:12 +00:00
isp->isp_osinfo.tmflags[bus] |= TM_TMODE_ENABLED;
Add 2 Gigabit Fibre Channel support (2300 && 2312 cards). This required some reworking (and consequent cleanup) of the interrupt service code. Also begin to start a cleanup of target mode support that will (eventually) not require more inforamtion routed with the ATIO to come back with the CTIO other than tag. MFC after: 4 weeks
2001-08-31 21:39:04 +00:00
isp_prt(isp, ISP_LOGINFO,
"Target Mode enabled on channel %d", bus);
Checkpoint of work in progress in cleaning up target mode. It actually seems to work well in RELENG_4. However, 5.X locking foo means that I'll have to do some quick redesign. Add ioctl handlers for ISP_GETROLE and ISP_SETROLE ioctls.
2004-02-07 03:47:33 +00:00
} else if (cel->enable == 0 && tm_on && wildcard) {
Add 2 Gigabit Fibre Channel support (2300 && 2312 cards). This required some reworking (and consequent cleanup) of the interrupt service code. Also begin to start a cleanup of target mode support that will (eventually) not require more inforamtion routed with the ATIO to come back with the CTIO other than tag. MFC after: 4 weeks
2001-08-31 21:39:04 +00:00
if (are_any_luns_enabled(isp, bus)) {
ccb->ccb_h.status = CAM_SCSI_BUSY;
Remove condition variables and status associated with target mode enabling. Instead, go to an interrupt/polled model. Fix get_lun_statep so we don't panic if there are no wildcard luns enabled. MFC after: 6 days
2004-02-08 19:17:56 +00:00
return (-1);
Add 2 Gigabit Fibre Channel support (2300 && 2312 cards). This required some reworking (and consequent cleanup) of the interrupt service code. Also begin to start a cleanup of target mode support that will (eventually) not require more inforamtion routed with the ATIO to come back with the CTIO other than tag. MFC after: 4 weeks
2001-08-31 21:39:04 +00:00
}
av = isp_control(isp, ISPCTL_TOGGLE_TMODE, &av);
if (av) {
ccb->ccb_h.status = CAM_FUNC_NOTAVAIL;
Remove condition variables and status associated with target mode enabling. Instead, go to an interrupt/polled model. Fix get_lun_statep so we don't panic if there are no wildcard luns enabled. MFC after: 6 days
2004-02-08 19:17:56 +00:00
return (-1);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
}
I don't know what I was thinking- if I have two separate busses on on SIM (as is true for the 1280 and the 12160), then I have to have separate flags && status for *both* busses. *Whap*. Implement condition variables for coordination with some target mode events. It's nice to use these and not panic in obscure little places in the kernel like 'propagate_priority' just because we went to sleep holding a mutex, or some other absurd thing. Remove some bogus ISP_UNLOCK calls. *Whap*. No longer require that somebody do a lun enable on the wildcard device to enable target mode. They are, in fact, orthogonal. A wildcard open is a statement that somebody upstream is willing to accept commands which are otherwise unrouteable. Now, for QLogic regular SCSI target mode, this won't matter for a damn because we'll never see ATIOs for luns we haven't enabled (are listening for, if you will). But for SCCLUN fibre channel SCSI, we get all kinds of ATIOs. We can either reflect them back here with minimal info (which is isp_target.c:isp_endcmd() is for), or the wildcard device (nominally targbh) can handle them. Do further checking against firmware attributes to see whether we can, in fact, support target mode in Fibre Channel. For now, require SCCLUN f/w to supoprt FC target mode. This is an awful lot of change, but target mode *still* isn't quite right. MFC after: 4 weeks
2001-09-04 21:53:12 +00:00
isp->isp_osinfo.tmflags[bus] &= ~TM_TMODE_ENABLED;
Add 2 Gigabit Fibre Channel support (2300 && 2312 cards). This required some reworking (and consequent cleanup) of the interrupt service code. Also begin to start a cleanup of target mode support that will (eventually) not require more inforamtion routed with the ATIO to come back with the CTIO other than tag. MFC after: 4 weeks
2001-08-31 21:39:04 +00:00
isp_prt(isp, ISP_LOGINFO,
"Target Mode disabled on channel %d", bus);
}
if (wildcard) {
I don't know what I was thinking- if I have two separate busses on on SIM (as is true for the 1280 and the 12160), then I have to have separate flags && status for *both* busses. *Whap*. Implement condition variables for coordination with some target mode events. It's nice to use these and not panic in obscure little places in the kernel like 'propagate_priority' just because we went to sleep holding a mutex, or some other absurd thing. Remove some bogus ISP_UNLOCK calls. *Whap*. No longer require that somebody do a lun enable on the wildcard device to enable target mode. They are, in fact, orthogonal. A wildcard open is a statement that somebody upstream is willing to accept commands which are otherwise unrouteable. Now, for QLogic regular SCSI target mode, this won't matter for a damn because we'll never see ATIOs for luns we haven't enabled (are listening for, if you will). But for SCCLUN fibre channel SCSI, we get all kinds of ATIOs. We can either reflect them back here with minimal info (which is isp_target.c:isp_endcmd() is for), or the wildcard device (nominally targbh) can handle them. Do further checking against firmware attributes to see whether we can, in fact, support target mode in Fibre Channel. For now, require SCCLUN f/w to supoprt FC target mode. This is an awful lot of change, but target mode *still* isn't quite right. MFC after: 4 weeks
2001-09-04 21:53:12 +00:00
ccb->ccb_h.status = CAM_REQ_CMP;
Remove condition variables and status associated with target mode enabling. Instead, go to an interrupt/polled model. Fix get_lun_statep so we don't panic if there are no wildcard luns enabled. MFC after: 6 days
2004-02-08 19:17:56 +00:00
return (-1);
}
/*
* Find an empty slot
*/
for (seq = 0; seq < NLEACT; seq++) {
if (isp->isp_osinfo.leact[seq] == 0) {
break;
}
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
}
Remove condition variables and status associated with target mode enabling. Instead, go to an interrupt/polled model. Fix get_lun_statep so we don't panic if there are no wildcard luns enabled. MFC after: 6 days
2004-02-08 19:17:56 +00:00
if (seq >= NLEACT) {
ccb->ccb_h.status = CAM_RESRC_UNAVAIL;
return (-1);
}
isp->isp_osinfo.leact[seq] = ccb;
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
if (cel->enable) {
ccb->ccb_h.status =
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
create_lun_state(isp, bus, ccb->ccb_h.path, &tptr);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
if (ccb->ccb_h.status != CAM_REQ_CMP) {
Remove condition variables and status associated with target mode enabling. Instead, go to an interrupt/polled model. Fix get_lun_statep so we don't panic if there are no wildcard luns enabled. MFC after: 6 days
2004-02-08 19:17:56 +00:00
isp->isp_osinfo.leact[seq] = 0;
return (-1);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
}
} else {
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
tptr = get_lun_statep(isp, bus, lun);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
if (tptr == NULL) {
ccb->ccb_h.status = CAM_LUN_INVALID;
Remove condition variables and status associated with target mode enabling. Instead, go to an interrupt/polled model. Fix get_lun_statep so we don't panic if there are no wildcard luns enabled. MFC after: 6 days
2004-02-08 19:17:56 +00:00
return (-1);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
}
}
if (cel->enable) {
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
int c, n, ulun = lun;
cmd = RQSTYPE_ENABLE_LUN;
c = DFLT_CMND_CNT;
n = DFLT_INOT_CNT;
if (IS_FC(isp) && lun != 0) {
cmd = RQSTYPE_MODIFY_LUN;
n = 0;
/*
* For SCC firmware, we only deal with setting
* (enabling or modifying) lun 0.
*/
ulun = 0;
}
Remove condition variables and status associated with target mode enabling. Instead, go to an interrupt/polled model. Fix get_lun_statep so we don't panic if there are no wildcard luns enabled. MFC after: 6 days
2004-02-08 19:17:56 +00:00
if (isp_lun_cmd(isp, cmd, bus, tgt, ulun, c, n, seq+1) == 0) {
rls_lun_statep(isp, tptr);
ccb->ccb_h.status = CAM_REQ_INPROG;
return (seq);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
}
} else {
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
int c, n, ulun = lun;
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
cmd = -RQSTYPE_MODIFY_LUN;
c = DFLT_CMND_CNT;
n = DFLT_INOT_CNT;
if (IS_FC(isp) && lun != 0) {
n = 0;
/*
* For SCC firmware, we only deal with setting
* (enabling or modifying) lun 0.
*/
ulun = 0;
}
Remove condition variables and status associated with target mode enabling. Instead, go to an interrupt/polled model. Fix get_lun_statep so we don't panic if there are no wildcard luns enabled. MFC after: 6 days
2004-02-08 19:17:56 +00:00
if (isp_lun_cmd(isp, cmd, bus, tgt, ulun, c, n, seq+1) == 0) {
rls_lun_statep(isp, tptr);
ccb->ccb_h.status = CAM_REQ_INPROG;
return (seq);
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
}
Remove condition variables and status associated with target mode enabling. Instead, go to an interrupt/polled model. Fix get_lun_statep so we don't panic if there are no wildcard luns enabled. MFC after: 6 days
2004-02-08 19:17:56 +00:00
}
rls_lun_statep(isp, tptr);
xpt_print_path(ccb->ccb_h.path);
printf("isp_lun_cmd failed\n");
isp->isp_osinfo.leact[seq] = 0;
ccb->ccb_h.status = CAM_REQ_CMP_ERR;
return (-1);
}
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
Remove condition variables and status associated with target mode enabling. Instead, go to an interrupt/polled model. Fix get_lun_statep so we don't panic if there are no wildcard luns enabled. MFC after: 6 days
2004-02-08 19:17:56 +00:00
static void
isp_ledone(struct ispsoftc *isp, lun_entry_t *lep)
{
const char lfmt[] = "lun %d now %sabled for target mode on channel %d";
union ccb *ccb;
a) clean up some declaration stuff (i.e., make more modern with respect to getting rid u_int for uint and so on). b) Turn back on 64 bit DAC support. Cheeze it a bit in that we have two DMA callback functions- one when we have bus_addr_t > 4 bits in width and the other which should be normal. Even Cheezier in that we turn off setting up DMA maps to be BUS_SPACE_MAXADDR if we're in ISP_TARGET_MODE. More work on this in a week or so. c) Tested under amd64 and 1MB DFLTPHYS, sparc64, i386 (PAE, but insufficient memory to really test > 4GB). LINT check under amd64. MFC after: 1 month
2006-02-15 00:31:48 +00:00
uint32_t seq;
Remove condition variables and status associated with target mode enabling. Instead, go to an interrupt/polled model. Fix get_lun_statep so we don't panic if there are no wildcard luns enabled. MFC after: 6 days
2004-02-08 19:17:56 +00:00
tstate_t *tptr;
int av;
struct ccb_en_lun *cel;
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
Remove condition variables and status associated with target mode enabling. Instead, go to an interrupt/polled model. Fix get_lun_statep so we don't panic if there are no wildcard luns enabled. MFC after: 6 days
2004-02-08 19:17:56 +00:00
seq = lep->le_reserved - 1;
if (seq >= NLEACT) {
isp_prt(isp, ISP_LOGERR,
"seq out of range (%u) in isp_ledone", seq);
return;
}
ccb = isp->isp_osinfo.leact[seq];
if (ccb == 0) {
isp_prt(isp, ISP_LOGERR,
"no ccb for seq %u in isp_ledone", seq);
return;
}
cel = &ccb->cel;
tptr = get_lun_statep(isp, XS_CHANNEL(ccb), XS_LUN(ccb));
if (tptr == NULL) {
xpt_print_path(ccb->ccb_h.path);
printf("null tptr in isp_ledone\n");
isp->isp_osinfo.leact[seq] = 0;
return;
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
}
Add 2 Gigabit Fibre Channel support (2300 && 2312 cards). This required some reworking (and consequent cleanup) of the interrupt service code. Also begin to start a cleanup of target mode support that will (eventually) not require more inforamtion routed with the ATIO to come back with the CTIO other than tag. MFC after: 4 weeks
2001-08-31 21:39:04 +00:00
Remove condition variables and status associated with target mode enabling. Instead, go to an interrupt/polled model. Fix get_lun_statep so we don't panic if there are no wildcard luns enabled. MFC after: 6 days
2004-02-08 19:17:56 +00:00
if (lep->le_status != LUN_OK) {
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
xpt_print_path(ccb->ccb_h.path);
Remove condition variables and status associated with target mode enabling. Instead, go to an interrupt/polled model. Fix get_lun_statep so we don't panic if there are no wildcard luns enabled. MFC after: 6 days
2004-02-08 19:17:56 +00:00
printf("ENABLE/MODIFY LUN returned 0x%x\n", lep->le_status);
err:
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
ccb->ccb_h.status = CAM_REQ_CMP_ERR;
Remove condition variables and status associated with target mode enabling. Instead, go to an interrupt/polled model. Fix get_lun_statep so we don't panic if there are no wildcard luns enabled. MFC after: 6 days
2004-02-08 19:17:56 +00:00
xpt_print_path(ccb->ccb_h.path);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
rls_lun_statep(isp, tptr);
Remove condition variables and status associated with target mode enabling. Instead, go to an interrupt/polled model. Fix get_lun_statep so we don't panic if there are no wildcard luns enabled. MFC after: 6 days
2004-02-08 19:17:56 +00:00
isp->isp_osinfo.leact[seq] = 0;
ISPLOCK_2_CAMLOCK(isp);
xpt_done(ccb);
CAMLOCK_2_ISPLOCK(isp);
return;
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
} else {
Remove condition variables and status associated with target mode enabling. Instead, go to an interrupt/polled model. Fix get_lun_statep so we don't panic if there are no wildcard luns enabled. MFC after: 6 days
2004-02-08 19:17:56 +00:00
isp_prt(isp, ISP_LOGTDEBUG0,
"isp_ledone: ENABLE/MODIFY done okay");
}
if (cel->enable) {
ccb->ccb_h.status = CAM_REQ_CMP;
isp_prt(isp, /* ISP_LOGINFO */ ISP_LOGALL, lfmt,
XS_LUN(ccb), "en", XS_CHANNEL(ccb));
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
rls_lun_statep(isp, tptr);
Remove condition variables and status associated with target mode enabling. Instead, go to an interrupt/polled model. Fix get_lun_statep so we don't panic if there are no wildcard luns enabled. MFC after: 6 days
2004-02-08 19:17:56 +00:00
isp->isp_osinfo.leact[seq] = 0;
ISPLOCK_2_CAMLOCK(isp);
xpt_done(ccb);
CAMLOCK_2_ISPLOCK(isp);
return;
}
if (lep->le_header.rqs_entry_type == RQSTYPE_MODIFY_LUN) {
if (isp_lun_cmd(isp, -RQSTYPE_ENABLE_LUN, XS_CHANNEL(ccb),
XS_TGT(ccb), XS_LUN(ccb), 0, 0, seq+1)) {
xpt_print_path(ccb->ccb_h.path);
printf("isp_ledone: isp_lun_cmd failed\n");
goto err;
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
}
Remove condition variables and status associated with target mode enabling. Instead, go to an interrupt/polled model. Fix get_lun_statep so we don't panic if there are no wildcard luns enabled. MFC after: 6 days
2004-02-08 19:17:56 +00:00
rls_lun_statep(isp, tptr);
return;
}
isp_prt(isp, ISP_LOGINFO, lfmt, XS_LUN(ccb), "dis", XS_CHANNEL(ccb));
rls_lun_statep(isp, tptr);
destroy_lun_state(isp, tptr);
ccb->ccb_h.status = CAM_REQ_CMP;
isp->isp_osinfo.leact[seq] = 0;
ISPLOCK_2_CAMLOCK(isp);
xpt_done(ccb);
CAMLOCK_2_ISPLOCK(isp);
if (are_any_luns_enabled(isp, XS_CHANNEL(ccb)) == 0) {
int bus = XS_CHANNEL(ccb);
av = bus << 31;
av = isp_control(isp, ISPCTL_TOGGLE_TMODE, &av);
if (av) {
isp_prt(isp, ISP_LOGWARN,
"disable target mode on channel %d failed", bus);
} else {
isp_prt(isp, ISP_LOGINFO,
"Target Mode disabled on channel %d", bus);
}
isp->isp_osinfo.tmflags[bus] &= ~TM_TMODE_ENABLED;
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
}
}
Remove condition variables and status associated with target mode enabling. Instead, go to an interrupt/polled model. Fix get_lun_statep so we don't panic if there are no wildcard luns enabled. MFC after: 6 days
2004-02-08 19:17:56 +00:00
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
static cam_status
isp_abort_tgt_ccb(struct ispsoftc *isp, union ccb *ccb)
{
tstate_t *tptr;
struct ccb_hdr_slist *lp;
struct ccb_hdr *curelm;
Checkpoint of work in progress in cleaning up target mode. It actually seems to work well in RELENG_4. However, 5.X locking foo means that I'll have to do some quick redesign. Add ioctl handlers for ISP_GETROLE and ISP_SETROLE ioctls.
2004-02-07 03:47:33 +00:00
int found, *ctr;
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
union ccb *accb = ccb->cab.abort_ccb;
Checkpoint of work in progress in cleaning up target mode. It actually seems to work well in RELENG_4. However, 5.X locking foo means that I'll have to do some quick redesign. Add ioctl handlers for ISP_GETROLE and ISP_SETROLE ioctls.
2004-02-07 03:47:33 +00:00
isp_prt(isp, ISP_LOGTDEBUG0, "aborting ccb %p", accb);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
if (accb->ccb_h.target_id != CAM_TARGET_WILDCARD) {
Checkpoint of work in progress in cleaning up target mode. It actually seems to work well in RELENG_4. However, 5.X locking foo means that I'll have to do some quick redesign. Add ioctl handlers for ISP_GETROLE and ISP_SETROLE ioctls.
2004-02-07 03:47:33 +00:00
int badpath = 0;
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
if (IS_FC(isp) && (accb->ccb_h.target_id !=
((fcparam *) isp->isp_param)->isp_loopid)) {
Checkpoint of work in progress in cleaning up target mode. It actually seems to work well in RELENG_4. However, 5.X locking foo means that I'll have to do some quick redesign. Add ioctl handlers for ISP_GETROLE and ISP_SETROLE ioctls.
2004-02-07 03:47:33 +00:00
badpath = 1;
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
} else if (IS_SCSI(isp) && (accb->ccb_h.target_id !=
((sdparam *) isp->isp_param)->isp_initiator_id)) {
Checkpoint of work in progress in cleaning up target mode. It actually seems to work well in RELENG_4. However, 5.X locking foo means that I'll have to do some quick redesign. Add ioctl handlers for ISP_GETROLE and ISP_SETROLE ioctls.
2004-02-07 03:47:33 +00:00
badpath = 1;
}
if (badpath) {
/*
* Being restrictive about target ids is really about
* making sure we're aborting for the right multi-tid
* path. This doesn't really make much sense at present.
*/
#if 0
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
return (CAM_PATH_INVALID);
Checkpoint of work in progress in cleaning up target mode. It actually seems to work well in RELENG_4. However, 5.X locking foo means that I'll have to do some quick redesign. Add ioctl handlers for ISP_GETROLE and ISP_SETROLE ioctls.
2004-02-07 03:47:33 +00:00
#endif
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
}
}
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
tptr = get_lun_statep(isp, XS_CHANNEL(ccb), accb->ccb_h.target_lun);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
if (tptr == NULL) {
Checkpoint of work in progress in cleaning up target mode. It actually seems to work well in RELENG_4. However, 5.X locking foo means that I'll have to do some quick redesign. Add ioctl handlers for ISP_GETROLE and ISP_SETROLE ioctls.
2004-02-07 03:47:33 +00:00
isp_prt(isp, ISP_LOGTDEBUG0,
"isp_abort_tgt_ccb: can't get statep");
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
return (CAM_PATH_INVALID);
}
if (accb->ccb_h.func_code == XPT_ACCEPT_TARGET_IO) {
lp = &tptr->atios;
Checkpoint of work in progress in cleaning up target mode. It actually seems to work well in RELENG_4. However, 5.X locking foo means that I'll have to do some quick redesign. Add ioctl handlers for ISP_GETROLE and ISP_SETROLE ioctls.
2004-02-07 03:47:33 +00:00
ctr = &tptr->atio_count;
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
} else if (accb->ccb_h.func_code == XPT_IMMED_NOTIFY) {
lp = &tptr->inots;
Checkpoint of work in progress in cleaning up target mode. It actually seems to work well in RELENG_4. However, 5.X locking foo means that I'll have to do some quick redesign. Add ioctl handlers for ISP_GETROLE and ISP_SETROLE ioctls.
2004-02-07 03:47:33 +00:00
ctr = &tptr->inot_count;
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
} else {
rls_lun_statep(isp, tptr);
Checkpoint of work in progress in cleaning up target mode. It actually seems to work well in RELENG_4. However, 5.X locking foo means that I'll have to do some quick redesign. Add ioctl handlers for ISP_GETROLE and ISP_SETROLE ioctls.
2004-02-07 03:47:33 +00:00
isp_prt(isp, ISP_LOGTDEBUG0,
"isp_abort_tgt_ccb: bad func %d\n", accb->ccb_h.func_code);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
return (CAM_UA_ABORT);
}
curelm = SLIST_FIRST(lp);
found = 0;
if (curelm == &accb->ccb_h) {
found = 1;
SLIST_REMOVE_HEAD(lp, sim_links.sle);
} else {
while(curelm != NULL) {
struct ccb_hdr *nextelm;
nextelm = SLIST_NEXT(curelm, sim_links.sle);
if (nextelm == &accb->ccb_h) {
found = 1;
SLIST_NEXT(curelm, sim_links.sle) =
SLIST_NEXT(nextelm, sim_links.sle);
break;
}
curelm = nextelm;
}
}
rls_lun_statep(isp, tptr);
if (found) {
Checkpoint of work in progress in cleaning up target mode. It actually seems to work well in RELENG_4. However, 5.X locking foo means that I'll have to do some quick redesign. Add ioctl handlers for ISP_GETROLE and ISP_SETROLE ioctls.
2004-02-07 03:47:33 +00:00
*ctr--;
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
accb->ccb_h.status = CAM_REQ_ABORTED;
Checkpoint of work in progress in cleaning up target mode. It actually seems to work well in RELENG_4. However, 5.X locking foo means that I'll have to do some quick redesign. Add ioctl handlers for ISP_GETROLE and ISP_SETROLE ioctls.
2004-02-07 03:47:33 +00:00
xpt_done(accb);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
return (CAM_REQ_CMP);
}
Checkpoint of work in progress in cleaning up target mode. It actually seems to work well in RELENG_4. However, 5.X locking foo means that I'll have to do some quick redesign. Add ioctl handlers for ISP_GETROLE and ISP_SETROLE ioctls.
2004-02-07 03:47:33 +00:00
isp_prt(isp, ISP_LOGTDEBUG0,
"isp_abort_tgt_ccb: CCB %p not found\n", ccb);
return (CAM_PATH_INVALID);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
}
static cam_status
isp_target_start_ctio(struct ispsoftc *isp, union ccb *ccb)
{
void *qe;
a whale of a lot of target mode cleanup
2000-01-15 01:50:48 +00:00
struct ccb_scsiio *cso = &ccb->csio;
a) clean up some declaration stuff (i.e., make more modern with respect to getting rid u_int for uint and so on). b) Turn back on 64 bit DAC support. Cheeze it a bit in that we have two DMA callback functions- one when we have bus_addr_t > 4 bits in width and the other which should be normal. Even Cheezier in that we turn off setting up DMA maps to be BUS_SPACE_MAXADDR if we're in ISP_TARGET_MODE. More work on this in a week or so. c) Tested under amd64 and 1MB DFLTPHYS, sparc64, i386 (PAE, but insufficient memory to really test > 4GB). LINT check under amd64. MFC after: 1 month
2006-02-15 00:31:48 +00:00
uint16_t *hp, save_handle;
uint16_t nxti, optr;
uint8_t local[QENTRY_LEN];
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
Add a isp_target_putback_atio- we aren't using CCINCR at this time, so we need a function that tells the Qlogic f/w that a target mode command is done, so increase the resource count for that lun. Add in a timeout function to kick the putback again if we fail to do it the first time (we may not have the request queue space for ATIO push). Split the function isp_handle_platform_ctio into two parts so that the timeout function for the ATIO push or isp_handle_platform_ctio can inform CAM that the requested CTIO(s) are now done. Clean up (cough) residual handling. What we need for Fibre Channel is to preserve the at_datalen field from the original incoming ATIO so we can calculate a 'true' residual. Unfortunately, we're not guaranteed to get that back from CAM. We'll *try* to find it hiding in the periph_priv field (layering violation)- but if an ATIO was passed in from user land- forget it. This means that we'll probably get residuals wrong for Fibre Channel commands we're completing with an error. It's too late to 4.1 release to fix this- too bad. Luckily the only device we'd really care about this occurring on is a tape device and they're still so rare as FC attached devices that this can be considered an untested combination anyway. Remove all CCINCR usage (resource autoreplenish). When we've proved to ourself that things are working properly, we can add it back in. Make sure we propage 'suggested' sense data from the incoming ATIO into the created system ATIO- and set sense_len appropriately. Correctly propagate tag values. Fall back to the model of generating (well, the functions in isp_pci.c do the work) multiple CTIOs based upon what we get from XPT. Instead of being able to pair Qlogic generated ATIOs with CAM ATIOs, and then to pair CAM CTIOs with Qlogic CTIOs, we have to take the CTIO passed to us from XPT, and if it implies that we have to generate extra Qlogic CTIOs, so be it. This means that we have to wait until the last CTIO in a sequence we generated completes before calling xpt_done. Executive summary- target mode actually now pretty much works well enough to tell folks about.
2000-07-18 06:58:28 +00:00
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). It also has to handle the SBus cards (for platforms that have them) which, while on a Big Endian machine, do *not* require *most* of the request/response queue entry fields to be swizzled or unswizzled. 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. 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). MFC after: 2 weeks
2001-12-11 00:18:45 +00:00
if (isp_getrqentry(isp, &nxti, &optr, &qe)) {
Clean up some debug printing. Find the correct lun when SCCLUN is defined. If we complete with a check condition but no sense data, say we had an AUTOSENSE failure.
2000-01-04 03:45:49 +00:00
xpt_print_path(ccb->ccb_h.path);
printf("Request Queue Overflow in isp_target_start_ctio\n");
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
return (CAM_RESRC_UNAVAIL);
}
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). It also has to handle the SBus cards (for platforms that have them) which, while on a Big Endian machine, do *not* require *most* of the request/response queue entry fields to be swizzled or unswizzled. 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. 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). MFC after: 2 weeks
2001-12-11 00:18:45 +00:00
bzero(local, QENTRY_LEN);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
/*
* We're either moving data or completing a command here.
*/
if (IS_FC(isp)) {
Begin to implement target mode that for Fibre Channel has a private per-command component that we *don't* try and pass thru CAM. CAM just is too risky and too much of a pain- structures get copied, but not all info of interest can be considered safely transported thru all consumers (including user space) from the incoming ATIO to the outgoing CTIO- it's just much safer to have a buddy structure, identified by the command's tag which *does* make it thru safely. Pay attention to link speed and report 200MB/s xfer speed for a 23XX card in 2GPs mode. MFC after: 1 week
2001-10-01 03:48:42 +00:00
atio_private_data_t *atp;
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). It also has to handle the SBus cards (for platforms that have them) which, while on a Big Endian machine, do *not* require *most* of the request/response queue entry fields to be swizzled or unswizzled. 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. 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). MFC after: 2 weeks
2001-12-11 00:18:45 +00:00
ct2_entry_t *cto = (ct2_entry_t *) local;
a whale of a lot of target mode cleanup
2000-01-15 01:50:48 +00:00
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
cto->ct_header.rqs_entry_type = RQSTYPE_CTIO2;
cto->ct_header.rqs_entry_count = 1;
a whale of a lot of target mode cleanup
2000-01-15 01:50:48 +00:00
cto->ct_iid = cso->init_id;
I don't know what I was thinking- if I have two separate busses on on SIM (as is true for the 1280 and the 12160), then I have to have separate flags && status for *both* busses. *Whap*. Implement condition variables for coordination with some target mode events. It's nice to use these and not panic in obscure little places in the kernel like 'propagate_priority' just because we went to sleep holding a mutex, or some other absurd thing. Remove some bogus ISP_UNLOCK calls. *Whap*. No longer require that somebody do a lun enable on the wildcard device to enable target mode. They are, in fact, orthogonal. A wildcard open is a statement that somebody upstream is willing to accept commands which are otherwise unrouteable. Now, for QLogic regular SCSI target mode, this won't matter for a damn because we'll never see ATIOs for luns we haven't enabled (are listening for, if you will). But for SCCLUN fibre channel SCSI, we get all kinds of ATIOs. We can either reflect them back here with minimal info (which is isp_target.c:isp_endcmd() is for), or the wildcard device (nominally targbh) can handle them. Do further checking against firmware attributes to see whether we can, in fact, support target mode in Fibre Channel. For now, require SCCLUN f/w to supoprt FC target mode. This is an awful lot of change, but target mode *still* isn't quite right. MFC after: 4 weeks
2001-09-04 21:53:12 +00:00
if ((FCPARAM(isp)->isp_fwattr & ISP_FW_ATTR_SCCLUN) == 0) {
Remove all ISP2100_SCCLUN define protected code and replace it with runtime checks.
2000-06-18 04:50:26 +00:00
cto->ct_lun = ccb->ccb_h.target_lun;
}
Begin to implement target mode that for Fibre Channel has a private per-command component that we *don't* try and pass thru CAM. CAM just is too risky and too much of a pain- structures get copied, but not all info of interest can be considered safely transported thru all consumers (including user space) from the incoming ATIO to the outgoing CTIO- it's just much safer to have a buddy structure, identified by the command's tag which *does* make it thru safely. Pay attention to link speed and report 200MB/s xfer speed for a 23XX card in 2GPs mode. MFC after: 1 week
2001-10-01 03:48:42 +00:00
atp = isp_get_atpd(isp, cso->tag_id);
if (atp == NULL) {
Add support for ISP_FC_GETHINFO, which returns current connection topology, speed, loopid, WWPN/WWNN, etc. Beef up target mode. Add isp_handle_platform_notify_scsi and isp_handle_platform_notify_fc platform handlers to handle immediate notifies (isp_handle_platform_notify_scsi is still stubbed out). In implementation of isp_handle_platform_notify_fc, for IN_ABORT_TASK, peel off a pending XPT_IMMED_NOTIFY and call xpt_done on it and hope that somebody upstream is listening. Make sure on final CTIO2s that we set residual correctly. These are absolutely crucial. Make sure we set relative offset for each CTIO2 based upon bytes we've already xferred. This is what the private adjunct datat to the original ATIO is. Note state of command so we can figure out where to find it if we get an ABORT from the firmware. Make sure we *always* set CAM_TAG_ACTION_VALID for ATIO2s. Make sure we keep track of the original lun. If se sent status (or we're otherwise done with the command), don't forget to free the adjunct structure.
2002-06-16 05:08:02 +00:00
isp_prt(isp, ISP_LOGERR,
"cannot find private data adjunct for tag %x",
Begin to implement target mode that for Fibre Channel has a private per-command component that we *don't* try and pass thru CAM. CAM just is too risky and too much of a pain- structures get copied, but not all info of interest can be considered safely transported thru all consumers (including user space) from the incoming ATIO to the outgoing CTIO- it's just much safer to have a buddy structure, identified by the command's tag which *does* make it thru safely. Pay attention to link speed and report 200MB/s xfer speed for a 23XX card in 2GPs mode. MFC after: 1 week
2001-10-01 03:48:42 +00:00
cso->tag_id);
Add support for ISP_FC_GETHINFO, which returns current connection topology, speed, loopid, WWPN/WWNN, etc. Beef up target mode. Add isp_handle_platform_notify_scsi and isp_handle_platform_notify_fc platform handlers to handle immediate notifies (isp_handle_platform_notify_scsi is still stubbed out). In implementation of isp_handle_platform_notify_fc, for IN_ABORT_TASK, peel off a pending XPT_IMMED_NOTIFY and call xpt_done on it and hope that somebody upstream is listening. Make sure on final CTIO2s that we set residual correctly. These are absolutely crucial. Make sure we set relative offset for each CTIO2 based upon bytes we've already xferred. This is what the private adjunct datat to the original ATIO is. Note state of command so we can figure out where to find it if we get an ABORT from the firmware. Make sure we *always* set CAM_TAG_ACTION_VALID for ATIO2s. Make sure we keep track of the original lun. If se sent status (or we're otherwise done with the command), don't forget to free the adjunct structure.
2002-06-16 05:08:02 +00:00
return (-1);
Begin to implement target mode that for Fibre Channel has a private per-command component that we *don't* try and pass thru CAM. CAM just is too risky and too much of a pain- structures get copied, but not all info of interest can be considered safely transported thru all consumers (including user space) from the incoming ATIO to the outgoing CTIO- it's just much safer to have a buddy structure, identified by the command's tag which *does* make it thru safely. Pay attention to link speed and report 200MB/s xfer speed for a 23XX card in 2GPs mode. MFC after: 1 week
2001-10-01 03:48:42 +00:00
}
Add a isp_target_putback_atio- we aren't using CCINCR at this time, so we need a function that tells the Qlogic f/w that a target mode command is done, so increase the resource count for that lun. Add in a timeout function to kick the putback again if we fail to do it the first time (we may not have the request queue space for ATIO push). Split the function isp_handle_platform_ctio into two parts so that the timeout function for the ATIO push or isp_handle_platform_ctio can inform CAM that the requested CTIO(s) are now done. Clean up (cough) residual handling. What we need for Fibre Channel is to preserve the at_datalen field from the original incoming ATIO so we can calculate a 'true' residual. Unfortunately, we're not guaranteed to get that back from CAM. We'll *try* to find it hiding in the periph_priv field (layering violation)- but if an ATIO was passed in from user land- forget it. This means that we'll probably get residuals wrong for Fibre Channel commands we're completing with an error. It's too late to 4.1 release to fix this- too bad. Luckily the only device we'd really care about this occurring on is a tape device and they're still so rare as FC attached devices that this can be considered an untested combination anyway. Remove all CCINCR usage (resource autoreplenish). When we've proved to ourself that things are working properly, we can add it back in. Make sure we propage 'suggested' sense data from the incoming ATIO into the created system ATIO- and set sense_len appropriately. Correctly propagate tag values. Fall back to the model of generating (well, the functions in isp_pci.c do the work) multiple CTIOs based upon what we get from XPT. Instead of being able to pair Qlogic generated ATIOs with CAM ATIOs, and then to pair CAM CTIOs with Qlogic CTIOs, we have to take the CTIO passed to us from XPT, and if it implies that we have to generate extra Qlogic CTIOs, so be it. This means that we have to wait until the last CTIO in a sequence we generated completes before calling xpt_done. Executive summary- target mode actually now pretty much works well enough to tell folks about.
2000-07-18 06:58:28 +00:00
a whale of a lot of target mode cleanup
2000-01-15 01:50:48 +00:00
cto->ct_rxid = cso->tag_id;
if (cso->dxfer_len == 0) {
cto->ct_flags |= CT2_FLAG_MODE1 | CT2_NO_DATA;
Add a isp_target_putback_atio- we aren't using CCINCR at this time, so we need a function that tells the Qlogic f/w that a target mode command is done, so increase the resource count for that lun. Add in a timeout function to kick the putback again if we fail to do it the first time (we may not have the request queue space for ATIO push). Split the function isp_handle_platform_ctio into two parts so that the timeout function for the ATIO push or isp_handle_platform_ctio can inform CAM that the requested CTIO(s) are now done. Clean up (cough) residual handling. What we need for Fibre Channel is to preserve the at_datalen field from the original incoming ATIO so we can calculate a 'true' residual. Unfortunately, we're not guaranteed to get that back from CAM. We'll *try* to find it hiding in the periph_priv field (layering violation)- but if an ATIO was passed in from user land- forget it. This means that we'll probably get residuals wrong for Fibre Channel commands we're completing with an error. It's too late to 4.1 release to fix this- too bad. Luckily the only device we'd really care about this occurring on is a tape device and they're still so rare as FC attached devices that this can be considered an untested combination anyway. Remove all CCINCR usage (resource autoreplenish). When we've proved to ourself that things are working properly, we can add it back in. Make sure we propage 'suggested' sense data from the incoming ATIO into the created system ATIO- and set sense_len appropriately. Correctly propagate tag values. Fall back to the model of generating (well, the functions in isp_pci.c do the work) multiple CTIOs based upon what we get from XPT. Instead of being able to pair Qlogic generated ATIOs with CAM ATIOs, and then to pair CAM CTIOs with Qlogic CTIOs, we have to take the CTIO passed to us from XPT, and if it implies that we have to generate extra Qlogic CTIOs, so be it. This means that we have to wait until the last CTIO in a sequence we generated completes before calling xpt_done. Executive summary- target mode actually now pretty much works well enough to tell folks about.
2000-07-18 06:58:28 +00:00
if (ccb->ccb_h.flags & CAM_SEND_STATUS) {
cto->ct_flags |= CT2_SENDSTATUS;
cto->rsp.m1.ct_scsi_status = cso->scsi_status;
Begin to implement target mode that for Fibre Channel has a private per-command component that we *don't* try and pass thru CAM. CAM just is too risky and too much of a pain- structures get copied, but not all info of interest can be considered safely transported thru all consumers (including user space) from the incoming ATIO to the outgoing CTIO- it's just much safer to have a buddy structure, identified by the command's tag which *does* make it thru safely. Pay attention to link speed and report 200MB/s xfer speed for a 23XX card in 2GPs mode. MFC after: 1 week
2001-10-01 03:48:42 +00:00
cto->ct_resid =
atp->orig_datalen - atp->bytes_xfered;
Add support for ISP_FC_GETHINFO, which returns current connection topology, speed, loopid, WWPN/WWNN, etc. Beef up target mode. Add isp_handle_platform_notify_scsi and isp_handle_platform_notify_fc platform handlers to handle immediate notifies (isp_handle_platform_notify_scsi is still stubbed out). In implementation of isp_handle_platform_notify_fc, for IN_ABORT_TASK, peel off a pending XPT_IMMED_NOTIFY and call xpt_done on it and hope that somebody upstream is listening. Make sure on final CTIO2s that we set residual correctly. These are absolutely crucial. Make sure we set relative offset for each CTIO2 based upon bytes we've already xferred. This is what the private adjunct datat to the original ATIO is. Note state of command so we can figure out where to find it if we get an ABORT from the firmware. Make sure we *always* set CAM_TAG_ACTION_VALID for ATIO2s. Make sure we keep track of the original lun. If se sent status (or we're otherwise done with the command), don't forget to free the adjunct structure.
2002-06-16 05:08:02 +00:00
if (cto->ct_resid < 0) {
cto->rsp.m1.ct_scsi_status |=
CT2_DATA_OVER;
} else if (cto->ct_resid > 0) {
cto->rsp.m1.ct_scsi_status |=
CT2_DATA_UNDER;
}
Add a isp_target_putback_atio- we aren't using CCINCR at this time, so we need a function that tells the Qlogic f/w that a target mode command is done, so increase the resource count for that lun. Add in a timeout function to kick the putback again if we fail to do it the first time (we may not have the request queue space for ATIO push). Split the function isp_handle_platform_ctio into two parts so that the timeout function for the ATIO push or isp_handle_platform_ctio can inform CAM that the requested CTIO(s) are now done. Clean up (cough) residual handling. What we need for Fibre Channel is to preserve the at_datalen field from the original incoming ATIO so we can calculate a 'true' residual. Unfortunately, we're not guaranteed to get that back from CAM. We'll *try* to find it hiding in the periph_priv field (layering violation)- but if an ATIO was passed in from user land- forget it. This means that we'll probably get residuals wrong for Fibre Channel commands we're completing with an error. It's too late to 4.1 release to fix this- too bad. Luckily the only device we'd really care about this occurring on is a tape device and they're still so rare as FC attached devices that this can be considered an untested combination anyway. Remove all CCINCR usage (resource autoreplenish). When we've proved to ourself that things are working properly, we can add it back in. Make sure we propage 'suggested' sense data from the incoming ATIO into the created system ATIO- and set sense_len appropriately. Correctly propagate tag values. Fall back to the model of generating (well, the functions in isp_pci.c do the work) multiple CTIOs based upon what we get from XPT. Instead of being able to pair Qlogic generated ATIOs with CAM ATIOs, and then to pair CAM CTIOs with Qlogic CTIOs, we have to take the CTIO passed to us from XPT, and if it implies that we have to generate extra Qlogic CTIOs, so be it. This means that we have to wait until the last CTIO in a sequence we generated completes before calling xpt_done. Executive summary- target mode actually now pretty much works well enough to tell folks about.
2000-07-18 06:58:28 +00:00
}
a whale of a lot of target mode cleanup
2000-01-15 01:50:48 +00:00
if ((ccb->ccb_h.flags & CAM_SEND_SENSE) != 0) {
int m = min(cso->sense_len, MAXRESPLEN);
bcopy(&cso->sense_data, cto->rsp.m1.ct_resp, m);
cto->rsp.m1.ct_senselen = m;
cto->rsp.m1.ct_scsi_status |= CT2_SNSLEN_VALID;
}
} else {
cto->ct_flags |= CT2_FLAG_MODE0;
if ((cso->ccb_h.flags & CAM_DIR_MASK) == CAM_DIR_IN) {
cto->ct_flags |= CT2_DATA_IN;
} else {
cto->ct_flags |= CT2_DATA_OUT;
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
}
Add support for ISP_FC_GETHINFO, which returns current connection topology, speed, loopid, WWPN/WWNN, etc. Beef up target mode. Add isp_handle_platform_notify_scsi and isp_handle_platform_notify_fc platform handlers to handle immediate notifies (isp_handle_platform_notify_scsi is still stubbed out). In implementation of isp_handle_platform_notify_fc, for IN_ABORT_TASK, peel off a pending XPT_IMMED_NOTIFY and call xpt_done on it and hope that somebody upstream is listening. Make sure on final CTIO2s that we set residual correctly. These are absolutely crucial. Make sure we set relative offset for each CTIO2 based upon bytes we've already xferred. This is what the private adjunct datat to the original ATIO is. Note state of command so we can figure out where to find it if we get an ABORT from the firmware. Make sure we *always* set CAM_TAG_ACTION_VALID for ATIO2s. Make sure we keep track of the original lun. If se sent status (or we're otherwise done with the command), don't forget to free the adjunct structure.
2002-06-16 05:08:02 +00:00
cto->ct_reloff = atp->bytes_xfered;
a whale of a lot of target mode cleanup
2000-01-15 01:50:48 +00:00
if ((ccb->ccb_h.flags & CAM_SEND_STATUS) != 0) {
cto->ct_flags |= CT2_SENDSTATUS;
cto->rsp.m0.ct_scsi_status = cso->scsi_status;
Begin to implement target mode that for Fibre Channel has a private per-command component that we *don't* try and pass thru CAM. CAM just is too risky and too much of a pain- structures get copied, but not all info of interest can be considered safely transported thru all consumers (including user space) from the incoming ATIO to the outgoing CTIO- it's just much safer to have a buddy structure, identified by the command's tag which *does* make it thru safely. Pay attention to link speed and report 200MB/s xfer speed for a 23XX card in 2GPs mode. MFC after: 1 week
2001-10-01 03:48:42 +00:00
cto->ct_resid =
atp->orig_datalen -
(atp->bytes_xfered + cso->dxfer_len);
Add support for ISP_FC_GETHINFO, which returns current connection topology, speed, loopid, WWPN/WWNN, etc. Beef up target mode. Add isp_handle_platform_notify_scsi and isp_handle_platform_notify_fc platform handlers to handle immediate notifies (isp_handle_platform_notify_scsi is still stubbed out). In implementation of isp_handle_platform_notify_fc, for IN_ABORT_TASK, peel off a pending XPT_IMMED_NOTIFY and call xpt_done on it and hope that somebody upstream is listening. Make sure on final CTIO2s that we set residual correctly. These are absolutely crucial. Make sure we set relative offset for each CTIO2 based upon bytes we've already xferred. This is what the private adjunct datat to the original ATIO is. Note state of command so we can figure out where to find it if we get an ABORT from the firmware. Make sure we *always* set CAM_TAG_ACTION_VALID for ATIO2s. Make sure we keep track of the original lun. If se sent status (or we're otherwise done with the command), don't forget to free the adjunct structure.
2002-06-16 05:08:02 +00:00
if (cto->ct_resid < 0) {
cto->rsp.m0.ct_scsi_status |=
CT2_DATA_OVER;
} else if (cto->ct_resid > 0) {
cto->rsp.m0.ct_scsi_status |=
CT2_DATA_UNDER;
}
Begin to implement target mode that for Fibre Channel has a private per-command component that we *don't* try and pass thru CAM. CAM just is too risky and too much of a pain- structures get copied, but not all info of interest can be considered safely transported thru all consumers (including user space) from the incoming ATIO to the outgoing CTIO- it's just much safer to have a buddy structure, identified by the command's tag which *does* make it thru safely. Pay attention to link speed and report 200MB/s xfer speed for a 23XX card in 2GPs mode. MFC after: 1 week
2001-10-01 03:48:42 +00:00
} else {
atp->last_xframt = cso->dxfer_len;
Clean up some debug printing. Find the correct lun when SCCLUN is defined. If we complete with a check condition but no sense data, say we had an AUTOSENSE failure.
2000-01-04 03:45:49 +00:00
}
Add a isp_target_putback_atio- we aren't using CCINCR at this time, so we need a function that tells the Qlogic f/w that a target mode command is done, so increase the resource count for that lun. Add in a timeout function to kick the putback again if we fail to do it the first time (we may not have the request queue space for ATIO push). Split the function isp_handle_platform_ctio into two parts so that the timeout function for the ATIO push or isp_handle_platform_ctio can inform CAM that the requested CTIO(s) are now done. Clean up (cough) residual handling. What we need for Fibre Channel is to preserve the at_datalen field from the original incoming ATIO so we can calculate a 'true' residual. Unfortunately, we're not guaranteed to get that back from CAM. We'll *try* to find it hiding in the periph_priv field (layering violation)- but if an ATIO was passed in from user land- forget it. This means that we'll probably get residuals wrong for Fibre Channel commands we're completing with an error. It's too late to 4.1 release to fix this- too bad. Luckily the only device we'd really care about this occurring on is a tape device and they're still so rare as FC attached devices that this can be considered an untested combination anyway. Remove all CCINCR usage (resource autoreplenish). When we've proved to ourself that things are working properly, we can add it back in. Make sure we propage 'suggested' sense data from the incoming ATIO into the created system ATIO- and set sense_len appropriately. Correctly propagate tag values. Fall back to the model of generating (well, the functions in isp_pci.c do the work) multiple CTIOs based upon what we get from XPT. Instead of being able to pair Qlogic generated ATIOs with CAM ATIOs, and then to pair CAM CTIOs with Qlogic CTIOs, we have to take the CTIO passed to us from XPT, and if it implies that we have to generate extra Qlogic CTIOs, so be it. This means that we have to wait until the last CTIO in a sequence we generated completes before calling xpt_done. Executive summary- target mode actually now pretty much works well enough to tell folks about.
2000-07-18 06:58:28 +00:00
/*
* If we're sending data and status back together,
* we can't also send back sense data as well.
*/
a whale of a lot of target mode cleanup
2000-01-15 01:50:48 +00:00
ccb->ccb_h.flags &= ~CAM_SEND_SENSE;
}
Begin to implement target mode that for Fibre Channel has a private per-command component that we *don't* try and pass thru CAM. CAM just is too risky and too much of a pain- structures get copied, but not all info of interest can be considered safely transported thru all consumers (including user space) from the incoming ATIO to the outgoing CTIO- it's just much safer to have a buddy structure, identified by the command's tag which *does* make it thru safely. Pay attention to link speed and report 200MB/s xfer speed for a 23XX card in 2GPs mode. MFC after: 1 week
2001-10-01 03:48:42 +00:00
Check CT2_SENDSTATUS/CT_SENDSTATUS against cto->ct_flags, not CAM_SEND_STATUS. Set a timeout of 2 seconds per CTIO. Make sure that the 'real' tag value is being checked against- not the one that also carries the firmware handle.
