freebsd-dev/sys/dev/advansys/advansys.c

1430 lines
40 KiB
C
Raw Normal View History

/*
* Generic driver for the Advanced Systems Inc. SCSI controllers
* Product specific probe and attach routines can be found in:
*
* i386/isa/adv_isa.c ABP5140, ABP542, ABP5150, ABP842, ABP852
* i386/eisa/adv_eisa.c ABP742, ABP752
* pci/adv_pci.c ABP920, ABP930, ABP930U, ABP930UA, ABP940, ABP940U,
* ABP940UA, ABP950, ABP960, ABP960U, ABP960UA,
* ABP970, ABP970U
*
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
* Copyright (c) 1996-2000 Justin Gibbs.
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions, and the following disclaimer,
* without modification, immediately at the beginning of the file.
* 2. The name of the author may not be used to endorse or promote products
* derived from this software without specific prior written permission.
*
* 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.
*
1999-08-28 01:08:13 +00:00
* $FreeBSD$
*/
/*
* Ported from:
* advansys.c - Linux Host Driver for AdvanSys SCSI Adapters
*
* Copyright (c) 1995-1997 Advanced System Products, Inc.
* All Rights Reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that redistributions of source
* code retain the above copyright notice and this comment without
* modification.
*/
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/malloc.h>
#include <sys/kernel.h>
#include <machine/bus_pio.h>
#include <machine/bus.h>
2000-04-07 11:32:42 +00:00
#include <machine/resource.h>
#include <sys/bus.h>
#include <sys/rman.h>
#include <cam/cam.h>
#include <cam/cam_ccb.h>
#include <cam/cam_sim.h>
#include <cam/cam_xpt_sim.h>
#include <cam/cam_xpt_periph.h>
#include <cam/cam_debug.h>
#include <cam/scsi/scsi_all.h>
#include <cam/scsi/scsi_message.h>
#include <vm/vm.h>
#include <vm/vm_param.h>
#include <vm/pmap.h>
#include <dev/advansys/advansys.h>
static void adv_action(struct cam_sim *sim, union ccb *ccb);
static void adv_execute_ccb(void *arg, bus_dma_segment_t *dm_segs,
int nsegments, int error);
static void adv_poll(struct cam_sim *sim);
static void adv_run_doneq(struct adv_softc *adv);
static struct adv_ccb_info *
adv_alloc_ccb_info(struct adv_softc *adv);
static void adv_destroy_ccb_info(struct adv_softc *adv,
struct adv_ccb_info *cinfo);
static __inline struct adv_ccb_info *
adv_get_ccb_info(struct adv_softc *adv);
static __inline void adv_free_ccb_info(struct adv_softc *adv,
struct adv_ccb_info *cinfo);
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
static __inline void adv_set_state(struct adv_softc *adv, adv_state state);
static __inline void adv_clear_state(struct adv_softc *adv, union ccb* ccb);
static void adv_clear_state_really(struct adv_softc *adv, union ccb* ccb);
static __inline struct adv_ccb_info *
adv_get_ccb_info(struct adv_softc *adv)
{
struct adv_ccb_info *cinfo;
int opri;
opri = splcam();
if ((cinfo = SLIST_FIRST(&adv->free_ccb_infos)) != NULL) {
SLIST_REMOVE_HEAD(&adv->free_ccb_infos, links);
} else {
cinfo = adv_alloc_ccb_info(adv);
}
splx(opri);
return (cinfo);
}
static __inline void
adv_free_ccb_info(struct adv_softc *adv, struct adv_ccb_info *cinfo)
{
int opri;
opri = splcam();
cinfo->state = ACCB_FREE;
SLIST_INSERT_HEAD(&adv->free_ccb_infos, cinfo, links);
splx(opri);
}
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
static __inline void
adv_set_state(struct adv_softc *adv, adv_state state)
{
if (adv->state == 0)
xpt_freeze_simq(adv->sim, /*count*/1);
adv->state |= state;
}
static __inline void
adv_clear_state(struct adv_softc *adv, union ccb* ccb)
{
if (adv->state != 0)
adv_clear_state_really(adv, ccb);
}
static void
adv_clear_state_really(struct adv_softc *adv, union ccb* ccb)
{
if ((adv->state & ADV_BUSDMA_BLOCK_CLEARED) != 0)
adv->state &= ~(ADV_BUSDMA_BLOCK_CLEARED|ADV_BUSDMA_BLOCK);
if ((adv->state & ADV_RESOURCE_SHORTAGE) != 0) {
int openings;
openings = adv->max_openings - adv->cur_active - ADV_MIN_FREE_Q;
if (openings >= adv->openings_needed) {
adv->state &= ~ADV_RESOURCE_SHORTAGE;
adv->openings_needed = 0;
}
}
if ((adv->state & ADV_IN_TIMEOUT) != 0) {
struct adv_ccb_info *cinfo;
cinfo = (struct adv_ccb_info *)ccb->ccb_h.ccb_cinfo_ptr;
if ((cinfo->state & ACCB_RECOVERY_CCB) != 0) {
struct ccb_hdr *ccb_h;
/*
* We now traverse our list of pending CCBs
* and reinstate their timeouts.
*/
ccb_h = LIST_FIRST(&adv->pending_ccbs);
while (ccb_h != NULL) {
ccb_h->timeout_ch =
timeout(adv_timeout, (caddr_t)ccb_h,
(ccb_h->timeout * hz) / 1000);
ccb_h = LIST_NEXT(ccb_h, sim_links.le);
}
adv->state &= ~ADV_IN_TIMEOUT;
printf("%s: No longer in timeout\n", adv_name(adv));
}
}
if (adv->state == 0)
ccb->ccb_h.status |= CAM_RELEASE_SIMQ;
}
void
adv_map(void *arg, bus_dma_segment_t *segs, int nseg, int error)
{
bus_addr_t* physaddr;
physaddr = (bus_addr_t*)arg;
*physaddr = segs->ds_addr;
}
char *
adv_name(struct adv_softc *adv)
{
static char name[10];
snprintf(name, sizeof(name), "adv%d", adv->unit);
return (name);
}
static void
adv_action(struct cam_sim *sim, union ccb *ccb)
{
struct adv_softc *adv;
CAM_DEBUG(ccb->ccb_h.path, CAM_DEBUG_TRACE, ("adv_action\n"));
adv = (struct adv_softc *)cam_sim_softc(sim);
switch (ccb->ccb_h.func_code) {
/* Common cases first */
case XPT_SCSI_IO: /* Execute the requested I/O operation */
{
struct ccb_hdr *ccb_h;
struct ccb_scsiio *csio;
struct adv_ccb_info *cinfo;
ccb_h = &ccb->ccb_h;
csio = &ccb->csio;
cinfo = adv_get_ccb_info(adv);
if (cinfo == NULL)
panic("XXX Handle CCB info error!!!");
ccb_h->ccb_cinfo_ptr = cinfo;
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
cinfo->ccb = ccb;
/* Only use S/G if there is a transfer */
if ((ccb_h->flags & CAM_DIR_MASK) != CAM_DIR_NONE) {
if ((ccb_h->flags & CAM_SCATTER_VALID) == 0) {
/*
* We've been given a pointer
* to a single buffer
*/
if ((ccb_h->flags & CAM_DATA_PHYS) == 0) {
int s;
int error;
s = splsoftvm();
error =
bus_dmamap_load(adv->buffer_dmat,
cinfo->dmamap,
csio->data_ptr,
csio->dxfer_len,
adv_execute_ccb,
csio, /*flags*/0);
if (error == EINPROGRESS) {
/*
* So as to maintain ordering,
* freeze the controller queue
* until our mapping is
* returned.
