2000-09-16 20:02:28 +00:00
|
|
|
/*
|
|
|
|
|
* Bus independent FreeBSD shim for the aic7xxx based adaptec SCSI controllers
|
|
|
|
|
*
|
2001-01-27 20:54:24 +00:00
|
|
|
* Copyright (c) 1994-2001 Justin T. Gibbs.
|
2000-09-16 20:02:28 +00:00
|
|
|
* All rights reserved.
|
|
|
|
|
*
|
|
|
|
|
* Redistribution and use in source and binary forms, with or without
|
|
|
|
|
* modification, are permitted provided that the following conditions
|
|
|
|
|
* are met:
|
|
|
|
|
* 1. Redistributions of source code must retain the above copyright
|
|
|
|
|
* notice, this list of conditions, and the following disclaimer,
|
|
|
|
|
* without modification.
|
|
|
|
|
* 2. The name of the author may not be used to endorse or promote products
|
|
|
|
|
* derived from this software without specific prior written permission.
|
|
|
|
|
*
|
|
|
|
|
* Alternatively, this software may be distributed under the terms of the
|
|
|
|
|
* GNU Public License ("GPL").
|
|
|
|
|
*
|
|
|
|
|
* 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.
|
|
|
|
|
*
|
2002-09-26 21:48:17 +00:00
|
|
|
* $Id: //depot/aic7xxx/freebsd/dev/aic7xxx/aic7xxx_osm.c#11 $
|
2000-09-16 20:02:28 +00:00
|
|
|
*
|
|
|
|
|
* $FreeBSD$
|
|
|
|
|
*/
|
|
|
|
|
|
2002-04-24 16:58:51 +00:00
|
|
|
#include <dev/aic7xxx/aic7xxx_osm.h>
|
2000-09-16 20:02:28 +00:00
|
|
|
#include <dev/aic7xxx/aic7xxx_inline.h>
|
|
|
|
|
|
|
|
|
|
#ifndef AHC_TMODE_ENABLE
|
|
|
|
|
#define AHC_TMODE_ENABLE 0
|
|
|
|
|
#endif
|
|
|
|
|
|
|
|
|
|
#define ccb_scb_ptr spriv_ptr0
|
|
|
|
|
|
2002-09-22 23:03:31 +00:00
|
|
|
devclass_t ahc_devclass;
|
|
|
|
|
|
2000-09-16 20:02:28 +00:00
|
|
|
#if UNUSED
|
|
|
|
|
static void ahc_dump_targcmd(struct target_cmd *cmd);
|
|
|
|
|
#endif
|
2001-05-15 19:41:12 +00:00
|
|
|
static int ahc_modevent(module_t mod, int type, void *data);
|
2000-09-16 20:02:28 +00:00
|
|
|
static void ahc_action(struct cam_sim *sim, union ccb *ccb);
|
|
|
|
|
static void ahc_get_tran_settings(struct ahc_softc *ahc,
|
|
|
|
|
int our_id, char channel,
|
|
|
|
|
struct ccb_trans_settings *cts);
|
|
|
|
|
static void ahc_async(void *callback_arg, uint32_t code,
|
|
|
|
|
struct cam_path *path, void *arg);
|
|
|
|
|
static void ahc_execute_scb(void *arg, bus_dma_segment_t *dm_segs,
|
|
|
|
|
int nsegments, int error);
|
|
|
|
|
static void ahc_poll(struct cam_sim *sim);
|
|
|
|
|
static void ahc_setup_data(struct ahc_softc *ahc, struct cam_sim *sim,
|
|
|
|
|
struct ccb_scsiio *csio, struct scb *scb);
|
|
|
|
|
static void ahc_abort_ccb(struct ahc_softc *ahc, struct cam_sim *sim,
|
|
|
|
|
union ccb *ccb);
|
|
|
|
|
static int ahc_create_path(struct ahc_softc *ahc,
|
2000-09-22 22:18:05 +00:00
|
|
|
char channel, u_int target, u_int lun,
|
2000-09-16 20:02:28 +00:00
|
|
|
struct cam_path **path);
|
|
|
|
|
|
|
|
|
|
static void ahc_set_recoveryscb(struct ahc_softc *ahc, struct scb *scb);
|
|
|
|
|
|
|
|
|
|
static int
|
2000-09-22 22:18:05 +00:00
|
|
|
ahc_create_path(struct ahc_softc *ahc, char channel, u_int target,
|
|
|
|
|
u_int lun, struct cam_path **path)
|
2000-09-16 20:02:28 +00:00
|
|
|
{
|
|
|
|
|
path_id_t path_id;
|
|
|
|
|
|
2000-09-22 22:18:05 +00:00
|
|
|
if (channel == 'B')
|
2000-09-16 20:02:28 +00:00
|
|
|
path_id = cam_sim_path(ahc->platform_data->sim_b);
|
|
|
|
|
else
|
|
|
|
|
path_id = cam_sim_path(ahc->platform_data->sim);
|
|
|
|
|
|
|
|
|
|
return (xpt_create_path(path, /*periph*/NULL,
|
2000-09-22 22:18:05 +00:00
|
|
|
path_id, target, lun));
|
2000-09-16 20:02:28 +00:00
|
|
|
}
|
|
|
|
|
|
2002-04-24 16:58:51 +00:00
|
|
|
int
|
|
|
|
|
ahc_map_int(struct ahc_softc *ahc)
|
|
|
|
|
{
|
|
|
|
|
int error;
|
|
|
|
|
|
|
|
|
|
/* Hook up our interrupt handler */
|
|
|
|
|
error = bus_setup_intr(ahc->dev_softc, ahc->platform_data->irq,
|
|
|
|
|
INTR_TYPE_CAM, ahc_platform_intr, ahc,
|
|
|
|
|
&ahc->platform_data->ih);
|
|
|
|
|
|
|
|
|
|
if (error != 0)
|
|
|
|
|
device_printf(ahc->dev_softc, "bus_setup_intr() failed: %d\n",
|
|
|
|
|
error);
|
|
|
|
|
return (error);
|
|
|
|
|
}
|
|
|
|
|
|
2000-09-16 20:02:28 +00:00
|
|
|
/*
|
|
|
|
|
* Attach all the sub-devices we can find
|
|
|
|
|
*/
|
|
|
|
|
int
|
|
|
|
|
ahc_attach(struct ahc_softc *ahc)
|
|
|
|
|
{
|
|
|
|
|
char ahc_info[256];
|
|
|
|
|
struct ccb_setasync csa;
|
|
|
|
|
struct cam_devq *devq;
|
|
|
|
|
int bus_id;
|
|
|
|
|
int bus_id2;
|
|
|
|
|
struct cam_sim *sim;
|
|
|
|
|
struct cam_sim *sim2;
|
|
|
|
|
struct cam_path *path;
|
|
|
|
|
struct cam_path *path2;
|
|
|
|
|
long s;
|
|
|
|
|
int count;
|
|
|
|
|
|
|
|
|
|
count = 0;
|
|
|
|
|
sim = NULL;
|
|
|
|
|
sim2 = NULL;
|
|
|
|
|
|
|
|
|
|
ahc_controller_info(ahc, ahc_info);
|
|
|
|
|
printf("%s\n", ahc_info);
|
|
|
|
|
ahc_lock(ahc, &s);
|
|
|
|
|
/*
|
|
|
|
|
* Attach secondary channel first if the user has
|
|
|
|
|
* declared it the primary channel.
|
|
|
|
|
*/
|
2001-02-17 01:53:22 +00:00
|
|
|
if ((ahc->features & AHC_TWIN) != 0
|
This is an MFC candidate.
ahc_eisa.c:
Change aic7770_map_int to take an additional irq parameter.
Although we can get the irq from the eisa dev under FreeBSD,
we can't do this under linux, so the OSM interface must supply
this.
ahc_pci.c:
Move ahc_power_state_change() to the OSM. This allows us to
use a platform supplied function that does the same thing.
-current will move to the FreeBSD native API in the near
future.
aic7770.c:
Sync up with core changes to support Linux EISA.
We now store a 2 bit primary channel number rather
than a bit flag that only allows b to be the primary
channel. Adjust for this change.
aic7xxx.c:
Namespace and staticization cleanup. All exported symbols
use an "ahc_" prefix to avoid collisions with other modules.
Correct a logic bug that prevented us from dropping
ATN during some exceptional conditions during message
processing.
Take advantage of a new flag managed by the sequencer
that indicates if an SCB fetch is in progress. If so,
the currently selected SCB needs to be returned to the
free list to prevent an SCB leak. This leak is a rarity
and would only occur if a bus reset or timeout resulting
in a bus reset occurred in the middle of an SCB fetch.
Don't attempt to perform ULTRA transfers on ultra capable
adapters missing the external precision resistor required
for ultra speeds. I've never encountered an adapter
configured this way, but better safe than sorry.
Handle the case of 5MHz user sync rate set as "0" instead of 0x1c
in scratch ram.
If we lookup a period of 0 in our table (async), clear the scsi offset.
aic7xxx.h:
Adjust for the primary channel being represented as
a 2 bit integer in the flags member of the ahc softc.
Cleanup the flags definitions so that comment blocks are
not cramped.
Update seeprom definitions to correctly reflect the fact
that the primary channel is represented as a 2 bit integer.
Add AHC_ULTRA_DIASABLED softc flag to denote controllers
missing the external precision resistor.
aic7xxx.reg:
Add DFCACHETH to the definition of DFSTATUS for completness sake.
Add SEQ_FLAGS2 which currently only contains the SCB_DMA
(SCB DMA in progress) flag.
aic7xxx.seq:
Correct a problem when one lun has a disconnected untagged
transaction and another lun has disconnected tagged transactions.
Just because an entry is found in the untagged table doesn't
mean that it will match. If the match on the lun fails, cleanup
the SCB (return it to the disconnected list or free it), and snoop
for a tag message. Before this change, we reported an unsolicited
reselection. This bug was introduced about a month ago during an
overly aggressive optimization pass on the reselection code.
When cleaning up an SCB, we can't just blindly free the SCB. In
the paging case, if the SCB came off of the disconnected list, its
state may never have been updated in host memory. So, check the
disconnected bit in SCB_CONTROL and return the SCB to the disconnected
list if appropriate.
Manage the SCB_DMA flag of SEQ_FLAGS2.
More carefully shutdown the S/G dma engine in all cases by using
a subroutine. Supposedly not doing this can cause an arbiter hang
on some ULTRA2 chips.
Formatting cleanup.
On some chips, at least the aic7856, the transition from
MREQPEND to HDONE can take a full 4 clock cycles. Test
HDONE one more time to avoid this race. We only want our
FIFO hung recovery code to execute when the engine is
really hung.
aic7xxx_93cx6.c:
Sync perforce ids.
aic7xxx_freebsd.c:
Adjust for the primary channel being a 2 bit integer
rather than a flag for 'B' channel being the primary.
Namespace cleanup.
Unpause the sequencer in one error recovery path that
neglected to do so. This could have caused us to perform
a bus reset when a recovery message might have otherwise been
successful.
aic7xxx_freebsd.h:
Use AHC_PCI_CONFIG for controlling compilation of PCI
support consistently throughout the driver.
Move ahc_power_state_change() to OSM.
aic7xxx_inline.h
Namespace cleanup.
Adjust our interrupt handler so it will work in the edge
interrupt case. We must process all interrupt sources
when the interrupt fires or risk not ever getting an
interrupt again. This involves marking the fact
that we are relying on an edge interrupt in ahc->flags
and checking for this condition in addition to the
AHC_ALL_INTERRUPTS flag. This fixes hangs on the
284X and any other aic7770 installation where level
interrupts are not available.
aic7xxx_pci.c:
Move the powerstate manipulation code into the OSM. Several
OSes now provide this functionality natively.
Take another shot at using the data stored in scratch ram
if the SCB2 signature is correct and no SEEPROM data is
available. In the past this failed if external SCB ram
was configured because the memory port was locked. We
now release the memory port prior to testing the values
in SCB2 and re-acquire it prior to doing termination control.
Adjust for new 2 bit primary channel setting.
Trust the STPWLEVEL setting on v 3.X BIOSes too.
Configure any 785X ID in the same fashion and assume
that any device with a rev id of 1 or higher has the
PCI 2.1 retry bug.
2001-03-11 06:34:17 +00:00
|
|
|
&& (ahc->flags & AHC_PRIMARY_CHANNEL) != 0) {
|
2000-09-16 20:02:28 +00:00
|
|
|
bus_id = 1;
|
|
|
|
|
bus_id2 = 0;
|
|
|
|
|
} else {
|
|
|
|
|
bus_id = 0;
|
|
|
|
|
bus_id2 = 1;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Create the device queue for our SIM(s).
|
|
|
|
|
*/
|
2001-01-22 21:03:48 +00:00
|
|
|
devq = cam_simq_alloc(AHC_MAX_QUEUE);
|
2000-09-16 20:02:28 +00:00
|
|
|
if (devq == NULL)
|
|
|
|
|
goto fail;
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Construct our first channel SIM entry
|
|
|
|
|
*/
|
2000-09-20 04:46:15 +00:00
|
|
|
sim = cam_sim_alloc(ahc_action, ahc_poll, "ahc", ahc,
|
|
|
|
|
device_get_unit(ahc->dev_softc),
|
2001-01-22 21:03:48 +00:00
|
|
|
1, AHC_MAX_QUEUE, devq);
|
2000-09-16 20:02:28 +00:00
|
|
|
if (sim == NULL) {
|
|
|
|
|
cam_simq_free(devq);
|
|
|
|
|
goto fail;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (xpt_bus_register(sim, bus_id) != CAM_SUCCESS) {
|
|
|
|
|
cam_sim_free(sim, /*free_devq*/TRUE);
|
|
|
|
|
sim = NULL;
|
|
|
|
|
goto fail;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (xpt_create_path(&path, /*periph*/NULL,
|
|
|
|
|
cam_sim_path(sim), CAM_TARGET_WILDCARD,
|
|
|
|
|
CAM_LUN_WILDCARD) != CAM_REQ_CMP) {
|
|
|
|
|
xpt_bus_deregister(cam_sim_path(sim));
|
|
|
|
|
cam_sim_free(sim, /*free_devq*/TRUE);
|
|
|
|
|
sim = NULL;
|
|
|
|
|
goto fail;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
xpt_setup_ccb(&csa.ccb_h, path, /*priority*/5);
|
|
|
|
|
csa.ccb_h.func_code = XPT_SASYNC_CB;
|
|
|
|
|
csa.event_enable = AC_LOST_DEVICE;
|
|
|
|
|
csa.callback = ahc_async;
|
|
|
|
|
csa.callback_arg = sim;
|
|
|
|
|
xpt_action((union ccb *)&csa);
|
|
|
|
|
count++;
|
|
|
|
|
|
|
|
|
|
if (ahc->features & AHC_TWIN) {
|
|
|
|
|
sim2 = cam_sim_alloc(ahc_action, ahc_poll, "ahc",
|
2000-09-20 04:46:15 +00:00
|
|
|
ahc, device_get_unit(ahc->dev_softc), 1,
|
2001-01-22 21:03:48 +00:00
|
|
|
AHC_MAX_QUEUE, devq);
|
2000-09-16 20:02:28 +00:00
|
|
|
|
|
|
|
|
if (sim2 == NULL) {
|
|
|
|
|
printf("ahc_attach: Unable to attach second "
|
|
|
|
|
"bus due to resource shortage");
|
|
|
|
|
goto fail;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (xpt_bus_register(sim2, bus_id2) != CAM_SUCCESS) {
|
|
|
|
|
printf("ahc_attach: Unable to attach second "
|
|
|
|
|
"bus due to resource shortage");
|
|
|
|
|
/*
|
|
|
|
|
* We do not want to destroy the device queue
|
|
|
|
|
* because the first bus is using it.
|
|
|
|
|
*/
|
|
|
|
|
cam_sim_free(sim2, /*free_devq*/FALSE);
|
|
|
|
|
goto fail;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (xpt_create_path(&path2, /*periph*/NULL,
|
|
|
|
|
cam_sim_path(sim2),
|
|
|
|
|
CAM_TARGET_WILDCARD,
|
|
|
|
|
CAM_LUN_WILDCARD) != CAM_REQ_CMP) {
|
|
|
|
|
xpt_bus_deregister(cam_sim_path(sim2));
|
|
|
|
|
cam_sim_free(sim2, /*free_devq*/FALSE);
|
|
|
|
|
sim2 = NULL;
|
|
|
|
|
goto fail;
|
|
|
|
|
}
|
|
|
|
|
xpt_setup_ccb(&csa.ccb_h, path2, /*priority*/5);
|
|
|
|
|
csa.ccb_h.func_code = XPT_SASYNC_CB;
|
|
|
|
|
csa.event_enable = AC_LOST_DEVICE;
|
|
|
|
|
csa.callback = ahc_async;
|
|
|
|
|
csa.callback_arg = sim2;
|
|
|
|
|
xpt_action((union ccb *)&csa);
|
|
|
|
|
count++;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
fail:
|
2001-02-17 01:53:22 +00:00
|
|
|
if ((ahc->features & AHC_TWIN) != 0
|
This is an MFC candidate.
ahc_eisa.c:
Change aic7770_map_int to take an additional irq parameter.