2001-03-21 00:46:44 +00:00
if (cto->ct_flags & CT2_SENDSTATUS) {
I don't know what I was thinking- if I have two separate busses on on SIM (as is true for the 1280 and the 12160), then I have to have separate flags && status for *both* busses. *Whap*. Implement condition variables for coordination with some target mode events. It's nice to use these and not panic in obscure little places in the kernel like 'propagate_priority' just because we went to sleep holding a mutex, or some other absurd thing. Remove some bogus ISP_UNLOCK calls. *Whap*. No longer require that somebody do a lun enable on the wildcard device to enable target mode. They are, in fact, orthogonal. A wildcard open is a statement that somebody upstream is willing to accept commands which are otherwise unrouteable. Now, for QLogic regular SCSI target mode, this won't matter for a damn because we'll never see ATIOs for luns we haven't enabled (are listening for, if you will). But for SCCLUN fibre channel SCSI, we get all kinds of ATIOs. We can either reflect them back here with minimal info (which is isp_target.c:isp_endcmd() is for), or the wildcard device (nominally targbh) can handle them. Do further checking against firmware attributes to see whether we can, in fact, support target mode in Fibre Channel. For now, require SCCLUN f/w to supoprt FC target mode. This is an awful lot of change, but target mode *still* isn't quite right. MFC after: 4 weeks
2001-09-04 21:53:12 +00:00
isp_prt(isp, ISP_LOGTDEBUG0,
Begin to implement target mode that for Fibre Channel has a private per-command component that we *don't* try and pass thru CAM. CAM just is too risky and too much of a pain- structures get copied, but not all info of interest can be considered safely transported thru all consumers (including user space) from the incoming ATIO to the outgoing CTIO- it's just much safer to have a buddy structure, identified by the command's tag which *does* make it thru safely. Pay attention to link speed and report 200MB/s xfer speed for a 23XX card in 2GPs mode. MFC after: 1 week
2001-10-01 03:48:42 +00:00
"CTIO2[%x] STATUS %x origd %u curd %u resid %u",
cto->ct_rxid, cso->scsi_status, atp->orig_datalen,
cso->dxfer_len, cto->ct_resid);
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
cto->ct_flags |= CT2_CCINCR;
Add support for ISP_FC_GETHINFO, which returns current connection topology, speed, loopid, WWPN/WWNN, etc. Beef up target mode. Add isp_handle_platform_notify_scsi and isp_handle_platform_notify_fc platform handlers to handle immediate notifies (isp_handle_platform_notify_scsi is still stubbed out). In implementation of isp_handle_platform_notify_fc, for IN_ABORT_TASK, peel off a pending XPT_IMMED_NOTIFY and call xpt_done on it and hope that somebody upstream is listening. Make sure on final CTIO2s that we set residual correctly. These are absolutely crucial. Make sure we set relative offset for each CTIO2 based upon bytes we've already xferred. This is what the private adjunct datat to the original ATIO is. Note state of command so we can figure out where to find it if we get an ABORT from the firmware. Make sure we *always* set CAM_TAG_ACTION_VALID for ATIO2s. Make sure we keep track of the original lun. If se sent status (or we're otherwise done with the command), don't forget to free the adjunct structure.
2002-06-16 05:08:02 +00:00
atp->state = ATPD_STATE_LAST_CTIO;
} else
atp->state = ATPD_STATE_CTIO;
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
cto->ct_timeout = 10;
Switch to using 16 bit handles instead of 32 bit handles. This is a pretty invasive change, but there are three good reasons to do this: 1. We'll never have > 16 bits of handle. 2. We can (eventually) enable the RIO (Reduced Interrupt Operation) bits which return multiple completing 16 bit handles in mailbox registers. 3. The !)$*)$*~)@$*~)$* Qlogic target mode for parallel SCSI spec changed such that at_reserved (which was 32 bits) was split into two pieces- and one of which was a 16 bit handle id that functions like the at_rxid for Fibre Channel (a tag for the f/w to correlate CTIOs with a particular command). Since we had to muck with that and this changed the whole handler architecture, we might as well... Propagate new at_handle on through int ct_fwhandle. Follow implications of changing to 16 bit handles. These above changes at least get Qlogic 1040 cards working in target mode again. 1080/12160 cards don't work yet. In isp.c: Prepare for doing all loop management in outer layers.
2001-03-02 06:28:55 +00:00
hp = &cto->ct_syshandle;
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
} else {
Major restructuring for swizzling to the request queue and unswizzling from the response queue. Instead of the ad hoc ISP_SWIZZLE_REQUEST, we now have a complete set of inline functions in isp_inline.h. Each platform is responsible for providing just one of a set of ISP_IOX_{GET,PUT}{8,16,32} macros. The reason this needs to be done is that we need to have a single set of functions that will work correctly on multiple architectures for both little and big endian machines. It also needs to work correctly in the case that we have the request or response queues in memory that has to be treated specially (e.g., have ddi_dma_sync called on it for Solaris after we update it or before we read from it). It also has to handle the SBus cards (for platforms that have them) which, while on a Big Endian machine, do *not* require *most* of the request/response queue entry fields to be swizzled or unswizzled. 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. 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). MFC after: 2 weeks
2001-12-11 00:18:45 +00:00
ct_entry_t *cto = (ct_entry_t *) local;
a whale of a lot of target mode cleanup
2000-01-15 01:50:48 +00:00
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
cto->ct_header.rqs_entry_type = RQSTYPE_CTIO;
cto->ct_header.rqs_entry_count = 1;
a whale of a lot of target mode cleanup
2000-01-15 01:50:48 +00:00
cto->ct_iid = cso->init_id;
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
cto->ct_iid |= XS_CHANNEL(ccb) << 7;
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
cto->ct_tgt = ccb->ccb_h.target_id;
cto->ct_lun = ccb->ccb_h.target_lun;
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
cto->ct_fwhandle = AT_GET_HANDLE(cso->tag_id);
if (AT_HAS_TAG(cso->tag_id)) {
a) clean up some declaration stuff (i.e., make more modern with respect to getting rid u_int for uint and so on). b) Turn back on 64 bit DAC support. Cheeze it a bit in that we have two DMA callback functions- one when we have bus_addr_t > 4 bits in width and the other which should be normal. Even Cheezier in that we turn off setting up DMA maps to be BUS_SPACE_MAXADDR if we're in ISP_TARGET_MODE. More work on this in a week or so. c) Tested under amd64 and 1MB DFLTPHYS, sparc64, i386 (PAE, but insufficient memory to really test > 4GB). LINT check under amd64. MFC after: 1 month
2006-02-15 00:31:48 +00:00
cto->ct_tag_val = (uint8_t) AT_GET_TAG(cso->tag_id);
Add a isp_target_putback_atio- we aren't using CCINCR at this time, so we need a function that tells the Qlogic f/w that a target mode command is done, so increase the resource count for that lun. Add in a timeout function to kick the putback again if we fail to do it the first time (we may not have the request queue space for ATIO push). Split the function isp_handle_platform_ctio into two parts so that the timeout function for the ATIO push or isp_handle_platform_ctio can inform CAM that the requested CTIO(s) are now done. Clean up (cough) residual handling. What we need for Fibre Channel is to preserve the at_datalen field from the original incoming ATIO so we can calculate a 'true' residual. Unfortunately, we're not guaranteed to get that back from CAM. We'll *try* to find it hiding in the periph_priv field (layering violation)- but if an ATIO was passed in from user land- forget it. This means that we'll probably get residuals wrong for Fibre Channel commands we're completing with an error. It's too late to 4.1 release to fix this- too bad. Luckily the only device we'd really care about this occurring on is a tape device and they're still so rare as FC attached devices that this can be considered an untested combination anyway. Remove all CCINCR usage (resource autoreplenish). When we've proved to ourself that things are working properly, we can add it back in. Make sure we propage 'suggested' sense data from the incoming ATIO into the created system ATIO- and set sense_len appropriately. Correctly propagate tag values. Fall back to the model of generating (well, the functions in isp_pci.c do the work) multiple CTIOs based upon what we get from XPT. Instead of being able to pair Qlogic generated ATIOs with CAM ATIOs, and then to pair CAM CTIOs with Qlogic CTIOs, we have to take the CTIO passed to us from XPT, and if it implies that we have to generate extra Qlogic CTIOs, so be it. This means that we have to wait until the last CTIO in a sequence we generated completes before calling xpt_done. Executive summary- target mode actually now pretty much works well enough to tell folks about.
2000-07-18 06:58:28 +00:00
cto->ct_flags |= CT_TQAE;
}
if (ccb->ccb_h.flags & CAM_DIS_DISCONNECT) {
cto->ct_flags |= CT_NODISC;
}
if (cso->dxfer_len == 0) {
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
cto->ct_flags |= CT_NO_DATA;
a whale of a lot of target mode cleanup
2000-01-15 01:50:48 +00:00
} else if ((cso->ccb_h.flags & CAM_DIR_MASK) == CAM_DIR_IN) {
cto->ct_flags |= CT_DATA_IN;
} else {
cto->ct_flags |= CT_DATA_OUT;
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
}
Add a isp_target_putback_atio- we aren't using CCINCR at this time, so we need a function that tells the Qlogic f/w that a target mode command is done, so increase the resource count for that lun. Add in a timeout function to kick the putback again if we fail to do it the first time (we may not have the request queue space for ATIO push). Split the function isp_handle_platform_ctio into two parts so that the timeout function for the ATIO push or isp_handle_platform_ctio can inform CAM that the requested CTIO(s) are now done. Clean up (cough) residual handling. What we need for Fibre Channel is to preserve the at_datalen field from the original incoming ATIO so we can calculate a 'true' residual. Unfortunately, we're not guaranteed to get that back from CAM. We'll *try* to find it hiding in the periph_priv field (layering violation)- but if an ATIO was passed in from user land- forget it. This means that we'll probably get residuals wrong for Fibre Channel commands we're completing with an error. It's too late to 4.1 release to fix this- too bad. Luckily the only device we'd really care about this occurring on is a tape device and they're still so rare as FC attached devices that this can be considered an untested combination anyway. Remove all CCINCR usage (resource autoreplenish). When we've proved to ourself that things are working properly, we can add it back in. Make sure we propage 'suggested' sense data from the incoming ATIO into the created system ATIO- and set sense_len appropriately. Correctly propagate tag values. Fall back to the model of generating (well, the functions in isp_pci.c do the work) multiple CTIOs based upon what we get from XPT. Instead of being able to pair Qlogic generated ATIOs with CAM ATIOs, and then to pair CAM CTIOs with Qlogic CTIOs, we have to take the CTIO passed to us from XPT, and if it implies that we have to generate extra Qlogic CTIOs, so be it. This means that we have to wait until the last CTIO in a sequence we generated completes before calling xpt_done. Executive summary- target mode actually now pretty much works well enough to tell folks about.
2000-07-18 06:58:28 +00:00
if (ccb->ccb_h.flags & CAM_SEND_STATUS) {
Begin to implement target mode that for Fibre Channel has a private per-command component that we *don't* try and pass thru CAM. CAM just is too risky and too much of a pain- structures get copied, but not all info of interest can be considered safely transported thru all consumers (including user space) from the incoming ATIO to the outgoing CTIO- it's just much safer to have a buddy structure, identified by the command's tag which *does* make it thru safely. Pay attention to link speed and report 200MB/s xfer speed for a 23XX card in 2GPs mode. MFC after: 1 week
2001-10-01 03:48:42 +00:00
cto->ct_flags |= CT_SENDSTATUS|CT_CCINCR;
a whale of a lot of target mode cleanup
2000-01-15 01:50:48 +00:00
cto->ct_scsi_status = cso->scsi_status;
cto->ct_resid = cso->resid;
I don't know what I was thinking- if I have two separate busses on on SIM (as is true for the 1280 and the 12160), then I have to have separate flags && status for *both* busses. *Whap*. Implement condition variables for coordination with some target mode events. It's nice to use these and not panic in obscure little places in the kernel like 'propagate_priority' just because we went to sleep holding a mutex, or some other absurd thing. Remove some bogus ISP_UNLOCK calls. *Whap*. No longer require that somebody do a lun enable on the wildcard device to enable target mode. They are, in fact, orthogonal. A wildcard open is a statement that somebody upstream is willing to accept commands which are otherwise unrouteable. Now, for QLogic regular SCSI target mode, this won't matter for a damn because we'll never see ATIOs for luns we haven't enabled (are listening for, if you will). But for SCCLUN fibre channel SCSI, we get all kinds of ATIOs. We can either reflect them back here with minimal info (which is isp_target.c:isp_endcmd() is for), or the wildcard device (nominally targbh) can handle them. Do further checking against firmware attributes to see whether we can, in fact, support target mode in Fibre Channel. For now, require SCCLUN f/w to supoprt FC target mode. This is an awful lot of change, but target mode *still* isn't quite right. MFC after: 4 weeks
2001-09-04 21:53:12 +00:00
isp_prt(isp, ISP_LOGTDEBUG0,
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
"CTIO[%x] SCSI STATUS 0x%x resid %d tag_id %x",
cto->ct_fwhandle, cso->scsi_status, cso->resid,
cso->tag_id);
Clean up some debug printing. Find the correct lun when SCCLUN is defined. If we complete with a check condition but no sense data, say we had an AUTOSENSE failure.
2000-01-04 03:45:49 +00:00
}
I don't know what I was thinking- if I have two separate busses on on SIM (as is true for the 1280 and the 12160), then I have to have separate flags && status for *both* busses. *Whap*. Implement condition variables for coordination with some target mode events. It's nice to use these and not panic in obscure little places in the kernel like 'propagate_priority' just because we went to sleep holding a mutex, or some other absurd thing. Remove some bogus ISP_UNLOCK calls. *Whap*. No longer require that somebody do a lun enable on the wildcard device to enable target mode. They are, in fact, orthogonal. A wildcard open is a statement that somebody upstream is willing to accept commands which are otherwise unrouteable. Now, for QLogic regular SCSI target mode, this won't matter for a damn because we'll never see ATIOs for luns we haven't enabled (are listening for, if you will). But for SCCLUN fibre channel SCSI, we get all kinds of ATIOs. We can either reflect them back here with minimal info (which is isp_target.c:isp_endcmd() is for), or the wildcard device (nominally targbh) can handle them. Do further checking against firmware attributes to see whether we can, in fact, support target mode in Fibre Channel. For now, require SCCLUN f/w to supoprt FC target mode. This is an awful lot of change, but target mode *still* isn't quite right. MFC after: 4 weeks
2001-09-04 21:53:12 +00:00
ccb->ccb_h.flags &= ~CAM_SEND_SENSE;
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
cto->ct_timeout = 10;
Switch to using 16 bit handles instead of 32 bit handles. This is a pretty invasive change, but there are three good reasons to do this: 1. We'll never have > 16 bits of handle. 2. We can (eventually) enable the RIO (Reduced Interrupt Operation) bits which return multiple completing 16 bit handles in mailbox registers. 3. The !)$*)$*~)@$*~)$* Qlogic target mode for parallel SCSI spec changed such that at_reserved (which was 32 bits) was split into two pieces- and one of which was a 16 bit handle id that functions like the at_rxid for Fibre Channel (a tag for the f/w to correlate CTIOs with a particular command). Since we had to muck with that and this changed the whole handler architecture, we might as well... Propagate new at_handle on through int ct_fwhandle. Follow implications of changing to 16 bit handles. These above changes at least get Qlogic 1040 cards working in target mode again. 1080/12160 cards don't work yet. In isp.c: Prepare for doing all loop management in outer layers.
2001-03-02 06:28:55 +00:00
hp = &cto->ct_syshandle;
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
}
Store the target handles in a separate list from normal commands. Add a CTIO fast post routine to handle CTIO completions. Submitted by: mjacob
2004-05-24 07:02:25 +00:00
if (isp_save_xs_tgt(isp, ccb, hp)) {
Clean up some debug printing. Find the correct lun when SCCLUN is defined. If we complete with a check condition but no sense data, say we had an AUTOSENSE failure.
2000-01-04 03:45:49 +00:00
xpt_print_path(ccb->ccb_h.path);
printf("No XFLIST pointers for isp_target_start_ctio\n");
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
return (CAM_RESRC_UNAVAIL);
}
/*
* Call the dma setup routines for this entry (and any subsequent
* CTIOs) if there's data to move, and then tell the f/w it's got
Core version 2.0 rewrite. In this file we replace isp_tdebug with isp_prt calls. We now use an argument to the ISPCTL_FCLINK_TEST call. We change all IDPRINTF macros to isp_prt calls. We add the isp_prt function here.
2000-08-01 06:31:44 +00:00
* new things to play with. As with isp_start's usage of DMA setup,
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
* any swizzling is done in the machine dependent layer. Because
* of this, we put the request onto the queue area first in native
* format.
*/
save_handle = *hp;
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
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). It also has to handle the SBus cards (for platforms that have them) which, while on a Big Endian machine, do *not* require *most* of the request/response queue entry fields to be swizzled or unswizzled. 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. 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). MFC after: 2 weeks
2001-12-11 00:18:45 +00:00
switch (ISP_DMASETUP(isp, cso, (ispreq_t *) local, &nxti, optr)) {
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
case CMD_QUEUED:
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). It also has to handle the SBus cards (for platforms that have them) which, while on a Big Endian machine, do *not* require *most* of the request/response queue entry fields to be swizzled or unswizzled. 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. 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). MFC after: 2 weeks
2001-12-11 00:18:45 +00:00
ISP_ADD_REQUEST(isp, nxti);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
return (CAM_REQ_INPROG);
case CMD_EAGAIN:
ccb->ccb_h.status = CAM_RESRC_UNAVAIL;
Store the target handles in a separate list from normal commands. Add a CTIO fast post routine to handle CTIO completions. Submitted by: mjacob
2004-05-24 07:02:25 +00:00
isp_destroy_tgt_handle(isp, save_handle);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
return (CAM_RESRC_UNAVAIL);
default:
Store the target handles in a separate list from normal commands. Add a CTIO fast post routine to handle CTIO completions. Submitted by: mjacob
2004-05-24 07:02:25 +00:00
isp_destroy_tgt_handle(isp, save_handle);
Add in the enabling of interrupts (to isp_attach). Clean up a busted comment. Check against firmware state- not loop state when enabling target mode. Other changes have to do with no longer enabling/disabling interrupts at will. Rearchitect command watchdog timeouts- First of all, set the timeout period for a command that has a timeout (in isp_action) to the period of time requested *plus* two seconds. We don't want the Qlogic firmware and the host system to race each other to report a dead command (the watchdog is there to catch dead and/or broken firmware). Next, make sure that the command being watched isn't done yet. If it's not done yet, check for INT_PENDING and call isp_intr- if that said it serviced an interrupt, check to see whether the command is now done (this is what the "IN WATCHDOG" private flag is for- if isp_intr completes the command, it won't call xpt_done on it because isp_watchdog is still looking at the command). If no interrupt was pending, or the command wasn't completed, check to see if we've set the private 'grace period' flag. If so, the command really *is* dead, so report it as dead and complete it with a CAM_CMD_TIMEOUT value. If the grace period flag wasn't set, set it and issue a SYNCHRONIZE_ALL Marker Request Queue entry and re-set the timeout for one second from now (see Revision 1.45 isp.c notes for more on this) to give the firmware a final chance to complete this command.
2000-06-27 19:31:02 +00:00
return (XS_ERR(ccb));
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
}
}
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
static void
isp_refire_putback_atio(void *arg)
Add a isp_target_putback_atio- we aren't using CCINCR at this time, so we need a function that tells the Qlogic f/w that a target mode command is done, so increase the resource count for that lun. Add in a timeout function to kick the putback again if we fail to do it the first time (we may not have the request queue space for ATIO push). Split the function isp_handle_platform_ctio into two parts so that the timeout function for the ATIO push or isp_handle_platform_ctio can inform CAM that the requested CTIO(s) are now done. Clean up (cough) residual handling. What we need for Fibre Channel is to preserve the at_datalen field from the original incoming ATIO so we can calculate a 'true' residual. Unfortunately, we're not guaranteed to get that back from CAM. We'll *try* to find it hiding in the periph_priv field (layering violation)- but if an ATIO was passed in from user land- forget it. This means that we'll probably get residuals wrong for Fibre Channel commands we're completing with an error. It's too late to 4.1 release to fix this- too bad. Luckily the only device we'd really care about this occurring on is a tape device and they're still so rare as FC attached devices that this can be considered an untested combination anyway. Remove all CCINCR usage (resource autoreplenish). When we've proved to ourself that things are working properly, we can add it back in. Make sure we propage 'suggested' sense data from the incoming ATIO into the created system ATIO- and set sense_len appropriately. Correctly propagate tag values. Fall back to the model of generating (well, the functions in isp_pci.c do the work) multiple CTIOs based upon what we get from XPT. Instead of being able to pair Qlogic generated ATIOs with CAM ATIOs, and then to pair CAM CTIOs with Qlogic CTIOs, we have to take the CTIO passed to us from XPT, and if it implies that we have to generate extra Qlogic CTIOs, so be it. This means that we have to wait until the last CTIO in a sequence we generated completes before calling xpt_done. Executive summary- target mode actually now pretty much works well enough to tell folks about.
2000-07-18 06:58:28 +00:00
{
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
int s = splcam();
isp_target_putback_atio(arg);
splx(s);
}
static void
isp_target_putback_atio(union ccb *ccb)
{
struct ispsoftc *isp;
struct ccb_scsiio *cso;
a) clean up some declaration stuff (i.e., make more modern with respect to getting rid u_int for uint and so on). b) Turn back on 64 bit DAC support. Cheeze it a bit in that we have two DMA callback functions- one when we have bus_addr_t > 4 bits in width and the other which should be normal. Even Cheezier in that we turn off setting up DMA maps to be BUS_SPACE_MAXADDR if we're in ISP_TARGET_MODE. More work on this in a week or so. c) Tested under amd64 and 1MB DFLTPHYS, sparc64, i386 (PAE, but insufficient memory to really test > 4GB). LINT check under amd64. MFC after: 1 month
2006-02-15 00:31:48 +00:00
uint16_t nxti, optr;
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
void *qe;
isp = XS_ISP(ccb);
Add a isp_target_putback_atio- we aren't using CCINCR at this time, so we need a function that tells the Qlogic f/w that a target mode command is done, so increase the resource count for that lun. Add in a timeout function to kick the putback again if we fail to do it the first time (we may not have the request queue space for ATIO push). Split the function isp_handle_platform_ctio into two parts so that the timeout function for the ATIO push or isp_handle_platform_ctio can inform CAM that the requested CTIO(s) are now done. Clean up (cough) residual handling. What we need for Fibre Channel is to preserve the at_datalen field from the original incoming ATIO so we can calculate a 'true' residual. Unfortunately, we're not guaranteed to get that back from CAM. We'll *try* to find it hiding in the periph_priv field (layering violation)- but if an ATIO was passed in from user land- forget it. This means that we'll probably get residuals wrong for Fibre Channel commands we're completing with an error. It's too late to 4.1 release to fix this- too bad. Luckily the only device we'd really care about this occurring on is a tape device and they're still so rare as FC attached devices that this can be considered an untested combination anyway. Remove all CCINCR usage (resource autoreplenish). When we've proved to ourself that things are working properly, we can add it back in. Make sure we propage 'suggested' sense data from the incoming ATIO into the created system ATIO- and set sense_len appropriately. Correctly propagate tag values. Fall back to the model of generating (well, the functions in isp_pci.c do the work) multiple CTIOs based upon what we get from XPT. Instead of being able to pair Qlogic generated ATIOs with CAM ATIOs, and then to pair CAM CTIOs with Qlogic CTIOs, we have to take the CTIO passed to us from XPT, and if it implies that we have to generate extra Qlogic CTIOs, so be it. This means that we have to wait until the last CTIO in a sequence we generated completes before calling xpt_done. Executive summary- target mode actually now pretty much works well enough to tell folks about.
2000-07-18 06:58:28 +00:00
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). It also has to handle the SBus cards (for platforms that have them) which, while on a Big Endian machine, do *not* require *most* of the request/response queue entry fields to be swizzled or unswizzled. 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. 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). MFC after: 2 weeks
2001-12-11 00:18:45 +00:00
if (isp_getrqentry(isp, &nxti, &optr, &qe)) {
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
(void) timeout(isp_refire_putback_atio, ccb, 10);
isp_prt(isp, ISP_LOGWARN,
"isp_target_putback_atio: Request Queue Overflow");
return;
Add a isp_target_putback_atio- we aren't using CCINCR at this time, so we need a function that tells the Qlogic f/w that a target mode command is done, so increase the resource count for that lun. Add in a timeout function to kick the putback again if we fail to do it the first time (we may not have the request queue space for ATIO push). Split the function isp_handle_platform_ctio into two parts so that the timeout function for the ATIO push or isp_handle_platform_ctio can inform CAM that the requested CTIO(s) are now done. Clean up (cough) residual handling. What we need for Fibre Channel is to preserve the at_datalen field from the original incoming ATIO so we can calculate a 'true' residual. Unfortunately, we're not guaranteed to get that back from CAM. We'll *try* to find it hiding in the periph_priv field (layering violation)- but if an ATIO was passed in from user land- forget it. This means that we'll probably get residuals wrong for Fibre Channel commands we're completing with an error. It's too late to 4.1 release to fix this- too bad. Luckily the only device we'd really care about this occurring on is a tape device and they're still so rare as FC attached devices that this can be considered an untested combination anyway. Remove all CCINCR usage (resource autoreplenish). When we've proved to ourself that things are working properly, we can add it back in. Make sure we propage 'suggested' sense data from the incoming ATIO into the created system ATIO- and set sense_len appropriately. Correctly propagate tag values. Fall back to the model of generating (well, the functions in isp_pci.c do the work) multiple CTIOs based upon what we get from XPT. Instead of being able to pair Qlogic generated ATIOs with CAM ATIOs, and then to pair CAM CTIOs with Qlogic CTIOs, we have to take the CTIO passed to us from XPT, and if it implies that we have to generate extra Qlogic CTIOs, so be it. This means that we have to wait until the last CTIO in a sequence we generated completes before calling xpt_done. Executive summary- target mode actually now pretty much works well enough to tell folks about.
2000-07-18 06:58:28 +00:00
}
bzero(qe, QENTRY_LEN);
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
cso = &ccb->csio;
Add a isp_target_putback_atio- we aren't using CCINCR at this time, so we need a function that tells the Qlogic f/w that a target mode command is done, so increase the resource count for that lun. Add in a timeout function to kick the putback again if we fail to do it the first time (we may not have the request queue space for ATIO push). Split the function isp_handle_platform_ctio into two parts so that the timeout function for the ATIO push or isp_handle_platform_ctio can inform CAM that the requested CTIO(s) are now done. Clean up (cough) residual handling. What we need for Fibre Channel is to preserve the at_datalen field from the original incoming ATIO so we can calculate a 'true' residual. Unfortunately, we're not guaranteed to get that back from CAM. We'll *try* to find it hiding in the periph_priv field (layering violation)- but if an ATIO was passed in from user land- forget it. This means that we'll probably get residuals wrong for Fibre Channel commands we're completing with an error. It's too late to 4.1 release to fix this- too bad. Luckily the only device we'd really care about this occurring on is a tape device and they're still so rare as FC attached devices that this can be considered an untested combination anyway. Remove all CCINCR usage (resource autoreplenish). When we've proved to ourself that things are working properly, we can add it back in. Make sure we propage 'suggested' sense data from the incoming ATIO into the created system ATIO- and set sense_len appropriately. Correctly propagate tag values. Fall back to the model of generating (well, the functions in isp_pci.c do the work) multiple CTIOs based upon what we get from XPT. Instead of being able to pair Qlogic generated ATIOs with CAM ATIOs, and then to pair CAM CTIOs with Qlogic CTIOs, we have to take the CTIO passed to us from XPT, and if it implies that we have to generate extra Qlogic CTIOs, so be it. This means that we have to wait until the last CTIO in a sequence we generated completes before calling xpt_done. Executive summary- target mode actually now pretty much works well enough to tell folks about.
2000-07-18 06:58:28 +00:00
if (IS_FC(isp)) {
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). It also has to handle the SBus cards (for platforms that have them) which, while on a Big Endian machine, do *not* require *most* of the request/response queue entry fields to be swizzled or unswizzled. 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. 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). MFC after: 2 weeks
2001-12-11 00:18:45 +00:00
at2_entry_t local, *at = &local;
MEMZERO(at, sizeof (at2_entry_t));
Add a isp_target_putback_atio- we aren't using CCINCR at this time, so we need a function that tells the Qlogic f/w that a target mode command is done, so increase the resource count for that lun. Add in a timeout function to kick the putback again if we fail to do it the first time (we may not have the request queue space for ATIO push). Split the function isp_handle_platform_ctio into two parts so that the timeout function for the ATIO push or isp_handle_platform_ctio can inform CAM that the requested CTIO(s) are now done. Clean up (cough) residual handling. What we need for Fibre Channel is to preserve the at_datalen field from the original incoming ATIO so we can calculate a 'true' residual. Unfortunately, we're not guaranteed to get that back from CAM. We'll *try* to find it hiding in the periph_priv field (layering violation)- but if an ATIO was passed in from user land- forget it. This means that we'll probably get residuals wrong for Fibre Channel commands we're completing with an error. It's too late to 4.1 release to fix this- too bad. Luckily the only device we'd really care about this occurring on is a tape device and they're still so rare as FC attached devices that this can be considered an untested combination anyway. Remove all CCINCR usage (resource autoreplenish). When we've proved to ourself that things are working properly, we can add it back in. Make sure we propage 'suggested' sense data from the incoming ATIO into the created system ATIO- and set sense_len appropriately. Correctly propagate tag values. Fall back to the model of generating (well, the functions in isp_pci.c do the work) multiple CTIOs based upon what we get from XPT. Instead of being able to pair Qlogic generated ATIOs with CAM ATIOs, and then to pair CAM CTIOs with Qlogic CTIOs, we have to take the CTIO passed to us from XPT, and if it implies that we have to generate extra Qlogic CTIOs, so be it. This means that we have to wait until the last CTIO in a sequence we generated completes before calling xpt_done. Executive summary- target mode actually now pretty much works well enough to tell folks about.
2000-07-18 06:58:28 +00:00
at->at_header.rqs_entry_type = RQSTYPE_ATIO2;
at->at_header.rqs_entry_count = 1;
I don't know what I was thinking- if I have two separate busses on on SIM (as is true for the 1280 and the 12160), then I have to have separate flags && status for *both* busses. *Whap*. Implement condition variables for coordination with some target mode events. It's nice to use these and not panic in obscure little places in the kernel like 'propagate_priority' just because we went to sleep holding a mutex, or some other absurd thing. Remove some bogus ISP_UNLOCK calls. *Whap*. No longer require that somebody do a lun enable on the wildcard device to enable target mode. They are, in fact, orthogonal. A wildcard open is a statement that somebody upstream is willing to accept commands which are otherwise unrouteable. Now, for QLogic regular SCSI target mode, this won't matter for a damn because we'll never see ATIOs for luns we haven't enabled (are listening for, if you will). But for SCCLUN fibre channel SCSI, we get all kinds of ATIOs. We can either reflect them back here with minimal info (which is isp_target.c:isp_endcmd() is for), or the wildcard device (nominally targbh) can handle them. Do further checking against firmware attributes to see whether we can, in fact, support target mode in Fibre Channel. For now, require SCCLUN f/w to supoprt FC target mode. This is an awful lot of change, but target mode *still* isn't quite right. MFC after: 4 weeks
2001-09-04 21:53:12 +00:00
if ((FCPARAM(isp)->isp_fwattr & ISP_FW_ATTR_SCCLUN) != 0) {
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
at->at_scclun = (uint16_t) ccb->ccb_h.target_lun;
Add a isp_target_putback_atio- we aren't using CCINCR at this time, so we need a function that tells the Qlogic f/w that a target mode command is done, so increase the resource count for that lun. Add in a timeout function to kick the putback again if we fail to do it the first time (we may not have the request queue space for ATIO push). Split the function isp_handle_platform_ctio into two parts so that the timeout function for the ATIO push or isp_handle_platform_ctio can inform CAM that the requested CTIO(s) are now done. Clean up (cough) residual handling. What we need for Fibre Channel is to preserve the at_datalen field from the original incoming ATIO so we can calculate a 'true' residual. Unfortunately, we're not guaranteed to get that back from CAM. We'll *try* to find it hiding in the periph_priv field (layering violation)- but if an ATIO was passed in from user land- forget it. This means that we'll probably get residuals wrong for Fibre Channel commands we're completing with an error. It's too late to 4.1 release to fix this- too bad. Luckily the only device we'd really care about this occurring on is a tape device and they're still so rare as FC attached devices that this can be considered an untested combination anyway. Remove all CCINCR usage (resource autoreplenish). When we've proved to ourself that things are working properly, we can add it back in. Make sure we propage 'suggested' sense data from the incoming ATIO into the created system ATIO- and set sense_len appropriately. Correctly propagate tag values. Fall back to the model of generating (well, the functions in isp_pci.c do the work) multiple CTIOs based upon what we get from XPT. Instead of being able to pair Qlogic generated ATIOs with CAM ATIOs, and then to pair CAM CTIOs with Qlogic CTIOs, we have to take the CTIO passed to us from XPT, and if it implies that we have to generate extra Qlogic CTIOs, so be it. This means that we have to wait until the last CTIO in a sequence we generated completes before calling xpt_done. Executive summary- target mode actually now pretty much works well enough to tell folks about.
2000-07-18 06:58:28 +00:00
} else {
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
at->at_lun = (uint8_t) ccb->ccb_h.target_lun;
Add a isp_target_putback_atio- we aren't using CCINCR at this time, so we need a function that tells the Qlogic f/w that a target mode command is done, so increase the resource count for that lun. Add in a timeout function to kick the putback again if we fail to do it the first time (we may not have the request queue space for ATIO push). Split the function isp_handle_platform_ctio into two parts so that the timeout function for the ATIO push or isp_handle_platform_ctio can inform CAM that the requested CTIO(s) are now done. Clean up (cough) residual handling. What we need for Fibre Channel is to preserve the at_datalen field from the original incoming ATIO so we can calculate a 'true' residual. Unfortunately, we're not guaranteed to get that back from CAM. We'll *try* to find it hiding in the periph_priv field (layering violation)- but if an ATIO was passed in from user land- forget it. This means that we'll probably get residuals wrong for Fibre Channel commands we're completing with an error. It's too late to 4.1 release to fix this- too bad. Luckily the only device we'd really care about this occurring on is a tape device and they're still so rare as FC attached devices that this can be considered an untested combination anyway. Remove all CCINCR usage (resource autoreplenish). When we've proved to ourself that things are working properly, we can add it back in. Make sure we propage 'suggested' sense data from the incoming ATIO into the created system ATIO- and set sense_len appropriately. Correctly propagate tag values. Fall back to the model of generating (well, the functions in isp_pci.c do the work) multiple CTIOs based upon what we get from XPT. Instead of being able to pair Qlogic generated ATIOs with CAM ATIOs, and then to pair CAM CTIOs with Qlogic CTIOs, we have to take the CTIO passed to us from XPT, and if it implies that we have to generate extra Qlogic CTIOs, so be it. This means that we have to wait until the last CTIO in a sequence we generated completes before calling xpt_done. Executive summary- target mode actually now pretty much works well enough to tell folks about.
2000-07-18 06:58:28 +00:00
}
at->at_status = CT_OK;
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
at->at_rxid = cso->tag_id;
Add support for ISP_FC_GETHINFO, which returns current connection topology, speed, loopid, WWPN/WWNN, etc. Beef up target mode. Add isp_handle_platform_notify_scsi and isp_handle_platform_notify_fc platform handlers to handle immediate notifies (isp_handle_platform_notify_scsi is still stubbed out). In implementation of isp_handle_platform_notify_fc, for IN_ABORT_TASK, peel off a pending XPT_IMMED_NOTIFY and call xpt_done on it and hope that somebody upstream is listening. Make sure on final CTIO2s that we set residual correctly. These are absolutely crucial. Make sure we set relative offset for each CTIO2 based upon bytes we've already xferred. This is what the private adjunct datat to the original ATIO is. Note state of command so we can figure out where to find it if we get an ABORT from the firmware. Make sure we *always* set CAM_TAG_ACTION_VALID for ATIO2s. Make sure we keep track of the original lun. If se sent status (or we're otherwise done with the command), don't forget to free the adjunct structure.
2002-06-16 05:08:02 +00:00
at->at_iid = cso->ccb_h.target_id;
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). It also has to handle the SBus cards (for platforms that have them) which, while on a Big Endian machine, do *not* require *most* of the request/response queue entry fields to be swizzled or unswizzled. 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. 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). MFC after: 2 weeks
2001-12-11 00:18:45 +00:00
isp_put_atio2(isp, at, qe);
Add a isp_target_putback_atio- we aren't using CCINCR at this time, so we need a function that tells the Qlogic f/w that a target mode command is done, so increase the resource count for that lun. Add in a timeout function to kick the putback again if we fail to do it the first time (we may not have the request queue space for ATIO push). Split the function isp_handle_platform_ctio into two parts so that the timeout function for the ATIO push or isp_handle_platform_ctio can inform CAM that the requested CTIO(s) are now done. Clean up (cough) residual handling. What we need for Fibre Channel is to preserve the at_datalen field from the original incoming ATIO so we can calculate a 'true' residual. Unfortunately, we're not guaranteed to get that back from CAM. We'll *try* to find it hiding in the periph_priv field (layering violation)- but if an ATIO was passed in from user land- forget it. This means that we'll probably get residuals wrong for Fibre Channel commands we're completing with an error. It's too late to 4.1 release to fix this- too bad. Luckily the only device we'd really care about this occurring on is a tape device and they're still so rare as FC attached devices that this can be considered an untested combination anyway. Remove all CCINCR usage (resource autoreplenish). When we've proved to ourself that things are working properly, we can add it back in. Make sure we propage 'suggested' sense data from the incoming ATIO into the created system ATIO- and set sense_len appropriately. Correctly propagate tag values. Fall back to the model of generating (well, the functions in isp_pci.c do the work) multiple CTIOs based upon what we get from XPT. Instead of being able to pair Qlogic generated ATIOs with CAM ATIOs, and then to pair CAM CTIOs with Qlogic CTIOs, we have to take the CTIO passed to us from XPT, and if it implies that we have to generate extra Qlogic CTIOs, so be it. This means that we have to wait until the last CTIO in a sequence we generated completes before calling xpt_done. Executive summary- target mode actually now pretty much works well enough to tell folks about.
2000-07-18 06:58:28 +00:00
} else {
Major restructuring for swizzling to the request queue and unswizzling from the response queue. Instead of the ad hoc ISP_SWIZZLE_REQUEST, we now have a complete set of inline functions in isp_inline.h. Each platform is responsible for providing just one of a set of ISP_IOX_{GET,PUT}{8,16,32} macros. The reason this needs to be done is that we need to have a single set of functions that will work correctly on multiple architectures for both little and big endian machines. It also needs to work correctly in the case that we have the request or response queues in memory that has to be treated specially (e.g., have ddi_dma_sync called on it for Solaris after we update it or before we read from it). It also has to handle the SBus cards (for platforms that have them) which, while on a Big Endian machine, do *not* require *most* of the request/response queue entry fields to be swizzled or unswizzled. 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. 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). MFC after: 2 weeks
2001-12-11 00:18:45 +00:00
at_entry_t local, *at = &local;
MEMZERO(at, sizeof (at_entry_t));
Add a isp_target_putback_atio- we aren't using CCINCR at this time, so we need a function that tells the Qlogic f/w that a target mode command is done, so increase the resource count for that lun. Add in a timeout function to kick the putback again if we fail to do it the first time (we may not have the request queue space for ATIO push). Split the function isp_handle_platform_ctio into two parts so that the timeout function for the ATIO push or isp_handle_platform_ctio can inform CAM that the requested CTIO(s) are now done. Clean up (cough) residual handling. What we need for Fibre Channel is to preserve the at_datalen field from the original incoming ATIO so we can calculate a 'true' residual. Unfortunately, we're not guaranteed to get that back from CAM. We'll *try* to find it hiding in the periph_priv field (layering violation)- but if an ATIO was passed in from user land- forget it. This means that we'll probably get residuals wrong for Fibre Channel commands we're completing with an error. It's too late to 4.1 release to fix this- too bad. Luckily the only device we'd really care about this occurring on is a tape device and they're still so rare as FC attached devices that this can be considered an untested combination anyway. Remove all CCINCR usage (resource autoreplenish). When we've proved to ourself that things are working properly, we can add it back in. Make sure we propage 'suggested' sense data from the incoming ATIO into the created system ATIO- and set sense_len appropriately. Correctly propagate tag values. Fall back to the model of generating (well, the functions in isp_pci.c do the work) multiple CTIOs based upon what we get from XPT. Instead of being able to pair Qlogic generated ATIOs with CAM ATIOs, and then to pair CAM CTIOs with Qlogic CTIOs, we have to take the CTIO passed to us from XPT, and if it implies that we have to generate extra Qlogic CTIOs, so be it. This means that we have to wait until the last CTIO in a sequence we generated completes before calling xpt_done. Executive summary- target mode actually now pretty much works well enough to tell folks about.
2000-07-18 06:58:28 +00:00
at->at_header.rqs_entry_type = RQSTYPE_ATIO;
at->at_header.rqs_entry_count = 1;
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
at->at_iid = cso->init_id;
at->at_iid |= XS_CHANNEL(ccb) << 7;
at->at_tgt = cso->ccb_h.target_id;
at->at_lun = cso->ccb_h.target_lun;
Add a isp_target_putback_atio- we aren't using CCINCR at this time, so we need a function that tells the Qlogic f/w that a target mode command is done, so increase the resource count for that lun. Add in a timeout function to kick the putback again if we fail to do it the first time (we may not have the request queue space for ATIO push). Split the function isp_handle_platform_ctio into two parts so that the timeout function for the ATIO push or isp_handle_platform_ctio can inform CAM that the requested CTIO(s) are now done. Clean up (cough) residual handling. What we need for Fibre Channel is to preserve the at_datalen field from the original incoming ATIO so we can calculate a 'true' residual. Unfortunately, we're not guaranteed to get that back from CAM. We'll *try* to find it hiding in the periph_priv field (layering violation)- but if an ATIO was passed in from user land- forget it. This means that we'll probably get residuals wrong for Fibre Channel commands we're completing with an error. It's too late to 4.1 release to fix this- too bad. Luckily the only device we'd really care about this occurring on is a tape device and they're still so rare as FC attached devices that this can be considered an untested combination anyway. Remove all CCINCR usage (resource autoreplenish). When we've proved to ourself that things are working properly, we can add it back in. Make sure we propage 'suggested' sense data from the incoming ATIO into the created system ATIO- and set sense_len appropriately. Correctly propagate tag values. Fall back to the model of generating (well, the functions in isp_pci.c do the work) multiple CTIOs based upon what we get from XPT. Instead of being able to pair Qlogic generated ATIOs with CAM ATIOs, and then to pair CAM CTIOs with Qlogic CTIOs, we have to take the CTIO passed to us from XPT, and if it implies that we have to generate extra Qlogic CTIOs, so be it. This means that we have to wait until the last CTIO in a sequence we generated completes before calling xpt_done. Executive summary- target mode actually now pretty much works well enough to tell folks about.
2000-07-18 06:58:28 +00:00
at->at_status = CT_OK;
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
at->at_tag_val = AT_GET_TAG(cso->tag_id);
at->at_handle = AT_GET_HANDLE(cso->tag_id);
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). It also has to handle the SBus cards (for platforms that have them) which, while on a Big Endian machine, do *not* require *most* of the request/response queue entry fields to be swizzled or unswizzled. 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. 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). MFC after: 2 weeks
2001-12-11 00:18:45 +00:00
isp_put_atio(isp, at, qe);
Add a isp_target_putback_atio- we aren't using CCINCR at this time, so we need a function that tells the Qlogic f/w that a target mode command is done, so increase the resource count for that lun. Add in a timeout function to kick the putback again if we fail to do it the first time (we may not have the request queue space for ATIO push). Split the function isp_handle_platform_ctio into two parts so that the timeout function for the ATIO push or isp_handle_platform_ctio can inform CAM that the requested CTIO(s) are now done. Clean up (cough) residual handling. What we need for Fibre Channel is to preserve the at_datalen field from the original incoming ATIO so we can calculate a 'true' residual. Unfortunately, we're not guaranteed to get that back from CAM. We'll *try* to find it hiding in the periph_priv field (layering violation)- but if an ATIO was passed in from user land- forget it. This means that we'll probably get residuals wrong for Fibre Channel commands we're completing with an error. It's too late to 4.1 release to fix this- too bad. Luckily the only device we'd really care about this occurring on is a tape device and they're still so rare as FC attached devices that this can be considered an untested combination anyway. Remove all CCINCR usage (resource autoreplenish). When we've proved to ourself that things are working properly, we can add it back in. Make sure we propage 'suggested' sense data from the incoming ATIO into the created system ATIO- and set sense_len appropriately. Correctly propagate tag values. Fall back to the model of generating (well, the functions in isp_pci.c do the work) multiple CTIOs based upon what we get from XPT. Instead of being able to pair Qlogic generated ATIOs with CAM ATIOs, and then to pair CAM CTIOs with Qlogic CTIOs, we have to take the CTIO passed to us from XPT, and if it implies that we have to generate extra Qlogic CTIOs, so be it. This means that we have to wait until the last CTIO in a sequence we generated completes before calling xpt_done. Executive summary- target mode actually now pretty much works well enough to tell folks about.
2000-07-18 06:58:28 +00:00
}
ISP_TDQE(isp, "isp_target_putback_atio", (int) optr, qe);
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). It also has to handle the SBus cards (for platforms that have them) which, while on a Big Endian machine, do *not* require *most* of the request/response queue entry fields to be swizzled or unswizzled. 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. 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). MFC after: 2 weeks
2001-12-11 00:18:45 +00:00
ISP_ADD_REQUEST(isp, nxti);
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
isp_complete_ctio(ccb);
Add a isp_target_putback_atio- we aren't using CCINCR at this time, so we need a function that tells the Qlogic f/w that a target mode command is done, so increase the resource count for that lun. Add in a timeout function to kick the putback again if we fail to do it the first time (we may not have the request queue space for ATIO push). Split the function isp_handle_platform_ctio into two parts so that the timeout function for the ATIO push or isp_handle_platform_ctio can inform CAM that the requested CTIO(s) are now done. Clean up (cough) residual handling. What we need for Fibre Channel is to preserve the at_datalen field from the original incoming ATIO so we can calculate a 'true' residual. Unfortunately, we're not guaranteed to get that back from CAM. We'll *try* to find it hiding in the periph_priv field (layering violation)- but if an ATIO was passed in from user land- forget it. This means that we'll probably get residuals wrong for Fibre Channel commands we're completing with an error. It's too late to 4.1 release to fix this- too bad. Luckily the only device we'd really care about this occurring on is a tape device and they're still so rare as FC attached devices that this can be considered an untested combination anyway. Remove all CCINCR usage (resource autoreplenish). When we've proved to ourself that things are working properly, we can add it back in. Make sure we propage 'suggested' sense data from the incoming ATIO into the created system ATIO- and set sense_len appropriately. Correctly propagate tag values. Fall back to the model of generating (well, the functions in isp_pci.c do the work) multiple CTIOs based upon what we get from XPT. Instead of being able to pair Qlogic generated ATIOs with CAM ATIOs, and then to pair CAM CTIOs with Qlogic CTIOs, we have to take the CTIO passed to us from XPT, and if it implies that we have to generate extra Qlogic CTIOs, so be it. This means that we have to wait until the last CTIO in a sequence we generated completes before calling xpt_done. Executive summary- target mode actually now pretty much works well enough to tell folks about.