*/
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
adv_set_state(adv,
ADV_BUSDMA_BLOCK);
}
splx(s);
} else {
struct bus_dma_segment seg;
/* Pointer to physical buffer */
seg.ds_addr =
(bus_addr_t)csio->data_ptr;
seg.ds_len = csio->dxfer_len;
adv_execute_ccb(csio, &seg, 1, 0);
}
} else {
struct bus_dma_segment *segs;
if ((ccb_h->flags & CAM_DATA_PHYS) != 0)
panic("adv_setup_data - Physical "
"segment pointers unsupported");
if ((ccb_h->flags & CAM_SG_LIST_PHYS) == 0)
panic("adv_setup_data - Virtual "
"segment addresses unsupported");
/* Just use the segments provided */
segs = (struct bus_dma_segment *)csio->data_ptr;
adv_execute_ccb(ccb, segs, csio->sglist_cnt, 0);
}
} else {
adv_execute_ccb(ccb, NULL, 0, 0);
}
break;
}
case XPT_RESET_DEV: /* Bus Device Reset the specified SCSI device */
case XPT_TARGET_IO: /* Execute target I/O request */
case XPT_ACCEPT_TARGET_IO: /* Accept Host Target Mode CDB */
case XPT_CONT_TARGET_IO: /* Continue Host Target I/O Connection*/
case XPT_EN_LUN: /* Enable LUN as a target */
case XPT_ABORT: /* Abort the specified CCB */
/* XXX Implement */
ccb->ccb_h.status = CAM_REQ_INVALID;
xpt_done(ccb);
break;
case XPT_SET_TRAN_SETTINGS:
{
struct ccb_trans_settings *cts;
target_bit_vector targ_mask;
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
struct adv_transinfo *tconf;
u_int update_type;
int s;
cts = &ccb->cts;
targ_mask = ADV_TID_TO_TARGET_MASK(cts->ccb_h.target_id);
update_type = 0;
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
/*
* The user must specify which type of settings he wishes
* to change.
*/
if (((cts->flags & CCB_TRANS_CURRENT_SETTINGS) != 0)
&& ((cts->flags & CCB_TRANS_USER_SETTINGS) == 0)) {
tconf = &adv->tinfo[cts->ccb_h.target_id].current;
update_type |= ADV_TRANS_GOAL;
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
} else if (((cts->flags & CCB_TRANS_USER_SETTINGS) != 0)
&& ((cts->flags & CCB_TRANS_CURRENT_SETTINGS) == 0)) {
tconf = &adv->tinfo[cts->ccb_h.target_id].user;
update_type |= ADV_TRANS_USER;
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
} else {
ccb->ccb_h.status = CAM_REQ_INVALID;
break;
}
s = splcam();
if ((update_type & ADV_TRANS_GOAL) != 0) {
if ((cts->valid & CCB_TRANS_DISC_VALID) != 0) {
if ((cts->flags & CCB_TRANS_DISC_ENB) != 0)
adv->disc_enable |= targ_mask;
else
adv->disc_enable &= ~targ_mask;
adv_write_lram_8(adv, ADVV_DISC_ENABLE_B,
adv->disc_enable);
}
if ((cts->valid & CCB_TRANS_TQ_VALID) != 0) {
if ((cts->flags & CCB_TRANS_TAG_ENB) != 0)
adv->cmd_qng_enabled |= targ_mask;
else
adv->cmd_qng_enabled &= ~targ_mask;
}
}
if ((update_type & ADV_TRANS_USER) != 0) {
if ((cts->valid & CCB_TRANS_DISC_VALID) != 0) {
if ((cts->flags & CCB_TRANS_DISC_ENB) != 0)
adv->user_disc_enable |= targ_mask;
else
adv->user_disc_enable &= ~targ_mask;
}
if ((cts->valid & CCB_TRANS_TQ_VALID) != 0) {
if ((cts->flags & CCB_TRANS_TAG_ENB) != 0)
adv->user_cmd_qng_enabled |= targ_mask;
else
adv->user_cmd_qng_enabled &= ~targ_mask;
}
}
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
/*
* If the user specifies either the sync rate, or offset,
* but not both, the unspecified parameter defaults to its
* current value in transfer negotiations.
*/
if (((cts->valid & CCB_TRANS_SYNC_RATE_VALID) != 0)
|| ((cts->valid & CCB_TRANS_SYNC_OFFSET_VALID) != 0)) {
/*
* If the user provided a sync rate but no offset,
* use the current offset.
*/
if ((cts->valid & CCB_TRANS_SYNC_OFFSET_VALID) == 0)
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
cts->sync_offset = tconf->offset;
/*
* If the user provided an offset but no sync rate,
* use the current sync rate.