Although we can get the irq from the eisa dev under FreeBSD,
we can't do this under linux, so the OSM interface must supply
this.
ahc_pci.c:
Move ahc_power_state_change() to the OSM. This allows us to
use a platform supplied function that does the same thing.
-current will move to the FreeBSD native API in the near
future.
aic7770.c:
Sync up with core changes to support Linux EISA.
We now store a 2 bit primary channel number rather
than a bit flag that only allows b to be the primary
channel. Adjust for this change.
aic7xxx.c:
Namespace and staticization cleanup. All exported symbols
use an "ahc_" prefix to avoid collisions with other modules.
Correct a logic bug that prevented us from dropping
ATN during some exceptional conditions during message
processing.
Take advantage of a new flag managed by the sequencer
that indicates if an SCB fetch is in progress. If so,
the currently selected SCB needs to be returned to the
free list to prevent an SCB leak. This leak is a rarity
and would only occur if a bus reset or timeout resulting
in a bus reset occurred in the middle of an SCB fetch.
Don't attempt to perform ULTRA transfers on ultra capable
adapters missing the external precision resistor required
for ultra speeds. I've never encountered an adapter
configured this way, but better safe than sorry.
Handle the case of 5MHz user sync rate set as "0" instead of 0x1c
in scratch ram.
If we lookup a period of 0 in our table (async), clear the scsi offset.
aic7xxx.h:
Adjust for the primary channel being represented as
a 2 bit integer in the flags member of the ahc softc.
Cleanup the flags definitions so that comment blocks are
not cramped.
Update seeprom definitions to correctly reflect the fact
that the primary channel is represented as a 2 bit integer.
Add AHC_ULTRA_DIASABLED softc flag to denote controllers
missing the external precision resistor.
aic7xxx.reg:
Add DFCACHETH to the definition of DFSTATUS for completness sake.
Add SEQ_FLAGS2 which currently only contains the SCB_DMA
(SCB DMA in progress) flag.
aic7xxx.seq:
Correct a problem when one lun has a disconnected untagged
transaction and another lun has disconnected tagged transactions.
Just because an entry is found in the untagged table doesn't
mean that it will match. If the match on the lun fails, cleanup
the SCB (return it to the disconnected list or free it), and snoop
for a tag message. Before this change, we reported an unsolicited
reselection. This bug was introduced about a month ago during an
overly aggressive optimization pass on the reselection code.
When cleaning up an SCB, we can't just blindly free the SCB. In
the paging case, if the SCB came off of the disconnected list, its
state may never have been updated in host memory. So, check the
disconnected bit in SCB_CONTROL and return the SCB to the disconnected
list if appropriate.
Manage the SCB_DMA flag of SEQ_FLAGS2.
More carefully shutdown the S/G dma engine in all cases by using
a subroutine. Supposedly not doing this can cause an arbiter hang
on some ULTRA2 chips.
Formatting cleanup.
On some chips, at least the aic7856, the transition from
MREQPEND to HDONE can take a full 4 clock cycles. Test
HDONE one more time to avoid this race. We only want our
FIFO hung recovery code to execute when the engine is
really hung.
aic7xxx_93cx6.c:
Sync perforce ids.
aic7xxx_freebsd.c:
Adjust for the primary channel being a 2 bit integer
rather than a flag for 'B' channel being the primary.
Namespace cleanup.
Unpause the sequencer in one error recovery path that
neglected to do so. This could have caused us to perform
a bus reset when a recovery message might have otherwise been
successful.
aic7xxx_freebsd.h:
Use AHC_PCI_CONFIG for controlling compilation of PCI
support consistently throughout the driver.
Move ahc_power_state_change() to OSM.
aic7xxx_inline.h
Namespace cleanup.
Adjust our interrupt handler so it will work in the edge
interrupt case. We must process all interrupt sources
when the interrupt fires or risk not ever getting an
interrupt again. This involves marking the fact
that we are relying on an edge interrupt in ahc->flags
and checking for this condition in addition to the
AHC_ALL_INTERRUPTS flag. This fixes hangs on the
284X and any other aic7770 installation where level
interrupts are not available.
aic7xxx_pci.c:
Move the powerstate manipulation code into the OSM. Several
OSes now provide this functionality natively.
Take another shot at using the data stored in scratch ram
if the SCB2 signature is correct and no SEEPROM data is
available. In the past this failed if external SCB ram
was configured because the memory port was locked. We
now release the memory port prior to testing the values
in SCB2 and re-acquire it prior to doing termination control.
Adjust for new 2 bit primary channel setting.
Trust the STPWLEVEL setting on v 3.X BIOSes too.
Configure any 785X ID in the same fashion and assume
that any device with a rev id of 1 or higher has the
PCI 2.1 retry bug.
2001-03-11 06:34:17 +00:00
|
|
|
&& (ahc->flags & AHC_PRIMARY_CHANNEL) != 0) {
|
2000-09-16 20:02:28 +00:00
|
|
|
ahc->platform_data->sim_b = sim;
|
|
|
|
|
ahc->platform_data->path_b = path;
|
|
|
|
|
ahc->platform_data->sim = sim2;
|
|
|
|
|
ahc->platform_data->path = path2;
|
|
|
|
|
} else {
|
|
|
|
|
ahc->platform_data->sim = sim;
|
|
|
|
|
ahc->platform_data->path = path;
|
|
|
|
|
ahc->platform_data->sim_b = sim2;
|
|
|
|
|
ahc->platform_data->path_b = path2;
|
|
|
|
|
}
|
|
|
|
|
|
2002-08-31 06:43:55 +00:00
|
|
|
if (count != 0) {
|
2000-09-16 20:02:28 +00:00
|
|
|
/* We have to wait until after any system dumps... */
|
ahc_eisa.c:
ahc_pci.c:
Add detach support.
Make use of soft allocated on our behalf by newbus.
For PCI devices, disable the mapping type we aren't
using for extra protection from rogue code.
aic7xxx_93cx6.c:
aic7xxx_93cx6.h:
Sync perforce IDs.
aic7xxx_freebsd.c:
Capture the eventhandle returned by EVENTHANDER_REGISTER
so we can kill the handler off during detach.
Use AHC_* constants instead of hard coded numbers in a
few more places.
Test PPR option state when deciding to "really" negotiate
when the CAM_NEGOTIATE flag is passed in a CCB.
Make use of core "ahc_pause_and_flushwork" routine in our
timeout handler rather than re-inventing this code.
Cleanup all of our resources (really!) in ahc_platform_free().
We should be all set to become a module now.
Implement the core ahc_detach() routine shared by all of
the FreeBSD front-ends.
aic7xxx_freebsd.h:
Softc storage for our event handler.
Null implementation for the ahc_platform_flushwork() OSM
callback. FreeBSD doesn't need this as XPT callbacks are
safe from all contexts and are done directly in ahc_done().
aic7xxx_inline.h:
Implement new lazy interrupt scheme. To avoid an extra
PCI bus read, we first check our completion queues to
see if any work has completed. If work is available, we
assume that this is the source of the interrupt and skip
reading INTSTAT. Any remaining interrupt status will be
cleared by a second call to the interrupt handler should
the interrupt line still be asserted. This drops the
interrupt handler down to a single PCI bus read in the
common case of I/O completion. This is the same overhead
as in the not so distant past, but the extra sanity of
perforning a PCI read after clearing the command complete
interrupt and before running the completion queue to avoid
missing command complete interrupts added a cycle.
aic7xxx.c:
During initialization, be sure to initialize all scratch
ram locations before they are read to avoid parity errors.
In this case, we use a new function, ahc_unbusy_tcl() to
initialize the scratch ram busy target table.
Replace instances of ahc_index_busy_tcl() used to unbusy
a tcl without looking at the old value with ahc_unbusy_tcl().
Modify ahc_sent_msg so that it can find single byte messages.
ahc_sent_msg is now used to determine if a transfer negotiation
attempt resulted in a bus free.
Be more careful in filtering out only the SCSI interrupts
of interest in ahc_handle_scsiint.
Rearrange interrupt clearing code to ensure that at least
one PCI transaction occurrs after hitting CLRSINT1 and
writting to CLRINT. CLRSINT1 writes take a bit to
take effect, and the re-arrangement provides sufficient
delay to ensure the write to CLRINT is effective. The
old code might report a spurious interrupt on some "fast"
chipsets.
export ahc-update_target_msg_request for use by OSM code.
If a target does not respond to our ATN request, clear
it once we move to a non-message phase. This avoids
sending a MSG_NOOP in some later message out phase.
Use max lun and max target constants instead of
hard-coded values.
Use softc storage built into our device_t under FreeBSD.
Fix a bug in ahc_free() that caused us to delete
resources that were not allocated.
Clean up any tstate/lstate info in ahc_free().
Clear the powerdown state in ahc_reset() so that
registers can be accessed.
Add a preliminary function for pausing the chip and
processing any posted work.
Add a preliminary suspend and resume functions.
aic7xxx.h:
Limit the number of supported luns to 64. We don't
support information unit transfers, so this is the
maximum that makes sense for these chips.
Add a new flag AHC_ALL_INTERRUPTS that forces the
processing of all interrupt state in a single invokation
of ahc_intr(). When the flag is not set, we use the
lazy interrupt handling scheme.
Add data structures to store controller state while
we are suspended.
Use constants instead of hard coded values where appropriate.
Correct some harmless "unsigned/signed" conflicts.
aic7xxx.seq:
Only perform the SCSIBUSL fix on ULTRA2 or newer controllers.
Older controllers seem to be confused by this.
In target mode, ignore PHASEMIS during data phases.
This bit seems to be flakey on U160 controllers acting
in target mode.
aic7xxx_pci.c:
Add support for the 29160C CPCI adapter.
Add definitions for subvendor ID information
available for devices with the "9005" vendor id.
We currently use this information to determine
if a multi-function device doesn't have the second
channel hooked up on a board.
Add rudimentary power mode code so we can put the
controller into the D0 state. In the future this
will be an OSM callback so that in FreeBSD we don't
duplicate functionality provided by the PCI code.
The powerstate code was added after I'd completed
my regression tests on this code.
Only capture "left over BIOS state" if the POWRDN
setting is not set in HCNTRL.
In target mode, don't bother sending incremental
CRC data.
2000-12-20 01:11:37 +00:00
|
|
|
ahc->platform_data->eh =
|
|
|
|
|
EVENTHANDLER_REGISTER(shutdown_final, ahc_shutdown,
|
|
|
|
|
ahc, SHUTDOWN_PRI_DEFAULT);
|
2002-08-31 06:43:55 +00:00
|
|
|
ahc_intr_enable(ahc, TRUE);
|
|
|
|
|
}
|
2000-09-16 20:02:28 +00:00
|
|
|
|
2002-08-31 06:43:55 +00:00
|
|
|
ahc_unlock(ahc, &s);
|
2000-09-16 20:02:28 +00:00
|
|
|
return (count);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Catch an interrupt from the adapter
|
|
|
|
|
*/
|
|
|
|
|
void
|
ahc_eisa.c:
ahc_pci.c:
Add detach support.
Make use of soft allocated on our behalf by newbus.
For PCI devices, disable the mapping type we aren't
using for extra protection from rogue code.
aic7xxx_93cx6.c:
aic7xxx_93cx6.h:
Sync perforce IDs.
aic7xxx_freebsd.c:
Capture the eventhandle returned by EVENTHANDER_REGISTER
so we can kill the handler off during detach.
Use AHC_* constants instead of hard coded numbers in a
few more places.
Test PPR option state when deciding to "really" negotiate
when the CAM_NEGOTIATE flag is passed in a CCB.
Make use of core "ahc_pause_and_flushwork" routine in our
timeout handler rather than re-inventing this code.
Cleanup all of our resources (really!) in ahc_platform_free().
We should be all set to become a module now.
Implement the core ahc_detach() routine shared by all of
the FreeBSD front-ends.
aic7xxx_freebsd.h:
Softc storage for our event handler.
Null implementation for the ahc_platform_flushwork() OSM
callback. FreeBSD doesn't need this as XPT callbacks are
safe from all contexts and are done directly in ahc_done().
aic7xxx_inline.h:
Implement new lazy interrupt scheme. To avoid an extra
PCI bus read, we first check our completion queues to
see if any work has completed. If work is available, we
assume that this is the source of the interrupt and skip
reading INTSTAT. Any remaining interrupt status will be
cleared by a second call to the interrupt handler should
the interrupt line still be asserted. This drops the
interrupt handler down to a single PCI bus read in the
common case of I/O completion. This is the same overhead
as in the not so distant past, but the extra sanity of
perforning a PCI read after clearing the command complete
interrupt and before running the completion queue to avoid
missing command complete interrupts added a cycle.
aic7xxx.c:
During initialization, be sure to initialize all scratch
ram locations before they are read to avoid parity errors.
In this case, we use a new function, ahc_unbusy_tcl() to
initialize the scratch ram busy target table.
Replace instances of ahc_index_busy_tcl() used to unbusy
a tcl without looking at the old value with ahc_unbusy_tcl().
Modify ahc_sent_msg so that it can find single byte messages.
ahc_sent_msg is now used to determine if a transfer negotiation
attempt resulted in a bus free.
Be more careful in filtering out only the SCSI interrupts
of interest in ahc_handle_scsiint.
Rearrange interrupt clearing code to ensure that at least
one PCI transaction occurrs after hitting CLRSINT1 and
writting to CLRINT. CLRSINT1 writes take a bit to
take effect, and the re-arrangement provides sufficient
delay to ensure the write to CLRINT is effective. The
old code might report a spurious interrupt on some "fast"
chipsets.
export ahc-update_target_msg_request for use by OSM code.
If a target does not respond to our ATN request, clear
it once we move to a non-message phase. This avoids
sending a MSG_NOOP in some later message out phase.
Use max lun and max target constants instead of
hard-coded values.
Use softc storage built into our device_t under FreeBSD.
Fix a bug in ahc_free() that caused us to delete
resources that were not allocated.
Clean up any tstate/lstate info in ahc_free().
Clear the powerdown state in ahc_reset() so that
registers can be accessed.
Add a preliminary function for pausing the chip and
processing any posted work.
Add a preliminary suspend and resume functions.
aic7xxx.h:
Limit the number of supported luns to 64. We don't
support information unit transfers, so this is the
maximum that makes sense for these chips.
Add a new flag AHC_ALL_INTERRUPTS that forces the
processing of all interrupt state in a single invokation
of ahc_intr(). When the flag is not set, we use the
lazy interrupt handling scheme.
Add data structures to store controller state while
we are suspended.