2000-07-18 06:58:28 +00:00
}
static void
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
isp_complete_ctio(union ccb *ccb)
Add a isp_target_putback_atio- we aren't using CCINCR at this time, so we need a function that tells the Qlogic f/w that a target mode command is done, so increase the resource count for that lun. Add in a timeout function to kick the putback again if we fail to do it the first time (we may not have the request queue space for ATIO push). Split the function isp_handle_platform_ctio into two parts so that the timeout function for the ATIO push or isp_handle_platform_ctio can inform CAM that the requested CTIO(s) are now done. Clean up (cough) residual handling. What we need for Fibre Channel is to preserve the at_datalen field from the original incoming ATIO so we can calculate a 'true' residual. Unfortunately, we're not guaranteed to get that back from CAM. We'll *try* to find it hiding in the periph_priv field (layering violation)- but if an ATIO was passed in from user land- forget it. This means that we'll probably get residuals wrong for Fibre Channel commands we're completing with an error. It's too late to 4.1 release to fix this- too bad. Luckily the only device we'd really care about this occurring on is a tape device and they're still so rare as FC attached devices that this can be considered an untested combination anyway. Remove all CCINCR usage (resource autoreplenish). When we've proved to ourself that things are working properly, we can add it back in. Make sure we propage 'suggested' sense data from the incoming ATIO into the created system ATIO- and set sense_len appropriately. Correctly propagate tag values. Fall back to the model of generating (well, the functions in isp_pci.c do the work) multiple CTIOs based upon what we get from XPT. Instead of being able to pair Qlogic generated ATIOs with CAM ATIOs, and then to pair CAM CTIOs with Qlogic CTIOs, we have to take the CTIO passed to us from XPT, and if it implies that we have to generate extra Qlogic CTIOs, so be it. This means that we have to wait until the last CTIO in a sequence we generated completes before calling xpt_done. Executive summary- target mode actually now pretty much works well enough to tell folks about.
2000-07-18 06:58:28 +00:00
{
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
if ((ccb->ccb_h.status & CAM_STATUS_MASK) == CAM_REQ_INPROG) {
ccb->ccb_h.status |= CAM_REQ_CMP;
Add a isp_target_putback_atio- we aren't using CCINCR at this time, so we need a function that tells the Qlogic f/w that a target mode command is done, so increase the resource count for that lun. Add in a timeout function to kick the putback again if we fail to do it the first time (we may not have the request queue space for ATIO push). Split the function isp_handle_platform_ctio into two parts so that the timeout function for the ATIO push or isp_handle_platform_ctio can inform CAM that the requested CTIO(s) are now done. Clean up (cough) residual handling. What we need for Fibre Channel is to preserve the at_datalen field from the original incoming ATIO so we can calculate a 'true' residual. Unfortunately, we're not guaranteed to get that back from CAM. We'll *try* to find it hiding in the periph_priv field (layering violation)- but if an ATIO was passed in from user land- forget it. This means that we'll probably get residuals wrong for Fibre Channel commands we're completing with an error. It's too late to 4.1 release to fix this- too bad. Luckily the only device we'd really care about this occurring on is a tape device and they're still so rare as FC attached devices that this can be considered an untested combination anyway. Remove all CCINCR usage (resource autoreplenish). When we've proved to ourself that things are working properly, we can add it back in. Make sure we propage 'suggested' sense data from the incoming ATIO into the created system ATIO- and set sense_len appropriately. Correctly propagate tag values. Fall back to the model of generating (well, the functions in isp_pci.c do the work) multiple CTIOs based upon what we get from XPT. Instead of being able to pair Qlogic generated ATIOs with CAM ATIOs, and then to pair CAM CTIOs with Qlogic CTIOs, we have to take the CTIO passed to us from XPT, and if it implies that we have to generate extra Qlogic CTIOs, so be it. This means that we have to wait until the last CTIO in a sequence we generated completes before calling xpt_done. Executive summary- target mode actually now pretty much works well enough to tell folks about.
2000-07-18 06:58:28 +00:00
}
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
ccb->ccb_h.status &= ~CAM_SIM_QUEUED;
xpt_done(ccb);
Add a isp_target_putback_atio- we aren't using CCINCR at this time, so we need a function that tells the Qlogic f/w that a target mode command is done, so increase the resource count for that lun. Add in a timeout function to kick the putback again if we fail to do it the first time (we may not have the request queue space for ATIO push). Split the function isp_handle_platform_ctio into two parts so that the timeout function for the ATIO push or isp_handle_platform_ctio can inform CAM that the requested CTIO(s) are now done. Clean up (cough) residual handling. What we need for Fibre Channel is to preserve the at_datalen field from the original incoming ATIO so we can calculate a 'true' residual. Unfortunately, we're not guaranteed to get that back from CAM. We'll *try* to find it hiding in the periph_priv field (layering violation)- but if an ATIO was passed in from user land- forget it. This means that we'll probably get residuals wrong for Fibre Channel commands we're completing with an error. It's too late to 4.1 release to fix this- too bad. Luckily the only device we'd really care about this occurring on is a tape device and they're still so rare as FC attached devices that this can be considered an untested combination anyway. Remove all CCINCR usage (resource autoreplenish). When we've proved to ourself that things are working properly, we can add it back in. Make sure we propage 'suggested' sense data from the incoming ATIO into the created system ATIO- and set sense_len appropriately. Correctly propagate tag values. Fall back to the model of generating (well, the functions in isp_pci.c do the work) multiple CTIOs based upon what we get from XPT. Instead of being able to pair Qlogic generated ATIOs with CAM ATIOs, and then to pair CAM CTIOs with Qlogic CTIOs, we have to take the CTIO passed to us from XPT, and if it implies that we have to generate extra Qlogic CTIOs, so be it. This means that we have to wait until the last CTIO in a sequence we generated completes before calling xpt_done. Executive summary- target mode actually now pretty much works well enough to tell folks about.
2000-07-18 06:58:28 +00:00
}
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
/*
* Handle ATIO stuff that the generic code can't.
* This means handling CDBs.
*/
static int
isp_handle_platform_atio(struct ispsoftc *isp, at_entry_t *aep)
{
tstate_t *tptr;
I don't know what I was thinking- if I have two separate busses on on SIM (as is true for the 1280 and the 12160), then I have to have separate flags && status for *both* busses. *Whap*. Implement condition variables for coordination with some target mode events. It's nice to use these and not panic in obscure little places in the kernel like 'propagate_priority' just because we went to sleep holding a mutex, or some other absurd thing. Remove some bogus ISP_UNLOCK calls. *Whap*. No longer require that somebody do a lun enable on the wildcard device to enable target mode. They are, in fact, orthogonal. A wildcard open is a statement that somebody upstream is willing to accept commands which are otherwise unrouteable. Now, for QLogic regular SCSI target mode, this won't matter for a damn because we'll never see ATIOs for luns we haven't enabled (are listening for, if you will). But for SCCLUN fibre channel SCSI, we get all kinds of ATIOs. We can either reflect them back here with minimal info (which is isp_target.c:isp_endcmd() is for), or the wildcard device (nominally targbh) can handle them. Do further checking against firmware attributes to see whether we can, in fact, support target mode in Fibre Channel. For now, require SCCLUN f/w to supoprt FC target mode. This is an awful lot of change, but target mode *still* isn't quite right. MFC after: 4 weeks
2001-09-04 21:53:12 +00:00
int status, bus, iswildcard;
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
struct ccb_accept_tio *atiop;
/*
* The firmware status (except for the QLTM_SVALID bit)
* indicates why this ATIO was sent to us.
*
* If QLTM_SVALID is set, the firware has recommended Sense Data.
*
* If the DISCONNECTS DISABLED bit is set in the flags field,
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
* we're still connected on the SCSI bus.
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
*/
status = aep->at_status;
if ((status & ~QLTM_SVALID) == AT_PHASE_ERROR) {
/*
* Bus Phase Sequence error. We should have sense data
* suggested by the f/w. I'm not sure quite yet what
* to do about this for CAM.
*/
Finally eliminate as many of the printf calls as possible (still leaving ones where we have a CAM path) and replacing them with calls to isp_prt., Eliminate isp_unit references- we no longer have an isp_unit- we now have an isp_dev that device_get_unit can work with.
2001-03-01 02:14:54 +00:00
isp_prt(isp, ISP_LOGWARN, "PHASE ERROR");
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
isp_endcmd(isp, aep, SCSI_STATUS_BUSY, 0);
return (0);
}
if ((status & ~QLTM_SVALID) != AT_CDB) {
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
isp_prt(isp, ISP_LOGWARN, "bad atio (0x%x) leaked to platform",
Finally eliminate as many of the printf calls as possible (still leaving ones where we have a CAM path) and replacing them with calls to isp_prt., Eliminate isp_unit references- we no longer have an isp_unit- we now have an isp_dev that device_get_unit can work with.
2001-03-01 02:14:54 +00:00
status);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
isp_endcmd(isp, aep, SCSI_STATUS_BUSY, 0);
return (0);
}
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
bus = GET_BUS_VAL(aep->at_iid);
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
tptr = get_lun_statep(isp, bus, aep->at_lun);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
if (tptr == NULL) {
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
tptr = get_lun_statep(isp, bus, CAM_LUN_WILDCARD);
Checkpoint of work in progress in cleaning up target mode. It actually seems to work well in RELENG_4. However, 5.X locking foo means that I'll have to do some quick redesign. Add ioctl handlers for ISP_GETROLE and ISP_SETROLE ioctls.
2004-02-07 03:47:33 +00:00
if (tptr == NULL) {
isp_endcmd(isp, aep,
SCSI_STATUS_CHECK_COND | ECMD_SVALID |
(0x5 << 12) | (0x25 << 16), 0);
return (0);
}
I don't know what I was thinking- if I have two separate busses on on SIM (as is true for the 1280 and the 12160), then I have to have separate flags && status for *both* busses. *Whap*. Implement condition variables for coordination with some target mode events. It's nice to use these and not panic in obscure little places in the kernel like 'propagate_priority' just because we went to sleep holding a mutex, or some other absurd thing. Remove some bogus ISP_UNLOCK calls. *Whap*. No longer require that somebody do a lun enable on the wildcard device to enable target mode. They are, in fact, orthogonal. A wildcard open is a statement that somebody upstream is willing to accept commands which are otherwise unrouteable. Now, for QLogic regular SCSI target mode, this won't matter for a damn because we'll never see ATIOs for luns we haven't enabled (are listening for, if you will). But for SCCLUN fibre channel SCSI, we get all kinds of ATIOs. We can either reflect them back here with minimal info (which is isp_target.c:isp_endcmd() is for), or the wildcard device (nominally targbh) can handle them. Do further checking against firmware attributes to see whether we can, in fact, support target mode in Fibre Channel. For now, require SCCLUN f/w to supoprt FC target mode. This is an awful lot of change, but target mode *still* isn't quite right. MFC after: 4 weeks
2001-09-04 21:53:12 +00:00
iswildcard = 1;
} else {
iswildcard = 0;
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
}
if (tptr == NULL) {
/*
* Because we can't autofeed sense data back with
* a command for parallel SCSI, we can't give back
* a CHECK CONDITION. We'll give back a BUSY status
* instead. This works out okay because the only
* time we should, in fact, get this, is in the
* case that somebody configured us without the
* blackhole driver, so they get what they deserve.
*/
isp_endcmd(isp, aep, SCSI_STATUS_BUSY, 0);
return (0);
}
atiop = (struct ccb_accept_tio *) SLIST_FIRST(&tptr->atios);
if (atiop == NULL) {
/*
* Because we can't autofeed sense data back with
* a command for parallel SCSI, we can't give back
* a CHECK CONDITION. We'll give back a QUEUE FULL status
* instead. This works out okay because the only time we
* should, in fact, get this, is in the case that we've
* run out of ATIOS.
*/
xpt_print_path(tptr->owner);
Finally eliminate as many of the printf calls as possible (still leaving ones where we have a CAM path) and replacing them with calls to isp_prt., Eliminate isp_unit references- we no longer have an isp_unit- we now have an isp_dev that device_get_unit can work with.
2001-03-01 02:14:54 +00:00
isp_prt(isp, ISP_LOGWARN,
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
"no ATIOS for lun %d from initiator %d on channel %d",
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
aep->at_lun, GET_IID_VAL(aep->at_iid), bus);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
if (aep->at_flags & AT_TQAE)
isp_endcmd(isp, aep, SCSI_STATUS_QUEUE_FULL, 0);
else
isp_endcmd(isp, aep, SCSI_STATUS_BUSY, 0);
I don't know what I was thinking- if I have two separate busses on on SIM (as is true for the 1280 and the 12160), then I have to have separate flags && status for *both* busses. *Whap*. Implement condition variables for coordination with some target mode events. It's nice to use these and not panic in obscure little places in the kernel like 'propagate_priority' just because we went to sleep holding a mutex, or some other absurd thing. Remove some bogus ISP_UNLOCK calls. *Whap*. No longer require that somebody do a lun enable on the wildcard device to enable target mode. They are, in fact, orthogonal. A wildcard open is a statement that somebody upstream is willing to accept commands which are otherwise unrouteable. Now, for QLogic regular SCSI target mode, this won't matter for a damn because we'll never see ATIOs for luns we haven't enabled (are listening for, if you will). But for SCCLUN fibre channel SCSI, we get all kinds of ATIOs. We can either reflect them back here with minimal info (which is isp_target.c:isp_endcmd() is for), or the wildcard device (nominally targbh) can handle them. Do further checking against firmware attributes to see whether we can, in fact, support target mode in Fibre Channel. For now, require SCCLUN f/w to supoprt FC target mode. This is an awful lot of change, but target mode *still* isn't quite right. MFC after: 4 weeks
2001-09-04 21:53:12 +00:00
rls_lun_statep(isp, tptr);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
return (0);
}
SLIST_REMOVE_HEAD(&tptr->atios, sim_links.sle);
Checkpoint of work in progress in cleaning up target mode. It actually seems to work well in RELENG_4. However, 5.X locking foo means that I'll have to do some quick redesign. Add ioctl handlers for ISP_GETROLE and ISP_SETROLE ioctls.
2004-02-07 03:47:33 +00:00
tptr->atio_count--;
isp_prt(isp, ISP_LOGTDEBUG0, "Take FREE ATIO lun %d, count now %d",
aep->at_lun, tptr->atio_count);
I don't know what I was thinking- if I have two separate busses on on SIM (as is true for the 1280 and the 12160), then I have to have separate flags && status for *both* busses. *Whap*. Implement condition variables for coordination with some target mode events. It's nice to use these and not panic in obscure little places in the kernel like 'propagate_priority' just because we went to sleep holding a mutex, or some other absurd thing. Remove some bogus ISP_UNLOCK calls. *Whap*. No longer require that somebody do a lun enable on the wildcard device to enable target mode. They are, in fact, orthogonal. A wildcard open is a statement that somebody upstream is willing to accept commands which are otherwise unrouteable. Now, for QLogic regular SCSI target mode, this won't matter for a damn because we'll never see ATIOs for luns we haven't enabled (are listening for, if you will). But for SCCLUN fibre channel SCSI, we get all kinds of ATIOs. We can either reflect them back here with minimal info (which is isp_target.c:isp_endcmd() is for), or the wildcard device (nominally targbh) can handle them. Do further checking against firmware attributes to see whether we can, in fact, support target mode in Fibre Channel. For now, require SCCLUN f/w to supoprt FC target mode. This is an awful lot of change, but target mode *still* isn't quite right. MFC after: 4 weeks
2001-09-04 21:53:12 +00:00
if (iswildcard) {
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
atiop->ccb_h.target_id = aep->at_tgt;
atiop->ccb_h.target_lun = aep->at_lun;
}
if (aep->at_flags & AT_NODISC) {
Add a isp_target_putback_atio- we aren't using CCINCR at this time, so we need a function that tells the Qlogic f/w that a target mode command is done, so increase the resource count for that lun. Add in a timeout function to kick the putback again if we fail to do it the first time (we may not have the request queue space for ATIO push). Split the function isp_handle_platform_ctio into two parts so that the timeout function for the ATIO push or isp_handle_platform_ctio can inform CAM that the requested CTIO(s) are now done. Clean up (cough) residual handling. What we need for Fibre Channel is to preserve the at_datalen field from the original incoming ATIO so we can calculate a 'true' residual. Unfortunately, we're not guaranteed to get that back from CAM. We'll *try* to find it hiding in the periph_priv field (layering violation)- but if an ATIO was passed in from user land- forget it. This means that we'll probably get residuals wrong for Fibre Channel commands we're completing with an error. It's too late to 4.1 release to fix this- too bad. Luckily the only device we'd really care about this occurring on is a tape device and they're still so rare as FC attached devices that this can be considered an untested combination anyway. Remove all CCINCR usage (resource autoreplenish). When we've proved to ourself that things are working properly, we can add it back in. Make sure we propage 'suggested' sense data from the incoming ATIO into the created system ATIO- and set sense_len appropriately. Correctly propagate tag values. Fall back to the model of generating (well, the functions in isp_pci.c do the work) multiple CTIOs based upon what we get from XPT. Instead of being able to pair Qlogic generated ATIOs with CAM ATIOs, and then to pair CAM CTIOs with Qlogic CTIOs, we have to take the CTIO passed to us from XPT, and if it implies that we have to generate extra Qlogic CTIOs, so be it. This means that we have to wait until the last CTIO in a sequence we generated completes before calling xpt_done. Executive summary- target mode actually now pretty much works well enough to tell folks about.
2000-07-18 06:58:28 +00:00
atiop->ccb_h.flags = CAM_DIS_DISCONNECT;
} else {
atiop->ccb_h.flags = 0;
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
}
Add a isp_target_putback_atio- we aren't using CCINCR at this time, so we need a function that tells the Qlogic f/w that a target mode command is done, so increase the resource count for that lun. Add in a timeout function to kick the putback again if we fail to do it the first time (we may not have the request queue space for ATIO push). Split the function isp_handle_platform_ctio into two parts so that the timeout function for the ATIO push or isp_handle_platform_ctio can inform CAM that the requested CTIO(s) are now done. Clean up (cough) residual handling. What we need for Fibre Channel is to preserve the at_datalen field from the original incoming ATIO so we can calculate a 'true' residual. Unfortunately, we're not guaranteed to get that back from CAM. We'll *try* to find it hiding in the periph_priv field (layering violation)- but if an ATIO was passed in from user land- forget it. This means that we'll probably get residuals wrong for Fibre Channel commands we're completing with an error. It's too late to 4.1 release to fix this- too bad. Luckily the only device we'd really care about this occurring on is a tape device and they're still so rare as FC attached devices that this can be considered an untested combination anyway. Remove all CCINCR usage (resource autoreplenish). When we've proved to ourself that things are working properly, we can add it back in. Make sure we propage 'suggested' sense data from the incoming ATIO into the created system ATIO- and set sense_len appropriately. Correctly propagate tag values. Fall back to the model of generating (well, the functions in isp_pci.c do the work) multiple CTIOs based upon what we get from XPT. Instead of being able to pair Qlogic generated ATIOs with CAM ATIOs, and then to pair CAM CTIOs with Qlogic CTIOs, we have to take the CTIO passed to us from XPT, and if it implies that we have to generate extra Qlogic CTIOs, so be it. This means that we have to wait until the last CTIO in a sequence we generated completes before calling xpt_done. Executive summary- target mode actually now pretty much works well enough to tell folks about.
2000-07-18 06:58:28 +00:00
if (status & QLTM_SVALID) {
size_t amt = imin(QLTM_SENSELEN, sizeof (atiop->sense_data));
atiop->sense_len = amt;
MEMCPY(&atiop->sense_data, aep->at_sense, amt);
} else {
atiop->sense_len = 0;
}
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
atiop->init_id = GET_IID_VAL(aep->at_iid);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
atiop->cdb_len = aep->at_cdblen;
MEMCPY(atiop->cdb_io.cdb_bytes, aep->at_cdb, aep->at_cdblen);
atiop->ccb_h.status = CAM_CDB_RECVD;
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
/*
* Construct a tag 'id' based upon tag value (which may be 0..255)
* and the handle (which we have to preserve).
*/
AT_MAKE_TAGID needs an instance as the 2nd arg- not just a 0.
2005-07-31 23:21:19 +00:00
AT_MAKE_TAGID(atiop->tag_id, device_get_unit(isp->isp_dev), aep);
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
if (aep->at_flags & AT_TQAE) {
atiop->tag_action = aep->at_tag_type;
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
atiop->ccb_h.status |= CAM_TAG_ACTION_VALID;
}
xpt_done((union ccb*)atiop);
I don't know what I was thinking- if I have two separate busses on on SIM (as is true for the 1280 and the 12160), then I have to have separate flags && status for *both* busses. *Whap*. Implement condition variables for coordination with some target mode events. It's nice to use these and not panic in obscure little places in the kernel like 'propagate_priority' just because we went to sleep holding a mutex, or some other absurd thing. Remove some bogus ISP_UNLOCK calls. *Whap*. No longer require that somebody do a lun enable on the wildcard device to enable target mode. They are, in fact, orthogonal. A wildcard open is a statement that somebody upstream is willing to accept commands which are otherwise unrouteable. Now, for QLogic regular SCSI target mode, this won't matter for a damn because we'll never see ATIOs for luns we haven't enabled (are listening for, if you will). But for SCCLUN fibre channel SCSI, we get all kinds of ATIOs. We can either reflect them back here with minimal info (which is isp_target.c:isp_endcmd() is for), or the wildcard device (nominally targbh) can handle them. Do further checking against firmware attributes to see whether we can, in fact, support target mode in Fibre Channel. For now, require SCCLUN f/w to supoprt FC target mode. This is an awful lot of change, but target mode *still* isn't quite right. MFC after: 4 weeks
2001-09-04 21:53:12 +00:00
isp_prt(isp, ISP_LOGTDEBUG0,
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
"ATIO[%x] CDB=0x%x bus %d iid%d->lun%d tag 0x%x ttype 0x%x %s",
aep->at_handle, aep->at_cdb[0] & 0xff, GET_BUS_VAL(aep->at_iid),
GET_IID_VAL(aep->at_iid), aep->at_lun, aep->at_tag_val & 0xff,
aep->at_tag_type, (aep->at_flags & AT_NODISC)?
"nondisc" : "disconnecting");
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
rls_lun_statep(isp, tptr);
return (0);
}
static int
isp_handle_platform_atio2(struct ispsoftc *isp, at2_entry_t *aep)
{
Clean up some debug printing. Find the correct lun when SCCLUN is defined. If we complete with a check condition but no sense data, say we had an AUTOSENSE failure.
2000-01-04 03:45:49 +00:00
lun_id_t lun;
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
tstate_t *tptr;
struct ccb_accept_tio *atiop;
Begin to implement target mode that for Fibre Channel has a private per-command component that we *don't* try and pass thru CAM. CAM just is too risky and too much of a pain- structures get copied, but not all info of interest can be considered safely transported thru all consumers (including user space) from the incoming ATIO to the outgoing CTIO- it's just much safer to have a buddy structure, identified by the command's tag which *does* make it thru safely. Pay attention to link speed and report 200MB/s xfer speed for a 23XX card in 2GPs mode. MFC after: 1 week
2001-10-01 03:48:42 +00:00
atio_private_data_t *atp;
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
/*
* The firmware status (except for the QLTM_SVALID bit)
* indicates why this ATIO was sent to us.
*
* If QLTM_SVALID is set, the firware has recommended Sense Data.
*/
if ((aep->at_status & ~QLTM_SVALID) != AT_CDB) {
Finally eliminate as many of the printf calls as possible (still leaving ones where we have a CAM path) and replacing them with calls to isp_prt., Eliminate isp_unit references- we no longer have an isp_unit- we now have an isp_dev that device_get_unit can work with.
2001-03-01 02:14:54 +00:00
isp_prt(isp, ISP_LOGWARN,
"bogus atio (0x%x) leaked to platform", aep->at_status);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
isp_endcmd(isp, aep, SCSI_STATUS_BUSY, 0);
return (0);
}
I don't know what I was thinking- if I have two separate busses on on SIM (as is true for the 1280 and the 12160), then I have to have separate flags && status for *both* busses. *Whap*. Implement condition variables for coordination with some target mode events. It's nice to use these and not panic in obscure little places in the kernel like 'propagate_priority' just because we went to sleep holding a mutex, or some other absurd thing. Remove some bogus ISP_UNLOCK calls. *Whap*. No longer require that somebody do a lun enable on the wildcard device to enable target mode. They are, in fact, orthogonal. A wildcard open is a statement that somebody upstream is willing to accept commands which are otherwise unrouteable. Now, for QLogic regular SCSI target mode, this won't matter for a damn because we'll never see ATIOs for luns we haven't enabled (are listening for, if you will). But for SCCLUN fibre channel SCSI, we get all kinds of ATIOs. We can either reflect them back here with minimal info (which is isp_target.c:isp_endcmd() is for), or the wildcard device (nominally targbh) can handle them. Do further checking against firmware attributes to see whether we can, in fact, support target mode in Fibre Channel. For now, require SCCLUN f/w to supoprt FC target mode. This is an awful lot of change, but target mode *still* isn't quite right. MFC after: 4 weeks
2001-09-04 21:53:12 +00:00
if ((FCPARAM(isp)->isp_fwattr & ISP_FW_ATTR_SCCLUN) != 0) {
Remove all ISP2100_SCCLUN define protected code and replace it with runtime checks.
2000-06-18 04:50:26 +00:00
lun = aep->at_scclun;
} else {
lun = aep->at_lun;
}
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
tptr = get_lun_statep(isp, 0, lun);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
if (tptr == NULL) {
Checkpoint of work in progress in cleaning up target mode. It actually seems to work well in RELENG_4. However, 5.X locking foo means that I'll have to do some quick redesign. Add ioctl handlers for ISP_GETROLE and ISP_SETROLE ioctls.
2004-02-07 03:47:33 +00:00
isp_prt(isp, ISP_LOGTDEBUG0,
"[0x%x] no state pointer for lun %d", aep->at_rxid, lun);
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
tptr = get_lun_statep(isp, 0, CAM_LUN_WILDCARD);
Checkpoint of work in progress in cleaning up target mode. It actually seems to work well in RELENG_4. However, 5.X locking foo means that I'll have to do some quick redesign. Add ioctl handlers for ISP_GETROLE and ISP_SETROLE ioctls.
2004-02-07 03:47:33 +00:00
if (tptr == NULL) {
isp_endcmd(isp, aep,
SCSI_STATUS_CHECK_COND | ECMD_SVALID |
(0x5 << 12) | (0x25 << 16), 0);
return (0);
}
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
}
Begin to implement target mode that for Fibre Channel has a private per-command component that we *don't* try and pass thru CAM. CAM just is too risky and too much of a pain- structures get copied, but not all info of interest can be considered safely transported thru all consumers (including user space) from the incoming ATIO to the outgoing CTIO- it's just much safer to have a buddy structure, identified by the command's tag which *does* make it thru safely. Pay attention to link speed and report 200MB/s xfer speed for a 23XX card in 2GPs mode. MFC after: 1 week
2001-10-01 03:48:42 +00:00
atp = isp_get_atpd(isp, 0);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
atiop = (struct ccb_accept_tio *) SLIST_FIRST(&tptr->atios);
Begin to implement target mode that for Fibre Channel has a private per-command component that we *don't* try and pass thru CAM. CAM just is too risky and too much of a pain- structures get copied, but not all info of interest can be considered safely transported thru all consumers (including user space) from the incoming ATIO to the outgoing CTIO- it's just much safer to have a buddy structure, identified by the command's tag which *does* make it thru safely. Pay attention to link speed and report 200MB/s xfer speed for a 23XX card in 2GPs mode. MFC after: 1 week
2001-10-01 03:48:42 +00:00
if (atiop == NULL || atp == NULL) {
Checkpoint of work in progress in cleaning up target mode. It actually seems to work well in RELENG_4. However, 5.X locking foo means that I'll have to do some quick redesign. Add ioctl handlers for ISP_GETROLE and ISP_SETROLE ioctls.
2004-02-07 03:47:33 +00:00
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
/*
* Because we can't autofeed sense data back with
* a command for parallel SCSI, we can't give back
* a CHECK CONDITION. We'll give back a QUEUE FULL status
* instead. This works out okay because the only time we
* should, in fact, get this, is in the case that we've
* run out of ATIOS.
*/
xpt_print_path(tptr->owner);
Finally eliminate as many of the printf calls as possible (still leaving ones where we have a CAM path) and replacing them with calls to isp_prt., Eliminate isp_unit references- we no longer have an isp_unit- we now have an isp_dev that device_get_unit can work with.
2001-03-01 02:14:54 +00:00
isp_prt(isp, ISP_LOGWARN,
Add support for ISP_FC_GETHINFO, which returns current connection topology, speed, loopid, WWPN/WWNN, etc. Beef up target mode. Add isp_handle_platform_notify_scsi and isp_handle_platform_notify_fc platform handlers to handle immediate notifies (isp_handle_platform_notify_scsi is still stubbed out). In implementation of isp_handle_platform_notify_fc, for IN_ABORT_TASK, peel off a pending XPT_IMMED_NOTIFY and call xpt_done on it and hope that somebody upstream is listening. Make sure on final CTIO2s that we set residual correctly. These are absolutely crucial. Make sure we set relative offset for each CTIO2 based upon bytes we've already xferred. This is what the private adjunct datat to the original ATIO is. Note state of command so we can figure out where to find it if we get an ABORT from the firmware. Make sure we *always* set CAM_TAG_ACTION_VALID for ATIO2s. Make sure we keep track of the original lun. If se sent status (or we're otherwise done with the command), don't forget to free the adjunct structure.
2002-06-16 05:08:02 +00:00
"no %s for lun %d from initiator %d",
(atp == NULL && atiop == NULL)? "ATIO2s *or* ATPS" :
((atp == NULL)? "ATPs" : "ATIO2s"), lun, aep->at_iid);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
rls_lun_statep(isp, tptr);
Add support for ISP_FC_GETHINFO, which returns current connection topology, speed, loopid, WWPN/WWNN, etc. Beef up target mode. Add isp_handle_platform_notify_scsi and isp_handle_platform_notify_fc platform handlers to handle immediate notifies (isp_handle_platform_notify_scsi is still stubbed out). In implementation of isp_handle_platform_notify_fc, for IN_ABORT_TASK, peel off a pending XPT_IMMED_NOTIFY and call xpt_done on it and hope that somebody upstream is listening. Make sure on final CTIO2s that we set residual correctly. These are absolutely crucial. Make sure we set relative offset for each CTIO2 based upon bytes we've already xferred. This is what the private adjunct datat to the original ATIO is. Note state of command so we can figure out where to find it if we get an ABORT from the firmware. Make sure we *always* set CAM_TAG_ACTION_VALID for ATIO2s. Make sure we keep track of the original lun. If se sent status (or we're otherwise done with the command), don't forget to free the adjunct structure.
2002-06-16 05:08:02 +00:00
isp_endcmd(isp, aep, SCSI_STATUS_QUEUE_FULL, 0);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
return (0);
}
Add support for ISP_FC_GETHINFO, which returns current connection topology, speed, loopid, WWPN/WWNN, etc. Beef up target mode. Add isp_handle_platform_notify_scsi and isp_handle_platform_notify_fc platform handlers to handle immediate notifies (isp_handle_platform_notify_scsi is still stubbed out). In implementation of isp_handle_platform_notify_fc, for IN_ABORT_TASK, peel off a pending XPT_IMMED_NOTIFY and call xpt_done on it and hope that somebody upstream is listening. Make sure on final CTIO2s that we set residual correctly. These are absolutely crucial. Make sure we set relative offset for each CTIO2 based upon bytes we've already xferred. This is what the private adjunct datat to the original ATIO is. Note state of command so we can figure out where to find it if we get an ABORT from the firmware. Make sure we *always* set CAM_TAG_ACTION_VALID for ATIO2s. Make sure we keep track of the original lun. If se sent status (or we're otherwise done with the command), don't forget to free the adjunct structure.
2002-06-16 05:08:02 +00:00
atp->state = ATPD_STATE_ATIO;
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
SLIST_REMOVE_HEAD(&tptr->atios, sim_links.sle);
Add support for ISP_FC_GETHINFO, which returns current connection topology, speed, loopid, WWPN/WWNN, etc. Beef up target mode. Add isp_handle_platform_notify_scsi and isp_handle_platform_notify_fc platform handlers to handle immediate notifies (isp_handle_platform_notify_scsi is still stubbed out). In implementation of isp_handle_platform_notify_fc, for IN_ABORT_TASK, peel off a pending XPT_IMMED_NOTIFY and call xpt_done on it and hope that somebody upstream is listening. Make sure on final CTIO2s that we set residual correctly. These are absolutely crucial. Make sure we set relative offset for each CTIO2 based upon bytes we've already xferred. This is what the private adjunct datat to the original ATIO is. Note state of command so we can figure out where to find it if we get an ABORT from the firmware. Make sure we *always* set CAM_TAG_ACTION_VALID for ATIO2s. Make sure we keep track of the original lun. If se sent status (or we're otherwise done with the command), don't forget to free the adjunct structure.
2002-06-16 05:08:02 +00:00
tptr->atio_count--;
Checkpoint of work in progress in cleaning up target mode. It actually seems to work well in RELENG_4. However, 5.X locking foo means that I'll have to do some quick redesign. Add ioctl handlers for ISP_GETROLE and ISP_SETROLE ioctls.
2004-02-07 03:47:33 +00:00
isp_prt(isp, ISP_LOGTDEBUG0, "Take FREE ATIO lun %d, count now %d",
Add support for ISP_FC_GETHINFO, which returns current connection topology, speed, loopid, WWPN/WWNN, etc. Beef up target mode. Add isp_handle_platform_notify_scsi and isp_handle_platform_notify_fc platform handlers to handle immediate notifies (isp_handle_platform_notify_scsi is still stubbed out). In implementation of isp_handle_platform_notify_fc, for IN_ABORT_TASK, peel off a pending XPT_IMMED_NOTIFY and call xpt_done on it and hope that somebody upstream is listening. Make sure on final CTIO2s that we set residual correctly. These are absolutely crucial. Make sure we set relative offset for each CTIO2 based upon bytes we've already xferred. This is what the private adjunct datat to the original ATIO is. Note state of command so we can figure out where to find it if we get an ABORT from the firmware. Make sure we *always* set CAM_TAG_ACTION_VALID for ATIO2s. Make sure we keep track of the original lun. If se sent status (or we're otherwise done with the command), don't forget to free the adjunct structure.
2002-06-16 05:08:02 +00:00
lun, tptr->atio_count);
Add a isp_target_putback_atio- we aren't using CCINCR at this time, so we need a function that tells the Qlogic f/w that a target mode command is done, so increase the resource count for that lun. Add in a timeout function to kick the putback again if we fail to do it the first time (we may not have the request queue space for ATIO push). Split the function isp_handle_platform_ctio into two parts so that the timeout function for the ATIO push or isp_handle_platform_ctio can inform CAM that the requested CTIO(s) are now done. Clean up (cough) residual handling. What we need for Fibre Channel is to preserve the at_datalen field from the original incoming ATIO so we can calculate a 'true' residual. Unfortunately, we're not guaranteed to get that back from CAM. We'll *try* to find it hiding in the periph_priv field (layering violation)- but if an ATIO was passed in from user land- forget it. This means that we'll probably get residuals wrong for Fibre Channel commands we're completing with an error. It's too late to 4.1 release to fix this- too bad. Luckily the only device we'd really care about this occurring on is a tape device and they're still so rare as FC attached devices that this can be considered an untested combination anyway. Remove all CCINCR usage (resource autoreplenish). When we've proved to ourself that things are working properly, we can add it back in. Make sure we propage 'suggested' sense data from the incoming ATIO into the created system ATIO- and set sense_len appropriately. Correctly propagate tag values. Fall back to the model of generating (well, the functions in isp_pci.c do the work) multiple CTIOs based upon what we get from XPT. Instead of being able to pair Qlogic generated ATIOs with CAM ATIOs, and then to pair CAM CTIOs with Qlogic CTIOs, we have to take the CTIO passed to us from XPT, and if it implies that we have to generate extra Qlogic CTIOs, so be it. This means that we have to wait until the last CTIO in a sequence we generated completes before calling xpt_done. Executive summary- target mode actually now pretty much works well enough to tell folks about.
2000-07-18 06:58:28 +00:00
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
if (tptr == &isp->isp_osinfo.tsdflt[0]) {
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
atiop->ccb_h.target_id =
Add support for ISP_FC_GETHINFO, which returns current connection topology, speed, loopid, WWPN/WWNN, etc. Beef up target mode. Add isp_handle_platform_notify_scsi and isp_handle_platform_notify_fc platform handlers to handle immediate notifies (isp_handle_platform_notify_scsi is still stubbed out). In implementation of isp_handle_platform_notify_fc, for IN_ABORT_TASK, peel off a pending XPT_IMMED_NOTIFY and call xpt_done on it and hope that somebody upstream is listening. Make sure on final CTIO2s that we set residual correctly. These are absolutely crucial. Make sure we set relative offset for each CTIO2 based upon bytes we've already xferred. This is what the private adjunct datat to the original ATIO is. Note state of command so we can figure out where to find it if we get an ABORT from the firmware. Make sure we *always* set CAM_TAG_ACTION_VALID for ATIO2s. Make sure we keep track of the original lun. If se sent status (or we're otherwise done with the command), don't forget to free the adjunct structure.
2002-06-16 05:08:02 +00:00
((fcparam *)isp->isp_param)->isp_loopid;
Clean up some debug printing. Find the correct lun when SCCLUN is defined. If we complete with a check condition but no sense data, say we had an AUTOSENSE failure.
2000-01-04 03:45:49 +00:00
atiop->ccb_h.target_lun = lun;
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
}
Fix at2_entry_t to reflect what the firmware actually writes (instead of just deriving from SCSI at_entry_t). In this case, there is no 'suggested sense' for FC cards.
2001-02-27 00:14:39 +00:00
/*
* We don't get 'suggested' sense data as we do with SCSI cards.
*/
atiop->sense_len = 0;
Add a isp_target_putback_atio- we aren't using CCINCR at this time, so we need a function that tells the Qlogic f/w that a target mode command is done, so increase the resource count for that lun. Add in a timeout function to kick the putback again if we fail to do it the first time (we may not have the request queue space for ATIO push). Split the function isp_handle_platform_ctio into two parts so that the timeout function for the ATIO push or isp_handle_platform_ctio can inform CAM that the requested CTIO(s) are now done. Clean up (cough) residual handling. What we need for Fibre Channel is to preserve the at_datalen field from the original incoming ATIO so we can calculate a 'true' residual. Unfortunately, we're not guaranteed to get that back from CAM. We'll *try* to find it hiding in the periph_priv field (layering violation)- but if an ATIO was passed in from user land- forget it. This means that we'll probably get residuals wrong for Fibre Channel commands we're completing with an error. It's too late to 4.1 release to fix this- too bad. Luckily the only device we'd really care about this occurring on is a tape device and they're still so rare as FC attached devices that this can be considered an untested combination anyway. Remove all CCINCR usage (resource autoreplenish). When we've proved to ourself that things are working properly, we can add it back in. Make sure we propage 'suggested' sense data from the incoming ATIO into the created system ATIO- and set sense_len appropriately. Correctly propagate tag values. Fall back to the model of generating (well, the functions in isp_pci.c do the work) multiple CTIOs based upon what we get from XPT. Instead of being able to pair Qlogic generated ATIOs with CAM ATIOs, and then to pair CAM CTIOs with Qlogic CTIOs, we have to take the CTIO passed to us from XPT, and if it implies that we have to generate extra Qlogic CTIOs, so be it. This means that we have to wait until the last CTIO in a sequence we generated completes before calling xpt_done. Executive summary- target mode actually now pretty much works well enough to tell folks about.
2000-07-18 06:58:28 +00:00
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
atiop->init_id = aep->at_iid;
atiop->cdb_len = ATIO2_CDBLEN;
MEMCPY(atiop->cdb_io.cdb_bytes, aep->at_cdb, ATIO2_CDBLEN);
atiop->ccb_h.status = CAM_CDB_RECVD;
atiop->tag_id = aep->at_rxid;
switch (aep->at_taskflags & ATIO2_TC_ATTR_MASK) {
case ATIO2_TC_ATTR_SIMPLEQ:
atiop->tag_action = MSG_SIMPLE_Q_TAG;
break;
case ATIO2_TC_ATTR_HEADOFQ:
atiop->tag_action = MSG_HEAD_OF_Q_TAG;
break;
case ATIO2_TC_ATTR_ORDERED:
atiop->tag_action = MSG_ORDERED_Q_TAG;
break;
case ATIO2_TC_ATTR_ACAQ: /* ?? */
case ATIO2_TC_ATTR_UNTAGGED:
default:
atiop->tag_action = 0;
break;
}
Add support for ISP_FC_GETHINFO, which returns current connection topology, speed, loopid, WWPN/WWNN, etc. Beef up target mode. Add isp_handle_platform_notify_scsi and isp_handle_platform_notify_fc platform handlers to handle immediate notifies (isp_handle_platform_notify_scsi is still stubbed out). In implementation of isp_handle_platform_notify_fc, for IN_ABORT_TASK, peel off a pending XPT_IMMED_NOTIFY and call xpt_done on it and hope that somebody upstream is listening. Make sure on final CTIO2s that we set residual correctly. These are absolutely crucial. Make sure we set relative offset for each CTIO2 based upon bytes we've already xferred. This is what the private adjunct datat to the original ATIO is. Note state of command so we can figure out where to find it if we get an ABORT from the firmware. Make sure we *always* set CAM_TAG_ACTION_VALID for ATIO2s. Make sure we keep track of the original lun. If se sent status (or we're otherwise done with the command), don't forget to free the adjunct structure.
2002-06-16 05:08:02 +00:00
atiop->ccb_h.flags = CAM_TAG_ACTION_VALID;
Add a isp_target_putback_atio- we aren't using CCINCR at this time, so we need a function that tells the Qlogic f/w that a target mode command is done, so increase the resource count for that lun. Add in a timeout function to kick the putback again if we fail to do it the first time (we may not have the request queue space for ATIO push). Split the function isp_handle_platform_ctio into two parts so that the timeout function for the ATIO push or isp_handle_platform_ctio can inform CAM that the requested CTIO(s) are now done. Clean up (cough) residual handling. What we need for Fibre Channel is to preserve the at_datalen field from the original incoming ATIO so we can calculate a 'true' residual. Unfortunately, we're not guaranteed to get that back from CAM. We'll *try* to find it hiding in the periph_priv field (layering violation)- but if an ATIO was passed in from user land- forget it. This means that we'll probably get residuals wrong for Fibre Channel commands we're completing with an error. It's too late to 4.1 release to fix this- too bad. Luckily the only device we'd really care about this occurring on is a tape device and they're still so rare as FC attached devices that this can be considered an untested combination anyway. Remove all CCINCR usage (resource autoreplenish). When we've proved to ourself that things are working properly, we can add it back in. Make sure we propage 'suggested' sense data from the incoming ATIO into the created system ATIO- and set sense_len appropriately. Correctly propagate tag values. Fall back to the model of generating (well, the functions in isp_pci.c do the work) multiple CTIOs based upon what we get from XPT. Instead of being able to pair Qlogic generated ATIOs with CAM ATIOs, and then to pair CAM CTIOs with Qlogic CTIOs, we have to take the CTIO passed to us from XPT, and if it implies that we have to generate extra Qlogic CTIOs, so be it. This means that we have to wait until the last CTIO in a sequence we generated completes before calling xpt_done. Executive summary- target mode actually now pretty much works well enough to tell folks about.
2000-07-18 06:58:28 +00:00
Begin to implement target mode that for Fibre Channel has a private per-command component that we *don't* try and pass thru CAM. CAM just is too risky and too much of a pain- structures get copied, but not all info of interest can be considered safely transported thru all consumers (including user space) from the incoming ATIO to the outgoing CTIO- it's just much safer to have a buddy structure, identified by the command's tag which *does* make it thru safely. Pay attention to link speed and report 200MB/s xfer speed for a 23XX card in 2GPs mode. MFC after: 1 week
2001-10-01 03:48:42 +00:00
atp->tag = atiop->tag_id;
Add support for ISP_FC_GETHINFO, which returns current connection topology, speed, loopid, WWPN/WWNN, etc. Beef up target mode. Add isp_handle_platform_notify_scsi and isp_handle_platform_notify_fc platform handlers to handle immediate notifies (isp_handle_platform_notify_scsi is still stubbed out). In implementation of isp_handle_platform_notify_fc, for IN_ABORT_TASK, peel off a pending XPT_IMMED_NOTIFY and call xpt_done on it and hope that somebody upstream is listening. Make sure on final CTIO2s that we set residual correctly. These are absolutely crucial. Make sure we set relative offset for each CTIO2 based upon bytes we've already xferred. This is what the private adjunct datat to the original ATIO is. Note state of command so we can figure out where to find it if we get an ABORT from the firmware. Make sure we *always* set CAM_TAG_ACTION_VALID for ATIO2s. Make sure we keep track of the original lun. If se sent status (or we're otherwise done with the command), don't forget to free the adjunct structure.
2002-06-16 05:08:02 +00:00
atp->lun = lun;
Begin to implement target mode that for Fibre Channel has a private per-command component that we *don't* try and pass thru CAM. CAM just is too risky and too much of a pain- structures get copied, but not all info of interest can be considered safely transported thru all consumers (including user space) from the incoming ATIO to the outgoing CTIO- it's just much safer to have a buddy structure, identified by the command's tag which *does* make it thru safely. Pay attention to link speed and report 200MB/s xfer speed for a 23XX card in 2GPs mode. MFC after: 1 week
2001-10-01 03:48:42 +00:00
atp->orig_datalen = aep->at_datalen;
atp->last_xframt = 0;
atp->bytes_xfered = 0;
Add support for ISP_FC_GETHINFO, which returns current connection topology, speed, loopid, WWPN/WWNN, etc. Beef up target mode. Add isp_handle_platform_notify_scsi and isp_handle_platform_notify_fc platform handlers to handle immediate notifies (isp_handle_platform_notify_scsi is still stubbed out). In implementation of isp_handle_platform_notify_fc, for IN_ABORT_TASK, peel off a pending XPT_IMMED_NOTIFY and call xpt_done on it and hope that somebody upstream is listening. Make sure on final CTIO2s that we set residual correctly. These are absolutely crucial. Make sure we set relative offset for each CTIO2 based upon bytes we've already xferred. This is what the private adjunct datat to the original ATIO is. Note state of command so we can figure out where to find it if we get an ABORT from the firmware. Make sure we *always* set CAM_TAG_ACTION_VALID for ATIO2s. Make sure we keep track of the original lun. If se sent status (or we're otherwise done with the command), don't forget to free the adjunct structure.
2002-06-16 05:08:02 +00:00
atp->state = ATPD_STATE_CAM;
Remove condition variables and status associated with target mode enabling. Instead, go to an interrupt/polled model. Fix get_lun_statep so we don't panic if there are no wildcard luns enabled. MFC after: 6 days
2004-02-08 19:17:56 +00:00
ISPLOCK_2_CAMLOCK(siP);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
xpt_done((union ccb*)atiop);
Add support for ISP_FC_GETHINFO, which returns current connection topology, speed, loopid, WWPN/WWNN, etc. Beef up target mode. Add isp_handle_platform_notify_scsi and isp_handle_platform_notify_fc platform handlers to handle immediate notifies (isp_handle_platform_notify_scsi is still stubbed out). In implementation of isp_handle_platform_notify_fc, for IN_ABORT_TASK, peel off a pending XPT_IMMED_NOTIFY and call xpt_done on it and hope that somebody upstream is listening. Make sure on final CTIO2s that we set residual correctly. These are absolutely crucial. Make sure we set relative offset for each CTIO2 based upon bytes we've already xferred. This is what the private adjunct datat to the original ATIO is. Note state of command so we can figure out where to find it if we get an ABORT from the firmware. Make sure we *always* set CAM_TAG_ACTION_VALID for ATIO2s. Make sure we keep track of the original lun. If se sent status (or we're otherwise done with the command), don't forget to free the adjunct structure.
2002-06-16 05:08:02 +00:00
I don't know what I was thinking- if I have two separate busses on on SIM (as is true for the 1280 and the 12160), then I have to have separate flags && status for *both* busses. *Whap*. Implement condition variables for coordination with some target mode events. It's nice to use these and not panic in obscure little places in the kernel like 'propagate_priority' just because we went to sleep holding a mutex, or some other absurd thing. Remove some bogus ISP_UNLOCK calls. *Whap*. No longer require that somebody do a lun enable on the wildcard device to enable target mode. They are, in fact, orthogonal. A wildcard open is a statement that somebody upstream is willing to accept commands which are otherwise unrouteable. Now, for QLogic regular SCSI target mode, this won't matter for a damn because we'll never see ATIOs for luns we haven't enabled (are listening for, if you will). But for SCCLUN fibre channel SCSI, we get all kinds of ATIOs. We can either reflect them back here with minimal info (which is isp_target.c:isp_endcmd() is for), or the wildcard device (nominally targbh) can handle them. Do further checking against firmware attributes to see whether we can, in fact, support target mode in Fibre Channel. For now, require SCCLUN f/w to supoprt FC target mode. This is an awful lot of change, but target mode *still* isn't quite right. MFC after: 4 weeks
2001-09-04 21:53:12 +00:00
isp_prt(isp, ISP_LOGTDEBUG0,
Switch to using 16 bit handles instead of 32 bit handles. This is a pretty invasive change, but there are three good reasons to do this: 1. We'll never have > 16 bits of handle. 2. We can (eventually) enable the RIO (Reduced Interrupt Operation) bits which return multiple completing 16 bit handles in mailbox registers. 3. The !)$*)$*~)@$*~)$* Qlogic target mode for parallel SCSI spec changed such that at_reserved (which was 32 bits) was split into two pieces- and one of which was a 16 bit handle id that functions like the at_rxid for Fibre Channel (a tag for the f/w to correlate CTIOs with a particular command). Since we had to muck with that and this changed the whole handler architecture, we might as well... Propagate new at_handle on through int ct_fwhandle. Follow implications of changing to 16 bit handles. These above changes at least get Qlogic 1040 cards working in target mode again. 1080/12160 cards don't work yet. In isp.c: Prepare for doing all loop management in outer layers.