*/
if ((cts->valid & CCB_TRANS_SYNC_RATE_VALID) == 0)
cts->sync_period = tconf->period;
adv_period_offset_to_sdtr(adv, &cts->sync_period,
&cts->sync_offset,
cts->ccb_h.target_id);
adv_set_syncrate(adv, /*struct cam_path */NULL,
cts->ccb_h.target_id, cts->sync_period,
cts->sync_offset, update_type);
}
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
splx(s);
ccb->ccb_h.status = CAM_REQ_CMP;
xpt_done(ccb);
break;
}
case XPT_GET_TRAN_SETTINGS:
/* Get default/user set transfer settings for the target */
{
struct ccb_trans_settings *cts;
struct adv_transinfo *tconf;
target_bit_vector target_mask;
int s;
cts = &ccb->cts;
target_mask = ADV_TID_TO_TARGET_MASK(cts->ccb_h.target_id);
cts->flags &= ~(CCB_TRANS_DISC_ENB|CCB_TRANS_TAG_ENB);
s = splcam();
if ((cts->flags & CCB_TRANS_CURRENT_SETTINGS) != 0) {
tconf = &adv->tinfo[cts->ccb_h.target_id].current;
if ((adv->disc_enable & target_mask) != 0)
cts->flags |= CCB_TRANS_DISC_ENB;
if ((adv->cmd_qng_enabled & target_mask) != 0)
cts->flags |= CCB_TRANS_TAG_ENB;
} else {
tconf = &adv->tinfo[cts->ccb_h.target_id].user;
if ((adv->user_disc_enable & target_mask) != 0)
cts->flags |= CCB_TRANS_DISC_ENB;
if ((adv->user_cmd_qng_enabled & target_mask) != 0)
cts->flags |= CCB_TRANS_TAG_ENB;
}
cts->sync_period = tconf->period;
cts->sync_offset = tconf->offset;
splx(s);
cts->bus_width = MSG_EXT_WDTR_BUS_8_BIT;
cts->valid = CCB_TRANS_SYNC_RATE_VALID
| CCB_TRANS_SYNC_OFFSET_VALID
| CCB_TRANS_BUS_WIDTH_VALID
| CCB_TRANS_DISC_VALID
| CCB_TRANS_TQ_VALID;
ccb->ccb_h.status = CAM_REQ_CMP;
xpt_done(ccb);
break;
}
case XPT_CALC_GEOMETRY:
{
struct ccb_calc_geometry *ccg;
u_int32_t size_mb;
u_int32_t secs_per_cylinder;
int extended;
ccg = &ccb->ccg;
size_mb = ccg->volume_size
/ ((1024L * 1024L) / ccg->block_size);
extended = (adv->control & ADV_CNTL_BIOS_GT_1GB) != 0;
if (size_mb > 1024 && extended) {
ccg->heads = 255;
ccg->secs_per_track = 63;
} else {
ccg->heads = 64;
ccg->secs_per_track = 32;
}
secs_per_cylinder = ccg->heads * ccg->secs_per_track;
ccg->cylinders = ccg->volume_size / secs_per_cylinder;
ccb->ccb_h.status = CAM_REQ_CMP;
xpt_done(ccb);
break;
}
case XPT_RESET_BUS: /* Reset the specified SCSI bus */
{
int s;
s = splcam();
adv_stop_execution(adv);
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
adv_reset_bus(adv, /*initiate_reset*/TRUE);
adv_start_execution(adv);
splx(s);
ccb->ccb_h.status = CAM_REQ_CMP;
xpt_done(ccb);
break;
}
case XPT_TERM_IO: /* Terminate the I/O process */
/* XXX Implement */
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; /* XXX??? */
cpi->hba_inquiry = PI_SDTR_ABLE|PI_TAG_ABLE;
cpi->target_sprt = 0;
cpi->hba_misc = 0;
cpi->hba_eng_cnt = 0;
cpi->max_target = 7;
cpi->max_lun = 7;
cpi->initiator_id = adv->scsi_id;
cpi->bus_id = cam_sim_bus(sim);
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;
strncpy(cpi->sim_vid, "FreeBSD", SIM_IDLEN);
strncpy(cpi->hba_vid, "Advansys", 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;
}
}
/*
* Currently, the output of bus_dmammap_load suits our needs just
* fine, but should it change, we'd need to do something here.
*/
#define adv_fixup_dmasegs(adv, dm_segs) (struct adv_sg_entry *)(dm_segs)
static void
adv_execute_ccb(void *arg, bus_dma_segment_t *dm_segs,
int nsegments, int error)
{
struct ccb_scsiio *csio;
struct ccb_hdr *ccb_h;
struct cam_sim *sim;
struct adv_softc *adv;
struct adv_ccb_info *cinfo;
struct adv_scsi_q scsiq;
struct adv_sg_head sghead;
int s;
csio = (struct ccb_scsiio *)arg;
ccb_h = &csio->ccb_h;
sim = xpt_path_sim(ccb_h->path);
adv = (struct adv_softc *)cam_sim_softc(sim);
cinfo = (struct adv_ccb_info *)csio->ccb_h.ccb_cinfo_ptr;
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
/*
* Setup our done routine to release the simq on
* the next ccb that completes.
*/
if ((adv->state & ADV_BUSDMA_BLOCK) != 0)
adv->state |= ADV_BUSDMA_BLOCK_CLEARED;
if ((ccb_h->flags & CAM_CDB_POINTER) != 0) {
if ((ccb_h->flags & CAM_CDB_PHYS) == 0) {
/* XXX Need phystovirt!!!! */
/* How about pmap_kenter??? */
scsiq.cdbptr = csio->cdb_io.cdb_ptr;
} else {
scsiq.cdbptr = csio->cdb_io.cdb_ptr;
}
} else {
scsiq.cdbptr = csio->cdb_io.cdb_bytes;
}
/*
* Build up the request
*/
scsiq.q1.status = 0;
scsiq.q1.q_no = 0;
scsiq.q1.cntl = 0;
scsiq.q1.sg_queue_cnt = 0;
scsiq.q1.target_id = ADV_TID_TO_TARGET_MASK(ccb_h->target_id);
scsiq.q1.target_lun = ccb_h->target_lun;
scsiq.q1.sense_len = csio->sense_len;
scsiq.q1.extra_bytes = 0;
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
scsiq.q2.ccb_index = cinfo - adv->ccb_infos;
scsiq.q2.target_ix = ADV_TIDLUN_TO_IX(ccb_h->target_id,
ccb_h->target_lun);
scsiq.q2.flag = 0;
scsiq.q2.cdb_len = csio->cdb_len;
if ((ccb_h->flags & CAM_TAG_ACTION_VALID) != 0)
scsiq.q2.tag_code = csio->tag_action;
else
scsiq.q2.tag_code = 0;
scsiq.q2.vm_id = 0;
if (nsegments != 0) {
bus_dmasync_op_t op;
scsiq.q1.data_addr = dm_segs->ds_addr;
scsiq.q1.data_cnt = dm_segs->ds_len;
if (nsegments > 1) {
scsiq.q1.cntl |= QC_SG_HEAD;
sghead.entry_cnt
= sghead.entry_to_copy
= nsegments;
sghead.res = 0;
sghead.sg_list = adv_fixup_dmasegs(adv, dm_segs);
scsiq.sg_head = &sghead;
} else {
scsiq.sg_head = NULL;
}
if ((ccb_h->flags & CAM_DIR_MASK) == CAM_DIR_IN)
op = BUS_DMASYNC_PREREAD;
else
op = BUS_DMASYNC_PREWRITE;
bus_dmamap_sync(adv->buffer_dmat, cinfo->dmamap, op);
} else {
scsiq.q1.data_addr = 0;
scsiq.q1.data_cnt = 0;
scsiq.sg_head = NULL;
}
s = splcam();
/*
* Last time we need to check if this SCB needs to
* be aborted.