Use constants instead of hard coded values where appropriate.
Correct some harmless "unsigned/signed" conflicts.
aic7xxx.seq:
Only perform the SCSIBUSL fix on ULTRA2 or newer controllers.
Older controllers seem to be confused by this.
In target mode, ignore PHASEMIS during data phases.
This bit seems to be flakey on U160 controllers acting
in target mode.
aic7xxx_pci.c:
Add support for the 29160C CPCI adapter.
Add definitions for subvendor ID information
available for devices with the "9005" vendor id.
We currently use this information to determine
if a multi-function device doesn't have the second
channel hooked up on a board.
Add rudimentary power mode code so we can put the
controller into the D0 state. In the future this
will be an OSM callback so that in FreeBSD we don't
duplicate functionality provided by the PCI code.
The powerstate code was added after I'd completed
my regression tests on this code.
Only capture "left over BIOS state" if the POWRDN
setting is not set in HCNTRL.
In target mode, don't bother sending incremental
CRC data.
2000-12-20 01:11:37 +00:00
|
|
|
ahc_platform_intr(void *arg)
|
2000-09-16 20:02:28 +00:00
|
|
|
{
|
|
|
|
|
struct ahc_softc *ahc;
|
|
|
|
|
|
|
|
|
|
ahc = (struct ahc_softc *)arg;
|
|
|
|
|
ahc_intr(ahc);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* We have an scb which has been processed by the
|
|
|
|
|
* adaptor, now we look to see how the operation
|
|
|
|
|
* went.
|
|
|
|
|
*/
|
|
|
|
|
void
|
|
|
|
|
ahc_done(struct ahc_softc *ahc, struct scb *scb)
|
|
|
|
|
{
|
|
|
|
|
union ccb *ccb;
|
|
|
|
|
|
|
|
|
|
CAM_DEBUG(scb->io_ctx->ccb_h.path, CAM_DEBUG_TRACE,
|
|
|
|
|
("ahc_done - scb %d\n", scb->hscb->tag));
|
|
|
|
|
|
|
|
|
|
ccb = scb->io_ctx;
|
|
|
|
|
LIST_REMOVE(scb, pending_links);
|
2000-10-11 23:46:34 +00:00
|
|
|
if ((scb->flags & SCB_UNTAGGEDQ) != 0) {
|
2000-09-16 20:02:28 +00:00
|
|
|
struct scb_tailq *untagged_q;
|
2001-02-21 20:50:36 +00:00
|
|
|
int target_offset;
|
2000-09-16 20:02:28 +00:00
|
|
|
|
2001-02-21 20:50:36 +00:00
|
|
|
target_offset = SCB_GET_TARGET_OFFSET(ahc, scb);
|
|
|
|
|
untagged_q = &ahc->untagged_queues[target_offset];
|
2000-09-16 20:02:28 +00:00
|
|
|
TAILQ_REMOVE(untagged_q, scb, links.tqe);
|
2000-10-11 23:46:34 +00:00
|
|
|
scb->flags &= ~SCB_UNTAGGEDQ;
|
2000-09-16 20:02:28 +00:00
|
|
|
ahc_run_untagged_queue(ahc, untagged_q);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
untimeout(ahc_timeout, (caddr_t)scb, 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(ahc->buffer_dmat, scb->dmamap, op);
|
|
|
|
|
bus_dmamap_unload(ahc->buffer_dmat, scb->dmamap);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (ccb->ccb_h.func_code == XPT_CONT_TARGET_IO) {
|
2001-03-29 00:36:35 +00:00
|
|
|
struct cam_path *ccb_path;
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* If we have finally disconnected, clean up our
|
|
|
|
|
* pending device state.
|
|
|
|
|
* XXX - There may be error states that cause where
|
|
|
|
|
* we will remain connected.
|
|
|
|
|
*/
|
|
|
|
|
ccb_path = ccb->ccb_h.path;
|
|
|
|
|
if (ahc->pending_device != NULL
|
|
|
|
|
&& xpt_path_comp(ahc->pending_device->path, ccb_path) == 0) {
|
|
|
|
|
|
|
|
|
|
if ((ccb->ccb_h.flags & CAM_SEND_STATUS) != 0) {
|
|
|
|
|
ahc->pending_device = NULL;
|
|
|
|
|
} else {
|
2002-04-24 16:58:51 +00:00
|
|
|
if (bootverbose) {
|
|
|
|
|
xpt_print_path(ccb->ccb_h.path);
|
|
|
|
|
printf("Still connected\n");
|
|
|
|
|
}
|
2001-03-29 00:36:35 +00:00
|
|
|
ahc_freeze_ccb(ccb);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2000-09-16 20:02:28 +00:00
|
|
|
if (ahc_get_transaction_status(scb) == CAM_REQ_INPROG)
|
|
|
|
|
ccb->ccb_h.status |= CAM_REQ_CMP;
|
|
|
|
|
ccb->ccb_h.status &= ~CAM_SIM_QUEUED;
|
|
|
|
|
ahc_free_scb(ahc, scb);
|
|
|
|
|
xpt_done(ccb);
|
|
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* If the recovery SCB completes, we have to be
|
|
|
|
|
* out of our timeout.
|
|
|
|
|
*/
|
|
|
|
|
if ((scb->flags & SCB_RECOVERY_SCB) != 0) {
|
|
|
|
|
struct scb *list_scb;
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* We were able to complete the command successfully,
|
|
|
|
|
* so reinstate the timeouts for all other pending
|
|
|
|
|
* commands.
|
|
|
|
|
*/
|
|
|
|
|
LIST_FOREACH(list_scb, &ahc->pending_scbs, pending_links) {
|
|
|
|
|
union ccb *ccb;
|
2001-01-22 21:03:48 +00:00
|
|
|
uint64_t time;
|
2000-09-16 20:02:28 +00:00
|
|
|
|
|
|
|
|
ccb = list_scb->io_ctx;
|
2001-01-22 21:03:48 +00:00
|
|
|
if (ccb->ccb_h.timeout == CAM_TIME_INFINITY)
|
|
|
|
|
continue;
|
|
|
|
|
|
|
|
|
|
time = ccb->ccb_h.timeout;
|
|
|
|
|
time *= hz;
|
|
|
|
|
time /= 1000;
|
2000-09-16 20:02:28 +00:00
|
|
|
ccb->ccb_h.timeout_ch =
|
2001-01-22 21:03:48 +00:00
|
|
|
timeout(ahc_timeout, list_scb, time);
|
2000-09-16 20:02:28 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (ahc_get_transaction_status(scb) == CAM_BDR_SENT
|
|
|
|
|
|| ahc_get_transaction_status(scb) == CAM_REQ_ABORTED)
|
|
|
|
|
ahc_set_transaction_status(scb, CAM_CMD_TIMEOUT);
|
|
|
|
|
ahc_print_path(ahc, scb);
|
|
|
|
|
printf("no longer in timeout, status = %x\n",
|
|
|
|
|
ccb->ccb_h.status);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* Don't clobber any existing error state */
|
|
|
|
|
if (ahc_get_transaction_status(scb) == CAM_REQ_INPROG) {
|
|
|
|
|
ccb->ccb_h.status |= CAM_REQ_CMP;
|
|
|
|
|
} else if ((scb->flags & SCB_SENSE) != 0) {
|
|
|
|
|
/*
|
|
|
|
|
* We performed autosense retrieval.
|
|
|
|
|
*
|
|
|
|
|
* Zero any sense not transferred by the
|
|
|
|
|
* device. The SCSI spec mandates that any
|
|
|
|
|
* untransfered data should be assumed to be
|
|
|
|
|
* zero. Complete the 'bounce' of sense information
|
|
|
|
|
* through buffers accessible via bus-space by
|
|
|
|
|
* copying it into the clients csio.
|
|
|
|
|
*/
|
|
|
|
|
memset(&ccb->csio.sense_data, 0, sizeof(ccb->csio.sense_data));
|
|
|
|
|
memcpy(&ccb->csio.sense_data,
|
2001-01-09 00:40:38 +00:00
|
|
|
ahc_get_sense_buf(ahc, scb),
|
2002-04-24 16:58:51 +00:00
|
|
|
(ahc_le32toh(scb->sg_list->len) & AHC_SG_LEN_MASK)
|
2000-09-16 20:02:28 +00:00
|
|
|
- ccb->csio.sense_resid);
|
|
|
|
|
scb->io_ctx->ccb_h.status |= CAM_AUTOSNS_VALID;
|
|
|
|
|
}
|
|
|
|
|
ccb->ccb_h.status &= ~CAM_SIM_QUEUED;
|
|
|
|
|
ahc_free_scb(ahc, scb);
|
|
|
|
|
xpt_done(ccb);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static void
|
|
|
|
|
ahc_action(struct cam_sim *sim, union ccb *ccb)
|
|
|
|
|
{
|
|
|
|
|
struct ahc_softc *ahc;
|
2001-03-29 00:36:35 +00:00
|
|
|
struct ahc_tmode_lstate *lstate;
|
2000-09-16 20:02:28 +00:00
|
|
|
u_int target_id;
|
|
|
|
|
u_int our_id;
|
|
|
|
|
long s;
|
|
|
|
|
|
|
|
|
|
CAM_DEBUG(ccb->ccb_h.path, CAM_DEBUG_TRACE, ("ahc_action\n"));
|
|
|
|
|
|
|
|
|
|
ahc = (struct ahc_softc *)cam_sim_softc(sim);
|
|
|
|
|
|
|
|
|
|
target_id = ccb->ccb_h.target_id;
|
|
|
|
|
our_id = SIM_SCSI_ID(ahc, sim);
|
|
|
|
|
|
|
|
|
|
switch (ccb->ccb_h.func_code) {
|
|
|
|
|
/* Common cases first */
|
|
|
|
|
case XPT_ACCEPT_TARGET_IO: /* Accept Host Target Mode CDB */
|
|
|
|
|
case XPT_CONT_TARGET_IO:/* Continue Host Target I/O Connection*/
|
|
|
|
|
{
|
2001-03-29 00:36:35 +00:00
|
|
|
struct ahc_tmode_tstate *tstate;
|
2000-09-16 20:02:28 +00:00
|
|
|
cam_status status;
|
|
|
|
|
|
|
|
|
|
status = ahc_find_tmode_devs(ahc, sim, ccb, &tstate,
|
|
|
|
|
&lstate, TRUE);
|
|
|
|
|
|
|
|
|
|
if (status != CAM_REQ_CMP) {
|
|
|
|
|
if (ccb->ccb_h.func_code == XPT_CONT_TARGET_IO) {
|
|
|
|
|
/* Response from the black hole device */
|
|
|
|
|
tstate = NULL;
|
|
|
|
|
lstate = ahc->black_hole;
|
|
|
|
|
} else {
|
|
|
|
|
ccb->ccb_h.status = status;
|
|
|
|
|
xpt_done(ccb);
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
if (ccb->ccb_h.func_code == XPT_ACCEPT_TARGET_IO) {
|
|
|
|
|
|
|
|
|
|
ahc_lock(ahc, &s);
|
|
|
|
|
SLIST_INSERT_HEAD(&lstate->accept_tios, &ccb->ccb_h,
|
|
|
|
|
sim_links.sle);
|
|
|
|
|
ccb->ccb_h.status = CAM_REQ_INPROG;
|
|
|
|
|
if ((ahc->flags & AHC_TQINFIFO_BLOCKED) != 0)
|
|
|
|
|
ahc_run_tqinfifo(ahc, /*paused*/FALSE);
|
|
|
|
|
ahc_unlock(ahc, &s);
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* The target_id represents the target we attempt to
|
|
|
|
|
* select. In target mode, this is the initiator of
|
|
|
|
|
* the original command.
|
|
|
|
|
*/
|
|
|
|
|
our_id = target_id;
|
|
|
|
|
target_id = ccb->csio.init_id;
|
|
|
|
|
/* FALLTHROUGH */
|
|
|
|
|
}
|
|
|
|
|
case XPT_SCSI_IO: /* Execute the requested I/O operation */
|
|
|
|
|
case XPT_RESET_DEV: /* Bus Device Reset the specified SCSI device */
|
|
|
|
|
{
|
2000-10-06 04:01:06 +00:00
|
|
|
struct scb *scb;
|
|
|
|
|
struct hardware_scb *hscb;
|
2000-09-16 20:02:28 +00:00
|
|
|
|
aic7xxx.c:
Filter incoming transfer negotiation requests to ensure they
never exceed the settings specified by the user.
In restart sequencer attempt to deal with a bug in the aic7895.
If a third party reset occurs at just the right time, the
stack register can lock up. When restarting the sequencer
after handling the SCSI reset, poke SEQADDR1 before resting
the sequencers program counter.
When something strange happens, dump the card's transaction
state via ahc_dump_card_state(). This should aid in debugging.
Handle request sense transactions via the QINFIFO instead of
attaching them to the waiting queue directly. The waiting
queue consumes card SCB resources and, in the pathological case
of every target on the bus beating our selection attemps and
issuing a check condition, could have caused us to run out
of SCBs. I have never seen this happen, and only early
cards with 3 or 4 SCBs had any real chance of ever getting
into this state.
Add additional sequencer interrupt codes to support firmware
diagnostics. The diagnostic code is enabled with the
AHC_DEBUG_SEQUENCER kernel option.
Make it possible to switch into and out of target mode on
the fly. The card comes up by default as an initiator but
will switch into target mode as soon as an enable lun operation
is performed. As always, target mode behavior is gated
by the AHC_TMODE_ENABLE kernel option so most users will
not be affected by this change.
In ahc_update_target_msg_request(), also issue a new
request if the ppr_options have changed.
Never issue a PPR as a target. It is forbidden by the spec.
Correct a bug in ahc_parse_msg() that prevented us from
responding to PPR messages as a target.
Mark SCBs that are on the untagged queue with a flag instead
of checking several fields in the SCB to see if the SCB should
be on the queue. This makes it easier for things like automatic
request sense requests to be queued without touching the
untagged queues even though they are untagged requests.
When dealing with ignore wide residue messages that occur
in the middle of a transfer, reset HADDR, not SHADDR for
non-ultra2 chips. Although SHADDR is where the firmware
fetches the ending transfer address for a save data pointers
request, it is readonly. Setting HADDR has the side effect
of also updating SHADDR.
Cleanup the output of ahc_dump_card_state() by nulling out the
free scb list in the non-paging case. The free list is only
used if we must page SCBs.
Correct the transmission of cdbs > 12 bytes in length. When
swapping HSCBs prior to notifing the sequencer of the new
transaction, the bus address pointer for the cdb must also
be recalculated to reflect its new location. We now defer
the calculation of the cdb address until just before queing
it to the card.
When pulling transfer negotiation settings out of scratch
ram, convert 5MHz/clock doubled settings to 10MHz.
Add a new function ahc_qinfifo_requeue_tail() for use by
error recovery actions and auto-request sense operations.
These operations always occur when the sequencer is paused,
so we can avoid the extra expense incurred in the normal
SCB queue method.
Use the BMOV instruction for all single byte moves on
controllers that support it. The bmov instruction is
twice as fast as an AND with an immediate of 0xFF as
is used on older controllers.
Correct a few bugs in ahc_dump_card_state(). If we have
hardware assisted queue registers, use them to get the
sequencer's idea of the head of the queue. When enumerating
the untagged queue, it helps to use the correct index for
the queue.
aic7xxx.h:
Indicate via a feature flag, which controllers can take
on both the target and the initiator role at the same time.
Add the AHC_SEQUENCER_DEBUG flag.
Add the SCB_CDB32_PTR flag used for dealing with cdbs
with lengths between 13 and 32 bytes.
Add new prototypes.
aic7xxx.reg:
Allow the SCSIBUSL register to be written to. This is
required to fix a selection timeout problem on the 7892/99.
Cleanup the sequencer interrupt codes so that all debugging
codes are grouped at the end of the list.