2001-03-02 06:28:55 +00:00
"ATIO2[%x] CDB=0x%x iid%d->lun%d tattr 0x%x datalen %u",
aep->at_rxid, aep->at_cdb[0] & 0xff, aep->at_iid,
Core version 2.0 rewrite. In this file we replace isp_tdebug with isp_prt calls. We now use an argument to the ISPCTL_FCLINK_TEST call. We change all IDPRINTF macros to isp_prt calls. We add the isp_prt function here.
2000-08-01 06:31:44 +00:00
lun, aep->at_taskflags, aep->at_datalen);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
rls_lun_statep(isp, tptr);
return (0);
}
static int
Add a isp_target_putback_atio- we aren't using CCINCR at this time, so we need a function that tells the Qlogic f/w that a target mode command is done, so increase the resource count for that lun. Add in a timeout function to kick the putback again if we fail to do it the first time (we may not have the request queue space for ATIO push). Split the function isp_handle_platform_ctio into two parts so that the timeout function for the ATIO push or isp_handle_platform_ctio can inform CAM that the requested CTIO(s) are now done. Clean up (cough) residual handling. What we need for Fibre Channel is to preserve the at_datalen field from the original incoming ATIO so we can calculate a 'true' residual. Unfortunately, we're not guaranteed to get that back from CAM. We'll *try* to find it hiding in the periph_priv field (layering violation)- but if an ATIO was passed in from user land- forget it. This means that we'll probably get residuals wrong for Fibre Channel commands we're completing with an error. It's too late to 4.1 release to fix this- too bad. Luckily the only device we'd really care about this occurring on is a tape device and they're still so rare as FC attached devices that this can be considered an untested combination anyway. Remove all CCINCR usage (resource autoreplenish). When we've proved to ourself that things are working properly, we can add it back in. Make sure we propage 'suggested' sense data from the incoming ATIO into the created system ATIO- and set sense_len appropriately. Correctly propagate tag values. Fall back to the model of generating (well, the functions in isp_pci.c do the work) multiple CTIOs based upon what we get from XPT. Instead of being able to pair Qlogic generated ATIOs with CAM ATIOs, and then to pair CAM CTIOs with Qlogic CTIOs, we have to take the CTIO passed to us from XPT, and if it implies that we have to generate extra Qlogic CTIOs, so be it. This means that we have to wait until the last CTIO in a sequence we generated completes before calling xpt_done. Executive summary- target mode actually now pretty much works well enough to tell folks about.
2000-07-18 06:58:28 +00:00
isp_handle_platform_ctio(struct ispsoftc *isp, void *arg)
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
{
union ccb *ccb;
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
int sentstatus, ok, notify_cam, resid = 0;
a) clean up some declaration stuff (i.e., make more modern with respect to getting rid u_int for uint and so on). b) Turn back on 64 bit DAC support. Cheeze it a bit in that we have two DMA callback functions- one when we have bus_addr_t > 4 bits in width and the other which should be normal. Even Cheezier in that we turn off setting up DMA maps to be BUS_SPACE_MAXADDR if we're in ISP_TARGET_MODE. More work on this in a week or so. c) Tested under amd64 and 1MB DFLTPHYS, sparc64, i386 (PAE, but insufficient memory to really test > 4GB). LINT check under amd64. MFC after: 1 month
2006-02-15 00:31:48 +00:00
uint16_t tval;
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
/*
* CTIO and CTIO2 are close enough....
*/
Store the target handles in a separate list from normal commands. Add a CTIO fast post routine to handle CTIO completions. Submitted by: mjacob
2004-05-24 07:02:25 +00:00
ccb = isp_find_xs_tgt(isp, ((ct_entry_t *)arg)->ct_syshandle);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
KASSERT((ccb != NULL), ("null ccb in isp_handle_platform_ctio"));
Store the target handles in a separate list from normal commands. Add a CTIO fast post routine to handle CTIO completions. Submitted by: mjacob
2004-05-24 07:02:25 +00:00
isp_destroy_tgt_handle(isp, ((ct_entry_t *)arg)->ct_syshandle);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
if (IS_FC(isp)) {
ct2_entry_t *ct = arg;
Add support for ISP_FC_GETHINFO, which returns current connection topology, speed, loopid, WWPN/WWNN, etc. Beef up target mode. Add isp_handle_platform_notify_scsi and isp_handle_platform_notify_fc platform handlers to handle immediate notifies (isp_handle_platform_notify_scsi is still stubbed out). In implementation of isp_handle_platform_notify_fc, for IN_ABORT_TASK, peel off a pending XPT_IMMED_NOTIFY and call xpt_done on it and hope that somebody upstream is listening. Make sure on final CTIO2s that we set residual correctly. These are absolutely crucial. Make sure we set relative offset for each CTIO2 based upon bytes we've already xferred. This is what the private adjunct datat to the original ATIO is. Note state of command so we can figure out where to find it if we get an ABORT from the firmware. Make sure we *always* set CAM_TAG_ACTION_VALID for ATIO2s. Make sure we keep track of the original lun. If se sent status (or we're otherwise done with the command), don't forget to free the adjunct structure.
2002-06-16 05:08:02 +00:00
atio_private_data_t *atp = isp_get_atpd(isp, ct->ct_rxid);
if (atp == NULL) {
isp_prt(isp, ISP_LOGERR,
"cannot find adjunct for %x after I/O",
ct->ct_rxid);
return (0);
}
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
sentstatus = ct->ct_flags & CT2_SENDSTATUS;
ok = (ct->ct_status & ~QLTM_SVALID) == CT_OK;
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
if (ok && sentstatus && (ccb->ccb_h.flags & CAM_SEND_SENSE)) {
a whale of a lot of target mode cleanup
2000-01-15 01:50:48 +00:00
ccb->ccb_h.status |= CAM_SENT_SENSE;
}
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
notify_cam = ct->ct_header.rqs_seqno & 0x1;
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
if ((ct->ct_flags & CT2_DATAMASK) != CT2_NO_DATA) {
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
resid = ct->ct_resid;
Begin to implement target mode that for Fibre Channel has a private per-command component that we *don't* try and pass thru CAM. CAM just is too risky and too much of a pain- structures get copied, but not all info of interest can be considered safely transported thru all consumers (including user space) from the incoming ATIO to the outgoing CTIO- it's just much safer to have a buddy structure, identified by the command's tag which *does* make it thru safely. Pay attention to link speed and report 200MB/s xfer speed for a 23XX card in 2GPs mode. MFC after: 1 week
2001-10-01 03:48:42 +00:00
atp->bytes_xfered += (atp->last_xframt - resid);
atp->last_xframt = 0;
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
}
Add support for ISP_FC_GETHINFO, which returns current connection topology, speed, loopid, WWPN/WWNN, etc. Beef up target mode. Add isp_handle_platform_notify_scsi and isp_handle_platform_notify_fc platform handlers to handle immediate notifies (isp_handle_platform_notify_scsi is still stubbed out). In implementation of isp_handle_platform_notify_fc, for IN_ABORT_TASK, peel off a pending XPT_IMMED_NOTIFY and call xpt_done on it and hope that somebody upstream is listening. Make sure on final CTIO2s that we set residual correctly. These are absolutely crucial. Make sure we set relative offset for each CTIO2 based upon bytes we've already xferred. This is what the private adjunct datat to the original ATIO is. Note state of command so we can figure out where to find it if we get an ABORT from the firmware. Make sure we *always* set CAM_TAG_ACTION_VALID for ATIO2s. Make sure we keep track of the original lun. If se sent status (or we're otherwise done with the command), don't forget to free the adjunct structure.
2002-06-16 05:08:02 +00:00
if (sentstatus || !ok) {
atp->tag = 0;
}
isp_prt(isp, ok? ISP_LOGTDEBUG0 : ISP_LOGWARN,
I don't know what I was thinking- if I have two separate busses on on SIM (as is true for the 1280 and the 12160), then I have to have separate flags && status for *both* busses. *Whap*. Implement condition variables for coordination with some target mode events. It's nice to use these and not panic in obscure little places in the kernel like 'propagate_priority' just because we went to sleep holding a mutex, or some other absurd thing. Remove some bogus ISP_UNLOCK calls. *Whap*. No longer require that somebody do a lun enable on the wildcard device to enable target mode. They are, in fact, orthogonal. A wildcard open is a statement that somebody upstream is willing to accept commands which are otherwise unrouteable. Now, for QLogic regular SCSI target mode, this won't matter for a damn because we'll never see ATIOs for luns we haven't enabled (are listening for, if you will). But for SCCLUN fibre channel SCSI, we get all kinds of ATIOs. We can either reflect them back here with minimal info (which is isp_target.c:isp_endcmd() is for), or the wildcard device (nominally targbh) can handle them. Do further checking against firmware attributes to see whether we can, in fact, support target mode in Fibre Channel. For now, require SCCLUN f/w to supoprt FC target mode. This is an awful lot of change, but target mode *still* isn't quite right. MFC after: 4 weeks
2001-09-04 21:53:12 +00:00
"CTIO2[%x] sts 0x%x flg 0x%x sns %d resid %d %s",
ct->ct_rxid, ct->ct_status, ct->ct_flags,
(ccb->ccb_h.status & CAM_SENT_SENSE) != 0,
resid, sentstatus? "FIN" : "MID");
tval = ct->ct_rxid;
Add support for ISP_FC_GETHINFO, which returns current connection topology, speed, loopid, WWPN/WWNN, etc. Beef up target mode. Add isp_handle_platform_notify_scsi and isp_handle_platform_notify_fc platform handlers to handle immediate notifies (isp_handle_platform_notify_scsi is still stubbed out). In implementation of isp_handle_platform_notify_fc, for IN_ABORT_TASK, peel off a pending XPT_IMMED_NOTIFY and call xpt_done on it and hope that somebody upstream is listening. Make sure on final CTIO2s that we set residual correctly. These are absolutely crucial. Make sure we set relative offset for each CTIO2 based upon bytes we've already xferred. This is what the private adjunct datat to the original ATIO is. Note state of command so we can figure out where to find it if we get an ABORT from the firmware. Make sure we *always* set CAM_TAG_ACTION_VALID for ATIO2s. Make sure we keep track of the original lun. If se sent status (or we're otherwise done with the command), don't forget to free the adjunct structure.
2002-06-16 05:08:02 +00:00
/* XXX: should really come after isp_complete_ctio */
atp->state = ATPD_STATE_PDON;
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
} else {
ct_entry_t *ct = arg;
sentstatus = ct->ct_flags & CT_SENDSTATUS;
ok = (ct->ct_status & ~QLTM_SVALID) == CT_OK;
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
/*
* We *ought* to be able to get back to the original ATIO
* here, but for some reason this gets lost. It's just as
* well because it's squirrelled away as part of periph
* private data.
*
* We can live without it as long as we continue to use
* the auto-replenish feature for CTIOs.
*/
notify_cam = ct->ct_header.rqs_seqno & 0x1;
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
if (ct->ct_status & QLTM_SVALID) {
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
char *sp = (char *)ct;
sp += CTIO_SENSE_OFFSET;
ccb->csio.sense_len =
min(sizeof (ccb->csio.sense_data), QLTM_SENSELEN);
MEMCPY(&ccb->csio.sense_data, sp, ccb->csio.sense_len);
ccb->ccb_h.status |= CAM_AUTOSNS_VALID;
}
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
if ((ct->ct_flags & CT_DATAMASK) != CT_NO_DATA) {
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
resid = ct->ct_resid;
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
}
I don't know what I was thinking- if I have two separate busses on on SIM (as is true for the 1280 and the 12160), then I have to have separate flags && status for *both* busses. *Whap*. Implement condition variables for coordination with some target mode events. It's nice to use these and not panic in obscure little places in the kernel like 'propagate_priority' just because we went to sleep holding a mutex, or some other absurd thing. Remove some bogus ISP_UNLOCK calls. *Whap*. No longer require that somebody do a lun enable on the wildcard device to enable target mode. They are, in fact, orthogonal. A wildcard open is a statement that somebody upstream is willing to accept commands which are otherwise unrouteable. Now, for QLogic regular SCSI target mode, this won't matter for a damn because we'll never see ATIOs for luns we haven't enabled (are listening for, if you will). But for SCCLUN fibre channel SCSI, we get all kinds of ATIOs. We can either reflect them back here with minimal info (which is isp_target.c:isp_endcmd() is for), or the wildcard device (nominally targbh) can handle them. Do further checking against firmware attributes to see whether we can, in fact, support target mode in Fibre Channel. For now, require SCCLUN f/w to supoprt FC target mode. This is an awful lot of change, but target mode *still* isn't quite right. MFC after: 4 weeks
2001-09-04 21:53:12 +00:00
isp_prt(isp, ISP_LOGTDEBUG0,
"CTIO[%x] tag %x iid %d lun %d sts %x flg %x resid %d %s",
ct->ct_fwhandle, ct->ct_tag_val, ct->ct_iid, ct->ct_lun,
ct->ct_status, ct->ct_flags, resid,
sentstatus? "FIN" : "MID");
tval = ct->ct_fwhandle;
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
}
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
ccb->csio.resid += resid;
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
/*
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
* We're here either because intermediate data transfers are done
* and/or the final status CTIO (which may have joined with a
* Data Transfer) is done.
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
*
* In any case, for this platform, the upper layers figure out
* what to do next, so all we do here is collect status and
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
* pass information along. Any DMA handles have already been
* freed.
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
*/
Add a isp_target_putback_atio- we aren't using CCINCR at this time, so we need a function that tells the Qlogic f/w that a target mode command is done, so increase the resource count for that lun. Add in a timeout function to kick the putback again if we fail to do it the first time (we may not have the request queue space for ATIO push). Split the function isp_handle_platform_ctio into two parts so that the timeout function for the ATIO push or isp_handle_platform_ctio can inform CAM that the requested CTIO(s) are now done. Clean up (cough) residual handling. What we need for Fibre Channel is to preserve the at_datalen field from the original incoming ATIO so we can calculate a 'true' residual. Unfortunately, we're not guaranteed to get that back from CAM. We'll *try* to find it hiding in the periph_priv field (layering violation)- but if an ATIO was passed in from user land- forget it. This means that we'll probably get residuals wrong for Fibre Channel commands we're completing with an error. It's too late to 4.1 release to fix this- too bad. Luckily the only device we'd really care about this occurring on is a tape device and they're still so rare as FC attached devices that this can be considered an untested combination anyway. Remove all CCINCR usage (resource autoreplenish). When we've proved to ourself that things are working properly, we can add it back in. Make sure we propage 'suggested' sense data from the incoming ATIO into the created system ATIO- and set sense_len appropriately. Correctly propagate tag values. Fall back to the model of generating (well, the functions in isp_pci.c do the work) multiple CTIOs based upon what we get from XPT. Instead of being able to pair Qlogic generated ATIOs with CAM ATIOs, and then to pair CAM CTIOs with Qlogic CTIOs, we have to take the CTIO passed to us from XPT, and if it implies that we have to generate extra Qlogic CTIOs, so be it. This means that we have to wait until the last CTIO in a sequence we generated completes before calling xpt_done. Executive summary- target mode actually now pretty much works well enough to tell folks about.
2000-07-18 06:58:28 +00:00
if (notify_cam == 0) {
I don't know what I was thinking- if I have two separate busses on on SIM (as is true for the 1280 and the 12160), then I have to have separate flags && status for *both* busses. *Whap*. Implement condition variables for coordination with some target mode events. It's nice to use these and not panic in obscure little places in the kernel like 'propagate_priority' just because we went to sleep holding a mutex, or some other absurd thing. Remove some bogus ISP_UNLOCK calls. *Whap*. No longer require that somebody do a lun enable on the wildcard device to enable target mode. They are, in fact, orthogonal. A wildcard open is a statement that somebody upstream is willing to accept commands which are otherwise unrouteable. Now, for QLogic regular SCSI target mode, this won't matter for a damn because we'll never see ATIOs for luns we haven't enabled (are listening for, if you will). But for SCCLUN fibre channel SCSI, we get all kinds of ATIOs. We can either reflect them back here with minimal info (which is isp_target.c:isp_endcmd() is for), or the wildcard device (nominally targbh) can handle them. Do further checking against firmware attributes to see whether we can, in fact, support target mode in Fibre Channel. For now, require SCCLUN f/w to supoprt FC target mode. This is an awful lot of change, but target mode *still* isn't quite right. MFC after: 4 weeks
2001-09-04 21:53:12 +00:00
isp_prt(isp, ISP_LOGTDEBUG0, " INTER CTIO[0x%x] done", tval);
Add a isp_target_putback_atio- we aren't using CCINCR at this time, so we need a function that tells the Qlogic f/w that a target mode command is done, so increase the resource count for that lun. Add in a timeout function to kick the putback again if we fail to do it the first time (we may not have the request queue space for ATIO push). Split the function isp_handle_platform_ctio into two parts so that the timeout function for the ATIO push or isp_handle_platform_ctio can inform CAM that the requested CTIO(s) are now done. Clean up (cough) residual handling. What we need for Fibre Channel is to preserve the at_datalen field from the original incoming ATIO so we can calculate a 'true' residual. Unfortunately, we're not guaranteed to get that back from CAM. We'll *try* to find it hiding in the periph_priv field (layering violation)- but if an ATIO was passed in from user land- forget it. This means that we'll probably get residuals wrong for Fibre Channel commands we're completing with an error. It's too late to 4.1 release to fix this- too bad. Luckily the only device we'd really care about this occurring on is a tape device and they're still so rare as FC attached devices that this can be considered an untested combination anyway. Remove all CCINCR usage (resource autoreplenish). When we've proved to ourself that things are working properly, we can add it back in. Make sure we propage 'suggested' sense data from the incoming ATIO into the created system ATIO- and set sense_len appropriately. Correctly propagate tag values. Fall back to the model of generating (well, the functions in isp_pci.c do the work) multiple CTIOs based upon what we get from XPT. Instead of being able to pair Qlogic generated ATIOs with CAM ATIOs, and then to pair CAM CTIOs with Qlogic CTIOs, we have to take the CTIO passed to us from XPT, and if it implies that we have to generate extra Qlogic CTIOs, so be it. This means that we have to wait until the last CTIO in a sequence we generated completes before calling xpt_done. Executive summary- target mode actually now pretty much works well enough to tell folks about.
2000-07-18 06:58:28 +00:00
return (0);
}
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
Begin to implement target mode that for Fibre Channel has a private per-command component that we *don't* try and pass thru CAM. CAM just is too risky and too much of a pain- structures get copied, but not all info of interest can be considered safely transported thru all consumers (including user space) from the incoming ATIO to the outgoing CTIO- it's just much safer to have a buddy structure, identified by the command's tag which *does* make it thru safely. Pay attention to link speed and report 200MB/s xfer speed for a 23XX card in 2GPs mode. MFC after: 1 week
2001-10-01 03:48:42 +00:00
isp_prt(isp, ISP_LOGTDEBUG0, "%s CTIO[0x%x] done",
(sentstatus)? " FINAL " : "MIDTERM ", tval);
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
if (!ok) {
isp_target_putback_atio(ccb);
Add a isp_target_putback_atio- we aren't using CCINCR at this time, so we need a function that tells the Qlogic f/w that a target mode command is done, so increase the resource count for that lun. Add in a timeout function to kick the putback again if we fail to do it the first time (we may not have the request queue space for ATIO push). Split the function isp_handle_platform_ctio into two parts so that the timeout function for the ATIO push or isp_handle_platform_ctio can inform CAM that the requested CTIO(s) are now done. Clean up (cough) residual handling. What we need for Fibre Channel is to preserve the at_datalen field from the original incoming ATIO so we can calculate a 'true' residual. Unfortunately, we're not guaranteed to get that back from CAM. We'll *try* to find it hiding in the periph_priv field (layering violation)- but if an ATIO was passed in from user land- forget it. This means that we'll probably get residuals wrong for Fibre Channel commands we're completing with an error. It's too late to 4.1 release to fix this- too bad. Luckily the only device we'd really care about this occurring on is a tape device and they're still so rare as FC attached devices that this can be considered an untested combination anyway. Remove all CCINCR usage (resource autoreplenish). When we've proved to ourself that things are working properly, we can add it back in. Make sure we propage 'suggested' sense data from the incoming ATIO into the created system ATIO- and set sense_len appropriately. Correctly propagate tag values. Fall back to the model of generating (well, the functions in isp_pci.c do the work) multiple CTIOs based upon what we get from XPT. Instead of being able to pair Qlogic generated ATIOs with CAM ATIOs, and then to pair CAM CTIOs with Qlogic CTIOs, we have to take the CTIO passed to us from XPT, and if it implies that we have to generate extra Qlogic CTIOs, so be it. This means that we have to wait until the last CTIO in a sequence we generated completes before calling xpt_done. Executive summary- target mode actually now pretty much works well enough to tell folks about.
2000-07-18 06:58:28 +00:00
} else {
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
isp_complete_ctio(ccb);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
}
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
return (0);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
}
Add support for ISP_FC_GETHINFO, which returns current connection topology, speed, loopid, WWPN/WWNN, etc. Beef up target mode. Add isp_handle_platform_notify_scsi and isp_handle_platform_notify_fc platform handlers to handle immediate notifies (isp_handle_platform_notify_scsi is still stubbed out). In implementation of isp_handle_platform_notify_fc, for IN_ABORT_TASK, peel off a pending XPT_IMMED_NOTIFY and call xpt_done on it and hope that somebody upstream is listening. Make sure on final CTIO2s that we set residual correctly. These are absolutely crucial. Make sure we set relative offset for each CTIO2 based upon bytes we've already xferred. This is what the private adjunct datat to the original ATIO is. Note state of command so we can figure out where to find it if we get an ABORT from the firmware. Make sure we *always* set CAM_TAG_ACTION_VALID for ATIO2s. Make sure we keep track of the original lun. If se sent status (or we're otherwise done with the command), don't forget to free the adjunct structure.
2002-06-16 05:08:02 +00:00
static int
isp_handle_platform_notify_scsi(struct ispsoftc *isp, in_entry_t *inp)
{
return (0); /* XXXX */
}
static int
isp_handle_platform_notify_fc(struct ispsoftc *isp, in_fcentry_t *inp)
{
switch (inp->in_status) {
case IN_PORT_LOGOUT:
isp_prt(isp, ISP_LOGWARN, "port logout of iid %d",
inp->in_iid);
break;
case IN_PORT_CHANGED:
isp_prt(isp, ISP_LOGWARN, "port changed for iid %d",
inp->in_iid);
break;
case IN_GLOBAL_LOGO:
isp_prt(isp, ISP_LOGINFO, "all ports logged out");
break;
case IN_ABORT_TASK:
{
atio_private_data_t *atp = isp_get_atpd(isp, inp->in_seqid);
struct ccb_immed_notify *inot = NULL;
if (atp) {
tstate_t *tptr = get_lun_statep(isp, 0, atp->lun);
if (tptr) {
inot = (struct ccb_immed_notify *)
SLIST_FIRST(&tptr->inots);
if (inot) {
Checkpoint of work in progress in cleaning up target mode. It actually seems to work well in RELENG_4. However, 5.X locking foo means that I'll have to do some quick redesign. Add ioctl handlers for ISP_GETROLE and ISP_SETROLE ioctls.
2004-02-07 03:47:33 +00:00
tptr->inot_count--;
Add support for ISP_FC_GETHINFO, which returns current connection topology, speed, loopid, WWPN/WWNN, etc. Beef up target mode. Add isp_handle_platform_notify_scsi and isp_handle_platform_notify_fc platform handlers to handle immediate notifies (isp_handle_platform_notify_scsi is still stubbed out). In implementation of isp_handle_platform_notify_fc, for IN_ABORT_TASK, peel off a pending XPT_IMMED_NOTIFY and call xpt_done on it and hope that somebody upstream is listening. Make sure on final CTIO2s that we set residual correctly. These are absolutely crucial. Make sure we set relative offset for each CTIO2 based upon bytes we've already xferred. This is what the private adjunct datat to the original ATIO is. Note state of command so we can figure out where to find it if we get an ABORT from the firmware. Make sure we *always* set CAM_TAG_ACTION_VALID for ATIO2s. Make sure we keep track of the original lun. If se sent status (or we're otherwise done with the command), don't forget to free the adjunct structure.
2002-06-16 05:08:02 +00:00
SLIST_REMOVE_HEAD(&tptr->inots,
sim_links.sle);
Checkpoint of work in progress in cleaning up target mode. It actually seems to work well in RELENG_4. However, 5.X locking foo means that I'll have to do some quick redesign. Add ioctl handlers for ISP_GETROLE and ISP_SETROLE ioctls.
2004-02-07 03:47:33 +00:00
isp_prt(isp, ISP_LOGTDEBUG0,
"Take FREE INOT count now %d",
tptr->inot_count);
Add support for ISP_FC_GETHINFO, which returns current connection topology, speed, loopid, WWPN/WWNN, etc. Beef up target mode. Add isp_handle_platform_notify_scsi and isp_handle_platform_notify_fc platform handlers to handle immediate notifies (isp_handle_platform_notify_scsi is still stubbed out). In implementation of isp_handle_platform_notify_fc, for IN_ABORT_TASK, peel off a pending XPT_IMMED_NOTIFY and call xpt_done on it and hope that somebody upstream is listening. Make sure on final CTIO2s that we set residual correctly. These are absolutely crucial. Make sure we set relative offset for each CTIO2 based upon bytes we've already xferred. This is what the private adjunct datat to the original ATIO is. Note state of command so we can figure out where to find it if we get an ABORT from the firmware. Make sure we *always* set CAM_TAG_ACTION_VALID for ATIO2s. Make sure we keep track of the original lun. If se sent status (or we're otherwise done with the command), don't forget to free the adjunct structure.
2002-06-16 05:08:02 +00:00
}
}
isp_prt(isp, ISP_LOGWARN,
"abort task RX_ID %x IID %d state %d",
inp->in_seqid, inp->in_iid, atp->state);
} else {
isp_prt(isp, ISP_LOGWARN,
"abort task RX_ID %x from iid %d, state unknown",
inp->in_seqid, inp->in_iid);
}
if (inot) {
inot->initiator_id = inp->in_iid;
inot->sense_len = 0;
inot->message_args[0] = MSG_ABORT_TAG;
inot->message_args[1] = inp->in_seqid & 0xff;
inot->message_args[2] = (inp->in_seqid >> 8) & 0xff;
inot->ccb_h.status = CAM_MESSAGE_RECV|CAM_DEV_QFRZN;
xpt_done((union ccb *)inot);
}
break;
}
default:
break;
}
return (0);
}
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
#endif
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
static void
a) clean up some declaration stuff (i.e., make more modern with respect to getting rid u_int for uint and so on). b) Turn back on 64 bit DAC support. Cheeze it a bit in that we have two DMA callback functions- one when we have bus_addr_t > 4 bits in width and the other which should be normal. Even Cheezier in that we turn off setting up DMA maps to be BUS_SPACE_MAXADDR if we're in ISP_TARGET_MODE. More work on this in a week or so. c) Tested under amd64 and 1MB DFLTPHYS, sparc64, i386 (PAE, but insufficient memory to really test > 4GB). LINT check under amd64. MFC after: 1 month
2006-02-15 00:31:48 +00:00
isp_cam_async(void *cbarg, uint32_t code, struct cam_path *path, void *arg)
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
{
struct cam_sim *sim;
struct ispsoftc *isp;
sim = (struct cam_sim *)cbarg;
isp = (struct ispsoftc *) cam_sim_softc(sim);
switch (code) {
case AC_LOST_DEVICE:
Add in isp_debug variable. It defaults to zero unless CAMDEBUG is defined where it defaults to one. Change simq width allocation to the max number of commands supported by the HBA after f/w fires up- not the constant MAXISPREQUEST value. Do some stylistic changes.
1999-10-17 18:50:47 +00:00
if (IS_SCSI(isp)) {
a) clean up some declaration stuff (i.e., make more modern with respect to getting rid u_int for uint and so on). b) Turn back on 64 bit DAC support. Cheeze it a bit in that we have two DMA callback functions- one when we have bus_addr_t > 4 bits in width and the other which should be normal. Even Cheezier in that we turn off setting up DMA maps to be BUS_SPACE_MAXADDR if we're in ISP_TARGET_MODE. More work on this in a week or so. c) Tested under amd64 and 1MB DFLTPHYS, sparc64, i386 (PAE, but insufficient memory to really test > 4GB). LINT check under amd64. MFC after: 1 month
2006-02-15 00:31:48 +00:00
uint16_t oflags, nflags;
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
sdparam *sdp = isp->isp_param;
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
int tgt;
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
Clean up lun width determination based upon f/w revisions for the parallel SCSI cards (4.55..4.65 :: 8.55..8.65).
1999-12-20 01:35:04 +00:00
tgt = xpt_path_target_id(path);
From PHK's flex tool- we can get -1 returned from xpt_path_target_id, so don't index off of it. MFC after: 1 day
2002-10-10 17:29:05 +00:00
if (tgt >= 0) {
sdp += cam_sim_bus(sim);
ISP_LOCK(isp);
nflags = sdp->isp_devparam[tgt].nvrm_flags;
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
#ifndef ISP_TARGET_MODE
From PHK's flex tool- we can get -1 returned from xpt_path_target_id, so don't index off of it. MFC after: 1 day
2002-10-10 17:29:05 +00:00
nflags &= DPARM_SAFE_DFLT;
if (isp->isp_loaded_fw) {
nflags |= DPARM_NARROW | DPARM_ASYNC;
}
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
#else
From PHK's flex tool- we can get -1 returned from xpt_path_target_id, so don't index off of it. MFC after: 1 day
2002-10-10 17:29:05 +00:00
nflags = DPARM_DEFAULT;
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
#endif
From PHK's flex tool- we can get -1 returned from xpt_path_target_id, so don't index off of it. MFC after: 1 day
2002-10-10 17:29:05 +00:00
oflags = sdp->isp_devparam[tgt].goal_flags;
sdp->isp_devparam[tgt].goal_flags = nflags;
sdp->isp_devparam[tgt].dev_update = 1;
isp->isp_update |= (1 << cam_sim_bus(sim));
(void) isp_control(isp,
ISPCTL_UPDATE_PARAMS, NULL);
sdp->isp_devparam[tgt].goal_flags = oflags;
ISP_UNLOCK(isp);
}
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
}
break;
default:
Finally eliminate as many of the printf calls as possible (still leaving ones where we have a CAM path) and replacing them with calls to isp_prt., Eliminate isp_unit references- we no longer have an isp_unit- we now have an isp_dev that device_get_unit can work with.
2001-03-01 02:14:54 +00:00
isp_prt(isp, ISP_LOGWARN, "isp_cam_async: Code 0x%x", code);
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
break;
}
}
static void
clarify headers;move uninit to outer layer;remove watchdog
1998-12-28 19:22:27 +00:00
isp_poll(struct cam_sim *sim)
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
{
Make sure we do locking if we call isp_intr. Make sure we enter Giant for now if we call into cam for completion.
2000-12-29 19:06:32 +00:00
struct ispsoftc *isp = cam_sim_softc(sim);
a) clean up some declaration stuff (i.e., make more modern with respect to getting rid u_int for uint and so on). b) Turn back on 64 bit DAC support. Cheeze it a bit in that we have two DMA callback functions- one when we have bus_addr_t > 4 bits in width and the other which should be normal. Even Cheezier in that we turn off setting up DMA maps to be BUS_SPACE_MAXADDR if we're in ISP_TARGET_MODE. More work on this in a week or so. c) Tested under amd64 and 1MB DFLTPHYS, sparc64, i386 (PAE, but insufficient memory to really test > 4GB). LINT check under amd64. MFC after: 1 month
2006-02-15 00:31:48 +00:00
uint16_t isr, sema, mbox;
Add 2 Gigabit Fibre Channel support (2300 && 2312 cards). This required some reworking (and consequent cleanup) of the interrupt service code. Also begin to start a cleanup of target mode support that will (eventually) not require more inforamtion routed with the ATIO to come back with the CTIO other than tag. MFC after: 4 weeks
2001-08-31 21:39:04 +00:00
Make sure we do locking if we call isp_intr. Make sure we enter Giant for now if we call into cam for completion.
2000-12-29 19:06:32 +00:00
ISP_LOCK(isp);
I don't know what I was thinking- if I have two separate busses on on SIM (as is true for the 1280 and the 12160), then I have to have separate flags && status for *both* busses. *Whap*. Implement condition variables for coordination with some target mode events. It's nice to use these and not panic in obscure little places in the kernel like 'propagate_priority' just because we went to sleep holding a mutex, or some other absurd thing. Remove some bogus ISP_UNLOCK calls. *Whap*. No longer require that somebody do a lun enable on the wildcard device to enable target mode. They are, in fact, orthogonal. A wildcard open is a statement that somebody upstream is willing to accept commands which are otherwise unrouteable. Now, for QLogic regular SCSI target mode, this won't matter for a damn because we'll never see ATIOs for luns we haven't enabled (are listening for, if you will). But for SCCLUN fibre channel SCSI, we get all kinds of ATIOs. We can either reflect them back here with minimal info (which is isp_target.c:isp_endcmd() is for), or the wildcard device (nominally targbh) can handle them. Do further checking against firmware attributes to see whether we can, in fact, support target mode in Fibre Channel. For now, require SCCLUN f/w to supoprt FC target mode. This is an awful lot of change, but target mode *still* isn't quite right. MFC after: 4 weeks
2001-09-04 21:53:12 +00:00
if (ISP_READ_ISR(isp, &isr, &sema, &mbox)) {
Add 2 Gigabit Fibre Channel support (2300 && 2312 cards). This required some reworking (and consequent cleanup) of the interrupt service code. Also begin to start a cleanup of target mode support that will (eventually) not require more inforamtion routed with the ATIO to come back with the CTIO other than tag. MFC after: 4 weeks
2001-08-31 21:39:04 +00:00
isp_intr(isp, isr, sema, mbox);
I don't know what I was thinking- if I have two separate busses on on SIM (as is true for the 1280 and the 12160), then I have to have separate flags && status for *both* busses. *Whap*. Implement condition variables for coordination with some target mode events. It's nice to use these and not panic in obscure little places in the kernel like 'propagate_priority' just because we went to sleep holding a mutex, or some other absurd thing. Remove some bogus ISP_UNLOCK calls. *Whap*. No longer require that somebody do a lun enable on the wildcard device to enable target mode. They are, in fact, orthogonal. A wildcard open is a statement that somebody upstream is willing to accept commands which are otherwise unrouteable. Now, for QLogic regular SCSI target mode, this won't matter for a damn because we'll never see ATIOs for luns we haven't enabled (are listening for, if you will). But for SCCLUN fibre channel SCSI, we get all kinds of ATIOs. We can either reflect them back here with minimal info (which is isp_target.c:isp_endcmd() is for), or the wildcard device (nominally targbh) can handle them. Do further checking against firmware attributes to see whether we can, in fact, support target mode in Fibre Channel. For now, require SCCLUN f/w to supoprt FC target mode. This is an awful lot of change, but target mode *still* isn't quite right. MFC after: 4 weeks
2001-09-04 21:53:12 +00:00
}
Make sure we do locking if we call isp_intr. Make sure we enter Giant for now if we call into cam for completion.
2000-12-29 19:06:32 +00:00
ISP_UNLOCK(isp);
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
}
Add in isp_debug variable. It defaults to zero unless CAMDEBUG is defined where it defaults to one. Change simq width allocation to the max number of commands supported by the HBA after f/w fires up- not the constant MAXISPREQUEST value. Do some stylistic changes.
1999-10-17 18:50:47 +00:00
Add in a watchdog routine to catch cases where we've dropped the command. Apparently the f/w has finished the command, but somehow an interrupt is being lost. So, we just plain wedge when booting alphas. This is a general routine we've needed for a while.
2000-05-09 01:08:21 +00:00
static void
Add in the enabling of interrupts (to isp_attach). Clean up a busted comment. Check against firmware state- not loop state when enabling target mode. Other changes have to do with no longer enabling/disabling interrupts at will. Rearchitect command watchdog timeouts- First of all, set the timeout period for a command that has a timeout (in isp_action) to the period of time requested *plus* two seconds. We don't want the Qlogic firmware and the host system to race each other to report a dead command (the watchdog is there to catch dead and/or broken firmware). Next, make sure that the command being watched isn't done yet. If it's not done yet, check for INT_PENDING and call isp_intr- if that said it serviced an interrupt, check to see whether the command is now done (this is what the "IN WATCHDOG" private flag is for- if isp_intr completes the command, it won't call xpt_done on it because isp_watchdog is still looking at the command). If no interrupt was pending, or the command wasn't completed, check to see if we've set the private 'grace period' flag. If so, the command really *is* dead, so report it as dead and complete it with a CAM_CMD_TIMEOUT value. If the grace period flag wasn't set, set it and issue a SYNCHRONIZE_ALL Marker Request Queue entry and re-set the timeout for one second from now (see Revision 1.45 isp.c notes for more on this) to give the firmware a final chance to complete this command.
2000-06-27 19:31:02 +00:00
isp_watchdog(void *arg)
Add in a watchdog routine to catch cases where we've dropped the command. Apparently the f/w has finished the command, but somehow an interrupt is being lost. So, we just plain wedge when booting alphas. This is a general routine we've needed for a while.
2000-05-09 01:08:21 +00:00
{
Core version 2.0 rewrite. In this file we replace isp_tdebug with isp_prt calls. We now use an argument to the ISPCTL_FCLINK_TEST call. We change all IDPRINTF macros to isp_prt calls. We add the isp_prt function here.
2000-08-01 06:31:44 +00:00
XS_T *xs = arg;
Add in a watchdog routine to catch cases where we've dropped the command. Apparently the f/w has finished the command, but somehow an interrupt is being lost. So, we just plain wedge when booting alphas. This is a general routine we've needed for a while.
2000-05-09 01:08:21 +00:00
struct ispsoftc *isp = XS_ISP(xs);
a) clean up some declaration stuff (i.e., make more modern with respect to getting rid u_int for uint and so on). b) Turn back on 64 bit DAC support. Cheeze it a bit in that we have two DMA callback functions- one when we have bus_addr_t > 4 bits in width and the other which should be normal. Even Cheezier in that we turn off setting up DMA maps to be BUS_SPACE_MAXADDR if we're in ISP_TARGET_MODE. More work on this in a week or so. c) Tested under amd64 and 1MB DFLTPHYS, sparc64, i386 (PAE, but insufficient memory to really test > 4GB). LINT check under amd64. MFC after: 1 month
2006-02-15 00:31:48 +00:00
uint32_t handle;
Add get/set param ioctl support. Remove sim queue freezes for resource shortages. I've had too many strange race conditions where I freeze on a resource shortage but never get unfrozen. Consolidate the remaining sim queue freeze condition (for loopdown) into an inline with debug messages that allows us to track problems at ISP_LOGDEBUG0 level easier. Change a bunch of debug messages about loop down/up conditions to ISP_LOGDEBUG0 level. Remove dead isp_relsim code. Change some internal flag stuff for efficiency. Complain vociferously if we try and use our FC scratch area while it's busy being used already (I mean, if we don't have solaris' ability to sleep as an interrupt thread which would allow us to just use a p/v semaphore, at least *say* when you've just borked yourself). Add infrastructure to allow overrides of hard loopid && initiator id from boot variables. Fix the usual quota of silly bugs: + 'ktmature' needs to be per-instance. Argh. + When entering isp_watchdog, set intsok to zero, preserving old value to restore later. It's not nice to try and sleep from splsoftclock. + Fix tick overflow buglet in checking timeout value. MFC after: 1 week
2002-07-08 17:42:47 +00:00
int iok;
Add in the enabling of interrupts (to isp_attach). Clean up a busted comment. Check against firmware state- not loop state when enabling target mode. Other changes have to do with no longer enabling/disabling interrupts at will. Rearchitect command watchdog timeouts- First of all, set the timeout period for a command that has a timeout (in isp_action) to the period of time requested *plus* two seconds. We don't want the Qlogic firmware and the host system to race each other to report a dead command (the watchdog is there to catch dead and/or broken firmware). Next, make sure that the command being watched isn't done yet. If it's not done yet, check for INT_PENDING and call isp_intr- if that said it serviced an interrupt, check to see whether the command is now done (this is what the "IN WATCHDOG" private flag is for- if isp_intr completes the command, it won't call xpt_done on it because isp_watchdog is still looking at the command). If no interrupt was pending, or the command wasn't completed, check to see if we've set the private 'grace period' flag. If so, the command really *is* dead, so report it as dead and complete it with a CAM_CMD_TIMEOUT value. If the grace period flag wasn't set, set it and issue a SYNCHRONIZE_ALL Marker Request Queue entry and re-set the timeout for one second from now (see Revision 1.45 isp.c notes for more on this) to give the firmware a final chance to complete this command.
2000-06-27 19:31:02 +00:00
Add in a watchdog routine to catch cases where we've dropped the command. Apparently the f/w has finished the command, but somehow an interrupt is being lost. So, we just plain wedge when booting alphas. This is a general routine we've needed for a while.
2000-05-09 01:08:21 +00:00
/*
Add in the enabling of interrupts (to isp_attach). Clean up a busted comment. Check against firmware state- not loop state when enabling target mode. Other changes have to do with no longer enabling/disabling interrupts at will. Rearchitect command watchdog timeouts- First of all, set the timeout period for a command that has a timeout (in isp_action) to the period of time requested *plus* two seconds. We don't want the Qlogic firmware and the host system to race each other to report a dead command (the watchdog is there to catch dead and/or broken firmware). Next, make sure that the command being watched isn't done yet. If it's not done yet, check for INT_PENDING and call isp_intr- if that said it serviced an interrupt, check to see whether the command is now done (this is what the "IN WATCHDOG" private flag is for- if isp_intr completes the command, it won't call xpt_done on it because isp_watchdog is still looking at the command). If no interrupt was pending, or the command wasn't completed, check to see if we've set the private 'grace period' flag. If so, the command really *is* dead, so report it as dead and complete it with a CAM_CMD_TIMEOUT value. If the grace period flag wasn't set, set it and issue a SYNCHRONIZE_ALL Marker Request Queue entry and re-set the timeout for one second from now (see Revision 1.45 isp.c notes for more on this) to give the firmware a final chance to complete this command.
2000-06-27 19:31:02 +00:00
* We've decided this command is dead. Make sure we're not trying
* to kill a command that's already dead by getting it's handle and
* and seeing whether it's still alive.
Add in a watchdog routine to catch cases where we've dropped the command. Apparently the f/w has finished the command, but somehow an interrupt is being lost. So, we just plain wedge when booting alphas. This is a general routine we've needed for a while.
2000-05-09 01:08:21 +00:00
*/
Add in config_hook for catching when interrupts are safe- this allows us to not the ints are ok and also to (re)ENABLE isp interrupts. Remove all splcam()/splx() invocates and replace them with ISP_LOCK/ISP_UNLOCK macros.
2000-07-04 01:05:43 +00:00
ISP_LOCK(isp);
Add get/set param ioctl support. Remove sim queue freezes for resource shortages. I've had too many strange race conditions where I freeze on a resource shortage but never get unfrozen. Consolidate the remaining sim queue freeze condition (for loopdown) into an inline with debug messages that allows us to track problems at ISP_LOGDEBUG0 level easier. Change a bunch of debug messages about loop down/up conditions to ISP_LOGDEBUG0 level. Remove dead isp_relsim code. Change some internal flag stuff for efficiency. Complain vociferously if we try and use our FC scratch area while it's busy being used already (I mean, if we don't have solaris' ability to sleep as an interrupt thread which would allow us to just use a p/v semaphore, at least *say* when you've just borked yourself). Add infrastructure to allow overrides of hard loopid && initiator id from boot variables. Fix the usual quota of silly bugs: + 'ktmature' needs to be per-instance. Argh. + When entering isp_watchdog, set intsok to zero, preserving old value to restore later. It's not nice to try and sleep from splsoftclock. + Fix tick overflow buglet in checking timeout value. MFC after: 1 week
2002-07-08 17:42:47 +00:00
iok = isp->isp_osinfo.intsok;
isp->isp_osinfo.intsok = 0;
Add in a watchdog routine to catch cases where we've dropped the command. Apparently the f/w has finished the command, but somehow an interrupt is being lost. So, we just plain wedge when booting alphas. This is a general routine we've needed for a while.
2000-05-09 01:08:21 +00:00
handle = isp_find_handle(isp, xs);
if (handle) {
a) clean up some declaration stuff (i.e., make more modern with respect to getting rid u_int for uint and so on). b) Turn back on 64 bit DAC support. Cheeze it a bit in that we have two DMA callback functions- one when we have bus_addr_t > 4 bits in width and the other which should be normal. Even Cheezier in that we turn off setting up DMA maps to be BUS_SPACE_MAXADDR if we're in ISP_TARGET_MODE. More work on this in a week or so. c) Tested under amd64 and 1MB DFLTPHYS, sparc64, i386 (PAE, but insufficient memory to really test > 4GB). LINT check under amd64. MFC after: 1 month
2006-02-15 00:31:48 +00:00
uint16_t isr, sema, mbox;
Add 2 Gigabit Fibre Channel support (2300 && 2312 cards). This required some reworking (and consequent cleanup) of the interrupt service code. Also begin to start a cleanup of target mode support that will (eventually) not require more inforamtion routed with the ATIO to come back with the CTIO other than tag. MFC after: 4 weeks
2001-08-31 21:39:04 +00:00
Add in the enabling of interrupts (to isp_attach). Clean up a busted comment. Check against firmware state- not loop state when enabling target mode. Other changes have to do with no longer enabling/disabling interrupts at will. Rearchitect command watchdog timeouts- First of all, set the timeout period for a command that has a timeout (in isp_action) to the period of time requested *plus* two seconds. We don't want the Qlogic firmware and the host system to race each other to report a dead command (the watchdog is there to catch dead and/or broken firmware). Next, make sure that the command being watched isn't done yet. If it's not done yet, check for INT_PENDING and call isp_intr- if that said it serviced an interrupt, check to see whether the command is now done (this is what the "IN WATCHDOG" private flag is for- if isp_intr completes the command, it won't call xpt_done on it because isp_watchdog is still looking at the command). If no interrupt was pending, or the command wasn't completed, check to see if we've set the private 'grace period' flag. If so, the command really *is* dead, so report it as dead and complete it with a CAM_CMD_TIMEOUT value. If the grace period flag wasn't set, set it and issue a SYNCHRONIZE_ALL Marker Request Queue entry and re-set the timeout for one second from now (see Revision 1.45 isp.c notes for more on this) to give the firmware a final chance to complete this command.
2000-06-27 19:31:02 +00:00
if (XS_CMD_DONE_P(xs)) {
Core version 2.0 rewrite. In this file we replace isp_tdebug with isp_prt calls. We now use an argument to the ISPCTL_FCLINK_TEST call. We change all IDPRINTF macros to isp_prt calls. We add the isp_prt function here.
2000-08-01 06:31:44 +00:00
isp_prt(isp, ISP_LOGDEBUG1,
"watchdog found done cmd (handle 0x%x)", handle);
Add in config_hook for catching when interrupts are safe- this allows us to not the ints are ok and also to (re)ENABLE isp interrupts. Remove all splcam()/splx() invocates and replace them with ISP_LOCK/ISP_UNLOCK macros.