*/
if (ccb_h->status != CAM_REQ_INPROG) {
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
if (nsegments != 0)
bus_dmamap_unload(adv->buffer_dmat, cinfo->dmamap);
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
adv_clear_state(adv, (union ccb *)csio);
adv_free_ccb_info(adv, cinfo);
xpt_done((union ccb *)csio);
splx(s);
return;
}
if (adv_execute_scsi_queue(adv, &scsiq, csio->dxfer_len) != 0) {
/* Temporary resource shortage */
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
adv_set_state(adv, ADV_RESOURCE_SHORTAGE);
if (nsegments != 0)
bus_dmamap_unload(adv->buffer_dmat, cinfo->dmamap);
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
csio->ccb_h.status = CAM_REQUEUE_REQ;
adv_clear_state(adv, (union ccb *)csio);
adv_free_ccb_info(adv, cinfo);
xpt_done((union ccb *)csio);
splx(s);
return;
}
cinfo->state |= ACCB_ACTIVE;
ccb_h->status |= CAM_SIM_QUEUED;
LIST_INSERT_HEAD(&adv->pending_ccbs, ccb_h, sim_links.le);
/* Schedule our timeout */
ccb_h->timeout_ch =
timeout(adv_timeout, csio, (ccb_h->timeout * hz)/1000);
splx(s);
}
static struct adv_ccb_info *
adv_alloc_ccb_info(struct adv_softc *adv)
{
int error;
struct adv_ccb_info *cinfo;
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
cinfo = &adv->ccb_infos[adv->ccb_infos_allocated];
cinfo->state = ACCB_FREE;
error = bus_dmamap_create(adv->buffer_dmat, /*flags*/0,
&cinfo->dmamap);
if (error != 0) {
printf("%s: Unable to allocate CCB info "
"dmamap - error %d\n", adv_name(adv), error);
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
return (NULL);
}
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
adv->ccb_infos_allocated++;
return (cinfo);
}
static void
adv_destroy_ccb_info(struct adv_softc *adv, struct adv_ccb_info *cinfo)
{
bus_dmamap_destroy(adv->buffer_dmat, cinfo->dmamap);
}
void
adv_timeout(void *arg)
{
int s;
union ccb *ccb;
struct adv_softc *adv;
struct adv_ccb_info *cinfo;
ccb = (union ccb *)arg;
adv = (struct adv_softc *)xpt_path_sim(ccb->ccb_h.path)->softc;
cinfo = (struct adv_ccb_info *)ccb->ccb_h.ccb_cinfo_ptr;
xpt_print_path(ccb->ccb_h.path);
printf("Timed out\n");
s = splcam();
/* Have we been taken care of already?? */
if (cinfo == NULL || cinfo->state == ACCB_FREE) {
splx(s);
return;
}
adv_stop_execution(adv);
if ((cinfo->state & ACCB_ABORT_QUEUED) == 0) {
struct ccb_hdr *ccb_h;
/*
* In order to simplify the recovery process, we ask the XPT
* layer to halt the queue of new transactions and we traverse
* the list of pending CCBs and remove their timeouts. This
* means that the driver attempts to clear only one error
* condition at a time. In general, timeouts that occur
* close together are related anyway, so there is no benefit
* in attempting to handle errors in parrallel. Timeouts will
* be reinstated when the recovery process ends.
*/
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
adv_set_state(adv, ADV_IN_TIMEOUT);
/* This CCB is the CCB representing our recovery actions */
cinfo->state |= ACCB_RECOVERY_CCB|ACCB_ABORT_QUEUED;
ccb_h = LIST_FIRST(&adv->pending_ccbs);
while (ccb_h != NULL) {
untimeout(adv_timeout, ccb_h, ccb_h->timeout_ch);
ccb_h = LIST_NEXT(ccb_h, sim_links.le);
}
/* XXX Should send a BDR */
/* Attempt an abort as our first tact */
xpt_print_path(ccb->ccb_h.path);
printf("Attempting abort\n");
adv_abort_ccb(adv, ccb->ccb_h.target_id,
ccb->ccb_h.target_lun, ccb,
CAM_CMD_TIMEOUT, /*queued_only*/FALSE);
ccb->ccb_h.timeout_ch =
timeout(adv_timeout, ccb, 2 * hz);
} else {
/* Our attempt to perform an abort failed, go for a reset */
xpt_print_path(ccb->ccb_h.path);
printf("Resetting bus\n");
ccb->ccb_h.status &= ~CAM_STATUS_MASK;
ccb->ccb_h.status |= CAM_CMD_TIMEOUT;
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
adv_reset_bus(adv, /*initiate_reset*/TRUE);
}
adv_start_execution(adv);
splx(s);
}
struct adv_softc *
2000-04-07 11:32:42 +00:00
adv_alloc(device_t dev, bus_space_tag_t tag, bus_space_handle_t bsh)
{
2000-04-07 11:32:42 +00:00
struct adv_softc *adv = device_get_softc(dev);
/*
* Allocate a storage area for us
*/
LIST_INIT(&adv->pending_ccbs);
SLIST_INIT(&adv->free_ccb_infos);
2000-04-07 11:32:42 +00:00
adv->dev = dev;
adv->unit = device_get_unit(dev);
adv->tag = tag;
adv->bsh = bsh;
return(adv);
}
void
adv_free(struct adv_softc *adv)
{
switch (adv->init_level) {
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
case 6:
{
struct adv_ccb_info *cinfo;
while ((cinfo = SLIST_FIRST(&adv->free_ccb_infos)) != NULL) {
SLIST_REMOVE_HEAD(&adv->free_ccb_infos, links);
adv_destroy_ccb_info(adv, cinfo);
}
bus_dmamap_unload(adv->sense_dmat, adv->sense_dmamap);
}
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
case 5:
bus_dmamem_free(adv->sense_dmat, adv->sense_buffers,
adv->sense_dmamap);
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
case 4:
bus_dma_tag_destroy(adv->sense_dmat);
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
case 3:
bus_dma_tag_destroy(adv->buffer_dmat);
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
case 2:
bus_dma_tag_destroy(adv->parent_dmat);
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
case 1:
free(adv->ccb_infos, M_DEVBUF);
case 0:
break;
}
}
int
adv_init(struct adv_softc *adv)
{
struct adv_eeprom_config eeprom_config;
int checksum, i;
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
int max_sync;
u_int16_t config_lsw;
u_int16_t config_msw;
adv_lib_init(adv);
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
/*
* Stop script execution.
*/
adv_write_lram_16(adv, ADV_HALTCODE_W, 0x00FE);
adv_stop_execution(adv);
if (adv_stop_chip(adv) == 0 || adv_is_chip_halted(adv) == 0) {
printf("adv%d: Unable to halt adapter. Initialization"
"failed\n", adv->unit);
return (1);
}
ADV_OUTW(adv, ADV_REG_PROG_COUNTER, ADV_MCODE_START_ADDR);
if (ADV_INW(adv, ADV_REG_PROG_COUNTER) != ADV_MCODE_START_ADDR) {
printf("adv%d: Unable to set program counter. Initialization"
"failed\n", adv->unit);
return (1);
}
config_msw = ADV_INW(adv, ADV_CONFIG_MSW);
config_lsw = ADV_INW(adv, ADV_CONFIG_LSW);
if ((config_msw & ADV_CFG_MSW_CLR_MASK) != 0) {
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
config_msw &= ~ADV_CFG_MSW_CLR_MASK;
/*
* XXX The Linux code flags this as an error,
* but what should we report to the user???