Correct the definition of the ULTRA_ENB and DISC_DSB locations
in scratch ram. This prevented the driver from properly honoring
these settings when no serial eeprom was available.
Remove an unused sequencer flag.
aic7xxx.seq:
Just before a potential select-out, clear the SCSIBUSL
register. Occasionally, during a selection timeout, the
contents of the register may be presented on the bus,
causing much confusion.
Add sequencer diagnostic code to detect software and or
hardware bugs. The code attempts to verify most list
operations so any corruption is caught before it occurs.
We also track information about why a particular reconnection
request was rejected.
Don't clobber the digital REQ/ACK filter setting in SXFRCTL0
when clearing the channel.
Fix a target mode bug that would cause us to return busy
status instead of queue full in respnse to a tagged transaction.
Cleanup the overrun case. It turns out that by simply
butting the chip in bitbucket mode, it will ack any
bytes until the phase changes. This drasticaly simplifies
things.
Prior to leaving the data phase, make sure that the S/G
preload queue is empty.
Remove code to place a request sense request on the waiting
queue. This is all handled by the kernel now.
Change the semantics of "findSCB". In the past, findSCB
ensured that a freshly paged in SCB appeared on the disconnected
list. The problem with this is that there is no guarantee that
the paged in SCB is for a disconnected transation. We now
defer any list manipulation to the caller who usually discards
the SCB via the free list.
Inline some busy target table operations.
Add a critical section to protect adding an SCB to
the disconnected list.
aic7xxx_freebsd.c:
Handle changes in the transfer negotiation setting API
to filter incoming requests. No filtering is necessary
for "goal" requests from the XPT.
Set the SCB_CDB32_PTR flag when queing a transaction with
a large cdb.
In ahc_timeout, only take action if the active SCB is
the timedout SCB. This deals with the case of two
transactions to the same device with different timeout
values.
Use ahc_qinfifo_requeu_tail() instead of home grown
version.
aic7xxx_inline.h:
Honor SCB_CDB32_PTR when queuing a new request.
aic7xxx_pci.c:
Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
|
|
|
if ((ahc->flags & AHC_INITIATORROLE) == 0
|
|
|
|
|
&& (ccb->ccb_h.func_code == XPT_SCSI_IO
|
|
|
|
|
|| ccb->ccb_h.func_code == XPT_RESET_DEV)) {
|
|
|
|
|
ccb->ccb_h.status = CAM_PROVIDE_FAIL;
|
|
|
|
|
xpt_done(ccb);
|
2001-03-29 00:36:35 +00:00
|
|
|
return;
|
aic7xxx.c:
Filter incoming transfer negotiation requests to ensure they
never exceed the settings specified by the user.
In restart sequencer attempt to deal with a bug in the aic7895.
If a third party reset occurs at just the right time, the
stack register can lock up. When restarting the sequencer
after handling the SCSI reset, poke SEQADDR1 before resting
the sequencers program counter.
When something strange happens, dump the card's transaction
state via ahc_dump_card_state(). This should aid in debugging.
Handle request sense transactions via the QINFIFO instead of
attaching them to the waiting queue directly. The waiting
queue consumes card SCB resources and, in the pathological case
of every target on the bus beating our selection attemps and
issuing a check condition, could have caused us to run out
of SCBs. I have never seen this happen, and only early
cards with 3 or 4 SCBs had any real chance of ever getting
into this state.
Add additional sequencer interrupt codes to support firmware
diagnostics. The diagnostic code is enabled with the
AHC_DEBUG_SEQUENCER kernel option.
Make it possible to switch into and out of target mode on
the fly. The card comes up by default as an initiator but
will switch into target mode as soon as an enable lun operation
is performed. As always, target mode behavior is gated
by the AHC_TMODE_ENABLE kernel option so most users will
not be affected by this change.
In ahc_update_target_msg_request(), also issue a new
request if the ppr_options have changed.
Never issue a PPR as a target. It is forbidden by the spec.
Correct a bug in ahc_parse_msg() that prevented us from
responding to PPR messages as a target.
Mark SCBs that are on the untagged queue with a flag instead
of checking several fields in the SCB to see if the SCB should
be on the queue. This makes it easier for things like automatic
request sense requests to be queued without touching the
untagged queues even though they are untagged requests.
When dealing with ignore wide residue messages that occur
in the middle of a transfer, reset HADDR, not SHADDR for
non-ultra2 chips. Although SHADDR is where the firmware
fetches the ending transfer address for a save data pointers
request, it is readonly. Setting HADDR has the side effect
of also updating SHADDR.
Cleanup the output of ahc_dump_card_state() by nulling out the
free scb list in the non-paging case. The free list is only
used if we must page SCBs.
Correct the transmission of cdbs > 12 bytes in length. When
swapping HSCBs prior to notifing the sequencer of the new
transaction, the bus address pointer for the cdb must also
be recalculated to reflect its new location. We now defer
the calculation of the cdb address until just before queing
it to the card.
When pulling transfer negotiation settings out of scratch
ram, convert 5MHz/clock doubled settings to 10MHz.
Add a new function ahc_qinfifo_requeue_tail() for use by
error recovery actions and auto-request sense operations.
These operations always occur when the sequencer is paused,
so we can avoid the extra expense incurred in the normal
SCB queue method.
Use the BMOV instruction for all single byte moves on
controllers that support it. The bmov instruction is
twice as fast as an AND with an immediate of 0xFF as
is used on older controllers.
Correct a few bugs in ahc_dump_card_state(). If we have
hardware assisted queue registers, use them to get the
sequencer's idea of the head of the queue. When enumerating
the untagged queue, it helps to use the correct index for
the queue.
aic7xxx.h:
Indicate via a feature flag, which controllers can take
on both the target and the initiator role at the same time.
Add the AHC_SEQUENCER_DEBUG flag.
Add the SCB_CDB32_PTR flag used for dealing with cdbs
with lengths between 13 and 32 bytes.
Add new prototypes.
aic7xxx.reg:
Allow the SCSIBUSL register to be written to. This is
required to fix a selection timeout problem on the 7892/99.
Cleanup the sequencer interrupt codes so that all debugging
codes are grouped at the end of the list.
Correct the definition of the ULTRA_ENB and DISC_DSB locations
in scratch ram. This prevented the driver from properly honoring
these settings when no serial eeprom was available.
Remove an unused sequencer flag.
aic7xxx.seq:
Just before a potential select-out, clear the SCSIBUSL
register. Occasionally, during a selection timeout, the
contents of the register may be presented on the bus,
causing much confusion.
Add sequencer diagnostic code to detect software and or
hardware bugs. The code attempts to verify most list
operations so any corruption is caught before it occurs.
We also track information about why a particular reconnection
request was rejected.
Don't clobber the digital REQ/ACK filter setting in SXFRCTL0
when clearing the channel.
Fix a target mode bug that would cause us to return busy
status instead of queue full in respnse to a tagged transaction.
Cleanup the overrun case. It turns out that by simply
butting the chip in bitbucket mode, it will ack any
bytes until the phase changes. This drasticaly simplifies
things.
Prior to leaving the data phase, make sure that the S/G
preload queue is empty.
Remove code to place a request sense request on the waiting
queue. This is all handled by the kernel now.
Change the semantics of "findSCB". In the past, findSCB
ensured that a freshly paged in SCB appeared on the disconnected
list. The problem with this is that there is no guarantee that
the paged in SCB is for a disconnected transation. We now
defer any list manipulation to the caller who usually discards
the SCB via the free list.
Inline some busy target table operations.
Add a critical section to protect adding an SCB to
the disconnected list.
aic7xxx_freebsd.c:
Handle changes in the transfer negotiation setting API
to filter incoming requests. No filtering is necessary
for "goal" requests from the XPT.
Set the SCB_CDB32_PTR flag when queing a transaction with
a large cdb.
In ahc_timeout, only take action if the active SCB is
the timedout SCB. This deals with the case of two
transactions to the same device with different timeout
values.
Use ahc_qinfifo_requeu_tail() instead of home grown
version.
aic7xxx_inline.h:
Honor SCB_CDB32_PTR when queuing a new request.
aic7xxx_pci.c:
Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
|
|
|
}
|
|
|
|
|
|
2000-09-16 20:02:28 +00:00
|
|
|
/*
|
|
|
|
|
* get an scb to use.
|
|
|
|
|
*/
|
2000-10-06 04:01:06 +00:00
|
|
|
ahc_lock(ahc, &s);
|
2000-09-16 20:02:28 +00:00
|
|
|
if ((scb = ahc_get_scb(ahc)) == NULL) {
|
|
|
|
|
|
2000-10-09 19:54:54 +00:00
|
|
|
xpt_freeze_simq(sim, /*count*/1);
|
2000-09-16 20:02:28 +00:00
|
|
|
ahc->flags |= AHC_RESOURCE_SHORTAGE;
|
|
|
|
|
ahc_unlock(ahc, &s);
|
2000-10-01 20:56:44 +00:00
|
|
|
ccb->ccb_h.status = CAM_REQUEUE_REQ;
|
2000-09-16 20:02:28 +00:00
|
|
|
xpt_done(ccb);
|
|
|
|
|
return;
|
|
|
|
|
}
|
2000-10-06 04:01:06 +00:00
|
|
|
ahc_unlock(ahc, &s);
|
2000-09-16 20:02:28 +00:00
|
|
|
|
|
|
|
|
hscb = scb->hscb;
|
|
|
|
|
|
|
|
|
|
CAM_DEBUG(ccb->ccb_h.path, CAM_DEBUG_SUBTRACE,
|
|
|
|
|
("start scb(%p)\n", scb));
|
|
|
|
|
scb->io_ctx = ccb;
|
|
|
|
|
/*
|
|
|
|
|
* So we can find the SCB when an abort is requested
|
|
|
|
|
*/
|
|
|
|
|
ccb->ccb_h.ccb_scb_ptr = scb;
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Put all the arguments for the xfer in the scb
|
|
|
|
|
*/
|
|
|
|
|
hscb->control = 0;
|
|
|
|
|
hscb->scsiid = BUILD_SCSIID(ahc, sim, target_id, our_id);
|
|
|
|
|
hscb->lun = ccb->ccb_h.target_lun;
|
|
|
|
|
if (ccb->ccb_h.func_code == XPT_RESET_DEV) {
|
|
|
|
|
hscb->cdb_len = 0;
|
|
|
|
|
scb->flags |= SCB_DEVICE_RESET;
|
|
|
|
|
hscb->control |= MK_MESSAGE;
|
|
|
|
|
ahc_execute_scb(scb, NULL, 0, 0);
|
|
|
|
|
} else {
|
|
|
|
|
if (ccb->ccb_h.func_code == XPT_CONT_TARGET_IO) {
|
|
|
|
|
struct target_data *tdata;
|
|
|
|
|
|
|
|
|
|
tdata = &hscb->shared_data.tdata;
|
2001-03-29 00:36:35 +00:00
|
|
|
if (ahc->pending_device == lstate)
|
2000-09-16 20:02:28 +00:00
|
|
|
scb->flags |= SCB_TARGET_IMMEDIATE;
|
|
|
|
|
hscb->control |= TARGET_SCB;
|
|
|
|
|
tdata->target_phases = IDENTIFY_SEEN;
|
|
|
|
|
if ((ccb->ccb_h.flags & CAM_SEND_STATUS) != 0) {
|
|
|
|
|
tdata->target_phases |= SPHASE_PENDING;
|
|
|
|
|
tdata->scsi_status =
|
|
|
|
|
ccb->csio.scsi_status;
|
|
|
|
|
}
|
2001-03-29 00:36:35 +00:00
|
|
|
if (ccb->ccb_h.flags & CAM_DIS_DISCONNECT)
|
|
|
|
|
tdata->target_phases |= NO_DISCONNECT;
|
|
|
|
|
|
2000-09-16 20:02:28 +00:00
|
|
|
tdata->initiator_tag = ccb->csio.tag_id;
|
|
|
|
|
}
|
|
|
|
|
if (ccb->ccb_h.flags & CAM_TAG_ACTION_VALID)
|
|
|
|
|
hscb->control |= ccb->csio.tag_action;
|
|
|
|
|
|
|
|
|
|
ahc_setup_data(ahc, sim, &ccb->csio, scb);
|
|
|
|
|
}
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
case XPT_NOTIFY_ACK:
|
|
|
|
|
case XPT_IMMED_NOTIFY:
|
|
|
|
|
{
|
2001-03-29 00:36:35 +00:00
|
|
|
struct ahc_tmode_tstate *tstate;
|
|
|
|
|
struct ahc_tmode_lstate *lstate;
|
2000-09-16 20:02:28 +00:00
|
|
|
cam_status status;
|
|
|
|
|
|
|
|
|
|
status = ahc_find_tmode_devs(ahc, sim, ccb, &tstate,
|
|
|
|
|
&lstate, TRUE);
|
|
|
|
|
|
|
|
|
|
if (status != CAM_REQ_CMP) {
|
|
|
|
|
ccb->ccb_h.status = status;
|
|
|
|
|
xpt_done(ccb);
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
SLIST_INSERT_HEAD(&lstate->immed_notifies, &ccb->ccb_h,
|
|
|
|
|
sim_links.sle);
|
|
|
|
|
ccb->ccb_h.status = CAM_REQ_INPROG;
|
|
|
|
|
ahc_send_lstate_events(ahc, lstate);
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
case XPT_EN_LUN: /* Enable LUN as a target */
|
|
|
|
|
ahc_handle_en_lun(ahc, sim, ccb);
|
|
|
|
|
xpt_done(ccb);
|
|
|
|
|
break;
|
|
|
|
|
case XPT_ABORT: /* Abort the specified CCB */
|
|
|
|
|
{
|
|
|
|
|
ahc_abort_ccb(ahc, sim, ccb);
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
case XPT_SET_TRAN_SETTINGS:
|
|
|
|
|
{
|
|
|
|
|
#ifdef AHC_NEW_TRAN_SETTINGS
|
|
|
|
|
struct ahc_devinfo devinfo;
|
|
|
|
|
struct ccb_trans_settings *cts;
|
|
|
|
|
struct ccb_trans_settings_scsi *scsi;
|
|
|
|
|
struct ccb_trans_settings_spi *spi;
|
|
|
|
|
struct ahc_initiator_tinfo *tinfo;
|
2001-03-29 00:36:35 +00:00
|
|
|
struct ahc_tmode_tstate *tstate;
|
2000-09-16 20:02:28 +00:00
|
|
|
uint16_t *discenable;
|
|
|
|
|
uint16_t *tagenable;
|
|
|
|
|
u_int update_type;
|
|
|
|
|
|
|
|
|
|
cts = &ccb->cts;
|
|
|
|
|
scsi = &cts->proto_specific.scsi;
|
|
|
|
|
spi = &cts->xport_specific.spi;
|
|
|
|
|
ahc_compile_devinfo(&devinfo, SIM_SCSI_ID(ahc, sim),
|
|
|
|
|
cts->ccb_h.target_id,
|
|
|
|
|
cts->ccb_h.target_lun,
|
|
|
|
|
SIM_CHANNEL(ahc, sim),
|
|
|
|
|
ROLE_UNKNOWN);
|
|
|
|
|
tinfo = ahc_fetch_transinfo(ahc, devinfo.channel,
|
|
|
|
|
devinfo.our_scsiid,
|
|
|
|
|
devinfo.target, &tstate);
|
|
|
|
|
update_type = 0;
|
|
|
|
|
if (cts->type == CTS_TYPE_CURRENT_SETTINGS) {
|
|
|
|
|
update_type |= AHC_TRANS_GOAL;
|
|
|
|
|
discenable = &tstate->discenable;
|
|
|
|
|
tagenable = &tstate->tagenable;
|
2001-05-15 19:41:12 +00:00
|
|
|
tinfo->curr.protocol_version =
|
2000-09-16 20:02:28 +00:00
|
|
|
cts->protocol_version;
|
2001-05-15 19:41:12 +00:00
|
|
|
tinfo->curr.transport_version =
|
2000-09-16 20:02:28 +00:00
|
|
|
cts->transport_version;
|
|
|
|
|
tinfo->goal.protocol_version =
|
|
|
|
|
cts->protocol_version;
|
|
|
|
|
tinfo->goal.transport_version =
|
|
|
|
|
cts->transport_version;
|
|
|
|
|
} else if (cts->type == CTS_TYPE_USER_SETTINGS) {
|
|
|
|
|
update_type |= AHC_TRANS_USER;
|
|
|
|
|
discenable = &ahc->user_discenable;
|
|
|
|
|
tagenable = &ahc->user_tagenable;
|
|
|
|
|
tinfo->user.protocol_version =
|
|
|
|
|
cts->protocol_version;
|
|
|
|
|
tinfo->user.transport_version =
|
|
|
|
|
cts->transport_version;
|
|
|
|
|
} else {
|
|
|
|
|
ccb->ccb_h.status = CAM_REQ_INVALID;
|
|
|
|
|
xpt_done(ccb);
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
ahc_lock(ahc, &s);
|
|
|
|
|
|
|
|
|
|
if ((spi->valid & CTS_SPI_VALID_DISC) != 0) {
|
|
|
|
|
if ((spi->flags & CTS_SPI_FLAGS_DISC_ENB) != 0)
|
|
|
|
|
*discenable |= devinfo.target_mask;
|
|
|
|
|
else
|
|
|
|
|
*discenable &= ~devinfo.target_mask;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if ((scsi->valid & CTS_SCSI_VALID_TQ) != 0) {
|
|
|
|
|
if ((scsi->flags & CTS_SCSI_FLAGS_TAG_ENB) != 0)
|
|
|
|
|
*tagenable |= devinfo.target_mask;
|
|
|
|
|
else
|
|
|
|
|
*tagenable &= ~devinfo.target_mask;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if ((spi->valid & CTS_SPI_VALID_BUS_WIDTH) != 0) {
|
aic7xxx.c:
Filter incoming transfer negotiation requests to ensure they
never exceed the settings specified by the user.