2000-07-04 01:05:43 +00:00
ISP_UNLOCK(isp);
Add in the enabling of interrupts (to isp_attach). Clean up a busted comment. Check against firmware state- not loop state when enabling target mode. Other changes have to do with no longer enabling/disabling interrupts at will. Rearchitect command watchdog timeouts- First of all, set the timeout period for a command that has a timeout (in isp_action) to the period of time requested *plus* two seconds. We don't want the Qlogic firmware and the host system to race each other to report a dead command (the watchdog is there to catch dead and/or broken firmware). Next, make sure that the command being watched isn't done yet. If it's not done yet, check for INT_PENDING and call isp_intr- if that said it serviced an interrupt, check to see whether the command is now done (this is what the "IN WATCHDOG" private flag is for- if isp_intr completes the command, it won't call xpt_done on it because isp_watchdog is still looking at the command). If no interrupt was pending, or the command wasn't completed, check to see if we've set the private 'grace period' flag. If so, the command really *is* dead, so report it as dead and complete it with a CAM_CMD_TIMEOUT value. If the grace period flag wasn't set, set it and issue a SYNCHRONIZE_ALL Marker Request Queue entry and re-set the timeout for one second from now (see Revision 1.45 isp.c notes for more on this) to give the firmware a final chance to complete this command.
2000-06-27 19:31:02 +00:00
return;
}
if (XS_CMD_WDOG_P(xs)) {
Core version 2.0 rewrite. In this file we replace isp_tdebug with isp_prt calls. We now use an argument to the ISPCTL_FCLINK_TEST call. We change all IDPRINTF macros to isp_prt calls. We add the isp_prt function here.
2000-08-01 06:31:44 +00:00
isp_prt(isp, ISP_LOGDEBUG2,
"recursive watchdog (handle 0x%x)", handle);
Add in config_hook for catching when interrupts are safe- this allows us to not the ints are ok and also to (re)ENABLE isp interrupts. Remove all splcam()/splx() invocates and replace them with ISP_LOCK/ISP_UNLOCK macros.
2000-07-04 01:05:43 +00:00
ISP_UNLOCK(isp);
Add in the enabling of interrupts (to isp_attach). Clean up a busted comment. Check against firmware state- not loop state when enabling target mode. Other changes have to do with no longer enabling/disabling interrupts at will. Rearchitect command watchdog timeouts- First of all, set the timeout period for a command that has a timeout (in isp_action) to the period of time requested *plus* two seconds. We don't want the Qlogic firmware and the host system to race each other to report a dead command (the watchdog is there to catch dead and/or broken firmware). Next, make sure that the command being watched isn't done yet. If it's not done yet, check for INT_PENDING and call isp_intr- if that said it serviced an interrupt, check to see whether the command is now done (this is what the "IN WATCHDOG" private flag is for- if isp_intr completes the command, it won't call xpt_done on it because isp_watchdog is still looking at the command). If no interrupt was pending, or the command wasn't completed, check to see if we've set the private 'grace period' flag. If so, the command really *is* dead, so report it as dead and complete it with a CAM_CMD_TIMEOUT value. If the grace period flag wasn't set, set it and issue a SYNCHRONIZE_ALL Marker Request Queue entry and re-set the timeout for one second from now (see Revision 1.45 isp.c notes for more on this) to give the firmware a final chance to complete this command.
2000-06-27 19:31:02 +00:00
return;
}
XS_CMD_S_WDOG(xs);
Add 2 Gigabit Fibre Channel support (2300 && 2312 cards). This required some reworking (and consequent cleanup) of the interrupt service code. Also begin to start a cleanup of target mode support that will (eventually) not require more inforamtion routed with the ATIO to come back with the CTIO other than tag. MFC after: 4 weeks
2001-08-31 21:39:04 +00:00
if (ISP_READ_ISR(isp, &isr, &sema, &mbox)) {
isp_intr(isp, isr, sema, mbox);
}
if (XS_CMD_DONE_P(xs)) {
Core version 2.0 rewrite. In this file we replace isp_tdebug with isp_prt calls. We now use an argument to the ISPCTL_FCLINK_TEST call. We change all IDPRINTF macros to isp_prt calls. We add the isp_prt function here.
2000-08-01 06:31:44 +00:00
isp_prt(isp, ISP_LOGDEBUG2,
Add 2 Gigabit Fibre Channel support (2300 && 2312 cards). This required some reworking (and consequent cleanup) of the interrupt service code. Also begin to start a cleanup of target mode support that will (eventually) not require more inforamtion routed with the ATIO to come back with the CTIO other than tag. MFC after: 4 weeks
2001-08-31 21:39:04 +00:00
"watchdog cleanup for handle 0x%x", handle);
Add in the enabling of interrupts (to isp_attach). Clean up a busted comment. Check against firmware state- not loop state when enabling target mode. Other changes have to do with no longer enabling/disabling interrupts at will. Rearchitect command watchdog timeouts- First of all, set the timeout period for a command that has a timeout (in isp_action) to the period of time requested *plus* two seconds. We don't want the Qlogic firmware and the host system to race each other to report a dead command (the watchdog is there to catch dead and/or broken firmware). Next, make sure that the command being watched isn't done yet. If it's not done yet, check for INT_PENDING and call isp_intr- if that said it serviced an interrupt, check to see whether the command is now done (this is what the "IN WATCHDOG" private flag is for- if isp_intr completes the command, it won't call xpt_done on it because isp_watchdog is still looking at the command). If no interrupt was pending, or the command wasn't completed, check to see if we've set the private 'grace period' flag. If so, the command really *is* dead, so report it as dead and complete it with a CAM_CMD_TIMEOUT value. If the grace period flag wasn't set, set it and issue a SYNCHRONIZE_ALL Marker Request Queue entry and re-set the timeout for one second from now (see Revision 1.45 isp.c notes for more on this) to give the firmware a final chance to complete this command.
2000-06-27 19:31:02 +00:00
xpt_done((union ccb *) xs);
} else if (XS_CMD_GRACE_P(xs)) {
Oops! If we're deciding a command is now really dead, make *darned* sure that it really is by issuing a ISPCTL_ABORT_CMD just on the off chance the f/w will start it up again and, ha ha, start using the DMA resources we gave it but are now taking away.
2000-07-05 06:44:17 +00:00
/*
* Make sure the command is *really* dead before we
* release the handle (and DMA resources) for reuse.
*/
(void) isp_control(isp, ISPCTL_ABORT_CMD, arg);
/*
* After this point, the comamnd is really dead.
*/
Add in config_hook for catching when interrupts are safe- this allows us to not the ints are ok and also to (re)ENABLE isp interrupts. Remove all splcam()/splx() invocates and replace them with ISP_LOCK/ISP_UNLOCK macros.
2000-07-04 01:05:43 +00:00
if (XS_XFRLEN(xs)) {
ISP_DMAFREE(isp, xs, handle);
}
Add in the enabling of interrupts (to isp_attach). Clean up a busted comment. Check against firmware state- not loop state when enabling target mode. Other changes have to do with no longer enabling/disabling interrupts at will. Rearchitect command watchdog timeouts- First of all, set the timeout period for a command that has a timeout (in isp_action) to the period of time requested *plus* two seconds. We don't want the Qlogic firmware and the host system to race each other to report a dead command (the watchdog is there to catch dead and/or broken firmware). Next, make sure that the command being watched isn't done yet. If it's not done yet, check for INT_PENDING and call isp_intr- if that said it serviced an interrupt, check to see whether the command is now done (this is what the "IN WATCHDOG" private flag is for- if isp_intr completes the command, it won't call xpt_done on it because isp_watchdog is still looking at the command). If no interrupt was pending, or the command wasn't completed, check to see if we've set the private 'grace period' flag. If so, the command really *is* dead, so report it as dead and complete it with a CAM_CMD_TIMEOUT value. If the grace period flag wasn't set, set it and issue a SYNCHRONIZE_ALL Marker Request Queue entry and re-set the timeout for one second from now (see Revision 1.45 isp.c notes for more on this) to give the firmware a final chance to complete this command.
2000-06-27 19:31:02 +00:00
isp_destroy_handle(isp, handle);
xpt_print_path(xs->ccb_h.path);
Finally eliminate as many of the printf calls as possible (still leaving ones where we have a CAM path) and replacing them with calls to isp_prt., Eliminate isp_unit references- we no longer have an isp_unit- we now have an isp_dev that device_get_unit can work with.
2001-03-01 02:14:54 +00:00
isp_prt(isp, ISP_LOGWARN,
Implement REDUCED INTERRUPT OPERATION usage form 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. MFC after: 2 weeks
2002-01-03 20:43:22 +00:00
"watchdog timeout for handle 0x%x", handle);
Add in the enabling of interrupts (to isp_attach). Clean up a busted comment. Check against firmware state- not loop state when enabling target mode. Other changes have to do with no longer enabling/disabling interrupts at will. Rearchitect command watchdog timeouts- First of all, set the timeout period for a command that has a timeout (in isp_action) to the period of time requested *plus* two seconds. We don't want the Qlogic firmware and the host system to race each other to report a dead command (the watchdog is there to catch dead and/or broken firmware). Next, make sure that the command being watched isn't done yet. If it's not done yet, check for INT_PENDING and call isp_intr- if that said it serviced an interrupt, check to see whether the command is now done (this is what the "IN WATCHDOG" private flag is for- if isp_intr completes the command, it won't call xpt_done on it because isp_watchdog is still looking at the command). If no interrupt was pending, or the command wasn't completed, check to see if we've set the private 'grace period' flag. If so, the command really *is* dead, so report it as dead and complete it with a CAM_CMD_TIMEOUT value. If the grace period flag wasn't set, set it and issue a SYNCHRONIZE_ALL Marker Request Queue entry and re-set the timeout for one second from now (see Revision 1.45 isp.c notes for more on this) to give the firmware a final chance to complete this command.
2000-06-27 19:31:02 +00:00
XS_SETERR(xs, CAM_CMD_TIMEOUT);
XS_CMD_C_WDOG(xs);
isp_done(xs);
} else {
a) clean up some declaration stuff (i.e., make more modern with respect to getting rid u_int for uint and so on). b) Turn back on 64 bit DAC support. Cheeze it a bit in that we have two DMA callback functions- one when we have bus_addr_t > 4 bits in width and the other which should be normal. Even Cheezier in that we turn off setting up DMA maps to be BUS_SPACE_MAXADDR if we're in ISP_TARGET_MODE. More work on this in a week or so. c) Tested under amd64 and 1MB DFLTPHYS, sparc64, i386 (PAE, but insufficient memory to really test > 4GB). LINT check under amd64. MFC after: 1 month
2006-02-15 00:31:48 +00:00
uint16_t nxti, optr;
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). It also has to handle the SBus cards (for platforms that have them) which, while on a Big Endian machine, do *not* require *most* of the request/response queue entry fields to be swizzled or unswizzled. 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. 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). MFC after: 2 weeks
2001-12-11 00:18:45 +00:00
ispreq_t local, *mp= &local, *qe;
Add in the enabling of interrupts (to isp_attach). Clean up a busted comment. Check against firmware state- not loop state when enabling target mode. Other changes have to do with no longer enabling/disabling interrupts at will. Rearchitect command watchdog timeouts- First of all, set the timeout period for a command that has a timeout (in isp_action) to the period of time requested *plus* two seconds. We don't want the Qlogic firmware and the host system to race each other to report a dead command (the watchdog is there to catch dead and/or broken firmware). Next, make sure that the command being watched isn't done yet. If it's not done yet, check for INT_PENDING and call isp_intr- if that said it serviced an interrupt, check to see whether the command is now done (this is what the "IN WATCHDOG" private flag is for- if isp_intr completes the command, it won't call xpt_done on it because isp_watchdog is still looking at the command). If no interrupt was pending, or the command wasn't completed, check to see if we've set the private 'grace period' flag. If so, the command really *is* dead, so report it as dead and complete it with a CAM_CMD_TIMEOUT value. If the grace period flag wasn't set, set it and issue a SYNCHRONIZE_ALL Marker Request Queue entry and re-set the timeout for one second from now (see Revision 1.45 isp.c notes for more on this) to give the firmware a final chance to complete this command.
2000-06-27 19:31:02 +00:00
XS_CMD_C_WDOG(xs);
xs->ccb_h.timeout_ch = timeout(isp_watchdog, xs, hz);
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). It also has to handle the SBus cards (for platforms that have them) which, while on a Big Endian machine, do *not* require *most* of the request/response queue entry fields to be swizzled or unswizzled. 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. 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). MFC after: 2 weeks
2001-12-11 00:18:45 +00:00
if (isp_getrqentry(isp, &nxti, &optr, (void **) &qe)) {
Add in config_hook for catching when interrupts are safe- this allows us to not the ints are ok and also to (re)ENABLE isp interrupts. Remove all splcam()/splx() invocates and replace them with ISP_LOCK/ISP_UNLOCK macros.
2000-07-04 01:05:43 +00:00
ISP_UNLOCK(isp);
Add in the enabling of interrupts (to isp_attach). Clean up a busted comment. Check against firmware state- not loop state when enabling target mode. Other changes have to do with no longer enabling/disabling interrupts at will. Rearchitect command watchdog timeouts- First of all, set the timeout period for a command that has a timeout (in isp_action) to the period of time requested *plus* two seconds. We don't want the Qlogic firmware and the host system to race each other to report a dead command (the watchdog is there to catch dead and/or broken firmware). Next, make sure that the command being watched isn't done yet. If it's not done yet, check for INT_PENDING and call isp_intr- if that said it serviced an interrupt, check to see whether the command is now done (this is what the "IN WATCHDOG" private flag is for- if isp_intr completes the command, it won't call xpt_done on it because isp_watchdog is still looking at the command). If no interrupt was pending, or the command wasn't completed, check to see if we've set the private 'grace period' flag. If so, the command really *is* dead, so report it as dead and complete it with a CAM_CMD_TIMEOUT value. If the grace period flag wasn't set, set it and issue a SYNCHRONIZE_ALL Marker Request Queue entry and re-set the timeout for one second from now (see Revision 1.45 isp.c notes for more on this) to give the firmware a final chance to complete this command.
2000-06-27 19:31:02 +00:00
return;
}
XS_CMD_S_GRACE(xs);
MEMZERO((void *) mp, sizeof (*mp));
mp->req_header.rqs_entry_count = 1;
mp->req_header.rqs_entry_type = RQSTYPE_MARKER;
mp->req_modifier = SYNC_ALL;
mp->req_target = XS_CHANNEL(xs) << 7;
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). It also has to handle the SBus cards (for platforms that have them) which, while on a Big Endian machine, do *not* require *most* of the request/response queue entry fields to be swizzled or unswizzled. 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. 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). MFC after: 2 weeks
2001-12-11 00:18:45 +00:00
isp_put_request(isp, mp, qe);
ISP_ADD_REQUEST(isp, nxti);
Add in the enabling of interrupts (to isp_attach). Clean up a busted comment. Check against firmware state- not loop state when enabling target mode. Other changes have to do with no longer enabling/disabling interrupts at will. Rearchitect command watchdog timeouts- First of all, set the timeout period for a command that has a timeout (in isp_action) to the period of time requested *plus* two seconds. We don't want the Qlogic firmware and the host system to race each other to report a dead command (the watchdog is there to catch dead and/or broken firmware). Next, make sure that the command being watched isn't done yet. If it's not done yet, check for INT_PENDING and call isp_intr- if that said it serviced an interrupt, check to see whether the command is now done (this is what the "IN WATCHDOG" private flag is for- if isp_intr completes the command, it won't call xpt_done on it because isp_watchdog is still looking at the command). If no interrupt was pending, or the command wasn't completed, check to see if we've set the private 'grace period' flag. If so, the command really *is* dead, so report it as dead and complete it with a CAM_CMD_TIMEOUT value. If the grace period flag wasn't set, set it and issue a SYNCHRONIZE_ALL Marker Request Queue entry and re-set the timeout for one second from now (see Revision 1.45 isp.c notes for more on this) to give the firmware a final chance to complete this command.
2000-06-27 19:31:02 +00:00
}
} else {
Core version 2.0 rewrite. In this file we replace isp_tdebug with isp_prt calls. We now use an argument to the ISPCTL_FCLINK_TEST call. We change all IDPRINTF macros to isp_prt calls. We add the isp_prt function here.
2000-08-01 06:31:44 +00:00
isp_prt(isp, ISP_LOGDEBUG2, "watchdog with no command");
Add in a watchdog routine to catch cases where we've dropped the command. Apparently the f/w has finished the command, but somehow an interrupt is being lost. So, we just plain wedge when booting alphas. This is a general routine we've needed for a while.
2000-05-09 01:08:21 +00:00
}
Add get/set param ioctl support. Remove sim queue freezes for resource shortages. I've had too many strange race conditions where I freeze on a resource shortage but never get unfrozen. Consolidate the remaining sim queue freeze condition (for loopdown) into an inline with debug messages that allows us to track problems at ISP_LOGDEBUG0 level easier. Change a bunch of debug messages about loop down/up conditions to ISP_LOGDEBUG0 level. Remove dead isp_relsim code. Change some internal flag stuff for efficiency. Complain vociferously if we try and use our FC scratch area while it's busy being used already (I mean, if we don't have solaris' ability to sleep as an interrupt thread which would allow us to just use a p/v semaphore, at least *say* when you've just borked yourself). Add infrastructure to allow overrides of hard loopid && initiator id from boot variables. Fix the usual quota of silly bugs: + 'ktmature' needs to be per-instance. Argh. + When entering isp_watchdog, set intsok to zero, preserving old value to restore later. It's not nice to try and sleep from splsoftclock. + Fix tick overflow buglet in checking timeout value. MFC after: 1 week
2002-07-08 17:42:47 +00:00
isp->isp_osinfo.intsok = iok;
Add in config_hook for catching when interrupts are safe- this allows us to not the ints are ok and also to (re)ENABLE isp interrupts. Remove all splcam()/splx() invocates and replace them with ISP_LOCK/ISP_UNLOCK macros.
2000-07-04 01:05:43 +00:00
ISP_UNLOCK(isp);
Add in a watchdog routine to catch cases where we've dropped the command. Apparently the f/w has finished the command, but somehow an interrupt is being lost. So, we just plain wedge when booting alphas. This is a general routine we've needed for a while.
2000-05-09 01:08:21 +00:00
}
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
static void
isp_kthread(void *arg)
{
struct ispsoftc *isp = arg;
Turn off usage of SMP style locking until we sort out CAM.
2002-09-03 04:31:55 +00:00
#ifdef ISP_SMPLOCK
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
mtx_lock(&isp->isp_lock);
Tsk. Now that we're not using our own locks, we have to remember to grab Giant in isp_kthread so that msleep is *happy* that there's no lock being passed to it (as tsleep turns out to be...)
2002-09-06 18:20:59 +00:00
#else
mtx_lock(&Giant);
Turn off usage of SMP style locking until we sort out CAM.
2002-09-03 04:31:55 +00:00
#endif
Add get/set param ioctl support. Remove sim queue freezes for resource shortages. I've had too many strange race conditions where I freeze on a resource shortage but never get unfrozen. Consolidate the remaining sim queue freeze condition (for loopdown) into an inline with debug messages that allows us to track problems at ISP_LOGDEBUG0 level easier. Change a bunch of debug messages about loop down/up conditions to ISP_LOGDEBUG0 level. Remove dead isp_relsim code. Change some internal flag stuff for efficiency. Complain vociferously if we try and use our FC scratch area while it's busy being used already (I mean, if we don't have solaris' ability to sleep as an interrupt thread which would allow us to just use a p/v semaphore, at least *say* when you've just borked yourself). Add infrastructure to allow overrides of hard loopid && initiator id from boot variables. Fix the usual quota of silly bugs: + 'ktmature' needs to be per-instance. Argh. + When entering isp_watchdog, set intsok to zero, preserving old value to restore later. It's not nice to try and sleep from splsoftclock. + Fix tick overflow buglet in checking timeout value. MFC after: 1 week
2002-07-08 17:42:47 +00:00
/*
* The first loop is for our usage where we have yet to have
* gotten good fibre channel state.
*/
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
for (;;) {
Add get/set param ioctl support. Remove sim queue freezes for resource shortages. I've had too many strange race conditions where I freeze on a resource shortage but never get unfrozen. Consolidate the remaining sim queue freeze condition (for loopdown) into an inline with debug messages that allows us to track problems at ISP_LOGDEBUG0 level easier. Change a bunch of debug messages about loop down/up conditions to ISP_LOGDEBUG0 level. Remove dead isp_relsim code. Change some internal flag stuff for efficiency. Complain vociferously if we try and use our FC scratch area while it's busy being used already (I mean, if we don't have solaris' ability to sleep as an interrupt thread which would allow us to just use a p/v semaphore, at least *say* when you've just borked yourself). Add infrastructure to allow overrides of hard loopid && initiator id from boot variables. Fix the usual quota of silly bugs: + 'ktmature' needs to be per-instance. Argh. + When entering isp_watchdog, set intsok to zero, preserving old value to restore later. It's not nice to try and sleep from splsoftclock. + Fix tick overflow buglet in checking timeout value. MFC after: 1 week
2002-07-08 17:42:47 +00:00
int wasfrozen;
isp_prt(isp, ISP_LOGDEBUG0, "kthread: checking FC state");
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
while (isp_fc_runstate(isp, 2 * 1000000) != 0) {
Add get/set param ioctl support. Remove sim queue freezes for resource shortages. I've had too many strange race conditions where I freeze on a resource shortage but never get unfrozen. Consolidate the remaining sim queue freeze condition (for loopdown) into an inline with debug messages that allows us to track problems at ISP_LOGDEBUG0 level easier. Change a bunch of debug messages about loop down/up conditions to ISP_LOGDEBUG0 level. Remove dead isp_relsim code. Change some internal flag stuff for efficiency. Complain vociferously if we try and use our FC scratch area while it's busy being used already (I mean, if we don't have solaris' ability to sleep as an interrupt thread which would allow us to just use a p/v semaphore, at least *say* when you've just borked yourself). Add infrastructure to allow overrides of hard loopid && initiator id from boot variables. Fix the usual quota of silly bugs: + 'ktmature' needs to be per-instance. Argh. + When entering isp_watchdog, set intsok to zero, preserving old value to restore later. It's not nice to try and sleep from splsoftclock. + Fix tick overflow buglet in checking timeout value. MFC after: 1 week
2002-07-08 17:42:47 +00:00
isp_prt(isp, ISP_LOGDEBUG0, "kthread: FC state ungood");
Remove ISP_SMPLOCK stuff- we're just using locking now. Correctly reintroduce loop_seen_once semantics- that is, if we've never seen good link, start bouncing commands with CAM_SEL_TIMEOUT. But we have to be careful to have let ourselves try (in isp_kthread) to check for loop up at least once. PR: 28992 MFC after: 1 week
2001-07-25 04:23:52 +00:00
if (FCPARAM(isp)->isp_fwstate != FW_READY ||
FCPARAM(isp)->isp_loopstate < LOOP_PDB_RCVD) {
if (FCPARAM(isp)->loop_seen_once == 0 ||
Add get/set param ioctl support. Remove sim queue freezes for resource shortages. I've had too many strange race conditions where I freeze on a resource shortage but never get unfrozen. Consolidate the remaining sim queue freeze condition (for loopdown) into an inline with debug messages that allows us to track problems at ISP_LOGDEBUG0 level easier. Change a bunch of debug messages about loop down/up conditions to ISP_LOGDEBUG0 level. Remove dead isp_relsim code. Change some internal flag stuff for efficiency. Complain vociferously if we try and use our FC scratch area while it's busy being used already (I mean, if we don't have solaris' ability to sleep as an interrupt thread which would allow us to just use a p/v semaphore, at least *say* when you've just borked yourself). Add infrastructure to allow overrides of hard loopid && initiator id from boot variables. Fix the usual quota of silly bugs: + 'ktmature' needs to be per-instance. Argh. + When entering isp_watchdog, set intsok to zero, preserving old value to restore later. It's not nice to try and sleep from splsoftclock. + Fix tick overflow buglet in checking timeout value. MFC after: 1 week
2002-07-08 17:42:47 +00:00
isp->isp_osinfo.ktmature == 0) {
Remove ISP_SMPLOCK stuff- we're just using locking now. Correctly reintroduce loop_seen_once semantics- that is, if we've never seen good link, start bouncing commands with CAM_SEL_TIMEOUT. But we have to be careful to have let ourselves try (in isp_kthread) to check for loop up at least once. PR: 28992 MFC after: 1 week
2001-07-25 04:23:52 +00:00
break;
}
}
Turn off usage of SMP style locking until we sort out CAM.
2002-09-03 04:31:55 +00:00
#ifdef ISP_SMPLOCK
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
msleep(isp_kthread, &isp->isp_lock,
PRIBIO, "isp_fcthrd", hz);
Turn off usage of SMP style locking until we sort out CAM.
2002-09-03 04:31:55 +00:00
#else
(void) tsleep(isp_kthread, PRIBIO, "isp_fcthrd", hz);
#endif
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
}
Add get/set param ioctl support. Remove sim queue freezes for resource shortages. I've had too many strange race conditions where I freeze on a resource shortage but never get unfrozen. Consolidate the remaining sim queue freeze condition (for loopdown) into an inline with debug messages that allows us to track problems at ISP_LOGDEBUG0 level easier. Change a bunch of debug messages about loop down/up conditions to ISP_LOGDEBUG0 level. Remove dead isp_relsim code. Change some internal flag stuff for efficiency. Complain vociferously if we try and use our FC scratch area while it's busy being used already (I mean, if we don't have solaris' ability to sleep as an interrupt thread which would allow us to just use a p/v semaphore, at least *say* when you've just borked yourself). Add infrastructure to allow overrides of hard loopid && initiator id from boot variables. Fix the usual quota of silly bugs: + 'ktmature' needs to be per-instance. Argh. + When entering isp_watchdog, set intsok to zero, preserving old value to restore later. It's not nice to try and sleep from splsoftclock. + Fix tick overflow buglet in checking timeout value. MFC after: 1 week
2002-07-08 17:42:47 +00:00
Remove ISP_SMPLOCK stuff- we're just using locking now. Correctly reintroduce loop_seen_once semantics- that is, if we've never seen good link, start bouncing commands with CAM_SEL_TIMEOUT. But we have to be careful to have let ourselves try (in isp_kthread) to check for loop up at least once. PR: 28992 MFC after: 1 week
2001-07-25 04:23:52 +00:00
/*
* Even if we didn't get good loop state we may be
* unfreezing the SIMQ so that we can kill off
Add get/set param ioctl support. Remove sim queue freezes for resource shortages. I've had too many strange race conditions where I freeze on a resource shortage but never get unfrozen. Consolidate the remaining sim queue freeze condition (for loopdown) into an inline with debug messages that allows us to track problems at ISP_LOGDEBUG0 level easier. Change a bunch of debug messages about loop down/up conditions to ISP_LOGDEBUG0 level. Remove dead isp_relsim code. Change some internal flag stuff for efficiency. Complain vociferously if we try and use our FC scratch area while it's busy being used already (I mean, if we don't have solaris' ability to sleep as an interrupt thread which would allow us to just use a p/v semaphore, at least *say* when you've just borked yourself). Add infrastructure to allow overrides of hard loopid && initiator id from boot variables. Fix the usual quota of silly bugs: + 'ktmature' needs to be per-instance. Argh. + When entering isp_watchdog, set intsok to zero, preserving old value to restore later. It's not nice to try and sleep from splsoftclock. + Fix tick overflow buglet in checking timeout value. MFC after: 1 week
2002-07-08 17:42:47 +00:00
* commands (if we've never seen loop before, for example).
Remove ISP_SMPLOCK stuff- we're just using locking now. Correctly reintroduce loop_seen_once semantics- that is, if we've never seen good link, start bouncing commands with CAM_SEL_TIMEOUT. But we have to be careful to have let ourselves try (in isp_kthread) to check for loop up at least once. PR: 28992 MFC after: 1 week
2001-07-25 04:23:52 +00:00
*/
Add get/set param ioctl support. Remove sim queue freezes for resource shortages. I've had too many strange race conditions where I freeze on a resource shortage but never get unfrozen. Consolidate the remaining sim queue freeze condition (for loopdown) into an inline with debug messages that allows us to track problems at ISP_LOGDEBUG0 level easier. Change a bunch of debug messages about loop down/up conditions to ISP_LOGDEBUG0 level. Remove dead isp_relsim code. Change some internal flag stuff for efficiency. Complain vociferously if we try and use our FC scratch area while it's busy being used already (I mean, if we don't have solaris' ability to sleep as an interrupt thread which would allow us to just use a p/v semaphore, at least *say* when you've just borked yourself). Add infrastructure to allow overrides of hard loopid && initiator id from boot variables. Fix the usual quota of silly bugs: + 'ktmature' needs to be per-instance. Argh. + When entering isp_watchdog, set intsok to zero, preserving old value to restore later. It's not nice to try and sleep from splsoftclock. + Fix tick overflow buglet in checking timeout value. MFC after: 1 week
2002-07-08 17:42:47 +00:00
isp->isp_osinfo.ktmature = 1;
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
wasfrozen = isp->isp_osinfo.simqfrozen & SIMQFRZ_LOOPDOWN;
isp->isp_osinfo.simqfrozen &= ~SIMQFRZ_LOOPDOWN;
if (wasfrozen && isp->isp_osinfo.simqfrozen == 0) {
Add get/set param ioctl support. Remove sim queue freezes for resource shortages. I've had too many strange race conditions where I freeze on a resource shortage but never get unfrozen. Consolidate the remaining sim queue freeze condition (for loopdown) into an inline with debug messages that allows us to track problems at ISP_LOGDEBUG0 level easier. Change a bunch of debug messages about loop down/up conditions to ISP_LOGDEBUG0 level. Remove dead isp_relsim code. Change some internal flag stuff for efficiency. Complain vociferously if we try and use our FC scratch area while it's busy being used already (I mean, if we don't have solaris' ability to sleep as an interrupt thread which would allow us to just use a p/v semaphore, at least *say* when you've just borked yourself). Add infrastructure to allow overrides of hard loopid && initiator id from boot variables. Fix the usual quota of silly bugs: + 'ktmature' needs to be per-instance. Argh. + When entering isp_watchdog, set intsok to zero, preserving old value to restore later. It's not nice to try and sleep from splsoftclock. + Fix tick overflow buglet in checking timeout value. MFC after: 1 week
2002-07-08 17:42:47 +00:00
isp_prt(isp, ISP_LOGDEBUG0, "kthread: releasing simq");
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
ISPLOCK_2_CAMLOCK(isp);
xpt_release_simq(isp->isp_sim, 1);
CAMLOCK_2_ISPLOCK(isp);
}
Add get/set param ioctl support. Remove sim queue freezes for resource shortages. I've had too many strange race conditions where I freeze on a resource shortage but never get unfrozen. Consolidate the remaining sim queue freeze condition (for loopdown) into an inline with debug messages that allows us to track problems at ISP_LOGDEBUG0 level easier. Change a bunch of debug messages about loop down/up conditions to ISP_LOGDEBUG0 level. Remove dead isp_relsim code. Change some internal flag stuff for efficiency. Complain vociferously if we try and use our FC scratch area while it's busy being used already (I mean, if we don't have solaris' ability to sleep as an interrupt thread which would allow us to just use a p/v semaphore, at least *say* when you've just borked yourself). Add infrastructure to allow overrides of hard loopid && initiator id from boot variables. Fix the usual quota of silly bugs: + 'ktmature' needs to be per-instance. Argh. + When entering isp_watchdog, set intsok to zero, preserving old value to restore later. It's not nice to try and sleep from splsoftclock. + Fix tick overflow buglet in checking timeout value. MFC after: 1 week
2002-07-08 17:42:47 +00:00
isp_prt(isp, ISP_LOGDEBUG0, "kthread: waiting until called");
Turn off usage of SMP style locking until we sort out CAM.
2002-09-03 04:31:55 +00:00
#ifdef ISP_SMPLOCK
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
cv_wait(&isp->isp_osinfo.kthread_cv, &isp->isp_lock);
Turn off usage of SMP style locking until we sort out CAM.
2002-09-03 04:31:55 +00:00
#else
(void) tsleep(&isp->isp_osinfo.kthread_cv, PRIBIO, "fc_cv", 0);
#endif
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
}
}
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
static void
clarify headers;move uninit to outer layer;remove watchdog
1998-12-28 19:22:27 +00:00
isp_action(struct cam_sim *sim, union ccb *ccb)
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
{
Add in config_hook for catching when interrupts are safe- this allows us to not the ints are ok and also to (re)ENABLE isp interrupts. Remove all splcam()/splx() invocates and replace them with ISP_LOCK/ISP_UNLOCK macros.
2000-07-04 01:05:43 +00:00
int bus, tgt, error;
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
struct ispsoftc *isp;
-Wunused cleanup. Submitted by: Poul-Henning Kamp <phk@freebsd.org>
1998-10-15 23:46:33 +00:00
struct ccb_trans_settings *cts;
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
CAM_DEBUG(ccb->ccb_h.path, CAM_DEBUG_TRACE, ("isp_action\n"));
isp = (struct ispsoftc *)cam_sim_softc(sim);
ccb->ccb_h.sim_priv.entries[0].field = 0;
ccb->ccb_h.sim_priv.entries[1].ptr = isp;
Clarify and cleanup some CAM queueing breakages.
1999-08-16 20:11:45 +00:00
if (isp->isp_state != ISP_RUNSTATE &&
ccb->ccb_h.func_code == XPT_SCSI_IO) {
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
CAMLOCK_2_ISPLOCK(isp);
A wad of changes- prepping for 1080/1240 support (which caused a massive thwank in register layout goop). A different mboxcmd approach. Some PDB change infrastructure. Some better management of loopdown/loopup events (keep them distinct from resource starvation for simq freeze/unfreeze actions).
1999-03-17 05:04:39 +00:00
isp_init(isp);
if (isp->isp_state != ISP_INITSTATE) {
Add in config_hook for catching when interrupts are safe- this allows us to not the ints are ok and also to (re)ENABLE isp interrupts. Remove all splcam()/splx() invocates and replace them with ISP_LOCK/ISP_UNLOCK macros.
2000-07-04 01:05:43 +00:00
ISP_UNLOCK(isp);
A wad of changes- prepping for 1080/1240 support (which caused a massive thwank in register layout goop). A different mboxcmd approach. Some PDB change infrastructure. Some better management of loopdown/loopup events (keep them distinct from resource starvation for simq freeze/unfreeze actions).
1999-03-17 05:04:39 +00:00
/*
* Lie. Say it was a selection timeout.
*/
Add in the enabling of interrupts (to isp_attach). Clean up a busted comment. Check against firmware state- not loop state when enabling target mode. Other changes have to do with no longer enabling/disabling interrupts at will. Rearchitect command watchdog timeouts- First of all, set the timeout period for a command that has a timeout (in isp_action) to the period of time requested *plus* two seconds. We don't want the Qlogic firmware and the host system to race each other to report a dead command (the watchdog is there to catch dead and/or broken firmware). Next, make sure that the command being watched isn't done yet. If it's not done yet, check for INT_PENDING and call isp_intr- if that said it serviced an interrupt, check to see whether the command is now done (this is what the "IN WATCHDOG" private flag is for- if isp_intr completes the command, it won't call xpt_done on it because isp_watchdog is still looking at the command). If no interrupt was pending, or the command wasn't completed, check to see if we've set the private 'grace period' flag. If so, the command really *is* dead, so report it as dead and complete it with a CAM_CMD_TIMEOUT value. If the grace period flag wasn't set, set it and issue a SYNCHRONIZE_ALL Marker Request Queue entry and re-set the timeout for one second from now (see Revision 1.45 isp.c notes for more on this) to give the firmware a final chance to complete this command.
2000-06-27 19:31:02 +00:00
ccb->ccb_h.status = CAM_SEL_TIMEOUT | CAM_DEV_QFRZN;
Clarify and cleanup some CAM queueing breakages.
1999-08-16 20:11:45 +00:00
xpt_freeze_devq(ccb->ccb_h.path, 1);
A wad of changes- prepping for 1080/1240 support (which caused a massive thwank in register layout goop). A different mboxcmd approach. Some PDB change infrastructure. Some better management of loopdown/loopup events (keep them distinct from resource starvation for simq freeze/unfreeze actions).
1999-03-17 05:04:39 +00:00
xpt_done(ccb);
return;
}
isp->isp_state = ISP_RUNSTATE;
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
ISPLOCK_2_CAMLOCK(isp);
A wad of changes- prepping for 1080/1240 support (which caused a massive thwank in register layout goop). A different mboxcmd approach. Some PDB change infrastructure. Some better management of loopdown/loopup events (keep them distinct from resource starvation for simq freeze/unfreeze actions).
1999-03-17 05:04:39 +00:00
}
Core version 2.0 rewrite. In this file we replace isp_tdebug with isp_prt calls. We now use an argument to the ISPCTL_FCLINK_TEST call. We change all IDPRINTF macros to isp_prt calls. We add the isp_prt function here.
2000-08-01 06:31:44 +00:00
isp_prt(isp, ISP_LOGDEBUG2, "isp_action code %x", ccb->ccb_h.func_code);
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
switch (ccb->ccb_h.func_code) {
case XPT_SCSI_IO: /* Execute the requested I/O operation */
/*
* Do a couple of preliminary checks...
*/
if ((ccb->ccb_h.flags & CAM_CDB_POINTER) != 0) {
if ((ccb->ccb_h.flags & CAM_CDB_PHYS) != 0) {
ccb->ccb_h.status = CAM_REQ_INVALID;
xpt_done(ccb);
break;
}
}
Clarify and cleanup some CAM queueing breakages.
1999-08-16 20:11:45 +00:00
#ifdef DIAGNOSTIC
if (ccb->ccb_h.target_id > (ISP_MAX_TARGETS(isp) - 1)) {
ccb->ccb_h.status = CAM_PATH_INVALID;
} else if (ccb->ccb_h.target_lun > (ISP_MAX_LUNS(isp) - 1)) {
ccb->ccb_h.status = CAM_PATH_INVALID;
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
}
if (ccb->ccb_h.status == CAM_PATH_INVALID) {
Replace some more printfs with isp_prt's. Use isp_prt/ISP_LOGDEBUG0 for rate setting/getting printouts.
2000-12-05 07:39:54 +00:00
isp_prt(isp, ISP_LOGERR,
"invalid tgt/lun (%d.%d) in XPT_SCSI_IO",
ccb->ccb_h.target_id, ccb->ccb_h.target_lun);
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
xpt_done(ccb);
break;
}
Clarify and cleanup some CAM queueing breakages.
1999-08-16 20:11:45 +00:00
#endif
((struct ccb_scsiio *) ccb)->scsi_status = SCSI_STATUS_OK;
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
CAMLOCK_2_ISPLOCK(isp);
Core version 2.0 rewrite. In this file we replace isp_tdebug with isp_prt calls. We now use an argument to the ISPCTL_FCLINK_TEST call. We change all IDPRINTF macros to isp_prt calls. We add the isp_prt function here.
2000-08-01 06:31:44 +00:00
error = isp_start((XS_T *) ccb);
Clarify and cleanup some CAM queueing breakages.
1999-08-16 20:11:45 +00:00
switch (error) {
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
case CMD_QUEUED:
ccb->ccb_h.status |= CAM_SIM_QUEUED;
Add in a watchdog routine to catch cases where we've dropped the command. Apparently the f/w has finished the command, but somehow an interrupt is being lost. So, we just plain wedge when booting alphas. This is a general routine we've needed for a while.
2000-05-09 01:08:21 +00:00
if (ccb->ccb_h.timeout != CAM_TIME_INFINITY) {
a) clean up some declaration stuff (i.e., make more modern with respect to getting rid u_int for uint and so on). b) Turn back on 64 bit DAC support. Cheeze it a bit in that we have two DMA callback functions- one when we have bus_addr_t > 4 bits in width and the other which should be normal. Even Cheezier in that we turn off setting up DMA maps to be BUS_SPACE_MAXADDR if we're in ISP_TARGET_MODE. More work on this in a week or so. c) Tested under amd64 and 1MB DFLTPHYS, sparc64, i386 (PAE, but insufficient memory to really test > 4GB). LINT check under amd64. MFC after: 1 month
2006-02-15 00:31:48 +00:00
uint64_t ticks = (uint64_t) hz;
Add in a watchdog routine to catch cases where we've dropped the command. Apparently the f/w has finished the command, but somehow an interrupt is being lost. So, we just plain wedge when booting alphas. This is a general routine we've needed for a while.
2000-05-09 01:08:21 +00:00
if (ccb->ccb_h.timeout == CAM_TIME_DEFAULT)
Guard against overflow of the calculated timeout value.
2001-01-16 07:15:36 +00:00
ticks = 60 * 1000 * ticks;
Add in the enabling of interrupts (to isp_attach). Clean up a busted comment. Check against firmware state- not loop state when enabling target mode. Other changes have to do with no longer enabling/disabling interrupts at will. Rearchitect command watchdog timeouts- First of all, set the timeout period for a command that has a timeout (in isp_action) to the period of time requested *plus* two seconds. We don't want the Qlogic firmware and the host system to race each other to report a dead command (the watchdog is there to catch dead and/or broken firmware). Next, make sure that the command being watched isn't done yet. If it's not done yet, check for INT_PENDING and call isp_intr- if that said it serviced an interrupt, check to see whether the command is now done (this is what the "IN WATCHDOG" private flag is for- if isp_intr completes the command, it won't call xpt_done on it because isp_watchdog is still looking at the command). If no interrupt was pending, or the command wasn't completed, check to see if we've set the private 'grace period' flag. If so, the command really *is* dead, so report it as dead and complete it with a CAM_CMD_TIMEOUT value. If the grace period flag wasn't set, set it and issue a SYNCHRONIZE_ALL Marker Request Queue entry and re-set the timeout for one second from now (see Revision 1.45 isp.c notes for more on this) to give the firmware a final chance to complete this command.
2000-06-27 19:31:02 +00:00
else
ticks = ccb->ccb_h.timeout * hz;
ticks = ((ticks + 999) / 1000) + hz + hz;
Guard against overflow of the calculated timeout value.
2001-01-16 07:15:36 +00:00
if (ticks >= 0x80000000) {
isp_prt(isp, ISP_LOGERR,
"timeout overflow");
Add get/set param ioctl support. Remove sim queue freezes for resource shortages. I've had too many strange race conditions where I freeze on a resource shortage but never get unfrozen. Consolidate the remaining sim queue freeze condition (for loopdown) into an inline with debug messages that allows us to track problems at ISP_LOGDEBUG0 level easier. Change a bunch of debug messages about loop down/up conditions to ISP_LOGDEBUG0 level. Remove dead isp_relsim code. Change some internal flag stuff for efficiency. Complain vociferously if we try and use our FC scratch area while it's busy being used already (I mean, if we don't have solaris' ability to sleep as an interrupt thread which would allow us to just use a p/v semaphore, at least *say* when you've just borked yourself). Add infrastructure to allow overrides of hard loopid && initiator id from boot variables. Fix the usual quota of silly bugs: + 'ktmature' needs to be per-instance. Argh. + When entering isp_watchdog, set intsok to zero, preserving old value to restore later. It's not nice to try and sleep from splsoftclock. + Fix tick overflow buglet in checking timeout value. MFC after: 1 week
2002-07-08 17:42:47 +00:00
ticks = 0x7fffffff;
Guard against overflow of the calculated timeout value.
2001-01-16 07:15:36 +00:00
}
ccb->ccb_h.timeout_ch = timeout(isp_watchdog,
(caddr_t)ccb, (int)ticks);
Add in the enabling of interrupts (to isp_attach). Clean up a busted comment. Check against firmware state- not loop state when enabling target mode. Other changes have to do with no longer enabling/disabling interrupts at will. Rearchitect command watchdog timeouts- First of all, set the timeout period for a command that has a timeout (in isp_action) to the period of time requested *plus* two seconds. We don't want the Qlogic firmware and the host system to race each other to report a dead command (the watchdog is there to catch dead and/or broken firmware). Next, make sure that the command being watched isn't done yet. If it's not done yet, check for INT_PENDING and call isp_intr- if that said it serviced an interrupt, check to see whether the command is now done (this is what the "IN WATCHDOG" private flag is for- if isp_intr completes the command, it won't call xpt_done on it because isp_watchdog is still looking at the command). If no interrupt was pending, or the command wasn't completed, check to see if we've set the private 'grace period' flag. If so, the command really *is* dead, so report it as dead and complete it with a CAM_CMD_TIMEOUT value. If the grace period flag wasn't set, set it and issue a SYNCHRONIZE_ALL Marker Request Queue entry and re-set the timeout for one second from now (see Revision 1.45 isp.c notes for more on this) to give the firmware a final chance to complete this command.
2000-06-27 19:31:02 +00:00
} else {
callout_handle_init(&ccb->ccb_h.timeout_ch);
Add in a watchdog routine to catch cases where we've dropped the command. Apparently the f/w has finished the command, but somehow an interrupt is being lost. So, we just plain wedge when booting alphas. This is a general routine we've needed for a while.
2000-05-09 01:08:21 +00:00
}
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
ISPLOCK_2_CAMLOCK(isp);
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
break;
Clarify and cleanup some CAM queueing breakages.
1999-08-16 20:11:45 +00:00
case CMD_RQLATER:
Remove ISP_SMPLOCK stuff- we're just using locking now. Correctly reintroduce loop_seen_once semantics- that is, if we've never seen good link, start bouncing commands with CAM_SEL_TIMEOUT. But we have to be careful to have let ourselves try (in isp_kthread) to check for loop up at least once. PR: 28992 MFC after: 1 week
2001-07-25 04:23:52 +00:00
/*
* This can only happen for Fibre Channel
*/
KASSERT((IS_FC(isp)), ("CMD_RQLATER for FC only"));
Add get/set param ioctl support. Remove sim queue freezes for resource shortages. I've had too many strange race conditions where I freeze on a resource shortage but never get unfrozen. Consolidate the remaining sim queue freeze condition (for loopdown) into an inline with debug messages that allows us to track problems at ISP_LOGDEBUG0 level easier. Change a bunch of debug messages about loop down/up conditions to ISP_LOGDEBUG0 level. Remove dead isp_relsim code. Change some internal flag stuff for efficiency. Complain vociferously if we try and use our FC scratch area while it's busy being used already (I mean, if we don't have solaris' ability to sleep as an interrupt thread which would allow us to just use a p/v semaphore, at least *say* when you've just borked yourself). Add infrastructure to allow overrides of hard loopid && initiator id from boot variables. Fix the usual quota of silly bugs: + 'ktmature' needs to be per-instance. Argh. + When entering isp_watchdog, set intsok to zero, preserving old value to restore later. It's not nice to try and sleep from splsoftclock. + Fix tick overflow buglet in checking timeout value. MFC after: 1 week
2002-07-08 17:42:47 +00:00
if (FCPARAM(isp)->loop_seen_once == 0 &&
isp->isp_osinfo.ktmature) {
Remove ISP_SMPLOCK stuff- we're just using locking now. Correctly reintroduce loop_seen_once semantics- that is, if we've never seen good link, start bouncing commands with CAM_SEL_TIMEOUT. But we have to be careful to have let ourselves try (in isp_kthread) to check for loop up at least once. PR: 28992 MFC after: 1 week
2001-07-25 04:23:52 +00:00
ISPLOCK_2_CAMLOCK(isp);
XS_SETERR(ccb, CAM_SEL_TIMEOUT);
xpt_done(ccb);
break;
}
Turn off usage of SMP style locking until we sort out CAM.
2002-09-03 04:31:55 +00:00
#ifdef ISP_SMPLOCK
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
cv_signal(&isp->isp_osinfo.kthread_cv);
Turn off usage of SMP style locking until we sort out CAM.