* It seems that clearing the config register
* makes this error recoverable.
*/
ADV_OUTW(adv, ADV_CONFIG_MSW, config_msw);
}
/* Suck in the configuration from the EEProm */
checksum = adv_get_eeprom_config(adv, &eeprom_config);
if (ADV_INW(adv, ADV_CHIP_STATUS) & ADV_CSW_AUTO_CONFIG) {
/*
* XXX The Linux code sets a warning level for this
* condition, yet nothing of meaning is printed to
* the user. What does this mean???
*/
if (adv->chip_version == 3) {
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
if (eeprom_config.cfg_lsw != config_lsw)
eeprom_config.cfg_lsw = config_lsw;
if (eeprom_config.cfg_msw != config_msw) {
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
eeprom_config.cfg_msw = config_msw;
}
}
}
if (checksum == eeprom_config.chksum) {
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
/* Range/Sanity checking */
if (eeprom_config.max_total_qng < ADV_MIN_TOTAL_QNG) {
eeprom_config.max_total_qng = ADV_MIN_TOTAL_QNG;
}
if (eeprom_config.max_total_qng > ADV_MAX_TOTAL_QNG) {
eeprom_config.max_total_qng = ADV_MAX_TOTAL_QNG;
}
if (eeprom_config.max_tag_qng > eeprom_config.max_total_qng) {
eeprom_config.max_tag_qng = eeprom_config.max_total_qng;
}
if (eeprom_config.max_tag_qng < ADV_MIN_TAG_Q_PER_DVC) {
eeprom_config.max_tag_qng = ADV_MIN_TAG_Q_PER_DVC;
}
adv->max_openings = eeprom_config.max_total_qng;
adv->user_disc_enable = eeprom_config.disc_enable;
adv->user_cmd_qng_enabled = eeprom_config.use_cmd_qng;
adv->isa_dma_speed = EEPROM_DMA_SPEED(eeprom_config);
adv->scsi_id = EEPROM_SCSIID(eeprom_config) & ADV_MAX_TID;
EEPROM_SET_SCSIID(eeprom_config, adv->scsi_id);
adv->control = eeprom_config.cntl;
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
for (i = 0; i <= ADV_MAX_TID; i++) {
u_int8_t sync_data;
if ((eeprom_config.init_sdtr & (0x1 << i)) == 0)
sync_data = 0;
else
sync_data = eeprom_config.sdtr_data[i];
adv_sdtr_to_period_offset(adv,
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
sync_data,
&adv->tinfo[i].user.period,
&adv->tinfo[i].user.offset,
i);
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
}
config_lsw = eeprom_config.cfg_lsw;
eeprom_config.cfg_msw = config_msw;
} else {
u_int8_t sync_data;
printf("adv%d: Warning EEPROM Checksum mismatch. "
"Using default device parameters\n", adv->unit);
/* Set reasonable defaults since we can't read the EEPROM */
adv->isa_dma_speed = /*ADV_DEF_ISA_DMA_SPEED*/1;
adv->max_openings = ADV_DEF_MAX_TOTAL_QNG;
adv->disc_enable = TARGET_BIT_VECTOR_SET;
adv->user_disc_enable = TARGET_BIT_VECTOR_SET;
adv->cmd_qng_enabled = TARGET_BIT_VECTOR_SET;
adv->user_cmd_qng_enabled = TARGET_BIT_VECTOR_SET;
adv->scsi_id = 7;
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
adv->control = 0xFFFF;
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
if (adv->chip_version == ADV_CHIP_VER_PCI_ULTRA_3050)
/* Default to no Ultra to support the 3030 */
adv->control &= ~ADV_CNTL_SDTR_ENABLE_ULTRA;
sync_data = ADV_DEF_SDTR_OFFSET | (ADV_DEF_SDTR_INDEX << 4);
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
for (i = 0; i <= ADV_MAX_TID; i++) {
adv_sdtr_to_period_offset(adv, sync_data,
&adv->tinfo[i].user.period,
&adv->tinfo[i].user.offset,
i);
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
}
config_lsw |= ADV_CFG_LSW_SCSI_PARITY_ON;
}
config_msw &= ~ADV_CFG_MSW_CLR_MASK;
config_lsw |= ADV_CFG_LSW_HOST_INT_ON;
if ((adv->type & (ADV_PCI|ADV_ULTRA)) == (ADV_PCI|ADV_ULTRA)
&& (adv->control & ADV_CNTL_SDTR_ENABLE_ULTRA) == 0)
/* 25ns or 10MHz */
max_sync = 25;
else
/* Unlimited */
max_sync = 0;
for (i = 0; i <= ADV_MAX_TID; i++) {
if (adv->tinfo[i].user.period < max_sync)
adv->tinfo[i].user.period = max_sync;
}
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
if (adv_test_external_lram(adv) == 0) {
if ((adv->type & (ADV_PCI|ADV_ULTRA)) == (ADV_PCI|ADV_ULTRA)) {
eeprom_config.max_total_qng =
ADV_MAX_PCI_ULTRA_INRAM_TOTAL_QNG;
eeprom_config.max_tag_qng =
ADV_MAX_PCI_ULTRA_INRAM_TAG_QNG;
} else {
eeprom_config.cfg_msw |= 0x0800;
config_msw |= 0x0800;
eeprom_config.max_total_qng =
ADV_MAX_PCI_INRAM_TOTAL_QNG;
eeprom_config.max_tag_qng = ADV_MAX_INRAM_TAG_QNG;
}
adv->max_openings = eeprom_config.max_total_qng;
}
ADV_OUTW(adv, ADV_CONFIG_MSW, config_msw);
ADV_OUTW(adv, ADV_CONFIG_LSW, config_lsw);
#if 0
/*
* Don't write the eeprom data back for now.
* I'd rather not mess up the user's card. We also don't
* fully sanitize the eeprom settings above for the write-back
* to be 100% correct.
*/
if (adv_set_eeprom_config(adv, &eeprom_config) != 0)
printf("%s: WARNING! Failure writing to EEPROM.\n",
adv_name(adv));
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
#endif
adv_set_chip_scsiid(adv, adv->scsi_id);
if (adv_init_lram_and_mcode(adv))
return (1);
adv->disc_enable = adv->user_disc_enable;
adv_write_lram_8(adv, ADVV_DISC_ENABLE_B, adv->disc_enable);
for (i = 0; i <= ADV_MAX_TID; i++) {
/*
* Start off in async mode.