In restart sequencer attempt to deal with a bug in the aic7895.
If a third party reset occurs at just the right time, the
stack register can lock up. When restarting the sequencer
after handling the SCSI reset, poke SEQADDR1 before resting
the sequencers program counter.
When something strange happens, dump the card's transaction
state via ahc_dump_card_state(). This should aid in debugging.
Handle request sense transactions via the QINFIFO instead of
attaching them to the waiting queue directly. The waiting
queue consumes card SCB resources and, in the pathological case
of every target on the bus beating our selection attemps and
issuing a check condition, could have caused us to run out
of SCBs. I have never seen this happen, and only early
cards with 3 or 4 SCBs had any real chance of ever getting
into this state.
Add additional sequencer interrupt codes to support firmware
diagnostics. The diagnostic code is enabled with the
AHC_DEBUG_SEQUENCER kernel option.
Make it possible to switch into and out of target mode on
the fly. The card comes up by default as an initiator but
will switch into target mode as soon as an enable lun operation
is performed. As always, target mode behavior is gated
by the AHC_TMODE_ENABLE kernel option so most users will
not be affected by this change.
In ahc_update_target_msg_request(), also issue a new
request if the ppr_options have changed.
Never issue a PPR as a target. It is forbidden by the spec.
Correct a bug in ahc_parse_msg() that prevented us from
responding to PPR messages as a target.
Mark SCBs that are on the untagged queue with a flag instead
of checking several fields in the SCB to see if the SCB should
be on the queue. This makes it easier for things like automatic
request sense requests to be queued without touching the
untagged queues even though they are untagged requests.
When dealing with ignore wide residue messages that occur
in the middle of a transfer, reset HADDR, not SHADDR for
non-ultra2 chips. Although SHADDR is where the firmware
fetches the ending transfer address for a save data pointers
request, it is readonly. Setting HADDR has the side effect
of also updating SHADDR.
Cleanup the output of ahc_dump_card_state() by nulling out the
free scb list in the non-paging case. The free list is only
used if we must page SCBs.
Correct the transmission of cdbs > 12 bytes in length. When
swapping HSCBs prior to notifing the sequencer of the new
transaction, the bus address pointer for the cdb must also
be recalculated to reflect its new location. We now defer
the calculation of the cdb address until just before queing
it to the card.
When pulling transfer negotiation settings out of scratch
ram, convert 5MHz/clock doubled settings to 10MHz.
Add a new function ahc_qinfifo_requeue_tail() for use by
error recovery actions and auto-request sense operations.
These operations always occur when the sequencer is paused,
so we can avoid the extra expense incurred in the normal
SCB queue method.
Use the BMOV instruction for all single byte moves on
controllers that support it. The bmov instruction is
twice as fast as an AND with an immediate of 0xFF as
is used on older controllers.
Correct a few bugs in ahc_dump_card_state(). If we have
hardware assisted queue registers, use them to get the
sequencer's idea of the head of the queue. When enumerating
the untagged queue, it helps to use the correct index for
the queue.
aic7xxx.h:
Indicate via a feature flag, which controllers can take
on both the target and the initiator role at the same time.
Add the AHC_SEQUENCER_DEBUG flag.
Add the SCB_CDB32_PTR flag used for dealing with cdbs
with lengths between 13 and 32 bytes.
Add new prototypes.
aic7xxx.reg:
Allow the SCSIBUSL register to be written to. This is
required to fix a selection timeout problem on the 7892/99.
Cleanup the sequencer interrupt codes so that all debugging
codes are grouped at the end of the list.
Correct the definition of the ULTRA_ENB and DISC_DSB locations
in scratch ram. This prevented the driver from properly honoring
these settings when no serial eeprom was available.
Remove an unused sequencer flag.
aic7xxx.seq:
Just before a potential select-out, clear the SCSIBUSL
register. Occasionally, during a selection timeout, the
contents of the register may be presented on the bus,
causing much confusion.
Add sequencer diagnostic code to detect software and or
hardware bugs. The code attempts to verify most list
operations so any corruption is caught before it occurs.
We also track information about why a particular reconnection
request was rejected.
Don't clobber the digital REQ/ACK filter setting in SXFRCTL0
when clearing the channel.
Fix a target mode bug that would cause us to return busy
status instead of queue full in respnse to a tagged transaction.
Cleanup the overrun case. It turns out that by simply
butting the chip in bitbucket mode, it will ack any
bytes until the phase changes. This drasticaly simplifies
things.
Prior to leaving the data phase, make sure that the S/G
preload queue is empty.
Remove code to place a request sense request on the waiting
queue. This is all handled by the kernel now.
Change the semantics of "findSCB". In the past, findSCB
ensured that a freshly paged in SCB appeared on the disconnected
list. The problem with this is that there is no guarantee that
the paged in SCB is for a disconnected transation. We now
defer any list manipulation to the caller who usually discards
the SCB via the free list.
Inline some busy target table operations.
Add a critical section to protect adding an SCB to
the disconnected list.
aic7xxx_freebsd.c:
Handle changes in the transfer negotiation setting API
to filter incoming requests. No filtering is necessary
for "goal" requests from the XPT.
Set the SCB_CDB32_PTR flag when queing a transaction with
a large cdb.
In ahc_timeout, only take action if the active SCB is
the timedout SCB. This deals with the case of two
transactions to the same device with different timeout
values.
Use ahc_qinfifo_requeu_tail() instead of home grown
version.
aic7xxx_inline.h:
Honor SCB_CDB32_PTR when queuing a new request.
aic7xxx_pci.c:
Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
|
|
|
ahc_validate_width(ahc, /*tinfo limit*/NULL,
|
|
|
|
|
&spi->bus_width, ROLE_UNKNOWN);
|
2000-09-16 20:02:28 +00:00
|
|
|
ahc_set_width(ahc, &devinfo, spi->bus_width,
|
|
|
|
|
update_type, /*paused*/FALSE);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if ((spi->valid & CTS_SPI_VALID_PPR_OPTIONS) == 0) {
|
|
|
|
|
if (update_type == AHC_TRANS_USER)
|
|
|
|
|
spi->ppr_options = tinfo->user.ppr_options;
|
|
|
|
|
else
|
|
|
|
|
spi->ppr_options = tinfo->goal.ppr_options;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if ((spi->valid & CTS_SPI_VALID_SYNC_OFFSET) == 0) {
|
|
|
|
|
if (update_type == AHC_TRANS_USER)
|
|
|
|
|
spi->sync_offset = tinfo->user.offset;
|
|
|
|
|
else
|
|
|
|
|
spi->sync_offset = tinfo->goal.offset;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if ((spi->valid & CTS_SPI_VALID_SYNC_RATE) == 0) {
|
|
|
|
|
if (update_type == AHC_TRANS_USER)
|
|
|
|
|
spi->sync_period = tinfo->user.period;
|
|
|
|
|
else
|
|
|
|
|
spi->sync_period = tinfo->goal.period;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (((spi->valid & CTS_SPI_VALID_SYNC_RATE) != 0)
|
|
|
|
|
|| ((spi->valid & CTS_SPI_VALID_SYNC_OFFSET) != 0)) {
|
|
|
|
|
struct ahc_syncrate *syncrate;
|
|
|
|
|
u_int maxsync;
|
|
|
|
|
|
|
|
|
|
if ((ahc->features & AHC_ULTRA2) != 0)
|
|
|
|
|
maxsync = AHC_SYNCRATE_DT;
|
|
|
|
|
else if ((ahc->features & AHC_ULTRA) != 0)
|
|
|
|
|
maxsync = AHC_SYNCRATE_ULTRA;
|
|
|
|
|
else
|
|
|
|
|
maxsync = AHC_SYNCRATE_FAST;
|
|
|
|
|
|
2002-04-24 16:58:51 +00:00
|
|
|
if (spi->bus_width != MSG_EXT_WDTR_BUS_16_BIT)
|
|
|
|
|
spi->ppr_options &= ~MSG_EXT_PPR_DT_REQ;
|
|
|
|
|
|
2000-09-16 20:02:28 +00:00
|
|
|
syncrate = ahc_find_syncrate(ahc, &spi->sync_period,
|
|
|
|
|
&spi->ppr_options,
|
|
|
|
|
maxsync);
|
aic7xxx.c:
Filter incoming transfer negotiation requests to ensure they
never exceed the settings specified by the user.
In restart sequencer attempt to deal with a bug in the aic7895.
If a third party reset occurs at just the right time, the
stack register can lock up. When restarting the sequencer
after handling the SCSI reset, poke SEQADDR1 before resting
the sequencers program counter.
When something strange happens, dump the card's transaction
state via ahc_dump_card_state(). This should aid in debugging.
Handle request sense transactions via the QINFIFO instead of
attaching them to the waiting queue directly. The waiting
queue consumes card SCB resources and, in the pathological case
of every target on the bus beating our selection attemps and
issuing a check condition, could have caused us to run out
of SCBs. I have never seen this happen, and only early
cards with 3 or 4 SCBs had any real chance of ever getting
into this state.
Add additional sequencer interrupt codes to support firmware
diagnostics. The diagnostic code is enabled with the
AHC_DEBUG_SEQUENCER kernel option.
Make it possible to switch into and out of target mode on
the fly. The card comes up by default as an initiator but
will switch into target mode as soon as an enable lun operation
is performed. As always, target mode behavior is gated
by the AHC_TMODE_ENABLE kernel option so most users will
not be affected by this change.
In ahc_update_target_msg_request(), also issue a new
request if the ppr_options have changed.
Never issue a PPR as a target. It is forbidden by the spec.
Correct a bug in ahc_parse_msg() that prevented us from
responding to PPR messages as a target.
Mark SCBs that are on the untagged queue with a flag instead
of checking several fields in the SCB to see if the SCB should
be on the queue. This makes it easier for things like automatic
request sense requests to be queued without touching the
untagged queues even though they are untagged requests.
When dealing with ignore wide residue messages that occur
in the middle of a transfer, reset HADDR, not SHADDR for
non-ultra2 chips. Although SHADDR is where the firmware
fetches the ending transfer address for a save data pointers
request, it is readonly. Setting HADDR has the side effect
of also updating SHADDR.
Cleanup the output of ahc_dump_card_state() by nulling out the
free scb list in the non-paging case. The free list is only
used if we must page SCBs.
Correct the transmission of cdbs > 12 bytes in length. When
swapping HSCBs prior to notifing the sequencer of the new
transaction, the bus address pointer for the cdb must also
be recalculated to reflect its new location. We now defer
the calculation of the cdb address until just before queing
it to the card.
When pulling transfer negotiation settings out of scratch
ram, convert 5MHz/clock doubled settings to 10MHz.
Add a new function ahc_qinfifo_requeue_tail() for use by
error recovery actions and auto-request sense operations.
These operations always occur when the sequencer is paused,
so we can avoid the extra expense incurred in the normal
SCB queue method.
Use the BMOV instruction for all single byte moves on
controllers that support it. The bmov instruction is
twice as fast as an AND with an immediate of 0xFF as
is used on older controllers.
Correct a few bugs in ahc_dump_card_state(). If we have
hardware assisted queue registers, use them to get the
sequencer's idea of the head of the queue. When enumerating
the untagged queue, it helps to use the correct index for
the queue.
aic7xxx.h:
Indicate via a feature flag, which controllers can take
on both the target and the initiator role at the same time.
Add the AHC_SEQUENCER_DEBUG flag.
Add the SCB_CDB32_PTR flag used for dealing with cdbs
with lengths between 13 and 32 bytes.
Add new prototypes.
aic7xxx.reg:
Allow the SCSIBUSL register to be written to. This is
required to fix a selection timeout problem on the 7892/99.
Cleanup the sequencer interrupt codes so that all debugging
codes are grouped at the end of the list.
Correct the definition of the ULTRA_ENB and DISC_DSB locations
in scratch ram. This prevented the driver from properly honoring
these settings when no serial eeprom was available.
Remove an unused sequencer flag.
aic7xxx.seq:
Just before a potential select-out, clear the SCSIBUSL
register. Occasionally, during a selection timeout, the
contents of the register may be presented on the bus,
causing much confusion.
Add sequencer diagnostic code to detect software and or
hardware bugs. The code attempts to verify most list
operations so any corruption is caught before it occurs.
We also track information about why a particular reconnection
request was rejected.
Don't clobber the digital REQ/ACK filter setting in SXFRCTL0
when clearing the channel.
Fix a target mode bug that would cause us to return busy
status instead of queue full in respnse to a tagged transaction.
Cleanup the overrun case. It turns out that by simply
butting the chip in bitbucket mode, it will ack any
bytes until the phase changes. This drasticaly simplifies
things.
Prior to leaving the data phase, make sure that the S/G
preload queue is empty.
Remove code to place a request sense request on the waiting
queue. This is all handled by the kernel now.
Change the semantics of "findSCB". In the past, findSCB
ensured that a freshly paged in SCB appeared on the disconnected
list. The problem with this is that there is no guarantee that
the paged in SCB is for a disconnected transation. We now
defer any list manipulation to the caller who usually discards
the SCB via the free list.
Inline some busy target table operations.
Add a critical section to protect adding an SCB to
the disconnected list.
aic7xxx_freebsd.c:
Handle changes in the transfer negotiation setting API
to filter incoming requests. No filtering is necessary
for "goal" requests from the XPT.
Set the SCB_CDB32_PTR flag when queing a transaction with
a large cdb.
In ahc_timeout, only take action if the active SCB is
the timedout SCB. This deals with the case of two
transactions to the same device with different timeout
values.