2002-09-03 04:31:55 +00:00
#else
wakeup(&isp->isp_osinfo.kthread_cv);
#endif
Add get/set param ioctl support. Remove sim queue freezes for resource shortages. I've had too many strange race conditions where I freeze on a resource shortage but never get unfrozen. Consolidate the remaining sim queue freeze condition (for loopdown) into an inline with debug messages that allows us to track problems at ISP_LOGDEBUG0 level easier. Change a bunch of debug messages about loop down/up conditions to ISP_LOGDEBUG0 level. Remove dead isp_relsim code. Change some internal flag stuff for efficiency. Complain vociferously if we try and use our FC scratch area while it's busy being used already (I mean, if we don't have solaris' ability to sleep as an interrupt thread which would allow us to just use a p/v semaphore, at least *say* when you've just borked yourself). Add infrastructure to allow overrides of hard loopid && initiator id from boot variables. Fix the usual quota of silly bugs: + 'ktmature' needs to be per-instance. Argh. + When entering isp_watchdog, set intsok to zero, preserving old value to restore later. It's not nice to try and sleep from splsoftclock. + Fix tick overflow buglet in checking timeout value. MFC after: 1 week
2002-07-08 17:42:47 +00:00
isp_freeze_loopdown(isp, "isp_action(RQLATER)");
Add in the enabling of interrupts (to isp_attach). Clean up a busted comment. Check against firmware state- not loop state when enabling target mode. Other changes have to do with no longer enabling/disabling interrupts at will. Rearchitect command watchdog timeouts- First of all, set the timeout period for a command that has a timeout (in isp_action) to the period of time requested *plus* two seconds. We don't want the Qlogic firmware and the host system to race each other to report a dead command (the watchdog is there to catch dead and/or broken firmware). Next, make sure that the command being watched isn't done yet. If it's not done yet, check for INT_PENDING and call isp_intr- if that said it serviced an interrupt, check to see whether the command is now done (this is what the "IN WATCHDOG" private flag is for- if isp_intr completes the command, it won't call xpt_done on it because isp_watchdog is still looking at the command). If no interrupt was pending, or the command wasn't completed, check to see if we've set the private 'grace period' flag. If so, the command really *is* dead, so report it as dead and complete it with a CAM_CMD_TIMEOUT value. If the grace period flag wasn't set, set it and issue a SYNCHRONIZE_ALL Marker Request Queue entry and re-set the timeout for one second from now (see Revision 1.45 isp.c notes for more on this) to give the firmware a final chance to complete this command.
2000-06-27 19:31:02 +00:00
XS_SETERR(ccb, CAM_REQUEUE_REQ);
Add get/set param ioctl support. Remove sim queue freezes for resource shortages. I've had too many strange race conditions where I freeze on a resource shortage but never get unfrozen. Consolidate the remaining sim queue freeze condition (for loopdown) into an inline with debug messages that allows us to track problems at ISP_LOGDEBUG0 level easier. Change a bunch of debug messages about loop down/up conditions to ISP_LOGDEBUG0 level. Remove dead isp_relsim code. Change some internal flag stuff for efficiency. Complain vociferously if we try and use our FC scratch area while it's busy being used already (I mean, if we don't have solaris' ability to sleep as an interrupt thread which would allow us to just use a p/v semaphore, at least *say* when you've just borked yourself). Add infrastructure to allow overrides of hard loopid && initiator id from boot variables. Fix the usual quota of silly bugs: + 'ktmature' needs to be per-instance. Argh. + When entering isp_watchdog, set intsok to zero, preserving old value to restore later. It's not nice to try and sleep from splsoftclock. + Fix tick overflow buglet in checking timeout value. MFC after: 1 week
2002-07-08 17:42:47 +00:00
ISPLOCK_2_CAMLOCK(isp);
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
xpt_done(ccb);
break;
Clarify and cleanup some CAM queueing breakages.
1999-08-16 20:11:45 +00:00
case CMD_EAGAIN:
Add in the enabling of interrupts (to isp_attach). Clean up a busted comment. Check against firmware state- not loop state when enabling target mode. Other changes have to do with no longer enabling/disabling interrupts at will. Rearchitect command watchdog timeouts- First of all, set the timeout period for a command that has a timeout (in isp_action) to the period of time requested *plus* two seconds. We don't want the Qlogic firmware and the host system to race each other to report a dead command (the watchdog is there to catch dead and/or broken firmware). Next, make sure that the command being watched isn't done yet. If it's not done yet, check for INT_PENDING and call isp_intr- if that said it serviced an interrupt, check to see whether the command is now done (this is what the "IN WATCHDOG" private flag is for- if isp_intr completes the command, it won't call xpt_done on it because isp_watchdog is still looking at the command). If no interrupt was pending, or the command wasn't completed, check to see if we've set the private 'grace period' flag. If so, the command really *is* dead, so report it as dead and complete it with a CAM_CMD_TIMEOUT value. If the grace period flag wasn't set, set it and issue a SYNCHRONIZE_ALL Marker Request Queue entry and re-set the timeout for one second from now (see Revision 1.45 isp.c notes for more on this) to give the firmware a final chance to complete this command.
2000-06-27 19:31:02 +00:00
XS_SETERR(ccb, CAM_REQUEUE_REQ);
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
ISPLOCK_2_CAMLOCK(isp);
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
xpt_done(ccb);
break;
Clarify and cleanup some CAM queueing breakages.
1999-08-16 20:11:45 +00:00
case CMD_COMPLETE:
isp_done((struct ccb_scsiio *) ccb);
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
ISPLOCK_2_CAMLOCK(isp);
Clarify and cleanup some CAM queueing breakages.
1999-08-16 20:11:45 +00:00
break;
default:
Replace some more printfs with isp_prt's. Use isp_prt/ISP_LOGDEBUG0 for rate setting/getting printouts.
2000-12-05 07:39:54 +00:00
isp_prt(isp, ISP_LOGERR,
"What's this? 0x%x at %d in file %s",
Fix problems with incomplete conversions from printf to isp_prt.
2000-12-31 20:50:56 +00:00
error, __LINE__, __FILE__);
Add in the enabling of interrupts (to isp_attach). Clean up a busted comment. Check against firmware state- not loop state when enabling target mode. Other changes have to do with no longer enabling/disabling interrupts at will. Rearchitect command watchdog timeouts- First of all, set the timeout period for a command that has a timeout (in isp_action) to the period of time requested *plus* two seconds. We don't want the Qlogic firmware and the host system to race each other to report a dead command (the watchdog is there to catch dead and/or broken firmware). Next, make sure that the command being watched isn't done yet. If it's not done yet, check for INT_PENDING and call isp_intr- if that said it serviced an interrupt, check to see whether the command is now done (this is what the "IN WATCHDOG" private flag is for- if isp_intr completes the command, it won't call xpt_done on it because isp_watchdog is still looking at the command). If no interrupt was pending, or the command wasn't completed, check to see if we've set the private 'grace period' flag. If so, the command really *is* dead, so report it as dead and complete it with a CAM_CMD_TIMEOUT value. If the grace period flag wasn't set, set it and issue a SYNCHRONIZE_ALL Marker Request Queue entry and re-set the timeout for one second from now (see Revision 1.45 isp.c notes for more on this) to give the firmware a final chance to complete this command.
2000-06-27 19:31:02 +00:00
XS_SETERR(ccb, CAM_REQ_CMP_ERR);
Clarify and cleanup some CAM queueing breakages.
1999-08-16 20:11:45 +00:00
xpt_done(ccb);
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
ISPLOCK_2_CAMLOCK(isp);
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
}
break;
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
#ifdef ISP_TARGET_MODE
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
case XPT_EN_LUN: /* Enable LUN as a target */
I don't know what I was thinking- if I have two separate busses on on SIM (as is true for the 1280 and the 12160), then I have to have separate flags && status for *both* busses. *Whap*. Implement condition variables for coordination with some target mode events. It's nice to use these and not panic in obscure little places in the kernel like 'propagate_priority' just because we went to sleep holding a mutex, or some other absurd thing. Remove some bogus ISP_UNLOCK calls. *Whap*. No longer require that somebody do a lun enable on the wildcard device to enable target mode. They are, in fact, orthogonal. A wildcard open is a statement that somebody upstream is willing to accept commands which are otherwise unrouteable. Now, for QLogic regular SCSI target mode, this won't matter for a damn because we'll never see ATIOs for luns we haven't enabled (are listening for, if you will). But for SCCLUN fibre channel SCSI, we get all kinds of ATIOs. We can either reflect them back here with minimal info (which is isp_target.c:isp_endcmd() is for), or the wildcard device (nominally targbh) can handle them. Do further checking against firmware attributes to see whether we can, in fact, support target mode in Fibre Channel. For now, require SCCLUN f/w to supoprt FC target mode. This is an awful lot of change, but target mode *still* isn't quite right. MFC after: 4 weeks
2001-09-04 21:53:12 +00:00
{
Remove condition variables and status associated with target mode enabling. Instead, go to an interrupt/polled model. Fix get_lun_statep so we don't panic if there are no wildcard luns enabled. MFC after: 6 days
2004-02-08 19:17:56 +00:00
int seq, iok, i;
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
CAMLOCK_2_ISPLOCK(isp);
I don't know what I was thinking- if I have two separate busses on on SIM (as is true for the 1280 and the 12160), then I have to have separate flags && status for *both* busses. *Whap*. Implement condition variables for coordination with some target mode events. It's nice to use these and not panic in obscure little places in the kernel like 'propagate_priority' just because we went to sleep holding a mutex, or some other absurd thing. Remove some bogus ISP_UNLOCK calls. *Whap*. No longer require that somebody do a lun enable on the wildcard device to enable target mode. They are, in fact, orthogonal. A wildcard open is a statement that somebody upstream is willing to accept commands which are otherwise unrouteable. Now, for QLogic regular SCSI target mode, this won't matter for a damn because we'll never see ATIOs for luns we haven't enabled (are listening for, if you will). But for SCCLUN fibre channel SCSI, we get all kinds of ATIOs. We can either reflect them back here with minimal info (which is isp_target.c:isp_endcmd() is for), or the wildcard device (nominally targbh) can handle them. Do further checking against firmware attributes to see whether we can, in fact, support target mode in Fibre Channel. For now, require SCCLUN f/w to supoprt FC target mode. This is an awful lot of change, but target mode *still* isn't quite right. MFC after: 4 weeks
2001-09-04 21:53:12 +00:00
iok = isp->isp_osinfo.intsok;
isp->isp_osinfo.intsok = 0;
Remove condition variables and status associated with target mode enabling. Instead, go to an interrupt/polled model. Fix get_lun_statep so we don't panic if there are no wildcard luns enabled. MFC after: 6 days
2004-02-08 19:17:56 +00:00
seq = isp_en_lun(isp, ccb);
if (seq < 0) {
isp->isp_osinfo.intsok = iok;
ISPLOCK_2_CAMLOCK(isp);
xpt_done(ccb);
break;
}
for (i = 0; isp->isp_osinfo.leact[seq] && i < 30 * 1000; i++) {
a) clean up some declaration stuff (i.e., make more modern with respect to getting rid u_int for uint and so on). b) Turn back on 64 bit DAC support. Cheeze it a bit in that we have two DMA callback functions- one when we have bus_addr_t > 4 bits in width and the other which should be normal. Even Cheezier in that we turn off setting up DMA maps to be BUS_SPACE_MAXADDR if we're in ISP_TARGET_MODE. More work on this in a week or so. c) Tested under amd64 and 1MB DFLTPHYS, sparc64, i386 (PAE, but insufficient memory to really test > 4GB). LINT check under amd64. MFC after: 1 month
2006-02-15 00:31:48 +00:00
uint16_t isr, sema, mbox;
Remove condition variables and status associated with target mode enabling. Instead, go to an interrupt/polled model. Fix get_lun_statep so we don't panic if there are no wildcard luns enabled. MFC after: 6 days
2004-02-08 19:17:56 +00:00
if (ISP_READ_ISR(isp, &isr, &sema, &mbox)) {
isp_intr(isp, isr, sema, mbox);
}
DELAY(1000);
}
I don't know what I was thinking- if I have two separate busses on on SIM (as is true for the 1280 and the 12160), then I have to have separate flags && status for *both* busses. *Whap*. Implement condition variables for coordination with some target mode events. It's nice to use these and not panic in obscure little places in the kernel like 'propagate_priority' just because we went to sleep holding a mutex, or some other absurd thing. Remove some bogus ISP_UNLOCK calls. *Whap*. No longer require that somebody do a lun enable on the wildcard device to enable target mode. They are, in fact, orthogonal. A wildcard open is a statement that somebody upstream is willing to accept commands which are otherwise unrouteable. Now, for QLogic regular SCSI target mode, this won't matter for a damn because we'll never see ATIOs for luns we haven't enabled (are listening for, if you will). But for SCCLUN fibre channel SCSI, we get all kinds of ATIOs. We can either reflect them back here with minimal info (which is isp_target.c:isp_endcmd() is for), or the wildcard device (nominally targbh) can handle them. Do further checking against firmware attributes to see whether we can, in fact, support target mode in Fibre Channel. For now, require SCCLUN f/w to supoprt FC target mode. This is an awful lot of change, but target mode *still* isn't quite right. MFC after: 4 weeks
2001-09-04 21:53:12 +00:00
isp->isp_osinfo.intsok = iok;
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
ISPLOCK_2_CAMLOCK(isp);
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
break;
I don't know what I was thinking- if I have two separate busses on on SIM (as is true for the 1280 and the 12160), then I have to have separate flags && status for *both* busses. *Whap*. Implement condition variables for coordination with some target mode events. It's nice to use these and not panic in obscure little places in the kernel like 'propagate_priority' just because we went to sleep holding a mutex, or some other absurd thing. Remove some bogus ISP_UNLOCK calls. *Whap*. No longer require that somebody do a lun enable on the wildcard device to enable target mode. They are, in fact, orthogonal. A wildcard open is a statement that somebody upstream is willing to accept commands which are otherwise unrouteable. Now, for QLogic regular SCSI target mode, this won't matter for a damn because we'll never see ATIOs for luns we haven't enabled (are listening for, if you will). But for SCCLUN fibre channel SCSI, we get all kinds of ATIOs. We can either reflect them back here with minimal info (which is isp_target.c:isp_endcmd() is for), or the wildcard device (nominally targbh) can handle them. Do further checking against firmware attributes to see whether we can, in fact, support target mode in Fibre Channel. For now, require SCCLUN f/w to supoprt FC target mode. This is an awful lot of change, but target mode *still* isn't quite right. MFC after: 4 weeks
2001-09-04 21:53:12 +00:00
}
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
case XPT_NOTIFY_ACK: /* recycle notify ack */
case XPT_IMMED_NOTIFY: /* Add Immediate Notify Resource */
case XPT_ACCEPT_TARGET_IO: /* Add Accept Target IO Resource */
{
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
tstate_t *tptr =
get_lun_statep(isp, XS_CHANNEL(ccb), ccb->ccb_h.target_lun);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
if (tptr == NULL) {
ccb->ccb_h.status = CAM_LUN_INVALID;
xpt_done(ccb);
break;
}
Add a isp_target_putback_atio- we aren't using CCINCR at this time, so we need a function that tells the Qlogic f/w that a target mode command is done, so increase the resource count for that lun. Add in a timeout function to kick the putback again if we fail to do it the first time (we may not have the request queue space for ATIO push). Split the function isp_handle_platform_ctio into two parts so that the timeout function for the ATIO push or isp_handle_platform_ctio can inform CAM that the requested CTIO(s) are now done. Clean up (cough) residual handling. What we need for Fibre Channel is to preserve the at_datalen field from the original incoming ATIO so we can calculate a 'true' residual. Unfortunately, we're not guaranteed to get that back from CAM. We'll *try* to find it hiding in the periph_priv field (layering violation)- but if an ATIO was passed in from user land- forget it. This means that we'll probably get residuals wrong for Fibre Channel commands we're completing with an error. It's too late to 4.1 release to fix this- too bad. Luckily the only device we'd really care about this occurring on is a tape device and they're still so rare as FC attached devices that this can be considered an untested combination anyway. Remove all CCINCR usage (resource autoreplenish). When we've proved to ourself that things are working properly, we can add it back in. Make sure we propage 'suggested' sense data from the incoming ATIO into the created system ATIO- and set sense_len appropriately. Correctly propagate tag values. Fall back to the model of generating (well, the functions in isp_pci.c do the work) multiple CTIOs based upon what we get from XPT. Instead of being able to pair Qlogic generated ATIOs with CAM ATIOs, and then to pair CAM CTIOs with Qlogic CTIOs, we have to take the CTIO passed to us from XPT, and if it implies that we have to generate extra Qlogic CTIOs, so be it. This means that we have to wait until the last CTIO in a sequence we generated completes before calling xpt_done. Executive summary- target mode actually now pretty much works well enough to tell folks about.
2000-07-18 06:58:28 +00:00
ccb->ccb_h.sim_priv.entries[0].field = 0;
ccb->ccb_h.sim_priv.entries[1].ptr = isp;
Add support for ISP_FC_GETHINFO, which returns current connection topology, speed, loopid, WWPN/WWNN, etc. Beef up target mode. Add isp_handle_platform_notify_scsi and isp_handle_platform_notify_fc platform handlers to handle immediate notifies (isp_handle_platform_notify_scsi is still stubbed out). In implementation of isp_handle_platform_notify_fc, for IN_ABORT_TASK, peel off a pending XPT_IMMED_NOTIFY and call xpt_done on it and hope that somebody upstream is listening. Make sure on final CTIO2s that we set residual correctly. These are absolutely crucial. Make sure we set relative offset for each CTIO2 based upon bytes we've already xferred. This is what the private adjunct datat to the original ATIO is. Note state of command so we can figure out where to find it if we get an ABORT from the firmware. Make sure we *always* set CAM_TAG_ACTION_VALID for ATIO2s. Make sure we keep track of the original lun. If se sent status (or we're otherwise done with the command), don't forget to free the adjunct structure.
2002-06-16 05:08:02 +00:00
ccb->ccb_h.flags = 0;
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
CAMLOCK_2_ISPLOCK(isp);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
if (ccb->ccb_h.func_code == XPT_ACCEPT_TARGET_IO) {
Add support for ISP_FC_GETHINFO, which returns current connection topology, speed, loopid, WWPN/WWNN, etc. Beef up target mode. Add isp_handle_platform_notify_scsi and isp_handle_platform_notify_fc platform handlers to handle immediate notifies (isp_handle_platform_notify_scsi is still stubbed out). In implementation of isp_handle_platform_notify_fc, for IN_ABORT_TASK, peel off a pending XPT_IMMED_NOTIFY and call xpt_done on it and hope that somebody upstream is listening. Make sure on final CTIO2s that we set residual correctly. These are absolutely crucial. Make sure we set relative offset for each CTIO2 based upon bytes we've already xferred. This is what the private adjunct datat to the original ATIO is. Note state of command so we can figure out where to find it if we get an ABORT from the firmware. Make sure we *always* set CAM_TAG_ACTION_VALID for ATIO2s. Make sure we keep track of the original lun. If se sent status (or we're otherwise done with the command), don't forget to free the adjunct structure.
2002-06-16 05:08:02 +00:00
/*
* Note that the command itself may not be done-
* it may not even have had the first CTIO sent.
*/
tptr->atio_count++;
isp_prt(isp, ISP_LOGTDEBUG0,
Checkpoint of work in progress in cleaning up target mode. It actually seems to work well in RELENG_4. However, 5.X locking foo means that I'll have to do some quick redesign. Add ioctl handlers for ISP_GETROLE and ISP_SETROLE ioctls.
2004-02-07 03:47:33 +00:00
"Put FREE ATIO, lun %d, count now %d",
Add support for ISP_FC_GETHINFO, which returns current connection topology, speed, loopid, WWPN/WWNN, etc. Beef up target mode. Add isp_handle_platform_notify_scsi and isp_handle_platform_notify_fc platform handlers to handle immediate notifies (isp_handle_platform_notify_scsi is still stubbed out). In implementation of isp_handle_platform_notify_fc, for IN_ABORT_TASK, peel off a pending XPT_IMMED_NOTIFY and call xpt_done on it and hope that somebody upstream is listening. Make sure on final CTIO2s that we set residual correctly. These are absolutely crucial. Make sure we set relative offset for each CTIO2 based upon bytes we've already xferred. This is what the private adjunct datat to the original ATIO is. Note state of command so we can figure out where to find it if we get an ABORT from the firmware. Make sure we *always* set CAM_TAG_ACTION_VALID for ATIO2s. Make sure we keep track of the original lun. If se sent status (or we're otherwise done with the command), don't forget to free the adjunct structure.
2002-06-16 05:08:02 +00:00
ccb->ccb_h.target_lun, tptr->atio_count);
SLIST_INSERT_HEAD(&tptr->atios, &ccb->ccb_h,
sim_links.sle);
} else if (ccb->ccb_h.func_code == XPT_IMMED_NOTIFY) {
Checkpoint of work in progress in cleaning up target mode. It actually seems to work well in RELENG_4. However, 5.X locking foo means that I'll have to do some quick redesign. Add ioctl handlers for ISP_GETROLE and ISP_SETROLE ioctls.
2004-02-07 03:47:33 +00:00
tptr->inot_count++;
isp_prt(isp, ISP_LOGTDEBUG0,
"Put FREE INOT, lun %d, count now %d",
ccb->ccb_h.target_lun, tptr->inot_count);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
SLIST_INSERT_HEAD(&tptr->inots, &ccb->ccb_h,
sim_links.sle);
Add support for ISP_FC_GETHINFO, which returns current connection topology, speed, loopid, WWPN/WWNN, etc. Beef up target mode. Add isp_handle_platform_notify_scsi and isp_handle_platform_notify_fc platform handlers to handle immediate notifies (isp_handle_platform_notify_scsi is still stubbed out). In implementation of isp_handle_platform_notify_fc, for IN_ABORT_TASK, peel off a pending XPT_IMMED_NOTIFY and call xpt_done on it and hope that somebody upstream is listening. Make sure on final CTIO2s that we set residual correctly. These are absolutely crucial. Make sure we set relative offset for each CTIO2 based upon bytes we've already xferred. This is what the private adjunct datat to the original ATIO is. Note state of command so we can figure out where to find it if we get an ABORT from the firmware. Make sure we *always* set CAM_TAG_ACTION_VALID for ATIO2s. Make sure we keep track of the original lun. If se sent status (or we're otherwise done with the command), don't forget to free the adjunct structure.
2002-06-16 05:08:02 +00:00
} else {
Checkpoint of work in progress in cleaning up target mode. It actually seems to work well in RELENG_4. However, 5.X locking foo means that I'll have to do some quick redesign. Add ioctl handlers for ISP_GETROLE and ISP_SETROLE ioctls.
2004-02-07 03:47:33 +00:00
isp_prt(isp, ISP_LOGWARN, "Got Notify ACK");;
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
}
rls_lun_statep(isp, tptr);
ccb->ccb_h.status = CAM_REQ_INPROG;
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
ISPLOCK_2_CAMLOCK(isp);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
break;
}
case XPT_CONT_TARGET_IO:
{
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
CAMLOCK_2_ISPLOCK(isp);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
ccb->ccb_h.status = isp_target_start_ctio(isp, ccb);
if (ccb->ccb_h.status != CAM_REQ_INPROG) {
Add support for ISP_FC_GETHINFO, which returns current connection topology, speed, loopid, WWPN/WWNN, etc. Beef up target mode. Add isp_handle_platform_notify_scsi and isp_handle_platform_notify_fc platform handlers to handle immediate notifies (isp_handle_platform_notify_scsi is still stubbed out). In implementation of isp_handle_platform_notify_fc, for IN_ABORT_TASK, peel off a pending XPT_IMMED_NOTIFY and call xpt_done on it and hope that somebody upstream is listening. Make sure on final CTIO2s that we set residual correctly. These are absolutely crucial. Make sure we set relative offset for each CTIO2 based upon bytes we've already xferred. This is what the private adjunct datat to the original ATIO is. Note state of command so we can figure out where to find it if we get an ABORT from the firmware. Make sure we *always* set CAM_TAG_ACTION_VALID for ATIO2s. Make sure we keep track of the original lun. If se sent status (or we're otherwise done with the command), don't forget to free the adjunct structure.
2002-06-16 05:08:02 +00:00
isp_prt(isp, ISP_LOGWARN,
"XPT_CONT_TARGET_IO: status 0x%x",
ccb->ccb_h.status);
Add in the enabling of interrupts (to isp_attach). Clean up a busted comment. Check against firmware state- not loop state when enabling target mode. Other changes have to do with no longer enabling/disabling interrupts at will. Rearchitect command watchdog timeouts- First of all, set the timeout period for a command that has a timeout (in isp_action) to the period of time requested *plus* two seconds. We don't want the Qlogic firmware and the host system to race each other to report a dead command (the watchdog is there to catch dead and/or broken firmware). Next, make sure that the command being watched isn't done yet. If it's not done yet, check for INT_PENDING and call isp_intr- if that said it serviced an interrupt, check to see whether the command is now done (this is what the "IN WATCHDOG" private flag is for- if isp_intr completes the command, it won't call xpt_done on it because isp_watchdog is still looking at the command). If no interrupt was pending, or the command wasn't completed, check to see if we've set the private 'grace period' flag. If so, the command really *is* dead, so report it as dead and complete it with a CAM_CMD_TIMEOUT value. If the grace period flag wasn't set, set it and issue a SYNCHRONIZE_ALL Marker Request Queue entry and re-set the timeout for one second from now (see Revision 1.45 isp.c notes for more on this) to give the firmware a final chance to complete this command.
2000-06-27 19:31:02 +00:00
XS_SETERR(ccb, CAM_REQUEUE_REQ);
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
ISPLOCK_2_CAMLOCK(isp);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
xpt_done(ccb);
} else {
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
ISPLOCK_2_CAMLOCK(isp);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
ccb->ccb_h.status |= CAM_SIM_QUEUED;
}
break;
}
#endif
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
case XPT_RESET_DEV: /* BDR the specified SCSI device */
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
bus = cam_sim_bus(xpt_path_sim(ccb->ccb_h.path));
tgt = ccb->ccb_h.target_id;
tgt |= (bus << 16);
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
CAMLOCK_2_ISPLOCK(isp);
A large set of changes to handle dual bus adapters.
1999-05-11 05:10:06 +00:00
error = isp_control(isp, ISPCTL_RESET_DEV, &tgt);
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
ISPLOCK_2_CAMLOCK(isp);
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
if (error) {
ccb->ccb_h.status = CAM_REQ_CMP_ERR;
} else {
ccb->ccb_h.status = CAM_REQ_CMP;
}
xpt_done(ccb);
break;
case XPT_ABORT: /* Abort the specified CCB */
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
{
union ccb *accb = ccb->cab.abort_ccb;
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
CAMLOCK_2_ISPLOCK(isp);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
switch (accb->ccb_h.func_code) {
#ifdef ISP_TARGET_MODE
case XPT_ACCEPT_TARGET_IO:
case XPT_IMMED_NOTIFY:
ccb->ccb_h.status = isp_abort_tgt_ccb(isp, ccb);
break;
case XPT_CONT_TARGET_IO:
Core version 2.0 rewrite. In this file we replace isp_tdebug with isp_prt calls. We now use an argument to the ISPCTL_FCLINK_TEST call. We change all IDPRINTF macros to isp_prt calls. We add the isp_prt function here.
2000-08-01 06:31:44 +00:00
isp_prt(isp, ISP_LOGERR, "cannot abort CTIOs yet");
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
ccb->ccb_h.status = CAM_UA_ABORT;
break;
#endif
case XPT_SCSI_IO:
error = isp_control(isp, ISPCTL_ABORT_CMD, ccb);
if (error) {
ccb->ccb_h.status = CAM_UA_ABORT;
} else {
ccb->ccb_h.status = CAM_REQ_CMP;
}
break;
default:
ccb->ccb_h.status = CAM_REQ_INVALID;
break;
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
}
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
ISPLOCK_2_CAMLOCK(isp);
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
xpt_done(ccb);
break;
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
}
Add CAM_NEW_TRAN_CODE support. Use correct CAMLOCK_2_ISPLOCK macros. For fibre channel, start going for the gusto and using AC_FOUND_DEVICE and AC_LOST_DEVICE calls to xpt_async when devices appear and disappear as the loop or fabric changes. ISPASYNC_FW_CRASH is the async event code where the platform layer deals with a firmware crash.
2001-07-04 18:54:29 +00:00
#ifdef CAM_NEW_TRAN_CODE
#define IS_CURRENT_SETTINGS(c) (c->type == CTS_TYPE_CURRENT_SETTINGS)
#else
#define IS_CURRENT_SETTINGS(c) (c->flags & CCB_TRANS_CURRENT_SETTINGS)
#endif
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
case XPT_SET_TRAN_SETTINGS: /* Nexus Settings */
cts = &ccb->cts;
Redo how we manage SCSI device settings- have a 3rd flags (nvram) that records either what's in NVRAM or what the safe defaults would be if we lack NVRAM. Then we rename cur_XXXX to actv_XXXX (these are the currently active settings) and the dev_XXX settings to goal_XXXX (these are the settings which we want cur_XXXX to converge to). This probably isn't entirely final as yet- but it's a lot closer to now being what it should be, including allowing camcontrol to actually set specific settings.
2001-07-30 01:00:21 +00:00
if (!IS_CURRENT_SETTINGS(cts)) {
ccb->ccb_h.status = CAM_REQ_INVALID;
xpt_done(ccb);
break;
}
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
tgt = cts->ccb_h.target_id;
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
CAMLOCK_2_ISPLOCK(isp);
Add in isp_debug variable. It defaults to zero unless CAMDEBUG is defined where it defaults to one. Change simq width allocation to the max number of commands supported by the HBA after f/w fires up- not the constant MAXISPREQUEST value. Do some stylistic changes.
1999-10-17 18:50:47 +00:00
if (IS_SCSI(isp)) {
Add CAM_NEW_TRAN_CODE support. Use correct CAMLOCK_2_ISPLOCK macros. For fibre channel, start going for the gusto and using AC_FOUND_DEVICE and AC_LOST_DEVICE calls to xpt_async when devices appear and disappear as the loop or fabric changes. ISPASYNC_FW_CRASH is the async event code where the platform layer deals with a firmware crash.
2001-07-04 18:54:29 +00:00
#ifndef CAM_NEW_TRAN_CODE
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
sdparam *sdp = isp->isp_param;
a) clean up some declaration stuff (i.e., make more modern with respect to getting rid u_int for uint and so on). b) Turn back on 64 bit DAC support. Cheeze it a bit in that we have two DMA callback functions- one when we have bus_addr_t > 4 bits in width and the other which should be normal. Even Cheezier in that we turn off setting up DMA maps to be BUS_SPACE_MAXADDR if we're in ISP_TARGET_MODE. More work on this in a week or so. c) Tested under amd64 and 1MB DFLTPHYS, sparc64, i386 (PAE, but insufficient memory to really test > 4GB). LINT check under amd64. MFC after: 1 month
2006-02-15 00:31:48 +00:00
uint16_t *dptr;
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
bus = cam_sim_bus(xpt_path_sim(cts->ccb_h.path));
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
A large set of changes to handle dual bus adapters.
1999-05-11 05:10:06 +00:00
sdp += bus;
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
/*
Redo how we manage SCSI device settings- have a 3rd flags (nvram) that records either what's in NVRAM or what the safe defaults would be if we lack NVRAM. Then we rename cur_XXXX to actv_XXXX (these are the currently active settings) and the dev_XXX settings to goal_XXXX (these are the settings which we want cur_XXXX to converge to). This probably isn't entirely final as yet- but it's a lot closer to now being what it should be, including allowing camcontrol to actually set specific settings.
2001-07-30 01:00:21 +00:00
* We always update (internally) from goal_flags
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
* so any request to change settings just gets
* vectored to that location.
*/
Redo how we manage SCSI device settings- have a 3rd flags (nvram) that records either what's in NVRAM or what the safe defaults would be if we lack NVRAM. Then we rename cur_XXXX to actv_XXXX (these are the currently active settings) and the dev_XXX settings to goal_XXXX (these are the settings which we want cur_XXXX to converge to). This probably isn't entirely final as yet- but it's a lot closer to now being what it should be, including allowing camcontrol to actually set specific settings.
2001-07-30 01:00:21 +00:00
dptr = &sdp->isp_devparam[tgt].goal_flags;
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
/*
* Note that these operations affect the
Redo how we manage SCSI device settings- have a 3rd flags (nvram) that records either what's in NVRAM or what the safe defaults would be if we lack NVRAM. Then we rename cur_XXXX to actv_XXXX (these are the currently active settings) and the dev_XXX settings to goal_XXXX (these are the settings which we want cur_XXXX to converge to). This probably isn't entirely final as yet- but it's a lot closer to now being what it should be, including allowing camcontrol to actually set specific settings.
2001-07-30 01:00:21 +00:00
* the goal flags (goal_flags)- not
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
* the current state flags. Then we mark
* things so that the next operation to
* this HBA will cause the update to occur.
*/
if (cts->valid & CCB_TRANS_DISC_VALID) {
if ((cts->flags & CCB_TRANS_DISC_ENB) != 0) {
*dptr |= DPARM_DISC;
} else {
*dptr &= ~DPARM_DISC;
}
}
if (cts->valid & CCB_TRANS_TQ_VALID) {
if ((cts->flags & CCB_TRANS_TAG_ENB) != 0) {
*dptr |= DPARM_TQING;
} else {
*dptr &= ~DPARM_TQING;
}
}
if (cts->valid & CCB_TRANS_BUS_WIDTH_VALID) {
switch (cts->bus_width) {
case MSG_EXT_WDTR_BUS_16_BIT:
*dptr |= DPARM_WIDE;
break;
default:
*dptr &= ~DPARM_WIDE;
}
}
/*
* Any SYNC RATE of nonzero and SYNC_OFFSET
* of nonzero will cause us to go to the
* selected (from NVRAM) maximum value for
* this device. At a later point, we'll
* allow finer control.
*/
if ((cts->valid & CCB_TRANS_SYNC_RATE_VALID) &&
(cts->valid & CCB_TRANS_SYNC_OFFSET_VALID) &&
(cts->sync_offset > 0)) {
*dptr |= DPARM_SYNC;
} else {
*dptr &= ~DPARM_SYNC;
}
Clean up some target mode debug messages. Fix (finally, I believe) Andrew's problems with SCSI on some alphas- do not call isp_update directly to update parameters- just mark them as being ready to update for the next command- the system would just hang on a READ CAPACITY for a drive. Really annoying because it wouldn't even timeout (and it has a timeout) so either the SET PARAMETERS call was nuking things or the f/w was really dropping the ball. approved: jkh Reviewed by: gallatin@freebsd.org
2000-02-15 00:52:52 +00:00
*dptr |= DPARM_SAFE_DFLT;
Add CAM_NEW_TRAN_CODE support. Use correct CAMLOCK_2_ISPLOCK macros. For fibre channel, start going for the gusto and using AC_FOUND_DEVICE and AC_LOST_DEVICE calls to xpt_async when devices appear and disappear as the loop or fabric changes. ISPASYNC_FW_CRASH is the async event code where the platform layer deals with a firmware crash.
2001-07-04 18:54:29 +00:00
#else
struct ccb_trans_settings_scsi *scsi =
&cts->proto_specific.scsi;
struct ccb_trans_settings_spi *spi =
&cts->xport_specific.spi;
sdparam *sdp = isp->isp_param;
a) clean up some declaration stuff (i.e., make more modern with respect to getting rid u_int for uint and so on). b) Turn back on 64 bit DAC support. Cheeze it a bit in that we have two DMA callback functions- one when we have bus_addr_t > 4 bits in width and the other which should be normal. Even Cheezier in that we turn off setting up DMA maps to be BUS_SPACE_MAXADDR if we're in ISP_TARGET_MODE. More work on this in a week or so. c) Tested under amd64 and 1MB DFLTPHYS, sparc64, i386 (PAE, but insufficient memory to really test > 4GB). LINT check under amd64. MFC after: 1 month
2006-02-15 00:31:48 +00:00
uint16_t *dptr;
Add CAM_NEW_TRAN_CODE support. Use correct CAMLOCK_2_ISPLOCK macros. For fibre channel, start going for the gusto and using AC_FOUND_DEVICE and AC_LOST_DEVICE calls to xpt_async when devices appear and disappear as the loop or fabric changes. ISPASYNC_FW_CRASH is the async event code where the platform layer deals with a firmware crash.
2001-07-04 18:54:29 +00:00
bus = cam_sim_bus(xpt_path_sim(cts->ccb_h.path));
sdp += bus;
/*
Redo how we manage SCSI device settings- have a 3rd flags (nvram) that records either what's in NVRAM or what the safe defaults would be if we lack NVRAM. Then we rename cur_XXXX to actv_XXXX (these are the currently active settings) and the dev_XXX settings to goal_XXXX (these are the settings which we want cur_XXXX to converge to). This probably isn't entirely final as yet- but it's a lot closer to now being what it should be, including allowing camcontrol to actually set specific settings.
2001-07-30 01:00:21 +00:00
* We always update (internally) from goal_flags
Add CAM_NEW_TRAN_CODE support. Use correct CAMLOCK_2_ISPLOCK macros. For fibre channel, start going for the gusto and using AC_FOUND_DEVICE and AC_LOST_DEVICE calls to xpt_async when devices appear and disappear as the loop or fabric changes. ISPASYNC_FW_CRASH is the async event code where the platform layer deals with a firmware crash.
2001-07-04 18:54:29 +00:00
* so any request to change settings just gets
* vectored to that location.
*/
Redo how we manage SCSI device settings- have a 3rd flags (nvram) that records either what's in NVRAM or what the safe defaults would be if we lack NVRAM. Then we rename cur_XXXX to actv_XXXX (these are the currently active settings) and the dev_XXX settings to goal_XXXX (these are the settings which we want cur_XXXX to converge to). This probably isn't entirely final as yet- but it's a lot closer to now being what it should be, including allowing camcontrol to actually set specific settings.
2001-07-30 01:00:21 +00:00
dptr = &sdp->isp_devparam[tgt].goal_flags;
Add CAM_NEW_TRAN_CODE support. Use correct CAMLOCK_2_ISPLOCK macros. For fibre channel, start going for the gusto and using AC_FOUND_DEVICE and AC_LOST_DEVICE calls to xpt_async when devices appear and disappear as the loop or fabric changes. ISPASYNC_FW_CRASH is the async event code where the platform layer deals with a firmware crash.
2001-07-04 18:54:29 +00:00
if ((spi->valid & CTS_SPI_VALID_DISC) != 0) {
if ((spi->flags & CTS_SPI_FLAGS_DISC_ENB) != 0)
*dptr |= DPARM_DISC;
else
*dptr &= ~DPARM_DISC;
}
if ((scsi->valid & CTS_SCSI_VALID_TQ) != 0) {
if ((scsi->flags & CTS_SCSI_FLAGS_TAG_ENB) != 0)
*dptr |= DPARM_TQING;
else
*dptr &= ~DPARM_TQING;
}
if ((spi->valid & CTS_SPI_VALID_BUS_WIDTH) != 0) {
if (spi->bus_width == MSG_EXT_WDTR_BUS_16_BIT)
*dptr |= DPARM_WIDE;
else
*dptr &= ~DPARM_WIDE;
}
/*
* XXX: FIX ME
*/
if ((spi->valid & CTS_SPI_VALID_SYNC_OFFSET) &&
Redo how we manage SCSI device settings- have a 3rd flags (nvram) that records either what's in NVRAM or what the safe defaults would be if we lack NVRAM. Then we rename cur_XXXX to actv_XXXX (these are the currently active settings) and the dev_XXX settings to goal_XXXX (these are the settings which we want cur_XXXX to converge to). This probably isn't entirely final as yet- but it's a lot closer to now being what it should be, including allowing camcontrol to actually set specific settings.
2001-07-30 01:00:21 +00:00
(spi->valid & CTS_SPI_VALID_SYNC_RATE) &&
(spi->sync_period && spi->sync_offset)) {
Add CAM_NEW_TRAN_CODE support. Use correct CAMLOCK_2_ISPLOCK macros. For fibre channel, start going for the gusto and using AC_FOUND_DEVICE and AC_LOST_DEVICE calls to xpt_async when devices appear and disappear as the loop or fabric changes. ISPASYNC_FW_CRASH is the async event code where the platform layer deals with a firmware crash.
2001-07-04 18:54:29 +00:00
*dptr |= DPARM_SYNC;
Redo how we manage SCSI device settings- have a 3rd flags (nvram) that records either what's in NVRAM or what the safe defaults would be if we lack NVRAM. Then we rename cur_XXXX to actv_XXXX (these are the currently active settings) and the dev_XXX settings to goal_XXXX (these are the settings which we want cur_XXXX to converge to). This probably isn't entirely final as yet- but it's a lot closer to now being what it should be, including allowing camcontrol to actually set specific settings.
2001-07-30 01:00:21 +00:00
/*
* XXX: CHECK FOR LEGALITY
*/
sdp->isp_devparam[tgt].goal_period =
spi->sync_period;
sdp->isp_devparam[tgt].goal_offset =
spi->sync_offset;
Add CAM_NEW_TRAN_CODE support. Use correct CAMLOCK_2_ISPLOCK macros. For fibre channel, start going for the gusto and using AC_FOUND_DEVICE and AC_LOST_DEVICE calls to xpt_async when devices appear and disappear as the loop or fabric changes. ISPASYNC_FW_CRASH is the async event code where the platform layer deals with a firmware crash.
2001-07-04 18:54:29 +00:00
} else {
*dptr &= ~DPARM_SYNC;
}
#endif
Replace some more printfs with isp_prt's. Use isp_prt/ISP_LOGDEBUG0 for rate setting/getting printouts.
2000-12-05 07:39:54 +00:00
isp_prt(isp, ISP_LOGDEBUG0,
Redo how we manage SCSI device settings- have a 3rd flags (nvram) that records either what's in NVRAM or what the safe defaults would be if we lack NVRAM. Then we rename cur_XXXX to actv_XXXX (these are the currently active settings) and the dev_XXX settings to goal_XXXX (these are the settings which we want cur_XXXX to converge to). This probably isn't entirely final as yet- but it's a lot closer to now being what it should be, including allowing camcontrol to actually set specific settings.
2001-07-30 01:00:21 +00:00
"SET bus %d targ %d to flags %x off %x per %x",
bus, tgt, sdp->isp_devparam[tgt].goal_flags,
sdp->isp_devparam[tgt].goal_offset,
sdp->isp_devparam[tgt].goal_period);
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
sdp->isp_devparam[tgt].dev_update = 1;
A large set of changes to handle dual bus adapters.
1999-05-11 05:10:06 +00:00
isp->isp_update |= (1 << bus);
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
}
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
ISPLOCK_2_CAMLOCK(isp);
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
ccb->ccb_h.status = CAM_REQ_CMP;
xpt_done(ccb);
break;
case XPT_GET_TRAN_SETTINGS:
cts = &ccb->cts;
tgt = cts->ccb_h.target_id;
Add CAM_NEW_TRAN_CODE support. Use correct CAMLOCK_2_ISPLOCK macros. For fibre channel, start going for the gusto and using AC_FOUND_DEVICE and AC_LOST_DEVICE calls to xpt_async when devices appear and disappear as the loop or fabric changes. ISPASYNC_FW_CRASH is the async event code where the platform layer deals with a firmware crash.
2001-07-04 18:54:29 +00:00
CAMLOCK_2_ISPLOCK(isp);
Add in isp_debug variable. It defaults to zero unless CAMDEBUG is defined where it defaults to one. Change simq width allocation to the max number of commands supported by the HBA after f/w fires up- not the constant MAXISPREQUEST value. Do some stylistic changes.
1999-10-17 18:50:47 +00:00
if (IS_FC(isp)) {
Add CAM_NEW_TRAN_CODE support. Use correct CAMLOCK_2_ISPLOCK macros. For fibre channel, start going for the gusto and using AC_FOUND_DEVICE and AC_LOST_DEVICE calls to xpt_async when devices appear and disappear as the loop or fabric changes. ISPASYNC_FW_CRASH is the async event code where the platform layer deals with a firmware crash.
2001-07-04 18:54:29 +00:00
#ifndef CAM_NEW_TRAN_CODE
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
/*
* a lot of normal SCSI things don't make sense.
*/
cts->flags = CCB_TRANS_TAG_ENB | CCB_TRANS_DISC_ENB;
cts->valid = CCB_TRANS_DISC_VALID | CCB_TRANS_TQ_VALID;
/*
* How do you measure the width of a high
* speed serial bus? Well, in bytes.
*
* Offset and period make no sense, though, so we set
* (above) a 'base' transfer speed to be gigabit.
*/
cts->bus_width = MSG_EXT_WDTR_BUS_8_BIT;
Add CAM_NEW_TRAN_CODE support. Use correct CAMLOCK_2_ISPLOCK macros. For fibre channel, start going for the gusto and using AC_FOUND_DEVICE and AC_LOST_DEVICE calls to xpt_async when devices appear and disappear as the loop or fabric changes. ISPASYNC_FW_CRASH is the async event code where the platform layer deals with a firmware crash.
2001-07-04 18:54:29 +00:00
#else
fcparam *fcp = isp->isp_param;
struct ccb_trans_settings_fc *fc =
&cts->xport_specific.fc;
cts->protocol = PROTO_SCSI;
cts->protocol_version = SCSI_REV_2;
cts->transport = XPORT_FC;
cts->transport_version = 0;
fc->valid = CTS_FC_VALID_SPEED;
Begin to implement target mode that for Fibre Channel has a private per-command component that we *don't* try and pass thru CAM. CAM just is too risky and too much of a pain- structures get copied, but not all info of interest can be considered safely transported thru all consumers (including user space) from the incoming ATIO to the outgoing CTIO- it's just much safer to have a buddy structure, identified by the command's tag which *does* make it thru safely. Pay attention to link speed and report 200MB/s xfer speed for a 23XX card in 2GPs mode. MFC after: 1 week
2001-10-01 03:48:42 +00:00
if (fcp->isp_gbspeed == 2)
fc->bitrate = 200000;
else
fc->bitrate = 100000;
Add CAM_NEW_TRAN_CODE support. Use correct CAMLOCK_2_ISPLOCK macros. For fibre channel, start going for the gusto and using AC_FOUND_DEVICE and AC_LOST_DEVICE calls to xpt_async when devices appear and disappear as the loop or fabric changes. ISPASYNC_FW_CRASH is the async event code where the platform layer deals with a firmware crash.
2001-07-04 18:54:29 +00:00
if (tgt > 0 && tgt < MAX_FC_TARG) {
struct lportdb *lp = &fcp->portdb[tgt];
fc->wwnn = lp->node_wwn;
fc->wwpn = lp->port_wwn;
fc->port = lp->portid;
fc->valid |= CTS_FC_VALID_WWNN |
CTS_FC_VALID_WWPN | CTS_FC_VALID_PORT;
}
#endif
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
} else {
Add CAM_NEW_TRAN_CODE support. Use correct CAMLOCK_2_ISPLOCK macros. For fibre channel, start going for the gusto and using AC_FOUND_DEVICE and AC_LOST_DEVICE calls to xpt_async when devices appear and disappear as the loop or fabric changes. ISPASYNC_FW_CRASH is the async event code where the platform layer deals with a firmware crash.
2001-07-04 18:54:29 +00:00
#ifdef CAM_NEW_TRAN_CODE
struct ccb_trans_settings_scsi *scsi =
&cts->proto_specific.scsi;
struct ccb_trans_settings_spi *spi =
&cts->xport_specific.spi;
#endif
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
sdparam *sdp = isp->isp_param;
A large set of changes to handle dual bus adapters.
1999-05-11 05:10:06 +00:00
int bus = cam_sim_bus(xpt_path_sim(cts->ccb_h.path));
a) clean up some declaration stuff (i.e., make more modern with respect to getting rid u_int for uint and so on). b) Turn back on 64 bit DAC support. Cheeze it a bit in that we have two DMA callback functions- one when we have bus_addr_t > 4 bits in width and the other which should be normal. Even Cheezier in that we turn off setting up DMA maps to be BUS_SPACE_MAXADDR if we're in ISP_TARGET_MODE. More work on this in a week or so. c) Tested under amd64 and 1MB DFLTPHYS, sparc64, i386 (PAE, but insufficient memory to really test > 4GB). LINT check under amd64. MFC after: 1 month
2006-02-15 00:31:48 +00:00
uint16_t dval, pval, oval;
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
A large set of changes to handle dual bus adapters.
1999-05-11 05:10:06 +00:00
sdp += bus;
Add CAM_NEW_TRAN_CODE support. Use correct CAMLOCK_2_ISPLOCK macros. For fibre channel, start going for the gusto and using AC_FOUND_DEVICE and AC_LOST_DEVICE calls to xpt_async when devices appear and disappear as the loop or fabric changes. ISPASYNC_FW_CRASH is the async event code where the platform layer deals with a firmware crash.
2001-07-04 18:54:29 +00:00
if (IS_CURRENT_SETTINGS(cts)) {
When asked to get the current transfer settings go do a dev_refresh isp_update call to get a better chance at seeing whether a recent settings change has latched up.