*/
adv_set_syncrate(adv, /*struct cam_path */NULL,
i, /*period*/0, /*offset*/0,
ADV_TRANS_CUR);
/*
* Enable the use of tagged commands on all targets.
* This allows the kernel driver to make up it's own mind
* as it sees fit to tag queue instead of having the
* firmware try and second guess the tag_code settins.
*/
adv_write_lram_8(adv, ADVV_MAX_DVC_QNG_BEG + i,
adv->max_openings);
}
adv_write_lram_8(adv, ADVV_USE_TAGGED_QNG_B, TARGET_BIT_VECTOR_SET);
adv_write_lram_8(adv, ADVV_CAN_TAGGED_QNG_B, TARGET_BIT_VECTOR_SET);
printf("adv%d: AdvanSys %s Host Adapter, SCSI ID %d, queue depth %d\n",
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
adv->unit, (adv->type & ADV_ULTRA) && (max_sync == 0)
? "Ultra SCSI" : "SCSI",
adv->scsi_id, adv->max_openings);
return (0);
}
void
adv_intr(void *arg)
{
struct adv_softc *adv;
u_int16_t chipstat;
u_int16_t saved_ram_addr;
u_int8_t ctrl_reg;
u_int8_t saved_ctrl_reg;
u_int8_t host_flag;
adv = (struct adv_softc *)arg;
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
chipstat = ADV_INW(adv, ADV_CHIP_STATUS);
/* Is it for us? */
if ((chipstat & (ADV_CSW_INT_PENDING|ADV_CSW_SCSI_RESET_LATCH)) == 0)
return;
ctrl_reg = ADV_INB(adv, ADV_CHIP_CTRL);
saved_ctrl_reg = ctrl_reg & (~(ADV_CC_SCSI_RESET | ADV_CC_CHIP_RESET |
ADV_CC_SINGLE_STEP | ADV_CC_DIAG |
ADV_CC_TEST));
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
if ((chipstat & (ADV_CSW_SCSI_RESET_LATCH|ADV_CSW_SCSI_RESET_ACTIVE))) {
printf("Detected Bus Reset\n");
adv_reset_bus(adv, /*initiate_reset*/FALSE);
return;
}
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
if ((chipstat & ADV_CSW_INT_PENDING) != 0) {
saved_ram_addr = ADV_INW(adv, ADV_LRAM_ADDR);
host_flag = adv_read_lram_8(adv, ADVV_HOST_FLAG_B);
adv_write_lram_8(adv, ADVV_HOST_FLAG_B,
host_flag | ADV_HOST_FLAG_IN_ISR);
adv_ack_interrupt(adv);
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
if ((chipstat & ADV_CSW_HALTED) != 0
&& (ctrl_reg & ADV_CC_SINGLE_STEP) != 0) {
adv_isr_chip_halted(adv);
saved_ctrl_reg &= ~ADV_CC_HALT;
} else {
adv_run_doneq(adv);
}
ADV_OUTW(adv, ADV_LRAM_ADDR, saved_ram_addr);
#ifdef DIAGNOSTIC
if (ADV_INW(adv, ADV_LRAM_ADDR) != saved_ram_addr)
panic("adv_intr: Unable to set LRAM addr");
#endif
adv_write_lram_8(adv, ADVV_HOST_FLAG_B, host_flag);
}
ADV_OUTB(adv, ADV_CHIP_CTRL, saved_ctrl_reg);
}
void
adv_run_doneq(struct adv_softc *adv)
{
struct adv_q_done_info scsiq;
u_int doneq_head;
u_int done_qno;
doneq_head = adv_read_lram_16(adv, ADVV_DONE_Q_TAIL_W) & 0xFF;
done_qno = adv_read_lram_8(adv, ADV_QNO_TO_QADDR(doneq_head)
+ ADV_SCSIQ_B_FWD);
while (done_qno != ADV_QLINK_END) {
union ccb* ccb;
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
struct adv_ccb_info *cinfo;
u_int done_qaddr;
u_int sg_queue_cnt;
int aborted;
done_qaddr = ADV_QNO_TO_QADDR(done_qno);
/* Pull status from this request */
sg_queue_cnt = adv_copy_lram_doneq(adv, done_qaddr, &scsiq,
adv->max_dma_count);
/* Mark it as free */
adv_write_lram_8(adv, done_qaddr + ADV_SCSIQ_B_STATUS,
scsiq.q_status & ~(QS_READY|QS_ABORTED));
/* Process request based on retrieved info */
if ((scsiq.cntl & QC_SG_HEAD) != 0) {
u_int i;
/*
* S/G based request. Free all of the queue
* structures that contained S/G information.
*/
for (i = 0; i < sg_queue_cnt; i++) {
done_qno = adv_read_lram_8(adv, done_qaddr
+ ADV_SCSIQ_B_FWD);
#ifdef DIAGNOSTIC
if (done_qno == ADV_QLINK_END) {
panic("adv_qdone: Corrupted SG "
"list encountered");
}
#endif
done_qaddr = ADV_QNO_TO_QADDR(done_qno);
/* Mark SG queue as free */
adv_write_lram_8(adv, done_qaddr
+ ADV_SCSIQ_B_STATUS, QS_FREE);
}
} else
sg_queue_cnt = 0;
#ifdef DIAGNOSTIC
if (adv->cur_active < (sg_queue_cnt + 1))
panic("adv_qdone: Attempting to free more "
"queues than are active");
#endif
adv->cur_active -= sg_queue_cnt + 1;
aborted = (scsiq.q_status & QS_ABORTED) != 0;
if ((scsiq.q_status != QS_DONE)
&& (scsiq.q_status & QS_ABORTED) == 0)
panic("adv_qdone: completed scsiq with unknown status");
scsiq.remain_bytes += scsiq.extra_bytes;
if ((scsiq.d3.done_stat == QD_WITH_ERROR) &&
(scsiq.d3.host_stat == QHSTA_M_DATA_OVER_RUN)) {
if ((scsiq.cntl & (QC_DATA_IN|QC_DATA_OUT)) == 0) {
scsiq.d3.done_stat = QD_NO_ERROR;
scsiq.d3.host_stat = QHSTA_NO_ERROR;
}
}
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
cinfo = &adv->ccb_infos[scsiq.d2.ccb_index];
ccb = cinfo->ccb;
ccb->csio.resid = scsiq.remain_bytes;
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
adv_done(adv, ccb,
scsiq.d3.done_stat, scsiq.d3.host_stat,
scsiq.d3.scsi_stat, scsiq.q_no);
doneq_head = done_qno;
done_qno = adv_read_lram_8(adv, done_qaddr + ADV_SCSIQ_B_FWD);
}
adv_write_lram_16(adv, ADVV_DONE_Q_TAIL_W, doneq_head);
}
void
adv_done(struct adv_softc *adv, union ccb *ccb, u_int done_stat,
u_int host_stat, u_int scsi_status, u_int q_no)
{
struct adv_ccb_info *cinfo;
cinfo = (struct adv_ccb_info *)ccb->ccb_h.ccb_cinfo_ptr;
LIST_REMOVE(&ccb->ccb_h, sim_links.le);
untimeout(adv_timeout, ccb, ccb->ccb_h.timeout_ch);
if ((ccb->ccb_h.flags & CAM_DIR_MASK) != CAM_DIR_NONE) {
bus_dmasync_op_t op;
if ((ccb->ccb_h.flags & CAM_DIR_MASK) == CAM_DIR_IN)
op = BUS_DMASYNC_POSTREAD;
else
op = BUS_DMASYNC_POSTWRITE;
bus_dmamap_sync(adv->buffer_dmat, cinfo->dmamap, op);
bus_dmamap_unload(adv->buffer_dmat, cinfo->dmamap);
}
switch (done_stat) {
case QD_NO_ERROR:
if (host_stat == QHSTA_NO_ERROR) {
ccb->ccb_h.status = CAM_REQ_CMP;
break;
}
xpt_print_path(ccb->ccb_h.path);
printf("adv_done - queue done without error, "
"but host status non-zero(%x)\n", host_stat);
/*FALLTHROUGH*/
case QD_WITH_ERROR:
switch (host_stat) {
case QHSTA_M_TARGET_STATUS_BUSY:
case QHSTA_M_BAD_QUEUE_FULL_OR_BUSY:
/*
* Assume that if we were a tagged transaction
* the target reported queue full. Otherwise,
* report busy. The firmware really should just
* pass the original status back up to us even
* if it thinks the target was in error for
* returning this status as no other transactions
* from this initiator are in effect, but this
* ignores multi-initiator setups and there is
* evidence that the firmware gets its per-device
* transaction counts screwed up occassionally.