Use ahc_qinfifo_requeu_tail() instead of home grown
version.
aic7xxx_inline.h:
Honor SCB_CDB32_PTR when queuing a new request.
aic7xxx_pci.c:
Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
|
|
|
ahc_validate_offset(ahc, /*tinfo limit*/NULL,
|
|
|
|
|
syncrate, &spi->sync_offset,
|
|
|
|
|
spi->bus_width, ROLE_UNKNOWN);
|
2000-09-16 20:02:28 +00:00
|
|
|
|
|
|
|
|
/* We use a period of 0 to represent async */
|
|
|
|
|
if (spi->sync_offset == 0) {
|
|
|
|
|
spi->sync_period = 0;
|
|
|
|
|
spi->ppr_options = 0;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
ahc_set_syncrate(ahc, &devinfo, syncrate,
|
|
|
|
|
spi->sync_period, spi->sync_offset,
|
|
|
|
|
spi->ppr_options, update_type,
|
|
|
|
|
/*paused*/FALSE);
|
|
|
|
|
}
|
|
|
|
|
ahc_unlock(ahc, &s);
|
|
|
|
|
ccb->ccb_h.status = CAM_REQ_CMP;
|
|
|
|
|
xpt_done(ccb);
|
|
|
|
|
#else
|
|
|
|
|
struct ahc_devinfo devinfo;
|
|
|
|
|
struct ccb_trans_settings *cts;
|
|
|
|
|
struct ahc_initiator_tinfo *tinfo;
|
2001-03-29 00:36:35 +00:00
|
|
|
struct ahc_tmode_tstate *tstate;
|
2000-09-16 20:02:28 +00:00
|
|
|
uint16_t *discenable;
|
|
|
|
|
uint16_t *tagenable;
|
|
|
|
|
u_int update_type;
|
|
|
|
|
long s;
|
|
|
|
|
|
|
|
|
|
cts = &ccb->cts;
|
|
|
|
|
ahc_compile_devinfo(&devinfo, SIM_SCSI_ID(ahc, sim),
|
|
|
|
|
cts->ccb_h.target_id,
|
|
|
|
|
cts->ccb_h.target_lun,
|
|
|
|
|
SIM_CHANNEL(ahc, sim),
|
|
|
|
|
ROLE_UNKNOWN);
|
|
|
|
|
tinfo = ahc_fetch_transinfo(ahc, devinfo.channel,
|
|
|
|
|
devinfo.our_scsiid,
|
|
|
|
|
devinfo.target, &tstate);
|
|
|
|
|
update_type = 0;
|
|
|
|
|
if ((cts->flags & CCB_TRANS_CURRENT_SETTINGS) != 0) {
|
|
|
|
|
update_type |= AHC_TRANS_GOAL;
|
|
|
|
|
discenable = &tstate->discenable;
|
|
|
|
|
tagenable = &tstate->tagenable;
|
|
|
|
|
} else if ((cts->flags & CCB_TRANS_USER_SETTINGS) != 0) {
|
|
|
|
|
update_type |= AHC_TRANS_USER;
|
|
|
|
|
discenable = &ahc->user_discenable;
|
|
|
|
|
tagenable = &ahc->user_tagenable;
|
|
|
|
|
} else {
|
|
|
|
|
ccb->ccb_h.status = CAM_REQ_INVALID;
|
|
|
|
|
xpt_done(ccb);
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
ahc_lock(ahc, &s);
|
|
|
|
|
|
|
|
|
|
if ((cts->valid & CCB_TRANS_DISC_VALID) != 0) {
|
|
|
|
|
if ((cts->flags & CCB_TRANS_DISC_ENB) != 0)
|
|
|
|
|
*discenable |= devinfo.target_mask;
|
|
|
|
|
else
|
|
|
|
|
*discenable &= ~devinfo.target_mask;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if ((cts->valid & CCB_TRANS_TQ_VALID) != 0) {
|
|
|
|
|
if ((cts->flags & CCB_TRANS_TAG_ENB) != 0)
|
|
|
|
|
*tagenable |= devinfo.target_mask;
|
|
|
|
|
else
|
|
|
|
|
*tagenable &= ~devinfo.target_mask;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if ((cts->valid & CCB_TRANS_BUS_WIDTH_VALID) != 0) {
|
aic7xxx.c:
Filter incoming transfer negotiation requests to ensure they
never exceed the settings specified by the user.
In restart sequencer attempt to deal with a bug in the aic7895.
If a third party reset occurs at just the right time, the
stack register can lock up. When restarting the sequencer
after handling the SCSI reset, poke SEQADDR1 before resting
the sequencers program counter.
When something strange happens, dump the card's transaction
state via ahc_dump_card_state(). This should aid in debugging.
Handle request sense transactions via the QINFIFO instead of
attaching them to the waiting queue directly. The waiting
queue consumes card SCB resources and, in the pathological case
of every target on the bus beating our selection attemps and
issuing a check condition, could have caused us to run out
of SCBs. I have never seen this happen, and only early
cards with 3 or 4 SCBs had any real chance of ever getting
into this state.
Add additional sequencer interrupt codes to support firmware
diagnostics. The diagnostic code is enabled with the
AHC_DEBUG_SEQUENCER kernel option.
Make it possible to switch into and out of target mode on
the fly. The card comes up by default as an initiator but
will switch into target mode as soon as an enable lun operation
is performed. As always, target mode behavior is gated
by the AHC_TMODE_ENABLE kernel option so most users will
not be affected by this change.
In ahc_update_target_msg_request(), also issue a new
request if the ppr_options have changed.
Never issue a PPR as a target. It is forbidden by the spec.
Correct a bug in ahc_parse_msg() that prevented us from
responding to PPR messages as a target.
Mark SCBs that are on the untagged queue with a flag instead
of checking several fields in the SCB to see if the SCB should
be on the queue. This makes it easier for things like automatic
request sense requests to be queued without touching the
untagged queues even though they are untagged requests.
When dealing with ignore wide residue messages that occur
in the middle of a transfer, reset HADDR, not SHADDR for
non-ultra2 chips. Although SHADDR is where the firmware
fetches the ending transfer address for a save data pointers
request, it is readonly. Setting HADDR has the side effect
of also updating SHADDR.
Cleanup the output of ahc_dump_card_state() by nulling out the
free scb list in the non-paging case. The free list is only
used if we must page SCBs.
Correct the transmission of cdbs > 12 bytes in length. When
swapping HSCBs prior to notifing the sequencer of the new
transaction, the bus address pointer for the cdb must also
be recalculated to reflect its new location. We now defer
the calculation of the cdb address until just before queing
it to the card.
When pulling transfer negotiation settings out of scratch
ram, convert 5MHz/clock doubled settings to 10MHz.
Add a new function ahc_qinfifo_requeue_tail() for use by
error recovery actions and auto-request sense operations.
These operations always occur when the sequencer is paused,
so we can avoid the extra expense incurred in the normal
SCB queue method.
Use the BMOV instruction for all single byte moves on
controllers that support it. The bmov instruction is
twice as fast as an AND with an immediate of 0xFF as
is used on older controllers.
Correct a few bugs in ahc_dump_card_state(). If we have
hardware assisted queue registers, use them to get the
sequencer's idea of the head of the queue. When enumerating
the untagged queue, it helps to use the correct index for
the queue.
aic7xxx.h:
Indicate via a feature flag, which controllers can take
on both the target and the initiator role at the same time.
Add the AHC_SEQUENCER_DEBUG flag.
Add the SCB_CDB32_PTR flag used for dealing with cdbs
with lengths between 13 and 32 bytes.
Add new prototypes.
aic7xxx.reg:
Allow the SCSIBUSL register to be written to. This is
required to fix a selection timeout problem on the 7892/99.
Cleanup the sequencer interrupt codes so that all debugging
codes are grouped at the end of the list.
Correct the definition of the ULTRA_ENB and DISC_DSB locations
in scratch ram. This prevented the driver from properly honoring
these settings when no serial eeprom was available.
Remove an unused sequencer flag.
aic7xxx.seq:
Just before a potential select-out, clear the SCSIBUSL
register. Occasionally, during a selection timeout, the
contents of the register may be presented on the bus,
causing much confusion.
Add sequencer diagnostic code to detect software and or
hardware bugs. The code attempts to verify most list
operations so any corruption is caught before it occurs.
We also track information about why a particular reconnection
request was rejected.
Don't clobber the digital REQ/ACK filter setting in SXFRCTL0
when clearing the channel.
Fix a target mode bug that would cause us to return busy
status instead of queue full in respnse to a tagged transaction.
Cleanup the overrun case. It turns out that by simply
butting the chip in bitbucket mode, it will ack any
bytes until the phase changes. This drasticaly simplifies
things.
Prior to leaving the data phase, make sure that the S/G
preload queue is empty.
Remove code to place a request sense request on the waiting
queue. This is all handled by the kernel now.
Change the semantics of "findSCB". In the past, findSCB
ensured that a freshly paged in SCB appeared on the disconnected
list. The problem with this is that there is no guarantee that
the paged in SCB is for a disconnected transation. We now
defer any list manipulation to the caller who usually discards
the SCB via the free list.
Inline some busy target table operations.
Add a critical section to protect adding an SCB to
the disconnected list.
aic7xxx_freebsd.c:
Handle changes in the transfer negotiation setting API
to filter incoming requests. No filtering is necessary
for "goal" requests from the XPT.
Set the SCB_CDB32_PTR flag when queing a transaction with
a large cdb.
In ahc_timeout, only take action if the active SCB is
the timedout SCB. This deals with the case of two
transactions to the same device with different timeout
values.
Use ahc_qinfifo_requeu_tail() instead of home grown
version.
aic7xxx_inline.h:
Honor SCB_CDB32_PTR when queuing a new request.
aic7xxx_pci.c:
Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
|
|
|
ahc_validate_width(ahc, /*tinfo limit*/NULL,
|
|
|
|
|
&cts->bus_width, ROLE_UNKNOWN);
|
2000-09-16 20:02:28 +00:00
|
|
|
ahc_set_width(ahc, &devinfo, cts->bus_width,
|
|
|
|
|
update_type, /*paused*/FALSE);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if ((cts->valid & CCB_TRANS_SYNC_OFFSET_VALID) == 0) {
|
|
|
|
|
if (update_type == AHC_TRANS_USER)
|
|
|
|
|
cts->sync_offset = tinfo->user.offset;
|
|
|
|
|
else
|
|
|
|
|
cts->sync_offset = tinfo->goal.offset;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if ((cts->valid & CCB_TRANS_SYNC_RATE_VALID) == 0) {
|
|
|
|
|
if (update_type == AHC_TRANS_USER)
|
|
|
|
|
cts->sync_period = tinfo->user.period;
|
|
|
|
|
else
|
|
|
|
|
cts->sync_period = tinfo->goal.period;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (((cts->valid & CCB_TRANS_SYNC_RATE_VALID) != 0)
|
|
|
|
|
|| ((cts->valid & CCB_TRANS_SYNC_OFFSET_VALID) != 0)) {
|
|
|
|
|
struct ahc_syncrate *syncrate;
|
|
|
|
|
u_int ppr_options;
|
|
|
|
|
u_int maxsync;
|
|
|
|
|
|
|
|
|
|
if ((ahc->features & AHC_ULTRA2) != 0)
|
|
|
|
|
maxsync = AHC_SYNCRATE_DT;
|
|
|
|
|
else if ((ahc->features & AHC_ULTRA) != 0)
|
|
|
|
|
maxsync = AHC_SYNCRATE_ULTRA;
|
|
|
|
|
else
|
|
|
|
|
maxsync = AHC_SYNCRATE_FAST;
|
|
|
|
|
|
|
|
|
|
ppr_options = 0;
|
2002-04-24 16:58:51 +00:00
|
|
|
if (cts->sync_period <= 9
|
|
|
|
|
&& cts->bus_width == MSG_EXT_WDTR_BUS_16_BIT)
|
2000-09-16 20:02:28 +00:00
|
|
|
ppr_options = MSG_EXT_PPR_DT_REQ;
|
|
|
|
|
|
|
|
|
|
syncrate = ahc_find_syncrate(ahc, &cts->sync_period,
|
|
|
|
|
&ppr_options,
|
|
|
|
|
maxsync);
|
aic7xxx.c:
Filter incoming transfer negotiation requests to ensure they
never exceed the settings specified by the user.
In restart sequencer attempt to deal with a bug in the aic7895.
If a third party reset occurs at just the right time, the
stack register can lock up. When restarting the sequencer
after handling the SCSI reset, poke SEQADDR1 before resting
the sequencers program counter.
When something strange happens, dump the card's transaction
state via ahc_dump_card_state(). This should aid in debugging.
Handle request sense transactions via the QINFIFO instead of
attaching them to the waiting queue directly. The waiting
queue consumes card SCB resources and, in the pathological case
of every target on the bus beating our selection attemps and
issuing a check condition, could have caused us to run out
of SCBs. I have never seen this happen, and only early
cards with 3 or 4 SCBs had any real chance of ever getting
into this state.
Add additional sequencer interrupt codes to support firmware
diagnostics. The diagnostic code is enabled with the
AHC_DEBUG_SEQUENCER kernel option.
Make it possible to switch into and out of target mode on
the fly. The card comes up by default as an initiator but
will switch into target mode as soon as an enable lun operation
is performed. As always, target mode behavior is gated
by the AHC_TMODE_ENABLE kernel option so most users will
not be affected by this change.
In ahc_update_target_msg_request(), also issue a new
request if the ppr_options have changed.
Never issue a PPR as a target. It is forbidden by the spec.
Correct a bug in ahc_parse_msg() that prevented us from
responding to PPR messages as a target.
Mark SCBs that are on the untagged queue with a flag instead
of checking several fields in the SCB to see if the SCB should
be on the queue. This makes it easier for things like automatic
request sense requests to be queued without touching the
untagged queues even though they are untagged requests.
When dealing with ignore wide residue messages that occur
in the middle of a transfer, reset HADDR, not SHADDR for
non-ultra2 chips. Although SHADDR is where the firmware
fetches the ending transfer address for a save data pointers
request, it is readonly. Setting HADDR has the side effect
of also updating SHADDR.
Cleanup the output of ahc_dump_card_state() by nulling out the
free scb list in the non-paging case. The free list is only
used if we must page SCBs.
Correct the transmission of cdbs > 12 bytes in length. When
swapping HSCBs prior to notifing the sequencer of the new
transaction, the bus address pointer for the cdb must also
be recalculated to reflect its new location. We now defer
the calculation of the cdb address until just before queing
it to the card.
When pulling transfer negotiation settings out of scratch
ram, convert 5MHz/clock doubled settings to 10MHz.
Add a new function ahc_qinfifo_requeue_tail() for use by
error recovery actions and auto-request sense operations.
These operations always occur when the sequencer is paused,
so we can avoid the extra expense incurred in the normal
SCB queue method.
Use the BMOV instruction for all single byte moves on
controllers that support it. The bmov instruction is
twice as fast as an AND with an immediate of 0xFF as
is used on older controllers.
Correct a few bugs in ahc_dump_card_state(). If we have
hardware assisted queue registers, use them to get the
sequencer's idea of the head of the queue. When enumerating
the untagged queue, it helps to use the correct index for
the queue.
aic7xxx.h:
Indicate via a feature flag, which controllers can take
on both the target and the initiator role at the same time.
Add the AHC_SEQUENCER_DEBUG flag.
Add the SCB_CDB32_PTR flag used for dealing with cdbs
with lengths between 13 and 32 bytes.
Add new prototypes.
aic7xxx.reg:
Allow the SCSIBUSL register to be written to. This is
required to fix a selection timeout problem on the 7892/99.
Cleanup the sequencer interrupt codes so that all debugging
codes are grouped at the end of the list.
Correct the definition of the ULTRA_ENB and DISC_DSB locations
in scratch ram. This prevented the driver from properly honoring
these settings when no serial eeprom was available.
Remove an unused sequencer flag.
aic7xxx.seq:
Just before a potential select-out, clear the SCSIBUSL
register. Occasionally, during a selection timeout, the
contents of the register may be presented on the bus,
causing much confusion.
Add sequencer diagnostic code to detect software and or
hardware bugs. The code attempts to verify most list
operations so any corruption is caught before it occurs.
We also track information about why a particular reconnection
request was rejected.