1999-05-12 19:00:16 +00:00
sdp->isp_devparam[tgt].dev_refresh = 1;
isp->isp_update |= (1 << bus);
(void) isp_control(isp, ISPCTL_UPDATE_PARAMS,
NULL);
Redo how we manage SCSI device settings- have a 3rd flags (nvram) that records either what's in NVRAM or what the safe defaults would be if we lack NVRAM. Then we rename cur_XXXX to actv_XXXX (these are the currently active settings) and the dev_XXX settings to goal_XXXX (these are the settings which we want cur_XXXX to converge to). This probably isn't entirely final as yet- but it's a lot closer to now being what it should be, including allowing camcontrol to actually set specific settings.
2001-07-30 01:00:21 +00:00
dval = sdp->isp_devparam[tgt].actv_flags;
oval = sdp->isp_devparam[tgt].actv_offset;
pval = sdp->isp_devparam[tgt].actv_period;
Add in 1080 LVD support and some basis also for the 1240. The port database printout is now enabled.
1999-03-25 22:52:45 +00:00
} else {
Redo how we manage SCSI device settings- have a 3rd flags (nvram) that records either what's in NVRAM or what the safe defaults would be if we lack NVRAM. Then we rename cur_XXXX to actv_XXXX (these are the currently active settings) and the dev_XXX settings to goal_XXXX (these are the settings which we want cur_XXXX to converge to). This probably isn't entirely final as yet- but it's a lot closer to now being what it should be, including allowing camcontrol to actually set specific settings.
2001-07-30 01:00:21 +00:00
dval = sdp->isp_devparam[tgt].nvrm_flags;
oval = sdp->isp_devparam[tgt].nvrm_offset;
pval = sdp->isp_devparam[tgt].nvrm_period;
Add in 1080 LVD support and some basis also for the 1240. The port database printout is now enabled.
1999-03-25 22:52:45 +00:00
}
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
Add CAM_NEW_TRAN_CODE support. Use correct CAMLOCK_2_ISPLOCK macros. For fibre channel, start going for the gusto and using AC_FOUND_DEVICE and AC_LOST_DEVICE calls to xpt_async when devices appear and disappear as the loop or fabric changes. ISPASYNC_FW_CRASH is the async event code where the platform layer deals with a firmware crash.
2001-07-04 18:54:29 +00:00
#ifndef CAM_NEW_TRAN_CODE
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
cts->flags &= ~(CCB_TRANS_DISC_ENB|CCB_TRANS_TAG_ENB);
if (dval & DPARM_DISC) {
cts->flags |= CCB_TRANS_DISC_ENB;
}
if (dval & DPARM_TQING) {
cts->flags |= CCB_TRANS_TAG_ENB;
}
if (dval & DPARM_WIDE) {
cts->bus_width = MSG_EXT_WDTR_BUS_16_BIT;
} else {
cts->bus_width = MSG_EXT_WDTR_BUS_8_BIT;
}
cts->valid = CCB_TRANS_BUS_WIDTH_VALID |
CCB_TRANS_DISC_VALID | CCB_TRANS_TQ_VALID;
Add in 1080 LVD support and some basis also for the 1240. The port database printout is now enabled.
1999-03-25 22:52:45 +00:00
if ((dval & DPARM_SYNC) && oval != 0) {
cts->sync_period = pval;
cts->sync_offset = oval;
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
cts->valid |=
CCB_TRANS_SYNC_RATE_VALID |
CCB_TRANS_SYNC_OFFSET_VALID;
}
Add CAM_NEW_TRAN_CODE support. Use correct CAMLOCK_2_ISPLOCK macros. For fibre channel, start going for the gusto and using AC_FOUND_DEVICE and AC_LOST_DEVICE calls to xpt_async when devices appear and disappear as the loop or fabric changes. ISPASYNC_FW_CRASH is the async event code where the platform layer deals with a firmware crash.
2001-07-04 18:54:29 +00:00
#else
cts->protocol = PROTO_SCSI;
cts->protocol_version = SCSI_REV_2;
cts->transport = XPORT_SPI;
cts->transport_version = 2;
scsi->flags &= ~CTS_SCSI_FLAGS_TAG_ENB;
spi->flags &= ~CTS_SPI_FLAGS_DISC_ENB;
if (dval & DPARM_DISC) {
spi->flags |= CTS_SPI_FLAGS_DISC_ENB;
}
if (dval & DPARM_TQING) {
scsi->flags |= CTS_SCSI_FLAGS_TAG_ENB;
}
Redo how we manage SCSI device settings- have a 3rd flags (nvram) that records either what's in NVRAM or what the safe defaults would be if we lack NVRAM. Then we rename cur_XXXX to actv_XXXX (these are the currently active settings) and the dev_XXX settings to goal_XXXX (these are the settings which we want cur_XXXX to converge to). This probably isn't entirely final as yet- but it's a lot closer to now being what it should be, including allowing camcontrol to actually set specific settings.
2001-07-30 01:00:21 +00:00
if ((dval & DPARM_SYNC) && oval && pval) {
Add CAM_NEW_TRAN_CODE support. Use correct CAMLOCK_2_ISPLOCK macros. For fibre channel, start going for the gusto and using AC_FOUND_DEVICE and AC_LOST_DEVICE calls to xpt_async when devices appear and disappear as the loop or fabric changes. ISPASYNC_FW_CRASH is the async event code where the platform layer deals with a firmware crash.
2001-07-04 18:54:29 +00:00
spi->sync_offset = oval;
spi->sync_period = pval;
spi->valid |= CTS_SPI_VALID_SYNC_OFFSET;
spi->valid |= CTS_SPI_VALID_SYNC_RATE;
}
spi->valid |= CTS_SPI_VALID_BUS_WIDTH;
if (dval & DPARM_WIDE) {
spi->bus_width = MSG_EXT_WDTR_BUS_16_BIT;
} else {
spi->bus_width = MSG_EXT_WDTR_BUS_8_BIT;
}
if (cts->ccb_h.target_lun != CAM_LUN_WILDCARD) {
scsi->valid = CTS_SCSI_VALID_TQ;
spi->valid |= CTS_SPI_VALID_DISC;
} else {
scsi->valid = 0;
}
#endif
Replace some more printfs with isp_prt's. Use isp_prt/ISP_LOGDEBUG0 for rate setting/getting printouts.
2000-12-05 07:39:54 +00:00
isp_prt(isp, ISP_LOGDEBUG0,
Redo how we manage SCSI device settings- have a 3rd flags (nvram) that records either what's in NVRAM or what the safe defaults would be if we lack NVRAM. Then we rename cur_XXXX to actv_XXXX (these are the currently active settings) and the dev_XXX settings to goal_XXXX (these are the settings which we want cur_XXXX to converge to). This probably isn't entirely final as yet- but it's a lot closer to now being what it should be, including allowing camcontrol to actually set specific settings.
2001-07-30 01:00:21 +00:00
"GET %s bus %d targ %d to flags %x off %x per %x",
IS_CURRENT_SETTINGS(cts)? "ACTIVE" : "NVRAM",
bus, tgt, dval, oval, pval);
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
}
Add CAM_NEW_TRAN_CODE support. Use correct CAMLOCK_2_ISPLOCK macros. For fibre channel, start going for the gusto and using AC_FOUND_DEVICE and AC_LOST_DEVICE calls to xpt_async when devices appear and disappear as the loop or fabric changes. ISPASYNC_FW_CRASH is the async event code where the platform layer deals with a firmware crash.
2001-07-04 18:54:29 +00:00
ISPLOCK_2_CAMLOCK(isp);
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
ccb->ccb_h.status = CAM_REQ_CMP;
xpt_done(ccb);
break;
case XPT_CALC_GEOMETRY:
{
struct ccb_calc_geometry *ccg;
ccg = &ccb->ccg;
if (ccg->block_size == 0) {
Replace some more printfs with isp_prt's. Use isp_prt/ISP_LOGDEBUG0 for rate setting/getting printouts.
2000-12-05 07:39:54 +00:00
isp_prt(isp, ISP_LOGERR,
"%d.%d XPT_CALC_GEOMETRY block size 0?",
ccg->ccb_h.target_id, ccg->ccb_h.target_lun);
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
ccb->ccb_h.status = CAM_REQ_INVALID;
xpt_done(ccb);
break;
}
Merge common XPT_CALC_GEOMETRY functions into a single convenience function. Devices below may experience a change in geometry. * Due to a bug, aic(4) never used extended geometry. Changes all drives >1G to now use extended translation. * sbp(4) drives exactly 1 GB in size now no longer use extended geometry. * umass(4) drives exactly 1 GB in size now no longer use extended geometry. For all other controllers in this commit, this should be a no-op. Looked over by: scottl
2003-06-14 22:17:41 +00:00
cam_calc_geometry(ccg, /*extended*/1);
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
xpt_done(ccb);
break;
}
case XPT_RESET_BUS: /* Reset the specified bus */
Add in isp_debug variable. It defaults to zero unless CAMDEBUG is defined where it defaults to one. Change simq width allocation to the max number of commands supported by the HBA after f/w fires up- not the constant MAXISPREQUEST value. Do some stylistic changes.
1999-10-17 18:50:47 +00:00
bus = cam_sim_bus(sim);
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
CAMLOCK_2_ISPLOCK(isp);
Add in isp_debug variable. It defaults to zero unless CAMDEBUG is defined where it defaults to one. Change simq width allocation to the max number of commands supported by the HBA after f/w fires up- not the constant MAXISPREQUEST value. Do some stylistic changes.
1999-10-17 18:50:47 +00:00
error = isp_control(isp, ISPCTL_RESET_BUS, &bus);
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
ISPLOCK_2_CAMLOCK(isp);
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
if (error)
ccb->ccb_h.status = CAM_REQ_CMP_ERR;
Allow fibre channel 'bus resets' to go through. Handle Loop Down/Loop Up events by freezing/unfreezing the simq- nice to have such control at this level! Do bus resets in attach layer (non-CAM defined code).
1999-02-09 01:08:38 +00:00
else {
A large set of changes to handle dual bus adapters.
1999-05-11 05:10:06 +00:00
if (cam_sim_bus(sim) && isp->isp_path2 != NULL)
xpt_async(AC_BUS_RESET, isp->isp_path2, NULL);
else if (isp->isp_path != NULL)
Allow fibre channel 'bus resets' to go through. Handle Loop Down/Loop Up events by freezing/unfreezing the simq- nice to have such control at this level! Do bus resets in attach layer (non-CAM defined code).
1999-02-09 01:08:38 +00:00
xpt_async(AC_BUS_RESET, isp->isp_path, NULL);
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
ccb->ccb_h.status = CAM_REQ_CMP;
Allow fibre channel 'bus resets' to go through. Handle Loop Down/Loop Up events by freezing/unfreezing the simq- nice to have such control at this level! Do bus resets in attach layer (non-CAM defined code).
1999-02-09 01:08:38 +00:00
}
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
xpt_done(ccb);
break;
case XPT_TERM_IO: /* Terminate the I/O process */
ccb->ccb_h.status = CAM_REQ_INVALID;
xpt_done(ccb);
break;
case XPT_PATH_INQ: /* Path routing inquiry */
{
struct ccb_pathinq *cpi = &ccb->cpi;
cpi->version_num = 1;
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
#ifdef ISP_TARGET_MODE
Redo a lot of the target mode infrastructure to be cognizant of Dual Bus cards like the 1280 && the 12160. Cleanup isp_target_putback_atio. Make sure bus and correct tag ids and firmware handles get propagated as needed.
2001-04-04 21:58:29 +00:00
cpi->target_sprt = PIT_PROCESSOR | PIT_DISCONNECT | PIT_TERM_IO;
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
#else
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
cpi->target_sprt = 0;
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
#endif
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
cpi->hba_eng_cnt = 0;
Clarify and cleanup some CAM queueing breakages.
1999-08-16 20:11:45 +00:00
cpi->max_target = ISP_MAX_TARGETS(isp) - 1;
cpi->max_lun = ISP_MAX_LUNS(isp) - 1;
cpi->bus_id = cam_sim_bus(sim);
Add in 1080 LVD support and some basis also for the 1240. The port database printout is now enabled.
1999-03-25 22:52:45 +00:00
if (IS_FC(isp)) {
cpi->hba_misc = PIM_NOBUSRESET;
Clarify and cleanup some CAM queueing breakages.
1999-08-16 20:11:45 +00:00
/*
* Because our loop ID can shift from time to time,
* make our initiator ID out of range of our bus.
*/
cpi->initiator_id = cpi->max_target + 1;
Add a number of interrelated CAM feature enhancements and bug fixes. NOTE: These changes will require recompilation of any userland applications, like cdrecord, xmcd, etc., that use the CAM passthrough interface. A make world is recommended. camcontrol.[c8]: - We now support two new commands, "tags" and "negotiate". - The tags commands allows users to view the number of tagged openings for a device as well as a number of other related parameters, and it allows users to set tagged openings for a device. - The negotiate command allows users to enable and disable disconnection and tagged queueing, set sync rates, offsets and bus width. Note that not all of those features are available for all controllers. Only the adv, ahc, and ncr drivers fully support all of the features at this point. Some cards do not allow the setting of sync rates, offsets and the like, and some of the drivers don't have any facilities to do so. Some drivers, like the adw driver, only support enabling or disabling sync negotiation, but do not support setting sync rates. - new description in the camcontrol man page of how to format a disk - cleanup of the camcontrol inquiry command - add support in the 'devlist' command for skipping unconfigured devices if -v was not specified on the command line. - make use of the new base_transfer_speed in the path inquiry CCB. - fix CCB bzero cases cam_xpt.c, cam_sim.[ch], cam_ccb.h: - new flags on many CCB function codes to designate whether they're non-immediate, use a user-supplied CCB, and can only be passed from userland programs via the xpt device. Use these flags in the transport layer and pass driver to categorize CCBs. - new flag in the transport layer device matching code for device nodes that indicates whether a device is unconfigured - bump the CAM version from 0x10 to 0x11 - Change the CAM ioctls to use the version as their group code, so we can force users to recompile code even when the CCB size doesn't change. - add + fill in a new value in the path inquiry CCB, base_transfer_speed. Remove a corresponding field from the cam_sim structure, and add code to every SIM to set this field to the proper value. - Fix the set transfer settings code in the transport layer. scsi_cd.c: - make some variables volatile instead of just casting them in various places - fix a race condition in the changer code - attach unless we get a "logical unit not supported" error. This should fix all of the cases where people have devices that return weird errors when they don't have media in the drive. scsi_da.c: - attach unless we get a "logical unit not supported" error scsi_pass.c: - for immediate CCBs, just malloc a CCB to send the user request in. This gets rid of the 'held' count problem in camcontrol tags. scsi_pass.h: - change the CAM ioctls to use the CAM version as their group code. adv driver: - Allow changing the sync rate and offset separately. adw driver - Allow changing the sync rate and offset separately. aha driver: - Don't return CAM_REQ_CMP for SET_TRAN_SETTINGS CCBs. ahc driver: - Allow setting offset and sync rate separately bt driver: - Don't return CAM_REQ_CMP for SET_TRAN_SETTINGS CCBs. NCR driver: - Fix the ultra/ultra 2 negotiation bug - allow setting both the sync rate and offset separately Other HBA drivers: - Put code in to set the base_transfer_speed field for XPT_GET_TRAN_SETTINGS CCBs. Reviewed by: gibbs, mjacob (isp), imp (aha)
1999-05-06 20:16:39 +00:00
/*
* Set base transfer capabilities for Fibre Channel.
* Technically not correct because we don't know
* what media we're running on top of- but we'll
* look good if we always say 100MB/s.
*/
Begin to implement target mode that for Fibre Channel has a private per-command component that we *don't* try and pass thru CAM. CAM just is too risky and too much of a pain- structures get copied, but not all info of interest can be considered safely transported thru all consumers (including user space) from the incoming ATIO to the outgoing CTIO- it's just much safer to have a buddy structure, identified by the command's tag which *does* make it thru safely. Pay attention to link speed and report 200MB/s xfer speed for a 23XX card in 2GPs mode. MFC after: 1 week
2001-10-01 03:48:42 +00:00
if (FCPARAM(isp)->isp_gbspeed == 2)
cpi->base_transfer_speed = 200000;
else
cpi->base_transfer_speed = 100000;
Clarify and cleanup some CAM queueing breakages.
1999-08-16 20:11:45 +00:00
cpi->hba_inquiry = PI_TAG_ABLE;
Add CAM_NEW_TRAN_CODE support. Use correct CAMLOCK_2_ISPLOCK macros. For fibre channel, start going for the gusto and using AC_FOUND_DEVICE and AC_LOST_DEVICE calls to xpt_async when devices appear and disappear as the loop or fabric changes. ISPASYNC_FW_CRASH is the async event code where the platform layer deals with a firmware crash.
2001-07-04 18:54:29 +00:00
#ifdef CAM_NEW_TRAN_CODE
cpi->transport = XPORT_FC;
cpi->transport_version = 0; /* WHAT'S THIS FOR? */
#endif
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
} else {
A large set of changes to handle dual bus adapters.
1999-05-11 05:10:06 +00:00
sdparam *sdp = isp->isp_param;
sdp += cam_sim_bus(xpt_path_sim(cpi->ccb_h.path));
Clarify and cleanup some CAM queueing breakages.
1999-08-16 20:11:45 +00:00
cpi->hba_inquiry = PI_SDTR_ABLE|PI_TAG_ABLE|PI_WIDE_16;
Now that we fixed the isp_sendmarker botch, we can now do initial bus resets for ULTRA2/ULTRA3 cards again (which were turned off really because of a botch for dual bus configurations).
2000-04-21 19:18:06 +00:00
cpi->hba_misc = 0;
A large set of changes to handle dual bus adapters.
1999-05-11 05:10:06 +00:00
cpi->initiator_id = sdp->isp_initiator_id;
Add a number of interrelated CAM feature enhancements and bug fixes. NOTE: These changes will require recompilation of any userland applications, like cdrecord, xmcd, etc., that use the CAM passthrough interface. A make world is recommended. camcontrol.[c8]: - We now support two new commands, "tags" and "negotiate". - The tags commands allows users to view the number of tagged openings for a device as well as a number of other related parameters, and it allows users to set tagged openings for a device. - The negotiate command allows users to enable and disable disconnection and tagged queueing, set sync rates, offsets and bus width. Note that not all of those features are available for all controllers. Only the adv, ahc, and ncr drivers fully support all of the features at this point. Some cards do not allow the setting of sync rates, offsets and the like, and some of the drivers don't have any facilities to do so. Some drivers, like the adw driver, only support enabling or disabling sync negotiation, but do not support setting sync rates. - new description in the camcontrol man page of how to format a disk - cleanup of the camcontrol inquiry command - add support in the 'devlist' command for skipping unconfigured devices if -v was not specified on the command line. - make use of the new base_transfer_speed in the path inquiry CCB. - fix CCB bzero cases cam_xpt.c, cam_sim.[ch], cam_ccb.h: - new flags on many CCB function codes to designate whether they're non-immediate, use a user-supplied CCB, and can only be passed from userland programs via the xpt device. Use these flags in the transport layer and pass driver to categorize CCBs. - new flag in the transport layer device matching code for device nodes that indicates whether a device is unconfigured - bump the CAM version from 0x10 to 0x11 - Change the CAM ioctls to use the version as their group code, so we can force users to recompile code even when the CCB size doesn't change. - add + fill in a new value in the path inquiry CCB, base_transfer_speed. Remove a corresponding field from the cam_sim structure, and add code to every SIM to set this field to the proper value. - Fix the set transfer settings code in the transport layer. scsi_cd.c: - make some variables volatile instead of just casting them in various places - fix a race condition in the changer code - attach unless we get a "logical unit not supported" error. This should fix all of the cases where people have devices that return weird errors when they don't have media in the drive. scsi_da.c: - attach unless we get a "logical unit not supported" error scsi_pass.c: - for immediate CCBs, just malloc a CCB to send the user request in. This gets rid of the 'held' count problem in camcontrol tags. scsi_pass.h: - change the CAM ioctls to use the CAM version as their group code. adv driver: - Allow changing the sync rate and offset separately. adw driver - Allow changing the sync rate and offset separately. aha driver: - Don't return CAM_REQ_CMP for SET_TRAN_SETTINGS CCBs. ahc driver: - Allow setting offset and sync rate separately bt driver: - Don't return CAM_REQ_CMP for SET_TRAN_SETTINGS CCBs. NCR driver: - Fix the ultra/ultra 2 negotiation bug - allow setting both the sync rate and offset separately Other HBA drivers: - Put code in to set the base_transfer_speed field for XPT_GET_TRAN_SETTINGS CCBs. Reviewed by: gibbs, mjacob (isp), imp (aha)
1999-05-06 20:16:39 +00:00
cpi->base_transfer_speed = 3300;
Add CAM_NEW_TRAN_CODE support. Use correct CAMLOCK_2_ISPLOCK macros. For fibre channel, start going for the gusto and using AC_FOUND_DEVICE and AC_LOST_DEVICE calls to xpt_async when devices appear and disappear as the loop or fabric changes. ISPASYNC_FW_CRASH is the async event code where the platform layer deals with a firmware crash.
2001-07-04 18:54:29 +00:00
#ifdef CAM_NEW_TRAN_CODE
cpi->transport = XPORT_SPI;
cpi->transport_version = 2; /* WHAT'S THIS FOR? */
#endif
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
}
Add CAM_NEW_TRAN_CODE support. Use correct CAMLOCK_2_ISPLOCK macros. For fibre channel, start going for the gusto and using AC_FOUND_DEVICE and AC_LOST_DEVICE calls to xpt_async when devices appear and disappear as the loop or fabric changes. ISPASYNC_FW_CRASH is the async event code where the platform layer deals with a firmware crash.
2001-07-04 18:54:29 +00:00
#ifdef CAM_NEW_TRAN_CODE
cpi->protocol = PROTO_SCSI;
cpi->protocol_version = SCSI_REV_2;
#endif
Update QLogic ISP support for CAM. Add preliminary target mode support. Submitted by: Matthew Jacob <mjacob@feral.com>
1998-09-15 08:42:56 +00:00
strncpy(cpi->sim_vid, "FreeBSD", SIM_IDLEN);
strncpy(cpi->hba_vid, "Qlogic", HBA_IDLEN);
strncpy(cpi->dev_name, cam_sim_name(sim), DEV_IDLEN);
cpi->unit_number = cam_sim_unit(sim);
cpi->ccb_h.status = CAM_REQ_CMP;
xpt_done(ccb);
break;
}
default:
ccb->ccb_h.status = CAM_REQ_INVALID;
xpt_done(ccb);
break;
}
}
make a real "done" func; clarify some LUN widths
1998-12-05 01:31:21 +00:00
#define ISPDDB (CAM_DEBUG_INFO|CAM_DEBUG_TRACE|CAM_DEBUG_CDB)
void
clarify headers;move uninit to outer layer;remove watchdog
1998-12-28 19:22:27 +00:00
isp_done(struct ccb_scsiio *sccb)
make a real "done" func; clarify some LUN widths
1998-12-05 01:31:21 +00:00
{
struct ispsoftc *isp = XS_ISP(sccb);
if (XS_NOERR(sccb))
XS_SETERR(sccb, CAM_REQ_CMP);
Add in the enabling of interrupts (to isp_attach). Clean up a busted comment. Check against firmware state- not loop state when enabling target mode. Other changes have to do with no longer enabling/disabling interrupts at will. Rearchitect command watchdog timeouts- First of all, set the timeout period for a command that has a timeout (in isp_action) to the period of time requested *plus* two seconds. We don't want the Qlogic firmware and the host system to race each other to report a dead command (the watchdog is there to catch dead and/or broken firmware). Next, make sure that the command being watched isn't done yet. If it's not done yet, check for INT_PENDING and call isp_intr- if that said it serviced an interrupt, check to see whether the command is now done (this is what the "IN WATCHDOG" private flag is for- if isp_intr completes the command, it won't call xpt_done on it because isp_watchdog is still looking at the command). If no interrupt was pending, or the command wasn't completed, check to see if we've set the private 'grace period' flag. If so, the command really *is* dead, so report it as dead and complete it with a CAM_CMD_TIMEOUT value. If the grace period flag wasn't set, set it and issue a SYNCHRONIZE_ALL Marker Request Queue entry and re-set the timeout for one second from now (see Revision 1.45 isp.c notes for more on this) to give the firmware a final chance to complete this command.
2000-06-27 19:31:02 +00:00
make a real "done" func; clarify some LUN widths
1998-12-05 01:31:21 +00:00
if ((sccb->ccb_h.status & CAM_STATUS_MASK) == CAM_REQ_CMP &&
(sccb->scsi_status != SCSI_STATUS_OK)) {
sccb->ccb_h.status &= ~CAM_STATUS_MASK;
Clean up some debug printing. Find the correct lun when SCCLUN is defined. If we complete with a check condition but no sense data, say we had an AUTOSENSE failure.
2000-01-04 03:45:49 +00:00
if ((sccb->scsi_status == SCSI_STATUS_CHECK_COND) &&
(sccb->ccb_h.status & CAM_AUTOSNS_VALID) == 0) {
sccb->ccb_h.status |= CAM_AUTOSENSE_FAIL;
} else {
sccb->ccb_h.status |= CAM_SCSI_STATUS_ERROR;
}
make a real "done" func; clarify some LUN widths
1998-12-05 01:31:21 +00:00
}
Add in the enabling of interrupts (to isp_attach). Clean up a busted comment. Check against firmware state- not loop state when enabling target mode. Other changes have to do with no longer enabling/disabling interrupts at will. Rearchitect command watchdog timeouts- First of all, set the timeout period for a command that has a timeout (in isp_action) to the period of time requested *plus* two seconds. We don't want the Qlogic firmware and the host system to race each other to report a dead command (the watchdog is there to catch dead and/or broken firmware). Next, make sure that the command being watched isn't done yet. If it's not done yet, check for INT_PENDING and call isp_intr- if that said it serviced an interrupt, check to see whether the command is now done (this is what the "IN WATCHDOG" private flag is for- if isp_intr completes the command, it won't call xpt_done on it because isp_watchdog is still looking at the command). If no interrupt was pending, or the command wasn't completed, check to see if we've set the private 'grace period' flag. If so, the command really *is* dead, so report it as dead and complete it with a CAM_CMD_TIMEOUT value. If the grace period flag wasn't set, set it and issue a SYNCHRONIZE_ALL Marker Request Queue entry and re-set the timeout for one second from now (see Revision 1.45 isp.c notes for more on this) to give the firmware a final chance to complete this command.
2000-06-27 19:31:02 +00:00
Clarify and cleanup some CAM queueing breakages.
1999-08-16 20:11:45 +00:00
sccb->ccb_h.status &= ~CAM_SIM_QUEUED;
make a real "done" func; clarify some LUN widths
1998-12-05 01:31:21 +00:00
if ((sccb->ccb_h.status & CAM_STATUS_MASK) != CAM_REQ_CMP) {
if ((sccb->ccb_h.status & CAM_DEV_QFRZN) == 0) {
sccb->ccb_h.status |= CAM_DEV_QFRZN;
Clarify and cleanup some CAM queueing breakages.
1999-08-16 20:11:45 +00:00
xpt_freeze_devq(sccb->ccb_h.path, 1);
Add get/set param ioctl support. Remove sim queue freezes for resource shortages. I've had too many strange race conditions where I freeze on a resource shortage but never get unfrozen. Consolidate the remaining sim queue freeze condition (for loopdown) into an inline with debug messages that allows us to track problems at ISP_LOGDEBUG0 level easier. Change a bunch of debug messages about loop down/up conditions to ISP_LOGDEBUG0 level. Remove dead isp_relsim code. Change some internal flag stuff for efficiency. Complain vociferously if we try and use our FC scratch area while it's busy being used already (I mean, if we don't have solaris' ability to sleep as an interrupt thread which would allow us to just use a p/v semaphore, at least *say* when you've just borked yourself). Add infrastructure to allow overrides of hard loopid && initiator id from boot variables. Fix the usual quota of silly bugs: + 'ktmature' needs to be per-instance. Argh. + When entering isp_watchdog, set intsok to zero, preserving old value to restore later. It's not nice to try and sleep from splsoftclock. + Fix tick overflow buglet in checking timeout value. MFC after: 1 week
2002-07-08 17:42:47 +00:00
isp_prt(isp, ISP_LOGDEBUG0,
"freeze devq %d.%d cam sts %x scsi sts %x",
sccb->ccb_h.target_id, sccb->ccb_h.target_lun,
sccb->ccb_h.status, sccb->scsi_status);
make a real "done" func; clarify some LUN widths
1998-12-05 01:31:21 +00:00
}
}
Add in the enabling of interrupts (to isp_attach). Clean up a busted comment. Check against firmware state- not loop state when enabling target mode. Other changes have to do with no longer enabling/disabling interrupts at will. Rearchitect command watchdog timeouts- First of all, set the timeout period for a command that has a timeout (in isp_action) to the period of time requested *plus* two seconds. We don't want the Qlogic firmware and the host system to race each other to report a dead command (the watchdog is there to catch dead and/or broken firmware). Next, make sure that the command being watched isn't done yet. If it's not done yet, check for INT_PENDING and call isp_intr- if that said it serviced an interrupt, check to see whether the command is now done (this is what the "IN WATCHDOG" private flag is for- if isp_intr completes the command, it won't call xpt_done on it because isp_watchdog is still looking at the command). If no interrupt was pending, or the command wasn't completed, check to see if we've set the private 'grace period' flag. If so, the command really *is* dead, so report it as dead and complete it with a CAM_CMD_TIMEOUT value. If the grace period flag wasn't set, set it and issue a SYNCHRONIZE_ALL Marker Request Queue entry and re-set the timeout for one second from now (see Revision 1.45 isp.c notes for more on this) to give the firmware a final chance to complete this command.
2000-06-27 19:31:02 +00:00
if ((CAM_DEBUGGED(sccb->ccb_h.path, ISPDDB)) &&
make a real "done" func; clarify some LUN widths
1998-12-05 01:31:21 +00:00
(sccb->ccb_h.status & CAM_STATUS_MASK) != CAM_REQ_CMP) {
xpt_print_path(sccb->ccb_h.path);
Finally eliminate as many of the printf calls as possible (still leaving ones where we have a CAM path) and replacing them with calls to isp_prt., Eliminate isp_unit references- we no longer have an isp_unit- we now have an isp_dev that device_get_unit can work with.
2001-03-01 02:14:54 +00:00
isp_prt(isp, ISP_LOGINFO,
"cam completion status 0x%x", sccb->ccb_h.status);
make a real "done" func; clarify some LUN widths
1998-12-05 01:31:21 +00:00
}
Add in the enabling of interrupts (to isp_attach). Clean up a busted comment. Check against firmware state- not loop state when enabling target mode. Other changes have to do with no longer enabling/disabling interrupts at will. Rearchitect command watchdog timeouts- First of all, set the timeout period for a command that has a timeout (in isp_action) to the period of time requested *plus* two seconds. We don't want the Qlogic firmware and the host system to race each other to report a dead command (the watchdog is there to catch dead and/or broken firmware). Next, make sure that the command being watched isn't done yet. If it's not done yet, check for INT_PENDING and call isp_intr- if that said it serviced an interrupt, check to see whether the command is now done (this is what the "IN WATCHDOG" private flag is for- if isp_intr completes the command, it won't call xpt_done on it because isp_watchdog is still looking at the command). If no interrupt was pending, or the command wasn't completed, check to see if we've set the private 'grace period' flag. If so, the command really *is* dead, so report it as dead and complete it with a CAM_CMD_TIMEOUT value. If the grace period flag wasn't set, set it and issue a SYNCHRONIZE_ALL Marker Request Queue entry and re-set the timeout for one second from now (see Revision 1.45 isp.c notes for more on this) to give the firmware a final chance to complete this command.
2000-06-27 19:31:02 +00:00
XS_CMD_S_DONE(sccb);
if (XS_CMD_WDOG_P(sccb) == 0) {
untimeout(isp_watchdog, (caddr_t)sccb, sccb->ccb_h.timeout_ch);
if (XS_CMD_GRACE_P(sccb)) {
Core version 2.0 rewrite. In this file we replace isp_tdebug with isp_prt calls. We now use an argument to the ISPCTL_FCLINK_TEST call. We change all IDPRINTF macros to isp_prt calls. We add the isp_prt function here.
2000-08-01 06:31:44 +00:00
isp_prt(isp, ISP_LOGDEBUG2,
"finished command on borrowed time");
Add in the enabling of interrupts (to isp_attach). Clean up a busted comment. Check against firmware state- not loop state when enabling target mode. Other changes have to do with no longer enabling/disabling interrupts at will. Rearchitect command watchdog timeouts- First of all, set the timeout period for a command that has a timeout (in isp_action) to the period of time requested *plus* two seconds. We don't want the Qlogic firmware and the host system to race each other to report a dead command (the watchdog is there to catch dead and/or broken firmware). Next, make sure that the command being watched isn't done yet. If it's not done yet, check for INT_PENDING and call isp_intr- if that said it serviced an interrupt, check to see whether the command is now done (this is what the "IN WATCHDOG" private flag is for- if isp_intr completes the command, it won't call xpt_done on it because isp_watchdog is still looking at the command). If no interrupt was pending, or the command wasn't completed, check to see if we've set the private 'grace period' flag. If so, the command really *is* dead, so report it as dead and complete it with a CAM_CMD_TIMEOUT value. If the grace period flag wasn't set, set it and issue a SYNCHRONIZE_ALL Marker Request Queue entry and re-set the timeout for one second from now (see Revision 1.45 isp.c notes for more on this) to give the firmware a final chance to complete this command.
2000-06-27 19:31:02 +00:00
}
XS_CMD_S_CLEAR(sccb);
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
ISPLOCK_2_CAMLOCK(isp);
Make sure we do locking if we call isp_intr. Make sure we enter Giant for now if we call into cam for completion.
2000-12-29 19:06:32 +00:00
xpt_done((union ccb *) sccb);
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
CAMLOCK_2_ISPLOCK(isp);
Add in the enabling of interrupts (to isp_attach). Clean up a busted comment. Check against firmware state- not loop state when enabling target mode. Other changes have to do with no longer enabling/disabling interrupts at will. Rearchitect command watchdog timeouts- First of all, set the timeout period for a command that has a timeout (in isp_action) to the period of time requested *plus* two seconds. We don't want the Qlogic firmware and the host system to race each other to report a dead command (the watchdog is there to catch dead and/or broken firmware). Next, make sure that the command being watched isn't done yet. If it's not done yet, check for INT_PENDING and call isp_intr- if that said it serviced an interrupt, check to see whether the command is now done (this is what the "IN WATCHDOG" private flag is for- if isp_intr completes the command, it won't call xpt_done on it because isp_watchdog is still looking at the command). If no interrupt was pending, or the command wasn't completed, check to see if we've set the private 'grace period' flag. If so, the command really *is* dead, so report it as dead and complete it with a CAM_CMD_TIMEOUT value. If the grace period flag wasn't set, set it and issue a SYNCHRONIZE_ALL Marker Request Queue entry and re-set the timeout for one second from now (see Revision 1.45 isp.c notes for more on this) to give the firmware a final chance to complete this command.
2000-06-27 19:31:02 +00:00
}
make a real "done" func; clarify some LUN widths
1998-12-05 01:31:21 +00:00
}
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 07:29:00 +00:00
int
Clarify and cleanup some CAM queueing breakages.
1999-08-16 20:11:45 +00:00
isp_async(struct ispsoftc *isp, ispasync_t cmd, void *arg)
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 07:29:00 +00:00
{
A large set of changes to handle dual bus adapters.
1999-05-11 05:10:06 +00:00
int bus, rv = 0;
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 07:29:00 +00:00
switch (cmd) {
case ISPASYNC_NEW_TGT_PARAMS:
Clarify and cleanup some CAM queueing breakages.
1999-08-16 20:11:45 +00:00
{
Add CAM_NEW_TRAN_CODE support. Use correct CAMLOCK_2_ISPLOCK macros. For fibre channel, start going for the gusto and using AC_FOUND_DEVICE and AC_LOST_DEVICE calls to xpt_async when devices appear and disappear as the loop or fabric changes. ISPASYNC_FW_CRASH is the async event code where the platform layer deals with a firmware crash.
2001-07-04 18:54:29 +00:00
#ifdef CAM_NEW_TRAN_CODE
struct ccb_trans_settings_scsi *scsi;
struct ccb_trans_settings_spi *spi;
#endif
Clarify and cleanup some CAM queueing breakages.
1999-08-16 20:11:45 +00:00
int flags, tgt;
sdparam *sdp = isp->isp_param;
Add CAM_NEW_TRAN_CODE support. Use correct CAMLOCK_2_ISPLOCK macros. For fibre channel, start going for the gusto and using AC_FOUND_DEVICE and AC_LOST_DEVICE calls to xpt_async when devices appear and disappear as the loop or fabric changes. ISPASYNC_FW_CRASH is the async event code where the platform layer deals with a firmware crash.
2001-07-04 18:54:29 +00:00
struct ccb_trans_settings cts;
Clarify and cleanup some CAM queueing breakages.
1999-08-16 20:11:45 +00:00
struct cam_path *tmppath;
Add CAM_NEW_TRAN_CODE support. Use correct CAMLOCK_2_ISPLOCK macros. For fibre channel, start going for the gusto and using AC_FOUND_DEVICE and AC_LOST_DEVICE calls to xpt_async when devices appear and disappear as the loop or fabric changes. ISPASYNC_FW_CRASH is the async event code where the platform layer deals with a firmware crash.
2001-07-04 18:54:29 +00:00
bzero(&cts, sizeof (struct ccb_trans_settings));
Clarify and cleanup some CAM queueing breakages.
1999-08-16 20:11:45 +00:00
tgt = *((int *)arg);
bus = (tgt >> 16) & 0xffff;
tgt &= 0xffff;
sdp += bus;
Things have become cinched down more tightly about assertions for Giant. This uncovered some missing spots where I trade off between isp's lock and Giant as I enter CAM.
2001-07-05 17:14:57 +00:00
ISPLOCK_2_CAMLOCK(isp);
Clarify and cleanup some CAM queueing breakages.
1999-08-16 20:11:45 +00:00
if (xpt_create_path(&tmppath, NULL,
cam_sim_path(bus? isp->isp_sim2 : isp->isp_sim),
tgt, CAM_LUN_WILDCARD) != CAM_REQ_CMP) {
Things have become cinched down more tightly about assertions for Giant. This uncovered some missing spots where I trade off between isp's lock and Giant as I enter CAM.
2001-07-05 17:14:57 +00:00
CAMLOCK_2_ISPLOCK(isp);
Replace some more printfs with isp_prt's. Use isp_prt/ISP_LOGDEBUG0 for rate setting/getting printouts.
2000-12-05 07:39:54 +00:00
isp_prt(isp, ISP_LOGWARN,
"isp_async cannot make temp path for %d.%d",
tgt, bus);
Clarify and cleanup some CAM queueing breakages.
1999-08-16 20:11:45 +00:00
rv = -1;
break;
}
Things have become cinched down more tightly about assertions for Giant. This uncovered some missing spots where I trade off between isp's lock and Giant as I enter CAM.
2001-07-05 17:14:57 +00:00
CAMLOCK_2_ISPLOCK(isp);
Redo how we manage SCSI device settings- have a 3rd flags (nvram) that records either what's in NVRAM or what the safe defaults would be if we lack NVRAM. Then we rename cur_XXXX to actv_XXXX (these are the currently active settings) and the dev_XXX settings to goal_XXXX (these are the settings which we want cur_XXXX to converge to). This probably isn't entirely final as yet- but it's a lot closer to now being what it should be, including allowing camcontrol to actually set specific settings.
2001-07-30 01:00:21 +00:00
flags = sdp->isp_devparam[tgt].actv_flags;
Add CAM_NEW_TRAN_CODE support. Use correct CAMLOCK_2_ISPLOCK macros. For fibre channel, start going for the gusto and using AC_FOUND_DEVICE and AC_LOST_DEVICE calls to xpt_async when devices appear and disappear as the loop or fabric changes. ISPASYNC_FW_CRASH is the async event code where the platform layer deals with a firmware crash.
2001-07-04 18:54:29 +00:00
#ifdef CAM_NEW_TRAN_CODE
cts.type = CTS_TYPE_CURRENT_SETTINGS;
cts.protocol = PROTO_SCSI;
cts.transport = XPORT_SPI;
scsi = &cts.proto_specific.scsi;
spi = &cts.xport_specific.spi;
if (flags & DPARM_TQING) {
scsi->valid |= CTS_SCSI_VALID_TQ;
scsi->flags |= CTS_SCSI_FLAGS_TAG_ENB;
spi->flags |= CTS_SPI_FLAGS_TAG_ENB;
}
Clarify and cleanup some CAM queueing breakages.
1999-08-16 20:11:45 +00:00
if (flags & DPARM_DISC) {
Add CAM_NEW_TRAN_CODE support. Use correct CAMLOCK_2_ISPLOCK macros. For fibre channel, start going for the gusto and using AC_FOUND_DEVICE and AC_LOST_DEVICE calls to xpt_async when devices appear and disappear as the loop or fabric changes. ISPASYNC_FW_CRASH is the async event code where the platform layer deals with a firmware crash.
2001-07-04 18:54:29 +00:00
spi->valid |= CTS_SPI_VALID_DISC;
spi->flags |= CTS_SPI_FLAGS_DISC_ENB;
}
spi->flags |= CTS_SPI_VALID_BUS_WIDTH;
if (flags & DPARM_WIDE) {
spi->bus_width = MSG_EXT_WDTR_BUS_16_BIT;
} else {
spi->bus_width = MSG_EXT_WDTR_BUS_8_BIT;
}
if (flags & DPARM_SYNC) {
spi->valid |= CTS_SPI_VALID_SYNC_RATE;
spi->valid |= CTS_SPI_VALID_SYNC_OFFSET;
Redo how we manage SCSI device settings- have a 3rd flags (nvram) that records either what's in NVRAM or what the safe defaults would be if we lack NVRAM. Then we rename cur_XXXX to actv_XXXX (these are the currently active settings) and the dev_XXX settings to goal_XXXX (these are the settings which we want cur_XXXX to converge to). This probably isn't entirely final as yet- but it's a lot closer to now being what it should be, including allowing camcontrol to actually set specific settings.
2001-07-30 01:00:21 +00:00
spi->sync_period = sdp->isp_devparam[tgt].actv_period;
spi->sync_offset = sdp->isp_devparam[tgt].actv_offset;
Add CAM_NEW_TRAN_CODE support. Use correct CAMLOCK_2_ISPLOCK macros. For fibre channel, start going for the gusto and using AC_FOUND_DEVICE and AC_LOST_DEVICE calls to xpt_async when devices appear and disappear as the loop or fabric changes. ISPASYNC_FW_CRASH is the async event code where the platform layer deals with a firmware crash.
2001-07-04 18:54:29 +00:00
}
#else
cts.flags = CCB_TRANS_CURRENT_SETTINGS;
cts.valid = CCB_TRANS_DISC_VALID | CCB_TRANS_TQ_VALID;
if (flags & DPARM_DISC) {
cts.flags |= CCB_TRANS_DISC_ENB;
Clarify and cleanup some CAM queueing breakages.
1999-08-16 20:11:45 +00:00
}
if (flags & DPARM_TQING) {
Add CAM_NEW_TRAN_CODE support. Use correct CAMLOCK_2_ISPLOCK macros. For fibre channel, start going for the gusto and using AC_FOUND_DEVICE and AC_LOST_DEVICE calls to xpt_async when devices appear and disappear as the loop or fabric changes. ISPASYNC_FW_CRASH is the async event code where the platform layer deals with a firmware crash.
2001-07-04 18:54:29 +00:00
cts.flags |= CCB_TRANS_TAG_ENB;
Clarify and cleanup some CAM queueing breakages.
1999-08-16 20:11:45 +00:00
}
Add CAM_NEW_TRAN_CODE support. Use correct CAMLOCK_2_ISPLOCK macros. For fibre channel, start going for the gusto and using AC_FOUND_DEVICE and AC_LOST_DEVICE calls to xpt_async when devices appear and disappear as the loop or fabric changes. ISPASYNC_FW_CRASH is the async event code where the platform layer deals with a firmware crash.
2001-07-04 18:54:29 +00:00
cts.valid |= CCB_TRANS_BUS_WIDTH_VALID;
cts.bus_width = (flags & DPARM_WIDE)?
Clarify and cleanup some CAM queueing breakages.
1999-08-16 20:11:45 +00:00
MSG_EXT_WDTR_BUS_8_BIT : MSG_EXT_WDTR_BUS_16_BIT;
Redo how we manage SCSI device settings- have a 3rd flags (nvram) that records either what's in NVRAM or what the safe defaults would be if we lack NVRAM. Then we rename cur_XXXX to actv_XXXX (these are the currently active settings) and the dev_XXX settings to goal_XXXX (these are the settings which we want cur_XXXX to converge to). This probably isn't entirely final as yet- but it's a lot closer to now being what it should be, including allowing camcontrol to actually set specific settings.
2001-07-30 01:00:21 +00:00
cts.sync_period = sdp->isp_devparam[tgt].actv_period;
cts.sync_offset = sdp->isp_devparam[tgt].actv_offset;
Clarify and cleanup some CAM queueing breakages.
1999-08-16 20:11:45 +00:00
if (flags & DPARM_SYNC) {
Add CAM_NEW_TRAN_CODE support. Use correct CAMLOCK_2_ISPLOCK macros. For fibre channel, start going for the gusto and using AC_FOUND_DEVICE and AC_LOST_DEVICE calls to xpt_async when devices appear and disappear as the loop or fabric changes. ISPASYNC_FW_CRASH is the async event code where the platform layer deals with a firmware crash.
2001-07-04 18:54:29 +00:00
cts.valid |=
Clarify and cleanup some CAM queueing breakages.
1999-08-16 20:11:45 +00:00
CCB_TRANS_SYNC_RATE_VALID |
CCB_TRANS_SYNC_OFFSET_VALID;
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 07:29:00 +00:00
}
Add CAM_NEW_TRAN_CODE support. Use correct CAMLOCK_2_ISPLOCK macros. For fibre channel, start going for the gusto and using AC_FOUND_DEVICE and AC_LOST_DEVICE calls to xpt_async when devices appear and disappear as the loop or fabric changes. ISPASYNC_FW_CRASH is the async event code where the platform layer deals with a firmware crash.
2001-07-04 18:54:29 +00:00
#endif
Core version 2.0 rewrite. In this file we replace isp_tdebug with isp_prt calls. We now use an argument to the ISPCTL_FCLINK_TEST call. We change all IDPRINTF macros to isp_prt calls. We add the isp_prt function here.
2000-08-01 06:31:44 +00:00
isp_prt(isp, ISP_LOGDEBUG2,
"NEW_TGT_PARAMS bus %d tgt %d period %x offset %x flags %x",
Redo how we manage SCSI device settings- have a 3rd flags (nvram) that records either what's in NVRAM or what the safe defaults would be if we lack NVRAM. Then we rename cur_XXXX to actv_XXXX (these are the currently active settings) and the dev_XXX settings to goal_XXXX (these are the settings which we want cur_XXXX to converge to). This probably isn't entirely final as yet- but it's a lot closer to now being what it should be, including allowing camcontrol to actually set specific settings.
2001-07-30 01:00:21 +00:00
bus, tgt, sdp->isp_devparam[tgt].actv_period,
sdp->isp_devparam[tgt].actv_offset, flags);
Add CAM_NEW_TRAN_CODE support. Use correct CAMLOCK_2_ISPLOCK macros. For fibre channel, start going for the gusto and using AC_FOUND_DEVICE and AC_LOST_DEVICE calls to xpt_async when devices appear and disappear as the loop or fabric changes. ISPASYNC_FW_CRASH is the async event code where the platform layer deals with a firmware crash.
2001-07-04 18:54:29 +00:00
xpt_setup_ccb(&cts.ccb_h, tmppath, 1);
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
ISPLOCK_2_CAMLOCK(isp);
Add CAM_NEW_TRAN_CODE support. Use correct CAMLOCK_2_ISPLOCK macros. For fibre channel, start going for the gusto and using AC_FOUND_DEVICE and AC_LOST_DEVICE calls to xpt_async when devices appear and disappear as the loop or fabric changes. ISPASYNC_FW_CRASH is the async event code where the platform layer deals with a firmware crash.
2001-07-04 18:54:29 +00:00
xpt_async(AC_TRANSFER_NEG, tmppath, &cts);
Clarify and cleanup some CAM queueing breakages.