*/
ccb->ccb_h.status |= CAM_SCSI_STATUS_ERROR;
if ((ccb->ccb_h.flags & CAM_TAG_ACTION_VALID) != 0
&& host_stat != QHSTA_M_TARGET_STATUS_BUSY)
scsi_status = SCSI_STATUS_QUEUE_FULL;
else
scsi_status = SCSI_STATUS_BUSY;
adv_abort_ccb(adv, ccb->ccb_h.target_id,
ccb->ccb_h.target_lun,
/*ccb*/NULL, CAM_REQUEUE_REQ,
/*queued_only*/TRUE);
/*FALLTHROUGH*/
case QHSTA_M_NO_AUTO_REQ_SENSE:
case QHSTA_NO_ERROR:
ccb->csio.scsi_status = scsi_status;
switch (scsi_status) {
case SCSI_STATUS_CHECK_COND:
case SCSI_STATUS_CMD_TERMINATED:
ccb->ccb_h.status |= CAM_AUTOSNS_VALID;
/* Structure copy */
ccb->csio.sense_data =
adv->sense_buffers[q_no - 1];
/* FALLTHROUGH */
case SCSI_STATUS_BUSY:
case SCSI_STATUS_RESERV_CONFLICT:
case SCSI_STATUS_QUEUE_FULL:
case SCSI_STATUS_COND_MET:
case SCSI_STATUS_INTERMED:
case SCSI_STATUS_INTERMED_COND_MET:
ccb->ccb_h.status |= CAM_SCSI_STATUS_ERROR;
break;
case SCSI_STATUS_OK:
ccb->ccb_h.status |= CAM_REQ_CMP;
break;
}
break;
case QHSTA_M_SEL_TIMEOUT:
ccb->ccb_h.status = CAM_SEL_TIMEOUT;
break;
case QHSTA_M_DATA_OVER_RUN:
ccb->ccb_h.status = CAM_DATA_RUN_ERR;
break;
case QHSTA_M_UNEXPECTED_BUS_FREE:
ccb->ccb_h.status = CAM_UNEXP_BUSFREE;
break;
case QHSTA_M_BAD_BUS_PHASE_SEQ:
ccb->ccb_h.status = CAM_SEQUENCE_FAIL;
break;
case QHSTA_M_BAD_CMPL_STATUS_IN:
/* No command complete after a status message */
ccb->ccb_h.status = CAM_SEQUENCE_FAIL;
break;
case QHSTA_D_EXE_SCSI_Q_BUSY_TIMEOUT:
case QHSTA_M_WTM_TIMEOUT:
case QHSTA_M_HUNG_REQ_SCSI_BUS_RESET:
/* The SCSI bus hung in a phase */
ccb->ccb_h.status = CAM_SEQUENCE_FAIL;
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
adv_reset_bus(adv, /*initiate_reset*/TRUE);
break;
case QHSTA_M_AUTO_REQ_SENSE_FAIL:
ccb->ccb_h.status = CAM_AUTOSENSE_FAIL;
break;
case QHSTA_D_QDONE_SG_LIST_CORRUPTED:
case QHSTA_D_ASC_DVC_ERROR_CODE_SET:
case QHSTA_D_HOST_ABORT_FAILED:
case QHSTA_D_EXE_SCSI_Q_FAILED:
case QHSTA_D_ASPI_NO_BUF_POOL:
case QHSTA_M_BAD_TAG_CODE:
case QHSTA_D_LRAM_CMP_ERROR:
case QHSTA_M_MICRO_CODE_ERROR_HALT:
default:
panic("%s: Unhandled Host status error %x",
adv_name(adv), host_stat);
/* NOTREACHED */
}
break;
case QD_ABORTED_BY_HOST:
/* Don't clobber any, more explicit, error codes we've set */
if ((ccb->ccb_h.status & CAM_STATUS_MASK) == CAM_REQ_INPROG)
ccb->ccb_h.status = CAM_REQ_ABORTED;
break;
default:
xpt_print_path(ccb->ccb_h.path);
printf("adv_done - queue done with unknown status %x:%x\n",
done_stat, host_stat);
ccb->ccb_h.status = CAM_REQ_CMP_ERR;
break;
}
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
adv_clear_state(adv, ccb);
if ((ccb->ccb_h.status & CAM_STATUS_MASK) != CAM_REQ_CMP
&& (ccb->ccb_h.status & CAM_DEV_QFRZN) == 0) {
xpt_freeze_devq(ccb->ccb_h.path, /*count*/1);
ccb->ccb_h.status |= CAM_DEV_QFRZN;
}
adv_free_ccb_info(adv, cinfo);
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
/*
* Null this out so that we catch driver bugs that cause a
* ccb to be completed twice.