Don't clobber the digital REQ/ACK filter setting in SXFRCTL0
when clearing the channel.
Fix a target mode bug that would cause us to return busy
status instead of queue full in respnse to a tagged transaction.
Cleanup the overrun case. It turns out that by simply
butting the chip in bitbucket mode, it will ack any
bytes until the phase changes. This drasticaly simplifies
things.
Prior to leaving the data phase, make sure that the S/G
preload queue is empty.
Remove code to place a request sense request on the waiting
queue. This is all handled by the kernel now.
Change the semantics of "findSCB". In the past, findSCB
ensured that a freshly paged in SCB appeared on the disconnected
list. The problem with this is that there is no guarantee that
the paged in SCB is for a disconnected transation. We now
defer any list manipulation to the caller who usually discards
the SCB via the free list.
Inline some busy target table operations.
Add a critical section to protect adding an SCB to
the disconnected list.
aic7xxx_freebsd.c:
Handle changes in the transfer negotiation setting API
to filter incoming requests. No filtering is necessary
for "goal" requests from the XPT.
Set the SCB_CDB32_PTR flag when queing a transaction with
a large cdb.
In ahc_timeout, only take action if the active SCB is
the timedout SCB. This deals with the case of two
transactions to the same device with different timeout
values.
Use ahc_qinfifo_requeu_tail() instead of home grown
version.
aic7xxx_inline.h:
Honor SCB_CDB32_PTR when queuing a new request.
aic7xxx_pci.c:
Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
|
|
|
ahc_validate_offset(ahc, /*tinfo limit*/NULL,
|
|
|
|
|
syncrate, &cts->sync_offset,
|
|
|
|
|
MSG_EXT_WDTR_BUS_8_BIT,
|
|
|
|
|
ROLE_UNKNOWN);
|
2000-09-16 20:02:28 +00:00
|
|
|
|
|
|
|
|
/* We use a period of 0 to represent async */
|
|
|
|
|
if (cts->sync_offset == 0) {
|
|
|
|
|
cts->sync_period = 0;
|
|
|
|
|
ppr_options = 0;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (ppr_options == MSG_EXT_PPR_DT_REQ
|
|
|
|
|
&& tinfo->user.transport_version >= 3) {
|
|
|
|
|
tinfo->goal.transport_version =
|
|
|
|
|
tinfo->user.transport_version;
|
2001-05-15 19:41:12 +00:00
|
|
|
tinfo->curr.transport_version =
|
2000-09-16 20:02:28 +00:00
|
|
|
tinfo->user.transport_version;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
ahc_set_syncrate(ahc, &devinfo, syncrate,
|
|
|
|
|
cts->sync_period, cts->sync_offset,
|
|
|
|
|
ppr_options, update_type,
|
|
|
|
|
/*paused*/FALSE);
|
|
|
|
|
}
|
|
|
|
|
ahc_unlock(ahc, &s);
|
|
|
|
|
ccb->ccb_h.status = CAM_REQ_CMP;
|
|
|
|
|
xpt_done(ccb);
|
|
|
|
|
#endif
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
case XPT_GET_TRAN_SETTINGS:
|
|
|
|
|
/* Get default/user set transfer settings for the target */
|
|
|
|
|
{
|
|
|
|
|
|
|
|
|
|
ahc_lock(ahc, &s);
|
|
|
|
|
ahc_get_tran_settings(ahc, SIM_SCSI_ID(ahc, sim),
|
|
|
|
|
SIM_CHANNEL(ahc, sim), &ccb->cts);
|
|
|
|
|
ahc_unlock(ahc, &s);
|
|
|
|
|
xpt_done(ccb);
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
case XPT_CALC_GEOMETRY:
|
|
|
|
|
{
|
|
|
|
|
struct ccb_calc_geometry *ccg;
|
|
|
|
|
uint32_t size_mb;
|
|
|
|
|
uint32_t secs_per_cylinder;
|
|
|
|
|
int extended;
|
|
|
|
|
|
|
|
|
|
ccg = &ccb->ccg;
|
|
|
|
|
size_mb = ccg->volume_size
|
|
|
|
|
/ ((1024L * 1024L) / ccg->block_size);
|
|
|
|
|
extended = SIM_IS_SCSIBUS_B(ahc, sim)
|
|
|
|
|
? ahc->flags & AHC_EXTENDED_TRANS_B
|
|
|
|
|
: ahc->flags & AHC_EXTENDED_TRANS_A;
|
|
|
|
|
|
|
|
|
|
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 found;
|
|
|
|
|
|
|
|
|
|
ahc_lock(ahc, &s);
|
|
|
|
|
found = ahc_reset_channel(ahc, SIM_CHANNEL(ahc, sim),
|
|
|
|
|
/*initiate reset*/TRUE);
|
|
|
|
|
ahc_unlock(ahc, &s);
|
|
|
|
|
if (bootverbose) {
|
|
|
|
|
xpt_print_path(SIM_PATH(ahc, sim));
|
|
|
|
|
printf("SCSI bus reset delivered. "
|
|
|
|
|
"%d SCBs aborted.\n", found);
|
|
|
|
|
}
|
|
|
|
|
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;
|
|
|
|
|
if ((ahc->features & AHC_WIDE) != 0)
|
|
|
|
|
cpi->hba_inquiry |= PI_WIDE_16;
|
aic7xxx.c:
Filter incoming transfer negotiation requests to ensure they
never exceed the settings specified by the user.
In restart sequencer attempt to deal with a bug in the aic7895.
If a third party reset occurs at just the right time, the
stack register can lock up. When restarting the sequencer
after handling the SCSI reset, poke SEQADDR1 before resting
the sequencers program counter.
When something strange happens, dump the card's transaction
state via ahc_dump_card_state(). This should aid in debugging.
Handle request sense transactions via the QINFIFO instead of
attaching them to the waiting queue directly. The waiting
queue consumes card SCB resources and, in the pathological case
of every target on the bus beating our selection attemps and
issuing a check condition, could have caused us to run out
of SCBs. I have never seen this happen, and only early
cards with 3 or 4 SCBs had any real chance of ever getting
into this state.
Add additional sequencer interrupt codes to support firmware
diagnostics. The diagnostic code is enabled with the
AHC_DEBUG_SEQUENCER kernel option.
Make it possible to switch into and out of target mode on
the fly. The card comes up by default as an initiator but
will switch into target mode as soon as an enable lun operation
is performed. As always, target mode behavior is gated
by the AHC_TMODE_ENABLE kernel option so most users will
not be affected by this change.
In ahc_update_target_msg_request(), also issue a new
request if the ppr_options have changed.
Never issue a PPR as a target. It is forbidden by the spec.
Correct a bug in ahc_parse_msg() that prevented us from
responding to PPR messages as a target.
Mark SCBs that are on the untagged queue with a flag instead
of checking several fields in the SCB to see if the SCB should
be on the queue. This makes it easier for things like automatic
request sense requests to be queued without touching the
untagged queues even though they are untagged requests.
When dealing with ignore wide residue messages that occur
in the middle of a transfer, reset HADDR, not SHADDR for
non-ultra2 chips. Although SHADDR is where the firmware
fetches the ending transfer address for a save data pointers
request, it is readonly. Setting HADDR has the side effect
of also updating SHADDR.
Cleanup the output of ahc_dump_card_state() by nulling out the
free scb list in the non-paging case. The free list is only
used if we must page SCBs.
Correct the transmission of cdbs > 12 bytes in length. When
swapping HSCBs prior to notifing the sequencer of the new
transaction, the bus address pointer for the cdb must also
be recalculated to reflect its new location. We now defer
the calculation of the cdb address until just before queing
it to the card.
When pulling transfer negotiation settings out of scratch
ram, convert 5MHz/clock doubled settings to 10MHz.
Add a new function ahc_qinfifo_requeue_tail() for use by
error recovery actions and auto-request sense operations.
These operations always occur when the sequencer is paused,
so we can avoid the extra expense incurred in the normal
SCB queue method.
Use the BMOV instruction for all single byte moves on
controllers that support it. The bmov instruction is
twice as fast as an AND with an immediate of 0xFF as
is used on older controllers.
Correct a few bugs in ahc_dump_card_state(). If we have
hardware assisted queue registers, use them to get the
sequencer's idea of the head of the queue. When enumerating
the untagged queue, it helps to use the correct index for
the queue.
aic7xxx.h:
Indicate via a feature flag, which controllers can take
on both the target and the initiator role at the same time.
Add the AHC_SEQUENCER_DEBUG flag.
Add the SCB_CDB32_PTR flag used for dealing with cdbs
with lengths between 13 and 32 bytes.
Add new prototypes.
aic7xxx.reg:
Allow the SCSIBUSL register to be written to. This is
required to fix a selection timeout problem on the 7892/99.
Cleanup the sequencer interrupt codes so that all debugging
codes are grouped at the end of the list.
Correct the definition of the ULTRA_ENB and DISC_DSB locations
in scratch ram. This prevented the driver from properly honoring
these settings when no serial eeprom was available.
Remove an unused sequencer flag.
aic7xxx.seq:
Just before a potential select-out, clear the SCSIBUSL
register. Occasionally, during a selection timeout, the
contents of the register may be presented on the bus,
causing much confusion.
Add sequencer diagnostic code to detect software and or
hardware bugs. The code attempts to verify most list
operations so any corruption is caught before it occurs.
We also track information about why a particular reconnection
request was rejected.
Don't clobber the digital REQ/ACK filter setting in SXFRCTL0
when clearing the channel.
Fix a target mode bug that would cause us to return busy
status instead of queue full in respnse to a tagged transaction.
Cleanup the overrun case. It turns out that by simply
butting the chip in bitbucket mode, it will ack any
bytes until the phase changes. This drasticaly simplifies
things.
Prior to leaving the data phase, make sure that the S/G
preload queue is empty.
Remove code to place a request sense request on the waiting
queue. This is all handled by the kernel now.
Change the semantics of "findSCB". In the past, findSCB
ensured that a freshly paged in SCB appeared on the disconnected
list. The problem with this is that there is no guarantee that
the paged in SCB is for a disconnected transation. We now
defer any list manipulation to the caller who usually discards
the SCB via the free list.
Inline some busy target table operations.
Add a critical section to protect adding an SCB to
the disconnected list.
aic7xxx_freebsd.c:
Handle changes in the transfer negotiation setting API
to filter incoming requests. No filtering is necessary
for "goal" requests from the XPT.
Set the SCB_CDB32_PTR flag when queing a transaction with
a large cdb.
In ahc_timeout, only take action if the active SCB is
the timedout SCB. This deals with the case of two
transactions to the same device with different timeout
values.
Use ahc_qinfifo_requeu_tail() instead of home grown
version.
aic7xxx_inline.h:
Honor SCB_CDB32_PTR when queuing a new request.
aic7xxx_pci.c:
Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
|
|
|
if ((ahc->features & AHC_TARGETMODE) != 0) {
|
2000-09-16 20:02:28 +00:00
|
|
|
cpi->target_sprt = PIT_PROCESSOR
|
|
|
|
|
| PIT_DISCONNECT
|
|
|
|
|
| PIT_TERM_IO;
|
|
|
|
|
} else {
|
|
|
|
|
cpi->target_sprt = 0;
|
|
|
|
|
}
|
aic7xxx.c:
Filter incoming transfer negotiation requests to ensure they
never exceed the settings specified by the user.
In restart sequencer attempt to deal with a bug in the aic7895.
If a third party reset occurs at just the right time, the
stack register can lock up. When restarting the sequencer
after handling the SCSI reset, poke SEQADDR1 before resting
the sequencers program counter.
When something strange happens, dump the card's transaction
state via ahc_dump_card_state(). This should aid in debugging.
Handle request sense transactions via the QINFIFO instead of
attaching them to the waiting queue directly. The waiting
queue consumes card SCB resources and, in the pathological case
of every target on the bus beating our selection attemps and
issuing a check condition, could have caused us to run out
of SCBs. I have never seen this happen, and only early
cards with 3 or 4 SCBs had any real chance of ever getting
into this state.
Add additional sequencer interrupt codes to support firmware
diagnostics. The diagnostic code is enabled with the
AHC_DEBUG_SEQUENCER kernel option.
Make it possible to switch into and out of target mode on
the fly. The card comes up by default as an initiator but
will switch into target mode as soon as an enable lun operation
is performed. As always, target mode behavior is gated
by the AHC_TMODE_ENABLE kernel option so most users will
not be affected by this change.
In ahc_update_target_msg_request(), also issue a new
request if the ppr_options have changed.
Never issue a PPR as a target. It is forbidden by the spec.
Correct a bug in ahc_parse_msg() that prevented us from
responding to PPR messages as a target.
Mark SCBs that are on the untagged queue with a flag instead
of checking several fields in the SCB to see if the SCB should
be on the queue. This makes it easier for things like automatic
request sense requests to be queued without touching the
untagged queues even though they are untagged requests.
When dealing with ignore wide residue messages that occur
in the middle of a transfer, reset HADDR, not SHADDR for
non-ultra2 chips. Although SHADDR is where the firmware
fetches the ending transfer address for a save data pointers
request, it is readonly. Setting HADDR has the side effect
of also updating SHADDR.
Cleanup the output of ahc_dump_card_state() by nulling out the
free scb list in the non-paging case. The free list is only
used if we must page SCBs.
Correct the transmission of cdbs > 12 bytes in length. When
swapping HSCBs prior to notifing the sequencer of the new
transaction, the bus address pointer for the cdb must also
be recalculated to reflect its new location. We now defer
the calculation of the cdb address until just before queing
it to the card.
When pulling transfer negotiation settings out of scratch
ram, convert 5MHz/clock doubled settings to 10MHz.
Add a new function ahc_qinfifo_requeue_tail() for use by
error recovery actions and auto-request sense operations.
These operations always occur when the sequencer is paused,
so we can avoid the extra expense incurred in the normal
SCB queue method.
Use the BMOV instruction for all single byte moves on
controllers that support it. The bmov instruction is
twice as fast as an AND with an immediate of 0xFF as
is used on older controllers.
Correct a few bugs in ahc_dump_card_state(). If we have
hardware assisted queue registers, use them to get the
sequencer's idea of the head of the queue. When enumerating
the untagged queue, it helps to use the correct index for
the queue.
aic7xxx.h:
Indicate via a feature flag, which controllers can take
on both the target and the initiator role at the same time.
Add the AHC_SEQUENCER_DEBUG flag.
Add the SCB_CDB32_PTR flag used for dealing with cdbs
with lengths between 13 and 32 bytes.
Add new prototypes.
aic7xxx.reg:
Allow the SCSIBUSL register to be written to. This is
required to fix a selection timeout problem on the 7892/99.
Cleanup the sequencer interrupt codes so that all debugging
codes are grouped at the end of the list.
Correct the definition of the ULTRA_ENB and DISC_DSB locations
in scratch ram. This prevented the driver from properly honoring
these settings when no serial eeprom was available.
Remove an unused sequencer flag.
aic7xxx.seq:
Just before a potential select-out, clear the SCSIBUSL
register. Occasionally, during a selection timeout, the
contents of the register may be presented on the bus,
causing much confusion.
Add sequencer diagnostic code to detect software and or
hardware bugs. The code attempts to verify most list
operations so any corruption is caught before it occurs.
We also track information about why a particular reconnection
request was rejected.
Don't clobber the digital REQ/ACK filter setting in SXFRCTL0
when clearing the channel.
Fix a target mode bug that would cause us to return busy
status instead of queue full in respnse to a tagged transaction.
Cleanup the overrun case. It turns out that by simply
butting the chip in bitbucket mode, it will ack any
bytes until the phase changes. This drasticaly simplifies
things.
Prior to leaving the data phase, make sure that the S/G
preload queue is empty.
Remove code to place a request sense request on the waiting
queue. This is all handled by the kernel now.