1999-08-16 20:11:45 +00:00
xpt_free_path(tmppath);
Remove ISP_SMPLOCK stuff- we're just using locking now. Correctly reintroduce loop_seen_once semantics- that is, if we've never seen good link, start bouncing commands with CAM_SEL_TIMEOUT. But we have to be careful to have let ourselves try (in isp_kthread) to check for loop up at least once. PR: 28992 MFC after: 1 week
2001-07-25 04:23:52 +00:00
CAMLOCK_2_ISPLOCK(isp);
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 07:29:00 +00:00
break;
Clarify and cleanup some CAM queueing breakages.
1999-08-16 20:11:45 +00:00
}
A wad of changes- prepping for 1080/1240 support (which caused a massive thwank in register layout goop). A different mboxcmd approach. Some PDB change infrastructure. Some better management of loopdown/loopup events (keep them distinct from resource starvation for simq freeze/unfreeze actions).
1999-03-17 05:04:39 +00:00
case ISPASYNC_BUS_RESET:
A large set of changes to handle dual bus adapters.
1999-05-11 05:10:06 +00:00
bus = *((int *)arg);
Core version 2.0 rewrite. In this file we replace isp_tdebug with isp_prt calls. We now use an argument to the ISPCTL_FCLINK_TEST call. We change all IDPRINTF macros to isp_prt calls. We add the isp_prt function here.
2000-08-01 06:31:44 +00:00
isp_prt(isp, ISP_LOGINFO, "SCSI bus reset on bus %d detected",
bus);
A large set of changes to handle dual bus adapters.
1999-05-11 05:10:06 +00:00
if (bus > 0 && isp->isp_path2) {
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
ISPLOCK_2_CAMLOCK(isp);
A large set of changes to handle dual bus adapters.
1999-05-11 05:10:06 +00:00
xpt_async(AC_BUS_RESET, isp->isp_path2, NULL);
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
CAMLOCK_2_ISPLOCK(isp);
A large set of changes to handle dual bus adapters.
1999-05-11 05:10:06 +00:00
} else if (isp->isp_path) {
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
ISPLOCK_2_CAMLOCK(isp);
A wad of changes- prepping for 1080/1240 support (which caused a massive thwank in register layout goop). A different mboxcmd approach. Some PDB change infrastructure. Some better management of loopdown/loopup events (keep them distinct from resource starvation for simq freeze/unfreeze actions).
1999-03-17 05:04:39 +00:00
xpt_async(AC_BUS_RESET, isp->isp_path, NULL);
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
CAMLOCK_2_ISPLOCK(isp);
A wad of changes- prepping for 1080/1240 support (which caused a massive thwank in register layout goop). A different mboxcmd approach. Some PDB change infrastructure. Some better management of loopdown/loopup events (keep them distinct from resource starvation for simq freeze/unfreeze actions).
1999-03-17 05:04:39 +00:00
}
break;
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
case ISPASYNC_LIP:
if (isp->isp_path) {
Add get/set param ioctl support. Remove sim queue freezes for resource shortages. I've had too many strange race conditions where I freeze on a resource shortage but never get unfrozen. Consolidate the remaining sim queue freeze condition (for loopdown) into an inline with debug messages that allows us to track problems at ISP_LOGDEBUG0 level easier. Change a bunch of debug messages about loop down/up conditions to ISP_LOGDEBUG0 level. Remove dead isp_relsim code. Change some internal flag stuff for efficiency. Complain vociferously if we try and use our FC scratch area while it's busy being used already (I mean, if we don't have solaris' ability to sleep as an interrupt thread which would allow us to just use a p/v semaphore, at least *say* when you've just borked yourself). Add infrastructure to allow overrides of hard loopid && initiator id from boot variables. Fix the usual quota of silly bugs: + 'ktmature' needs to be per-instance. Argh. + When entering isp_watchdog, set intsok to zero, preserving old value to restore later. It's not nice to try and sleep from splsoftclock. + Fix tick overflow buglet in checking timeout value. MFC after: 1 week
2002-07-08 17:42:47 +00:00
isp_freeze_loopdown(isp, "ISPASYNC_LIP");
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
}
isp_prt(isp, ISP_LOGINFO, "LIP Received");
break;
case ISPASYNC_LOOP_RESET:
if (isp->isp_path) {
Add get/set param ioctl support. Remove sim queue freezes for resource shortages. I've had too many strange race conditions where I freeze on a resource shortage but never get unfrozen. Consolidate the remaining sim queue freeze condition (for loopdown) into an inline with debug messages that allows us to track problems at ISP_LOGDEBUG0 level easier. Change a bunch of debug messages about loop down/up conditions to ISP_LOGDEBUG0 level. Remove dead isp_relsim code. Change some internal flag stuff for efficiency. Complain vociferously if we try and use our FC scratch area while it's busy being used already (I mean, if we don't have solaris' ability to sleep as an interrupt thread which would allow us to just use a p/v semaphore, at least *say* when you've just borked yourself). Add infrastructure to allow overrides of hard loopid && initiator id from boot variables. Fix the usual quota of silly bugs: + 'ktmature' needs to be per-instance. Argh. + When entering isp_watchdog, set intsok to zero, preserving old value to restore later. It's not nice to try and sleep from splsoftclock. + Fix tick overflow buglet in checking timeout value. MFC after: 1 week
2002-07-08 17:42:47 +00:00
isp_freeze_loopdown(isp, "ISPASYNC_LOOP_RESET");
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
}
isp_prt(isp, ISP_LOGINFO, "Loop Reset Received");
break;
A wad of changes- prepping for 1080/1240 support (which caused a massive thwank in register layout goop). A different mboxcmd approach. Some PDB change infrastructure. Some better management of loopdown/loopup events (keep them distinct from resource starvation for simq freeze/unfreeze actions).
1999-03-17 05:04:39 +00:00
case ISPASYNC_LOOP_DOWN:
if (isp->isp_path) {
Add get/set param ioctl support. Remove sim queue freezes for resource shortages. I've had too many strange race conditions where I freeze on a resource shortage but never get unfrozen. Consolidate the remaining sim queue freeze condition (for loopdown) into an inline with debug messages that allows us to track problems at ISP_LOGDEBUG0 level easier. Change a bunch of debug messages about loop down/up conditions to ISP_LOGDEBUG0 level. Remove dead isp_relsim code. Change some internal flag stuff for efficiency. Complain vociferously if we try and use our FC scratch area while it's busy being used already (I mean, if we don't have solaris' ability to sleep as an interrupt thread which would allow us to just use a p/v semaphore, at least *say* when you've just borked yourself). Add infrastructure to allow overrides of hard loopid && initiator id from boot variables. Fix the usual quota of silly bugs: + 'ktmature' needs to be per-instance. Argh. + When entering isp_watchdog, set intsok to zero, preserving old value to restore later. It's not nice to try and sleep from splsoftclock. + Fix tick overflow buglet in checking timeout value. MFC after: 1 week
2002-07-08 17:42:47 +00:00
isp_freeze_loopdown(isp, "ISPASYNC_LOOP_DOWN");
A wad of changes- prepping for 1080/1240 support (which caused a massive thwank in register layout goop). A different mboxcmd approach. Some PDB change infrastructure. Some better management of loopdown/loopup events (keep them distinct from resource starvation for simq freeze/unfreeze actions).
1999-03-17 05:04:39 +00:00
}
Core version 2.0 rewrite. In this file we replace isp_tdebug with isp_prt calls. We now use an argument to the ISPCTL_FCLINK_TEST call. We change all IDPRINTF macros to isp_prt calls. We add the isp_prt function here.
2000-08-01 06:31:44 +00:00
isp_prt(isp, ISP_LOGINFO, "Loop DOWN");
A wad of changes- prepping for 1080/1240 support (which caused a massive thwank in register layout goop). A different mboxcmd approach. Some PDB change infrastructure. Some better management of loopdown/loopup events (keep them distinct from resource starvation for simq freeze/unfreeze actions).
1999-03-17 05:04:39 +00:00
break;
case ISPASYNC_LOOP_UP:
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
/*
* Now we just note that Loop has come up. We don't
* actually do anything because we're waiting for a
* Change Notify before activating the FC cleanup
* thread to look at the state of the loop again.
*/
Core version 2.0 rewrite. In this file we replace isp_tdebug with isp_prt calls. We now use an argument to the ISPCTL_FCLINK_TEST call. We change all IDPRINTF macros to isp_prt calls. We add the isp_prt function here.
2000-08-01 06:31:44 +00:00
isp_prt(isp, ISP_LOGINFO, "Loop UP");
A wad of changes- prepping for 1080/1240 support (which caused a massive thwank in register layout goop). A different mboxcmd approach. Some PDB change infrastructure. Some better management of loopdown/loopup events (keep them distinct from resource starvation for simq freeze/unfreeze actions).
1999-03-17 05:04:39 +00:00
break;
Do some cleanup based upon adapter role- mainly not enabling interrupts if we're ISP_ROLE_NONE. Change ISPASYNC_LOGGED_INOUT to ISPASYNC_PROMENADE. Make sure we note if something is a fabric device. Target mode: Finally fix (to a first approximation) SCSI Target Mode again- we needed to correctly check against CAM_TARGET_WILDCARD and CAM_LUN_WILDCARD so that targbh won't confuse us. Comment out the drainqueue stuff for now. Use isp_fc_runstate instead if isp_control/ISPCTL_FCLINK_TEST.
2001-02-11 03:47:39 +00:00
case ISPASYNC_PROMENADE:
Clarify and cleanup some CAM queueing breakages.
1999-08-16 20:11:45 +00:00
{
Add CAM_NEW_TRAN_CODE support. Use correct CAMLOCK_2_ISPLOCK macros. For fibre channel, start going for the gusto and using AC_FOUND_DEVICE and AC_LOST_DEVICE calls to xpt_async when devices appear and disappear as the loop or fabric changes. ISPASYNC_FW_CRASH is the async event code where the platform layer deals with a firmware crash.
2001-07-04 18:54:29 +00:00
struct cam_path *tmppath;
Core version 2.0 rewrite. In this file we replace isp_tdebug with isp_prt calls. We now use an argument to the ISPCTL_FCLINK_TEST call. We change all IDPRINTF macros to isp_prt calls. We add the isp_prt function here.
2000-08-01 06:31:44 +00:00
const char *fmt = "Target %d (Loop 0x%x) Port ID 0x%x "
Do some cleanup based upon adapter role- mainly not enabling interrupts if we're ISP_ROLE_NONE. Change ISPASYNC_LOGGED_INOUT to ISPASYNC_PROMENADE. Make sure we note if something is a fabric device. Target mode: Finally fix (to a first approximation) SCSI Target Mode again- we needed to correctly check against CAM_TARGET_WILDCARD and CAM_LUN_WILDCARD so that targbh won't confuse us. Comment out the drainqueue stuff for now. Use isp_fc_runstate instead if isp_control/ISPCTL_FCLINK_TEST.
2001-02-11 03:47:39 +00:00
"(role %s) %s\n Port WWN 0x%08x%08x\n Node WWN 0x%08x%08x";
static const char *roles[4] = {
Clarify and cleanup some CAM queueing breakages.
1999-08-16 20:11:45 +00:00
"(none)", "Target", "Initiator", "Target/Initiator"
A wad of changes- prepping for 1080/1240 support (which caused a massive thwank in register layout goop). A different mboxcmd approach. Some PDB change infrastructure. Some better management of loopdown/loopup events (keep them distinct from resource starvation for simq freeze/unfreeze actions).
1999-03-17 05:04:39 +00:00
};
Remove all pre-CAM code. Fix breakage for SCCLUN when it is in fact ISP2100_SCCLUN. Make changes for how ISPASYNC_PDB_CHANGE_COMPLETE is now ISPASYNC_PDB_CHANGED. Add in ISPASYNC_FABRIC_DEV case.
1999-07-02 23:12:58 +00:00
fcparam *fcp = isp->isp_param;
int tgt = *((int *) arg);
Remove ISP_SMPLOCK stuff- we're just using locking now. Correctly reintroduce loop_seen_once semantics- that is, if we've never seen good link, start bouncing commands with CAM_SEL_TIMEOUT. But we have to be careful to have let ourselves try (in isp_kthread) to check for loop up at least once. PR: 28992 MFC after: 1 week
2001-07-25 04:23:52 +00:00
int is_tgt_mask = (SVC3_TGT_ROLE >> SVC3_ROLE_SHIFT);
Remove all pre-CAM code. Fix breakage for SCCLUN when it is in fact ISP2100_SCCLUN. Make changes for how ISPASYNC_PDB_CHANGE_COMPLETE is now ISPASYNC_PDB_CHANGED. Add in ISPASYNC_FABRIC_DEV case.
1999-07-02 23:12:58 +00:00
struct lportdb *lp = &fcp->portdb[tgt];
Core version 2.0 rewrite. In this file we replace isp_tdebug with isp_prt calls. We now use an argument to the ISPCTL_FCLINK_TEST call. We change all IDPRINTF macros to isp_prt calls. We add the isp_prt function here.
2000-08-01 06:31:44 +00:00
isp_prt(isp, ISP_LOGINFO, fmt, tgt, lp->loopid, lp->portid,
Do some cleanup based upon adapter role- mainly not enabling interrupts if we're ISP_ROLE_NONE. Change ISPASYNC_LOGGED_INOUT to ISPASYNC_PROMENADE. Make sure we note if something is a fabric device. Target mode: Finally fix (to a first approximation) SCSI Target Mode again- we needed to correctly check against CAM_TARGET_WILDCARD and CAM_LUN_WILDCARD so that targbh won't confuse us. Comment out the drainqueue stuff for now. Use isp_fc_runstate instead if isp_control/ISPCTL_FCLINK_TEST.
2001-02-11 03:47:39 +00:00
roles[lp->roles & 0x3],
(lp->valid)? "Arrived" : "Departed",
a) clean up some declaration stuff (i.e., make more modern with respect to getting rid u_int for uint and so on). b) Turn back on 64 bit DAC support. Cheeze it a bit in that we have two DMA callback functions- one when we have bus_addr_t > 4 bits in width and the other which should be normal. Even Cheezier in that we turn off setting up DMA maps to be BUS_SPACE_MAXADDR if we're in ISP_TARGET_MODE. More work on this in a week or so. c) Tested under amd64 and 1MB DFLTPHYS, sparc64, i386 (PAE, but insufficient memory to really test > 4GB). LINT check under amd64. MFC after: 1 month
2006-02-15 00:31:48 +00:00
(uint32_t) (lp->port_wwn >> 32),
(uint32_t) (lp->port_wwn & 0xffffffffLL),
(uint32_t) (lp->node_wwn >> 32),
(uint32_t) (lp->node_wwn & 0xffffffffLL));
Add CAM_NEW_TRAN_CODE support. Use correct CAMLOCK_2_ISPLOCK macros. For fibre channel, start going for the gusto and using AC_FOUND_DEVICE and AC_LOST_DEVICE calls to xpt_async when devices appear and disappear as the loop or fabric changes. ISPASYNC_FW_CRASH is the async event code where the platform layer deals with a firmware crash.
2001-07-04 18:54:29 +00:00
Things have become cinched down more tightly about assertions for Giant. This uncovered some missing spots where I trade off between isp's lock and Giant as I enter CAM.
2001-07-05 17:14:57 +00:00
ISPLOCK_2_CAMLOCK(isp);
Add CAM_NEW_TRAN_CODE support. Use correct CAMLOCK_2_ISPLOCK macros. For fibre channel, start going for the gusto and using AC_FOUND_DEVICE and AC_LOST_DEVICE calls to xpt_async when devices appear and disappear as the loop or fabric changes. ISPASYNC_FW_CRASH is the async event code where the platform layer deals with a firmware crash.
2001-07-04 18:54:29 +00:00
if (xpt_create_path(&tmppath, NULL, cam_sim_path(isp->isp_sim),
(target_id_t)tgt, CAM_LUN_WILDCARD) != CAM_REQ_CMP) {
Things have become cinched down more tightly about assertions for Giant. This uncovered some missing spots where I trade off between isp's lock and Giant as I enter CAM.
2001-07-05 17:14:57 +00:00
CAMLOCK_2_ISPLOCK(isp);
Add CAM_NEW_TRAN_CODE support. Use correct CAMLOCK_2_ISPLOCK macros. For fibre channel, start going for the gusto and using AC_FOUND_DEVICE and AC_LOST_DEVICE calls to xpt_async when devices appear and disappear as the loop or fabric changes. ISPASYNC_FW_CRASH is the async event code where the platform layer deals with a firmware crash.
2001-07-04 18:54:29 +00:00
break;
}
Remove ISP_SMPLOCK stuff- we're just using locking now. Correctly reintroduce loop_seen_once semantics- that is, if we've never seen good link, start bouncing commands with CAM_SEL_TIMEOUT. But we have to be careful to have let ourselves try (in isp_kthread) to check for loop up at least once. PR: 28992 MFC after: 1 week
2001-07-25 04:23:52 +00:00
/*
* Policy: only announce targets.
*/
if (lp->roles & is_tgt_mask) {
if (lp->valid) {
xpt_async(AC_FOUND_DEVICE, tmppath, NULL);
} else {
xpt_async(AC_LOST_DEVICE, tmppath, NULL);
}
Add CAM_NEW_TRAN_CODE support. Use correct CAMLOCK_2_ISPLOCK macros. For fibre channel, start going for the gusto and using AC_FOUND_DEVICE and AC_LOST_DEVICE calls to xpt_async when devices appear and disappear as the loop or fabric changes. ISPASYNC_FW_CRASH is the async event code where the platform layer deals with a firmware crash.
2001-07-04 18:54:29 +00:00
}
xpt_free_path(tmppath);
Remove ISP_SMPLOCK stuff- we're just using locking now. Correctly reintroduce loop_seen_once semantics- that is, if we've never seen good link, start bouncing commands with CAM_SEL_TIMEOUT. But we have to be careful to have let ourselves try (in isp_kthread) to check for loop up at least once. PR: 28992 MFC after: 1 week
2001-07-25 04:23:52 +00:00
CAMLOCK_2_ISPLOCK(isp);
A wad of changes- prepping for 1080/1240 support (which caused a massive thwank in register layout goop). A different mboxcmd approach. Some PDB change infrastructure. Some better management of loopdown/loopup events (keep them distinct from resource starvation for simq freeze/unfreeze actions).
1999-03-17 05:04:39 +00:00
break;
}
case ISPASYNC_CHANGE_NOTIFY:
Remove ISP_SMPLOCK stuff- we're just using locking now. Correctly reintroduce loop_seen_once semantics- that is, if we've never seen good link, start bouncing commands with CAM_SEL_TIMEOUT. But we have to be careful to have let ourselves try (in isp_kthread) to check for loop up at least once. PR: 28992 MFC after: 1 week
2001-07-25 04:23:52 +00:00
if (arg == ISPASYNC_CHANGE_PDB) {
Follow the ISPASYNC_PDB_CHANGED -> ISPASYNC_LOGGED_INOUT change. Also, ISPASYNC_NOTIFY_CHANGE now is for both local loop && fabric changes.
2001-01-09 02:47:15 +00:00
isp_prt(isp, ISP_LOGINFO,
Remove ISP_SMPLOCK stuff- we're just using locking now. Correctly reintroduce loop_seen_once semantics- that is, if we've never seen good link, start bouncing commands with CAM_SEL_TIMEOUT. But we have to be careful to have let ourselves try (in isp_kthread) to check for loop up at least once. PR: 28992 MFC after: 1 week
2001-07-25 04:23:52 +00:00
"Port Database Changed");
} else if (arg == ISPASYNC_CHANGE_SNS) {
Follow the ISPASYNC_PDB_CHANGED -> ISPASYNC_LOGGED_INOUT change. Also, ISPASYNC_NOTIFY_CHANGE now is for both local loop && fabric changes.
2001-01-09 02:47:15 +00:00
isp_prt(isp, ISP_LOGINFO,
"Name Server Database Changed");
}
Turn off usage of SMP style locking until we sort out CAM.
2002-09-03 04:31:55 +00:00
#ifdef ISP_SMPLOCK
Spring MegaChange #1. ---- Make a device for each ISP- really usable only with devfs and add an ioctl entry point (this can be used to (re)set debug levels, reset the HBA, rescan the fabric, issue lips, etc). ---- Add in a kernel thread for Fibre Channel cards. The purpose of this thread is to be woken up to clean up after Fibre Channel events block things. Basically, any FC event that casts doubt on the location or identify of FC devices blocks the queues. When, and if, we get the PORT DATABASE CHANGED or NAME SERVER DATABASE CHANGED async event, we activate the kthread which will then, in full thread context, re-evaluate the local loop and/or the fabric. When it's satisfied that things are stable, it can then release the blocked queues and let commands flow again. The prior mechanism was a lazy evaluation. That is, the next command to come down the pipe after change events would pay the full price for re-evaluation. And if this was done off of a softcall, it really could hang up the system. These changes brings the FreeBSD port more in line with the Solaris, Linux and NetBSD ports. It also, more importantly, gets us being more proactive about topology changes which could then be reflected upwards to CAM so that the periph driver can be informed sooner rather than later when things arrive or depart. --- Add in the (correct) usage of locking macros- we now have lock transition macros which allow us to transition from holding the CAM lock (Giant) and grabbing the softc lock and vice versa. Switch over to having this HBA do real locking. Some folks claim this won't be a win. They're right. But you have to start somewhere, and this will begin to teach us how to DTRT for HBAs, etc. -- Start putting in prototype 2300 support. Add back in LIP and Loop Reset as async events that each platform will handle. Add in another int_bogus instrumentation point. Do some more substantial target mode cleanups. MFC after: 8 weeks
2001-05-28 21:20:43 +00:00
cv_signal(&isp->isp_osinfo.kthread_cv);
Turn off usage of SMP style locking until we sort out CAM.
2002-09-03 04:31:55 +00:00
#else
wakeup(&isp->isp_osinfo.kthread_cv);
#endif
A wad of changes- prepping for 1080/1240 support (which caused a massive thwank in register layout goop). A different mboxcmd approach. Some PDB change infrastructure. Some better management of loopdown/loopup events (keep them distinct from resource starvation for simq freeze/unfreeze actions).
1999-03-17 05:04:39 +00:00
break;
Remove all pre-CAM code. Fix breakage for SCCLUN when it is in fact ISP2100_SCCLUN. Make changes for how ISPASYNC_PDB_CHANGE_COMPLETE is now ISPASYNC_PDB_CHANGED. Add in ISPASYNC_FABRIC_DEV case.
1999-07-02 23:12:58 +00:00
case ISPASYNC_FABRIC_DEV:
{
Fix bus dma segment count to be based off of MAXPHYS, not BUS_SPACE_MAXSIZE. Grumble. I've seen better documented architectures out of Redmond. 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. MFC after: 1 week
2002-04-04 23:46:01 +00:00
int target, base, lim;
fcparam *fcp = isp->isp_param;
Do more cleanup of the usage of 0..125 for F-port topologies.
2001-01-15 18:34:49 +00:00
struct lportdb *lp = NULL;
Fix bus dma segment count to be based off of MAXPHYS, not BUS_SPACE_MAXSIZE. Grumble. I've seen better documented architectures out of Redmond. 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. MFC after: 1 week
2002-04-04 23:46:01 +00:00
struct lportdb *clp = (struct lportdb *) arg;
Prettier print of fabric devices being attached- say what kind of port they are (e.g., F_Port vs. N_Port). Approved: jkh
2000-02-29 05:53:10 +00:00
char *pt;
Fix bus dma segment count to be based off of MAXPHYS, not BUS_SPACE_MAXSIZE. Grumble. I've seen better documented architectures out of Redmond. 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. MFC after: 1 week
2002-04-04 23:46:01 +00:00
switch (clp->port_type) {
Prettier print of fabric devices being attached- say what kind of port they are (e.g., F_Port vs. N_Port). Approved: jkh
2000-02-29 05:53:10 +00:00
case 1:
pt = " N_Port";
break;
case 2:
pt = " NL_Port";
break;
case 3:
pt = "F/NL_Port";
break;
case 0x7f:
pt = " Nx_Port";
break;
case 0x81:
pt = " F_port";
break;
case 0x82:
pt = " FL_Port";
break;
case 0x84:
pt = " E_port";
break;
default:
Fix bus dma segment count to be based off of MAXPHYS, not BUS_SPACE_MAXSIZE. Grumble. I've seen better documented architectures out of Redmond. 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. MFC after: 1 week
2002-04-04 23:46:01 +00:00
pt = " ";
Prettier print of fabric devices being attached- say what kind of port they are (e.g., F_Port vs. N_Port). Approved: jkh
2000-02-29 05:53:10 +00:00
break;
}
Fix bus dma segment count to be based off of MAXPHYS, not BUS_SPACE_MAXSIZE. Grumble. I've seen better documented architectures out of Redmond. 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. MFC after: 1 week
2002-04-04 23:46:01 +00:00
Core version 2.0 rewrite. In this file we replace isp_tdebug with isp_prt calls. We now use an argument to the ISPCTL_FCLINK_TEST call. We change all IDPRINTF macros to isp_prt calls. We add the isp_prt function here.
2000-08-01 06:31:44 +00:00
isp_prt(isp, ISP_LOGINFO,
Fix bus dma segment count to be based off of MAXPHYS, not BUS_SPACE_MAXSIZE. Grumble. I've seen better documented architectures out of Redmond. 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. MFC after: 1 week
2002-04-04 23:46:01 +00:00
"%s Fabric Device @ PortID 0x%x", pt, clp->portid);
/*
* If we don't have an initiator role we bail.
*
* We just use ISPASYNC_FABRIC_DEV for announcement purposes.
*/
if ((isp->isp_role & ISP_ROLE_INITIATOR) == 0) {
break;
}
Do more cleanup of the usage of 0..125 for F-port topologies.
2001-01-15 18:34:49 +00:00
/*
Fix bus dma segment count to be based off of MAXPHYS, not BUS_SPACE_MAXSIZE. Grumble. I've seen better documented architectures out of Redmond. 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. MFC after: 1 week
2002-04-04 23:46:01 +00:00
* Is this entry for us? If so, we bail.
Do more cleanup of the usage of 0..125 for F-port topologies.
2001-01-15 18:34:49 +00:00
*/
Fix bus dma segment count to be based off of MAXPHYS, not BUS_SPACE_MAXSIZE. Grumble. I've seen better documented architectures out of Redmond. 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. MFC after: 1 week
2002-04-04 23:46:01 +00:00
if (fcp->isp_portid == clp->portid) {
Do more cleanup of the usage of 0..125 for F-port topologies.
2001-01-15 18:34:49 +00:00
break;
}
Fix bus dma segment count to be based off of MAXPHYS, not BUS_SPACE_MAXSIZE. Grumble. I've seen better documented architectures out of Redmond. 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. MFC after: 1 week
2002-04-04 23:46:01 +00:00
/*
* Else, the default policy is to find room for it in
* our local port database. Later, when we execute
* the call to isp_pdb_sync either this newly arrived
* or already logged in device will be (re)announced.
*/
if (fcp->isp_topo == TOPO_FL_PORT)
base = FC_SNS_ID+1;
Do more cleanup of the usage of 0..125 for F-port topologies.
2001-01-15 18:34:49 +00:00
else
Fix bus dma segment count to be based off of MAXPHYS, not BUS_SPACE_MAXSIZE. Grumble. I've seen better documented architectures out of Redmond. 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. MFC after: 1 week
2002-04-04 23:46:01 +00:00
base = 0;
if (fcp->isp_topo == TOPO_N_PORT)
lim = 1;
else
lim = MAX_FC_TARG;
Do more cleanup of the usage of 0..125 for F-port topologies.
2001-01-15 18:34:49 +00:00
/*
* Is it already in our list?
*/
Fix bus dma segment count to be based off of MAXPHYS, not BUS_SPACE_MAXSIZE. Grumble. I've seen better documented architectures out of Redmond. 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. MFC after: 1 week
2002-04-04 23:46:01 +00:00
for (target = base; target < lim; target++) {
Do more cleanup of the usage of 0..125 for F-port topologies.
2001-01-15 18:34:49 +00:00
if (target >= FL_PORT_ID && target <= FC_SNS_ID) {
continue;
}
Remove all pre-CAM code. Fix breakage for SCCLUN when it is in fact ISP2100_SCCLUN. Make changes for how ISPASYNC_PDB_CHANGE_COMPLETE is now ISPASYNC_PDB_CHANGED. Add in ISPASYNC_FABRIC_DEV case.
1999-07-02 23:12:58 +00:00
lp = &fcp->portdb[target];
Fix bus dma segment count to be based off of MAXPHYS, not BUS_SPACE_MAXSIZE. Grumble. I've seen better documented architectures out of Redmond. 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. MFC after: 1 week
2002-04-04 23:46:01 +00:00
if (lp->port_wwn == clp->port_wwn &&
lp->node_wwn == clp->node_wwn) {
Do more cleanup of the usage of 0..125 for F-port topologies.
2001-01-15 18:34:49 +00:00
lp->fabric_dev = 1;
Remove all pre-CAM code. Fix breakage for SCCLUN when it is in fact ISP2100_SCCLUN. Make changes for how ISPASYNC_PDB_CHANGE_COMPLETE is now ISPASYNC_PDB_CHANGED. Add in ISPASYNC_FABRIC_DEV case.
1999-07-02 23:12:58 +00:00
break;
Do more cleanup of the usage of 0..125 for F-port topologies.
2001-01-15 18:34:49 +00:00
}
F/W revisions now a tuple (not a duple). Fix pre-CAM code.
1999-04-04 01:35:03 +00:00
}
Fix bus dma segment count to be based off of MAXPHYS, not BUS_SPACE_MAXSIZE. Grumble. I've seen better documented architectures out of Redmond. 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. MFC after: 1 week
2002-04-04 23:46:01 +00:00
if (target < lim) {
clarify headers;move uninit to outer layer;remove watchdog
1998-12-28 19:22:27 +00:00
break;
}
Fix bus dma segment count to be based off of MAXPHYS, not BUS_SPACE_MAXSIZE. Grumble. I've seen better documented architectures out of Redmond. 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. MFC after: 1 week
2002-04-04 23:46:01 +00:00
for (target = base; target < lim; target++) {
Do more cleanup of the usage of 0..125 for F-port topologies.
2001-01-15 18:34:49 +00:00
if (target >= FL_PORT_ID && target <= FC_SNS_ID) {
continue;
}
Remove all pre-CAM code. Fix breakage for SCCLUN when it is in fact ISP2100_SCCLUN. Make changes for how ISPASYNC_PDB_CHANGE_COMPLETE is now ISPASYNC_PDB_CHANGED. Add in ISPASYNC_FABRIC_DEV case.
1999-07-02 23:12:58 +00:00
lp = &fcp->portdb[target];
Do more cleanup of the usage of 0..125 for F-port topologies.
2001-01-15 18:34:49 +00:00
if (lp->port_wwn == 0) {
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 07:29:00 +00:00
break;
Do more cleanup of the usage of 0..125 for F-port topologies.
2001-01-15 18:34:49 +00:00
}
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 07:29:00 +00:00
}
Fix bus dma segment count to be based off of MAXPHYS, not BUS_SPACE_MAXSIZE. Grumble. I've seen better documented architectures out of Redmond. 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. MFC after: 1 week
2002-04-04 23:46:01 +00:00
if (target == lim) {
Replace some more printfs with isp_prt's. Use isp_prt/ISP_LOGDEBUG0 for rate setting/getting printouts.
2000-12-05 07:39:54 +00:00
isp_prt(isp, ISP_LOGWARN,
Fix bus dma segment count to be based off of MAXPHYS, not BUS_SPACE_MAXSIZE. Grumble. I've seen better documented architectures out of Redmond. 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. MFC after: 1 week
2002-04-04 23:46:01 +00:00
"out of space for fabric devices");
Remove all pre-CAM code. Fix breakage for SCCLUN when it is in fact ISP2100_SCCLUN. Make changes for how ISPASYNC_PDB_CHANGE_COMPLETE is now ISPASYNC_PDB_CHANGED. Add in ISPASYNC_FABRIC_DEV case.
1999-07-02 23:12:58 +00:00
break;
A wad of changes- prepping for 1080/1240 support (which caused a massive thwank in register layout goop). A different mboxcmd approach. Some PDB change infrastructure. Some better management of loopdown/loopup events (keep them distinct from resource starvation for simq freeze/unfreeze actions).
1999-03-17 05:04:39 +00:00
}
Fix bus dma segment count to be based off of MAXPHYS, not BUS_SPACE_MAXSIZE. Grumble. I've seen better documented architectures out of Redmond. 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. MFC after: 1 week
2002-04-04 23:46:01 +00:00
lp->port_type = clp->port_type;
lp->fc4_type = clp->fc4_type;
lp->node_wwn = clp->node_wwn;
lp->port_wwn = clp->port_wwn;
lp->portid = clp->portid;
Do more cleanup of the usage of 0..125 for F-port topologies.
2001-01-15 18:34:49 +00:00
lp->fabric_dev = 1;
A wad of changes- prepping for 1080/1240 support (which caused a massive thwank in register layout goop). A different mboxcmd approach. Some PDB change infrastructure. Some better management of loopdown/loopup events (keep them distinct from resource starvation for simq freeze/unfreeze actions).
1999-03-17 05:04:39 +00:00
break;
}
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
#ifdef ISP_TARGET_MODE
First of several commits as this driver is dusted off and maybe brought up to date. Principle changes for this reelase is to support 2K Port Login firmware. This allows us to support the 2322 (and 2422 4Gb) cards which only come with the 2K Port Login firmware. The 2322 should now work- but we don't have firmware sets for it in ispfw (as the change to load 2K Port Login f/w hasn't been made- that f/w is so big it has to be loaded in more than one chunk). Other changes are the beginnings of cleaning up some long standing target mode issues. The next changes here will incorporate a lot of bug fixes from others. Finally, some copyright cleanup and attempts to make the parts of the driver that are FreeBSD specific start conforming more to FreeBSD style. MFC after: 1 month
2006-01-23 06:23:37 +00:00
case ISPASYNC_TARGET_NOTIFY:
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
{
First of several commits as this driver is dusted off and maybe brought up to date. Principle changes for this reelase is to support 2K Port Login firmware. This allows us to support the 2322 (and 2422 4Gb) cards which only come with the 2K Port Login firmware. The 2322 should now work- but we don't have firmware sets for it in ispfw (as the change to load 2K Port Login f/w hasn't been made- that f/w is so big it has to be loaded in more than one chunk). Other changes are the beginnings of cleaning up some long standing target mode issues. The next changes here will incorporate a lot of bug fixes from others. Finally, some copyright cleanup and attempts to make the parts of the driver that are FreeBSD specific start conforming more to FreeBSD style. MFC after: 1 month
2006-01-23 06:23:37 +00:00
tmd_notify_t *nt = arg;
I don't know what I was thinking- if I have two separate busses on on SIM (as is true for the 1280 and the 12160), then I have to have separate flags && status for *both* busses. *Whap*. Implement condition variables for coordination with some target mode events. It's nice to use these and not panic in obscure little places in the kernel like 'propagate_priority' just because we went to sleep holding a mutex, or some other absurd thing. Remove some bogus ISP_UNLOCK calls. *Whap*. No longer require that somebody do a lun enable on the wildcard device to enable target mode. They are, in fact, orthogonal. A wildcard open is a statement that somebody upstream is willing to accept commands which are otherwise unrouteable. Now, for QLogic regular SCSI target mode, this won't matter for a damn because we'll never see ATIOs for luns we haven't enabled (are listening for, if you will). But for SCCLUN fibre channel SCSI, we get all kinds of ATIOs. We can either reflect them back here with minimal info (which is isp_target.c:isp_endcmd() is for), or the wildcard device (nominally targbh) can handle them. Do further checking against firmware attributes to see whether we can, in fact, support target mode in Fibre Channel. For now, require SCCLUN f/w to supoprt FC target mode. This is an awful lot of change, but target mode *still* isn't quite right. MFC after: 4 weeks
2001-09-04 21:53:12 +00:00
isp_prt(isp, ISP_LOGALL,
First of several commits as this driver is dusted off and maybe brought up to date. Principle changes for this reelase is to support 2K Port Login firmware. This allows us to support the 2322 (and 2422 4Gb) cards which only come with the 2K Port Login firmware. The 2322 should now work- but we don't have firmware sets for it in ispfw (as the change to load 2K Port Login f/w hasn't been made- that f/w is so big it has to be loaded in more than one chunk). Other changes are the beginnings of cleaning up some long standing target mode issues. The next changes here will incorporate a lot of bug fixes from others. Finally, some copyright cleanup and attempts to make the parts of the driver that are FreeBSD specific start conforming more to FreeBSD style. MFC after: 1 month
2006-01-23 06:23:37 +00:00
"target notify code 0x%x", nt->nt_ncode);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
break;
}
case ISPASYNC_TARGET_ACTION:
switch (((isphdr_t *)arg)->rqs_entry_type) {
default:
Replace some more printfs with isp_prt's. Use isp_prt/ISP_LOGDEBUG0 for rate setting/getting printouts.
2000-12-05 07:39:54 +00:00
isp_prt(isp, ISP_LOGWARN,
"event 0x%x for unhandled target action",
((isphdr_t *)arg)->rqs_entry_type);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
break;
Add support for ISP_FC_GETHINFO, which returns current connection topology, speed, loopid, WWPN/WWNN, etc. Beef up target mode. Add isp_handle_platform_notify_scsi and isp_handle_platform_notify_fc platform handlers to handle immediate notifies (isp_handle_platform_notify_scsi is still stubbed out). In implementation of isp_handle_platform_notify_fc, for IN_ABORT_TASK, peel off a pending XPT_IMMED_NOTIFY and call xpt_done on it and hope that somebody upstream is listening. Make sure on final CTIO2s that we set residual correctly. These are absolutely crucial. Make sure we set relative offset for each CTIO2 based upon bytes we've already xferred. This is what the private adjunct datat to the original ATIO is. Note state of command so we can figure out where to find it if we get an ABORT from the firmware. Make sure we *always* set CAM_TAG_ACTION_VALID for ATIO2s. Make sure we keep track of the original lun. If se sent status (or we're otherwise done with the command), don't forget to free the adjunct structure.
2002-06-16 05:08:02 +00:00
case RQSTYPE_NOTIFY:
if (IS_SCSI(isp)) {
rv = isp_handle_platform_notify_scsi(isp,
(in_entry_t *) arg);
} else {
rv = isp_handle_platform_notify_fc(isp,
(in_fcentry_t *) arg);
}
break;
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
case RQSTYPE_ATIO:
rv = isp_handle_platform_atio(isp, (at_entry_t *) arg);
break;
case RQSTYPE_ATIO2:
rv = isp_handle_platform_atio2(isp, (at2_entry_t *)arg);
break;
case RQSTYPE_CTIO2:
case RQSTYPE_CTIO:
rv = isp_handle_platform_ctio(isp, arg);
break;
case RQSTYPE_ENABLE_LUN:
case RQSTYPE_MODIFY_LUN:
Remove condition variables and status associated with target mode enabling. Instead, go to an interrupt/polled model. Fix get_lun_statep so we don't panic if there are no wildcard luns enabled. MFC after: 6 days
2004-02-08 19:17:56 +00:00
isp_ledone(isp, (lun_entry_t *) arg);
Make a static chain of isp softcs- gdb usage becomes a lot easier. Add in a very large amount of target mode support code- this is just a first pass at this. It's a difficult thing because some of the code can be in platform independent areas (see isp_target.?) but a lot has to be in platform dependent areas because of not only the tight coupling of received commands/events and the specific OS subsystem but because the platform independent code has (deliberately) no event/wait mechanisms.
2000-01-03 23:55:30 +00:00
break;
}
break;
Remove all pre-CAM code. Fix breakage for SCCLUN when it is in fact ISP2100_SCCLUN. Make changes for how ISPASYNC_PDB_CHANGE_COMPLETE is now ISPASYNC_PDB_CHANGED. Add in ISPASYNC_FABRIC_DEV case.
1999-07-02 23:12:58 +00:00
#endif
Add CAM_NEW_TRAN_CODE support. Use correct CAMLOCK_2_ISPLOCK macros. For fibre channel, start going for the gusto and using AC_FOUND_DEVICE and AC_LOST_DEVICE calls to xpt_async when devices appear and disappear as the loop or fabric changes. ISPASYNC_FW_CRASH is the async event code where the platform layer deals with a firmware crash.
2001-07-04 18:54:29 +00:00
case ISPASYNC_FW_CRASH:
{
a) clean up some declaration stuff (i.e., make more modern with respect to getting rid u_int for uint and so on). b) Turn back on 64 bit DAC support. Cheeze it a bit in that we have two DMA callback functions- one when we have bus_addr_t > 4 bits in width and the other which should be normal. Even Cheezier in that we turn off setting up DMA maps to be BUS_SPACE_MAXADDR if we're in ISP_TARGET_MODE. More work on this in a week or so. c) Tested under amd64 and 1MB DFLTPHYS, sparc64, i386 (PAE, but insufficient memory to really test > 4GB). LINT check under amd64. MFC after: 1 month
2006-02-15 00:31:48 +00:00
uint16_t mbox1, mbox6;
Add CAM_NEW_TRAN_CODE support. Use correct CAMLOCK_2_ISPLOCK macros. For fibre channel, start going for the gusto and using AC_FOUND_DEVICE and AC_LOST_DEVICE calls to xpt_async when devices appear and disappear as the loop or fabric changes. ISPASYNC_FW_CRASH is the async event code where the platform layer deals with a firmware crash.
2001-07-04 18:54:29 +00:00
mbox1 = ISP_READ(isp, OUTMAILBOX1);
if (IS_DUALBUS(isp)) {
mbox6 = ISP_READ(isp, OUTMAILBOX6);
} else {
mbox6 = 0;
}
isp_prt(isp, ISP_LOGERR,
Add support for ISP_FC_GETHINFO, which returns current connection topology, speed, loopid, WWPN/WWNN, etc. Beef up target mode. Add isp_handle_platform_notify_scsi and isp_handle_platform_notify_fc platform handlers to handle immediate notifies (isp_handle_platform_notify_scsi is still stubbed out). In implementation of isp_handle_platform_notify_fc, for IN_ABORT_TASK, peel off a pending XPT_IMMED_NOTIFY and call xpt_done on it and hope that somebody upstream is listening. Make sure on final CTIO2s that we set residual correctly. These are absolutely crucial. Make sure we set relative offset for each CTIO2 based upon bytes we've already xferred. This is what the private adjunct datat to the original ATIO is. Note state of command so we can figure out where to find it if we get an ABORT from the firmware. Make sure we *always* set CAM_TAG_ACTION_VALID for ATIO2s. Make sure we keep track of the original lun. If se sent status (or we're otherwise done with the command), don't forget to free the adjunct structure.
2002-06-16 05:08:02 +00:00
"Internal Firmware Error on bus %d @ RISC Address 0x%x",
Add CAM_NEW_TRAN_CODE support. Use correct CAMLOCK_2_ISPLOCK macros. For fibre channel, start going for the gusto and using AC_FOUND_DEVICE and AC_LOST_DEVICE calls to xpt_async when devices appear and disappear as the loop or fabric changes. ISPASYNC_FW_CRASH is the async event code where the platform layer deals with a firmware crash.
2001-07-04 18:54:29 +00:00
mbox6, mbox1);
We only do isp_reinit from isp_async if ISP_FW_CRASH_DUMP is defined- otherwise the core code does this.
2002-08-17 17:23:15 +00:00
#ifdef ISP_FW_CRASH_DUMP
/*
* XXX: really need a thread to do this right.
*/
if (IS_FC(isp)) {
FCPARAM(isp)->isp_fwstate = FW_CONFIG_WAIT;
FCPARAM(isp)->isp_loopstate = LOOP_NIL;
isp_freeze_loopdown(isp, "f/w crash");
isp_fw_dump(isp);
}
Add CAM_NEW_TRAN_CODE support. Use correct CAMLOCK_2_ISPLOCK macros. For fibre channel, start going for the gusto and using AC_FOUND_DEVICE and AC_LOST_DEVICE calls to xpt_async when devices appear and disappear as the loop or fabric changes. ISPASYNC_FW_CRASH is the async event code where the platform layer deals with a firmware crash.
2001-07-04 18:54:29 +00:00
isp_reinit(isp);
We only do isp_reinit from isp_async if ISP_FW_CRASH_DUMP is defined- otherwise the core code does this.
2002-08-17 17:23:15 +00:00
isp_async(isp, ISPASYNC_FW_RESTARTED, NULL);
#endif
Add CAM_NEW_TRAN_CODE support. Use correct CAMLOCK_2_ISPLOCK macros. For fibre channel, start going for the gusto and using AC_FOUND_DEVICE and AC_LOST_DEVICE calls to xpt_async when devices appear and disappear as the loop or fabric changes. ISPASYNC_FW_CRASH is the async event code where the platform layer deals with a firmware crash.
2001-07-04 18:54:29 +00:00
break;
}
Thanks to PHK for spotting: ISPASYNC_UNHANDLED_RESPONSE not handle in isp_async.
2001-08-16 17:25:41 +00:00
case ISPASYNC_UNHANDLED_RESPONSE:
break;
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 07:29:00 +00:00
default:
Core version 2.0 rewrite. In this file we replace isp_tdebug with isp_prt calls. We now use an argument to the ISPCTL_FCLINK_TEST call. We change all IDPRINTF macros to isp_prt calls. We add the isp_prt function here.
2000-08-01 06:31:44 +00:00
isp_prt(isp, ISP_LOGERR, "unknown isp_async event %d", cmd);
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 07:29:00 +00:00
break;
}
Remove all pre-CAM code. Fix breakage for SCCLUN when it is in fact ISP2100_SCCLUN. Make changes for how ISPASYNC_PDB_CHANGE_COMPLETE is now ISPASYNC_PDB_CHANGED. Add in ISPASYNC_FABRIC_DEV case.
1999-07-02 23:12:58 +00:00
return (rv);
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 07:29:00 +00:00
}
clarify headers;move uninit to outer layer;remove watchdog
1998-12-28 19:22:27 +00:00
Remove all pre-CAM code. Fix breakage for SCCLUN when it is in fact ISP2100_SCCLUN. Make changes for how ISPASYNC_PDB_CHANGE_COMPLETE is now ISPASYNC_PDB_CHANGED. Add in ISPASYNC_FABRIC_DEV case.
1999-07-02 23:12:58 +00:00
clarify headers;move uninit to outer layer;remove watchdog
1998-12-28 19:22:27 +00:00
/*
* Locks are held before coming here.
*/
void
Remove all pre-CAM code. Fix breakage for SCCLUN when it is in fact ISP2100_SCCLUN. Make changes for how ISPASYNC_PDB_CHANGE_COMPLETE is now ISPASYNC_PDB_CHANGED. Add in ISPASYNC_FABRIC_DEV case.
1999-07-02 23:12:58 +00:00
isp_uninit(struct ispsoftc *isp)
clarify headers;move uninit to outer layer;remove watchdog
1998-12-28 19:22:27 +00:00
{
A large set of changes to handle dual bus adapters.
1999-05-11 05:10:06 +00:00
ISP_WRITE(isp, HCCR, HCCR_CMD_RESET);
clarify headers;move uninit to outer layer;remove watchdog
1998-12-28 19:22:27 +00:00
DISABLE_INTS(isp);
}
Core version 2.0 rewrite. In this file we replace isp_tdebug with isp_prt calls. We now use an argument to the ISPCTL_FCLINK_TEST call. We change all IDPRINTF macros to isp_prt calls. We add the isp_prt function here.
2000-08-01 06:31:44 +00:00
void
isp_prt(struct ispsoftc *isp, int level, const char *fmt, ...)
{
va_list ap;
if (level != ISP_LOGALL && (level & isp->isp_dblev) == 0) {
return;
}
Finally eliminate as many of the printf calls as possible (still leaving ones where we have a CAM path) and replacing them with calls to isp_prt., Eliminate isp_unit references- we no longer have an isp_unit- we now have an isp_dev that device_get_unit can work with.
2001-03-01 02:14:54 +00:00
printf("%s: ", device_get_nameunit(isp->isp_dev));
Core version 2.0 rewrite. In this file we replace isp_tdebug with isp_prt calls. We now use an argument to the ISPCTL_FCLINK_TEST call. We change all IDPRINTF macros to isp_prt calls. We add the isp_prt function here.
2000-08-01 06:31:44 +00:00
va_start(ap, fmt);
vprintf(fmt, ap);
va_end(ap);
printf("\n");
}
Reference in New Issue Copy Permalink