*/
ccb->ccb_h.ccb_cinfo_ptr = NULL;
ccb->ccb_h.status &= ~CAM_SIM_QUEUED;
xpt_done(ccb);
}
/*
* Function to poll for command completion when
* interrupts are disabled (crash dumps)
*/
static void
adv_poll(struct cam_sim *sim)
{
adv_intr(cam_sim_softc(sim));
}
/*
* Attach all the sub-devices we can find
*/
int
adv_attach(adv)
struct adv_softc *adv;
{
struct ccb_setasync csa;
struct cam_devq *devq;
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
int max_sg;
/*
* Allocate an array of ccb mapping structures. We put the
* index of the ccb_info structure into the queue representing
* a transaction and use it for mapping the queue to the
* upper level SCSI transaction it represents.
*/
adv->ccb_infos = malloc(sizeof(*adv->ccb_infos) * adv->max_openings,
M_DEVBUF, M_NOWAIT);
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
if (adv->ccb_infos == NULL)
return (ENOMEM);
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
adv->init_level++;
/*
* Create our DMA tags. These tags define the kinds of device
* accessible memory allocations and memory mappings we will
* need to perform during normal operation.
*
* Unless we need to further restrict the allocation, we rely
* on the restrictions of the parent dmat, hence the common
* use of MAXADDR and MAXSIZE.
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
*
* The ASC boards use chains of "queues" (the transactional
* resources on the board) to represent long S/G lists.
* The first queue represents the command and holds a
* single address and data pair. The queues that follow
* can each hold ADV_SG_LIST_PER_Q entries. Given the
* total number of queues, we can express the largest
* transaction we can map. We reserve a few queues for
* error recovery. Take those into account as well.
*
* There is a way to take an interrupt to download the
* next batch of S/G entries if there are more than 255
* of them (the counter in the queue structure is a u_int8_t).
* We don't use this feature, so limit the S/G list size
* accordingly.
*/
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
max_sg = (adv->max_openings - ADV_MIN_FREE_Q - 1) * ADV_SG_LIST_PER_Q;
if (max_sg > 255)
max_sg = 255;
/* DMA tag for mapping buffers into device visible space. */
if (bus_dma_tag_create(adv->parent_dmat, /*alignment*/1, /*boundary*/0,
/*lowaddr*/BUS_SPACE_MAXADDR,
/*highaddr*/BUS_SPACE_MAXADDR,
/*filter*/NULL, /*filterarg*/NULL,
adv_pci.c: Update list of supported products. Adjust probe message to include the ASC3030. advansys.c: Fix a long standing bug in the error recovery strategy. In order to keep recovery simple, we freeze the SIMQ, stopping the XPT from submitting new requests. Unfortunately, we also will freeze the SIMQ if bus_dmamap_load blocks or we run out of controller resources. On cards with limited resources it was possible to freeze the SIM a second time and never unfreeze it. Now we more carefully track our exception state so we never freeze the SIMQ more than once. Don't rely on pointers fitting in a 32bit field stored in the per-transaction data structures on the card. Use an index to an array of transaction mapping structures instead. This should allow this driver to work on the Alpha. Deal with the ASC3030 which is almost idistinguishable from the ASC3050. Unfortunately the ASC3030 does not work at Ultra speeds, so if we can't find an eeprom, we must assume that ultra is disabled. The SIIG cards using the 3030 do not have eeproms. As a side effect, we now honor the ultra disable bit in the eeprom if it is present. Don't bother attempting to write corrected eeprom data back to the eeprom. We can function just fine if the data is corrupted and I'd rather not risk messing up the user's eeprom. Modify the interrupt handler to catch latched external bus rests. Dynamically determine the maximum number of S/G elements we can map at a single time. The nature of the firmware interface for these cards makes this value dependent on the number of "queues" the card can support. advlib.c: advlib.h: advmcode.c: advmcode.h: Synchronize with the latest firmware image released in the Linux Advansys driver.
2000-01-14 03:33:38 +00:00
/*maxsize*/MAXPHYS,
/*nsegments*/max_sg,
/*maxsegsz*/BUS_SPACE_MAXSIZE_32BIT,
/*flags*/BUS_DMA_ALLOCNOW,
&adv->buffer_dmat) != 0) {
return (ENXIO);
}
adv->init_level++;
/* DMA tag for our sense buffers */
if (bus_dma_tag_create(adv->parent_dmat, /*alignment*/1, /*boundary*/0,
/*lowaddr*/BUS_SPACE_MAXADDR,
/*highaddr*/BUS_SPACE_MAXADDR,
/*filter*/NULL, /*filterarg*/NULL,
sizeof(struct scsi_sense_data)*adv->max_openings,
/*nsegments*/1,
/*maxsegsz*/BUS_SPACE_MAXSIZE_32BIT,
/*flags*/0, &adv->sense_dmat) != 0) {
return (ENXIO);
}
adv->init_level++;
/* Allocation for our sense buffers */
if (bus_dmamem_alloc(adv->sense_dmat, (void **)&adv->sense_buffers,
BUS_DMA_NOWAIT, &adv->sense_dmamap) != 0) {
return (ENOMEM);
}
adv->init_level++;
/* And permanently map them */
bus_dmamap_load(adv->sense_dmat, adv->sense_dmamap,
adv->sense_buffers,
sizeof(struct scsi_sense_data)*adv->max_openings,
adv_map, &adv->sense_physbase, /*flags*/0);
adv->init_level++;
/*
* Fire up the chip
*/
if (adv_start_chip(adv) != 1) {
printf("adv%d: Unable to start on board processor. Aborting.\n",
adv->unit);
return (ENXIO);
}
/*
* Create the device queue for our SIM.
*/
devq = cam_simq_alloc(adv->max_openings);
if (devq == NULL)
return (ENOMEM);
/*
* Construct our SIM entry.
*/
adv->sim = cam_sim_alloc(adv_action, adv_poll, "adv", adv, adv->unit,
1, adv->max_openings, devq);
if (adv->sim == NULL)
return (ENOMEM);
/*
* Register the bus.
*
* XXX Twin Channel EISA Cards???
*/
if (xpt_bus_register(adv->sim, 0) != CAM_SUCCESS) {
cam_sim_free(adv->sim, /*free devq*/TRUE);
return (ENXIO);
}
if (xpt_create_path(&adv->path, /*periph*/NULL, cam_sim_path(adv->sim),
CAM_TARGET_WILDCARD, CAM_LUN_WILDCARD)
!= CAM_REQ_CMP) {
xpt_bus_deregister(cam_sim_path(adv->sim));
cam_sim_free(adv->sim, /*free devq*/TRUE);
return (ENXIO);
}
xpt_setup_ccb(&csa.ccb_h, adv->path, /*priority*/5);
csa.ccb_h.func_code = XPT_SASYNC_CB;
csa.event_enable = AC_FOUND_DEVICE|AC_LOST_DEVICE;
csa.callback = advasync;
csa.callback_arg = adv;
xpt_action((union ccb *)&csa);
return (0);
}