Change the semantics of "findSCB". In the past, findSCB
ensured that a freshly paged in SCB appeared on the disconnected
list. The problem with this is that there is no guarantee that
the paged in SCB is for a disconnected transation. We now
defer any list manipulation to the caller who usually discards
the SCB via the free list.
Inline some busy target table operations.
Add a critical section to protect adding an SCB to
the disconnected list.
aic7xxx_freebsd.c:
Handle changes in the transfer negotiation setting API
to filter incoming requests. No filtering is necessary
for "goal" requests from the XPT.
Set the SCB_CDB32_PTR flag when queing a transaction with
a large cdb.
In ahc_timeout, only take action if the active SCB is
the timedout SCB. This deals with the case of two
transactions to the same device with different timeout
values.
Use ahc_qinfifo_requeu_tail() instead of home grown
version.
aic7xxx_inline.h:
Honor SCB_CDB32_PTR when queuing a new request.
aic7xxx_pci.c:
Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
|
|
|
cpi->hba_misc = 0;
|
2000-09-16 20:02:28 +00:00
|
|
|
cpi->hba_eng_cnt = 0;
|
|
|
|
|
cpi->max_target = (ahc->features & AHC_WIDE) ? 15 : 7;
|
ahc_eisa.c:
ahc_pci.c:
Add detach support.
Make use of soft allocated on our behalf by newbus.
For PCI devices, disable the mapping type we aren't
using for extra protection from rogue code.
aic7xxx_93cx6.c:
aic7xxx_93cx6.h:
Sync perforce IDs.
aic7xxx_freebsd.c:
Capture the eventhandle returned by EVENTHANDER_REGISTER
so we can kill the handler off during detach.
Use AHC_* constants instead of hard coded numbers in a
few more places.
Test PPR option state when deciding to "really" negotiate
when the CAM_NEGOTIATE flag is passed in a CCB.
Make use of core "ahc_pause_and_flushwork" routine in our
timeout handler rather than re-inventing this code.
Cleanup all of our resources (really!) in ahc_platform_free().
We should be all set to become a module now.
Implement the core ahc_detach() routine shared by all of
the FreeBSD front-ends.
aic7xxx_freebsd.h:
Softc storage for our event handler.
Null implementation for the ahc_platform_flushwork() OSM
callback. FreeBSD doesn't need this as XPT callbacks are
safe from all contexts and are done directly in ahc_done().
aic7xxx_inline.h:
Implement new lazy interrupt scheme. To avoid an extra
PCI bus read, we first check our completion queues to
see if any work has completed. If work is available, we
assume that this is the source of the interrupt and skip
reading INTSTAT. Any remaining interrupt status will be
cleared by a second call to the interrupt handler should
the interrupt line still be asserted. This drops the
interrupt handler down to a single PCI bus read in the
common case of I/O completion. This is the same overhead
as in the not so distant past, but the extra sanity of
perforning a PCI read after clearing the command complete
interrupt and before running the completion queue to avoid
missing command complete interrupts added a cycle.
aic7xxx.c:
During initialization, be sure to initialize all scratch
ram locations before they are read to avoid parity errors.
In this case, we use a new function, ahc_unbusy_tcl() to
initialize the scratch ram busy target table.
Replace instances of ahc_index_busy_tcl() used to unbusy
a tcl without looking at the old value with ahc_unbusy_tcl().
Modify ahc_sent_msg so that it can find single byte messages.
ahc_sent_msg is now used to determine if a transfer negotiation
attempt resulted in a bus free.
Be more careful in filtering out only the SCSI interrupts
of interest in ahc_handle_scsiint.
Rearrange interrupt clearing code to ensure that at least
one PCI transaction occurrs after hitting CLRSINT1 and
writting to CLRINT. CLRSINT1 writes take a bit to
take effect, and the re-arrangement provides sufficient
delay to ensure the write to CLRINT is effective. The
old code might report a spurious interrupt on some "fast"
chipsets.
export ahc-update_target_msg_request for use by OSM code.
If a target does not respond to our ATN request, clear
it once we move to a non-message phase. This avoids
sending a MSG_NOOP in some later message out phase.
Use max lun and max target constants instead of
hard-coded values.
Use softc storage built into our device_t under FreeBSD.
Fix a bug in ahc_free() that caused us to delete
resources that were not allocated.
Clean up any tstate/lstate info in ahc_free().
Clear the powerdown state in ahc_reset() so that
registers can be accessed.
Add a preliminary function for pausing the chip and
processing any posted work.
Add a preliminary suspend and resume functions.
aic7xxx.h:
Limit the number of supported luns to 64. We don't
support information unit transfers, so this is the
maximum that makes sense for these chips.
Add a new flag AHC_ALL_INTERRUPTS that forces the
processing of all interrupt state in a single invokation
of ahc_intr(). When the flag is not set, we use the
lazy interrupt handling scheme.
Add data structures to store controller state while
we are suspended.
Use constants instead of hard coded values where appropriate.
Correct some harmless "unsigned/signed" conflicts.
aic7xxx.seq:
Only perform the SCSIBUSL fix on ULTRA2 or newer controllers.
Older controllers seem to be confused by this.
In target mode, ignore PHASEMIS during data phases.
This bit seems to be flakey on U160 controllers acting
in target mode.
aic7xxx_pci.c:
Add support for the 29160C CPCI adapter.
Add definitions for subvendor ID information
available for devices with the "9005" vendor id.
We currently use this information to determine
if a multi-function device doesn't have the second
channel hooked up on a board.
Add rudimentary power mode code so we can put the
controller into the D0 state. In the future this
will be an OSM callback so that in FreeBSD we don't
duplicate functionality provided by the PCI code.
The powerstate code was added after I'd completed
my regression tests on this code.
Only capture "left over BIOS state" if the POWRDN
setting is not set in HCNTRL.
In target mode, don't bother sending incremental
CRC data.
2000-12-20 01:11:37 +00:00
|
|
|
cpi->max_lun = AHC_NUM_LUNS - 1;
|
2000-09-16 20:02:28 +00:00
|
|
|
if (SIM_IS_SCSIBUS_B(ahc, sim)) {
|
|
|
|
|
cpi->initiator_id = ahc->our_id_b;
|
|
|
|
|
if ((ahc->flags & AHC_RESET_BUS_B) == 0)
|
|
|
|
|
cpi->hba_misc |= PIM_NOBUSRESET;
|
|
|
|
|
} else {
|
|
|
|
|
cpi->initiator_id = ahc->our_id;
|
|
|
|
|
if ((ahc->flags & AHC_RESET_BUS_A) == 0)
|
|
|
|
|
cpi->hba_misc |= PIM_NOBUSRESET;
|
|
|
|
|
}
|
|
|
|
|
cpi->bus_id = cam_sim_bus(sim);
|
|
|
|
|
cpi->base_transfer_speed = 3300;
|
|
|
|
|
strncpy(cpi->sim_vid, "FreeBSD", SIM_IDLEN);
|
|
|
|
|
strncpy(cpi->hba_vid, "Adaptec", HBA_IDLEN);
|
|
|
|
|
strncpy(cpi->dev_name, cam_sim_name(sim), DEV_IDLEN);
|
|
|
|
|
cpi->unit_number = cam_sim_unit(sim);
|
|
|
|
|
#ifdef AHC_NEW_TRAN_SETTINGS
|
|
|
|
|
cpi->protocol = PROTO_SCSI;
|
|
|
|
|
cpi->protocol_version = SCSI_REV_2;
|
|
|
|
|
cpi->transport = XPORT_SPI;
|
|
|
|
|
cpi->transport_version = 2;
|
|
|
|
|
cpi->xport_specific.spi.ppr_options = SID_SPI_CLOCK_ST;
|
|
|
|
|
if ((ahc->features & AHC_DT) != 0) {
|
|
|
|
|
cpi->transport_version = 3;
|
|
|
|
|
cpi->xport_specific.spi.ppr_options =
|
|
|
|
|
SID_SPI_CLOCK_DT_ST;
|
|
|
|
|
}
|
|
|
|
|
#endif
|
|
|
|
|
cpi->ccb_h.status = CAM_REQ_CMP;
|
|
|
|
|
xpt_done(ccb);
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
default:
|
aic7xxx.c:
Filter incoming transfer negotiation requests to ensure they
never exceed the settings specified by the user.
In restart sequencer attempt to deal with a bug in the aic7895.
If a third party reset occurs at just the right time, the
stack register can lock up. When restarting the sequencer
after handling the SCSI reset, poke SEQADDR1 before resting
the sequencers program counter.
When something strange happens, dump the card's transaction
state via ahc_dump_card_state(). This should aid in debugging.
Handle request sense transactions via the QINFIFO instead of
attaching them to the waiting queue directly. The waiting
queue consumes card SCB resources and, in the pathological case
of every target on the bus beating our selection attemps and
issuing a check condition, could have caused us to run out
of SCBs. I have never seen this happen, and only early
cards with 3 or 4 SCBs had any real chance of ever getting
into this state.
Add additional sequencer interrupt codes to support firmware
diagnostics. The diagnostic code is enabled with the
AHC_DEBUG_SEQUENCER kernel option.
Make it possible to switch into and out of target mode on
the fly. The card comes up by default as an initiator but
will switch into target mode as soon as an enable lun operation
is performed. As always, target mode behavior is gated
by the AHC_TMODE_ENABLE kernel option so most users will
not be affected by this change.
In ahc_update_target_msg_request(), also issue a new
request if the ppr_options have changed.
Never issue a PPR as a target. It is forbidden by the spec.
Correct a bug in ahc_parse_msg() that prevented us from
responding to PPR messages as a target.
Mark SCBs that are on the untagged queue with a flag instead
of checking several fields in the SCB to see if the SCB should
be on the queue. This makes it easier for things like automatic
request sense requests to be queued without touching the
untagged queues even though they are untagged requests.
When dealing with ignore wide residue messages that occur
in the middle of a transfer, reset HADDR, not SHADDR for
non-ultra2 chips. Although SHADDR is where the firmware
fetches the ending transfer address for a save data pointers
request, it is readonly. Setting HADDR has the side effect
of also updating SHADDR.
Cleanup the output of ahc_dump_card_state() by nulling out the
free scb list in the non-paging case. The free list is only
used if we must page SCBs.
Correct the transmission of cdbs > 12 bytes in length. When
swapping HSCBs prior to notifing the sequencer of the new
transaction, the bus address pointer for the cdb must also
be recalculated to reflect its new location. We now defer
the calculation of the cdb address until just before queing
it to the card.
When pulling transfer negotiation settings out of scratch
ram, convert 5MHz/clock doubled settings to 10MHz.
Add a new function ahc_qinfifo_requeue_tail() for use by
error recovery actions and auto-request sense operations.
These operations always occur when the sequencer is paused,
so we can avoid the extra expense incurred in the normal
SCB queue method.
Use the BMOV instruction for all single byte moves on
controllers that support it. The bmov instruction is
twice as fast as an AND with an immediate of 0xFF as
is used on older controllers.
Correct a few bugs in ahc_dump_card_state(). If we have
hardware assisted queue registers, use them to get the
sequencer's idea of the head of the queue. When enumerating
the untagged queue, it helps to use the correct index for
the queue.
aic7xxx.h:
Indicate via a feature flag, which controllers can take
on both the target and the initiator role at the same time.
Add the AHC_SEQUENCER_DEBUG flag.
Add the SCB_CDB32_PTR flag used for dealing with cdbs
with lengths between 13 and 32 bytes.
Add new prototypes.
aic7xxx.reg:
Allow the SCSIBUSL register to be written to. This is
required to fix a selection timeout problem on the 7892/99.
Cleanup the sequencer interrupt codes so that all debugging
codes are grouped at the end of the list.
Correct the definition of the ULTRA_ENB and DISC_DSB locations
in scratch ram. This prevented the driver from properly honoring
these settings when no serial eeprom was available.
Remove an unused sequencer flag.
aic7xxx.seq:
Just before a potential select-out, clear the SCSIBUSL
register. Occasionally, during a selection timeout, the
contents of the register may be presented on the bus,
causing much confusion.
Add sequencer diagnostic code to detect software and or
hardware bugs. The code attempts to verify most list
operations so any corruption is caught before it occurs.
We also track information about why a particular reconnection
request was rejected.
Don't clobber the digital REQ/ACK filter setting in SXFRCTL0
when clearing the channel.
Fix a target mode bug that would cause us to return busy
status instead of queue full in respnse to a tagged transaction.
Cleanup the overrun case. It turns out that by simply
butting the chip in bitbucket mode, it will ack any
bytes until the phase changes. This drasticaly simplifies
things.
Prior to leaving the data phase, make sure that the S/G
preload queue is empty.
Remove code to place a request sense request on the waiting
queue. This is all handled by the kernel now.
Change the semantics of "findSCB". In the past, findSCB
ensured that a freshly paged in SCB appeared on the disconnected
list. The problem with this is that there is no guarantee that
the paged in SCB is for a disconnected transation. We now
defer any list manipulation to the caller who usually discards
the SCB via the free list.
Inline some busy target table operations.
Add a critical section to protect adding an SCB to
the disconnected list.
aic7xxx_freebsd.c:
Handle changes in the transfer negotiation setting API
to filter incoming requests. No filtering is necessary
for "goal" requests from the XPT.
Set the SCB_CDB32_PTR flag when queing a transaction with
a large cdb.
In ahc_timeout, only take action if the active SCB is
the timedout SCB. This deals with the case of two
transactions to the same device with different timeout
values.
Use ahc_qinfifo_requeu_tail() instead of home grown
version.
aic7xxx_inline.h:
Honor SCB_CDB32_PTR when queuing a new request.
aic7xxx_pci.c:
Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
|
|
|
ccb->ccb_h.status = CAM_PROVIDE_FAIL;
|
2000-09-16 20:02:28 +00:00
|
|
|
xpt_done(ccb);
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static void
|
|
|
|
|
ahc_get_tran_settings(struct ahc_softc *ahc, int our_id, char channel,
|
|
|
|
|
struct ccb_trans_settings *cts)
|
|
|
|
|
{
|
|
|
|
|
#ifdef AHC_NEW_TRAN_SETTINGS
|
|
|
|
|
struct ahc_devinfo devinfo;
|
|
|
|
|
struct ccb_trans_settings_scsi *scsi;
|
|
|
|
|
struct ccb_trans_settings_spi *spi;
|
|
|
|
|
struct ahc_initiator_tinfo *targ_info;
|
2001-03-29 00:36:35 +00:00
|
|
|
struct ahc_tmode_tstate *tstate;
|
2000-09-16 20:02:28 +00:00
|
|
|
struct ahc_transinfo *tinfo;
|
|
|
|
|
|
|
|
|
|
scsi = &cts->proto_specific.scsi;
|
|
|
|
|
spi = &cts->xport_specific.spi;
|
|
|
|
|
ahc_compile_devinfo(&devinfo, our_id,
|
|
|
|
|
cts->ccb_h.target_id,
|
|
|
|
|
cts->ccb_h.target_lun,
|
|
|
|
|
channel, ROLE_UNKNOWN);
|
|
|
|
|
targ_info = ahc_fetch_transinfo(ahc, devinfo.channel,
|
|
|
|
|
devinfo.our_scsiid,
|
|
|
|
|
devinfo.target, &tstate);
|
|
|
|
|
|
|
|
|
|
if (cts->type == CTS_TYPE_CURRENT_SETTINGS)
|
2001-05-15 19:41:12 +00:00
|
|
|
tinfo = &targ_info->curr;
|
2000-09-16 20:02:28 +00:00
|
|
|
else
|
|
|
|
|
tinfo = &targ_info->user;
|
|
|
|
|
|
|
|
|
|
scsi->flags &= ~CTS_SCSI_FLAGS_TAG_ENB;
|
|
|
|
|
spi->flags &= ~CTS_SPI_FLAGS_DISC_ENB;
|
|
|
|
|
if (cts->type == CTS_TYPE_USER_SETTINGS) {
|
|
|
|
|
if ((ahc->user_discenable & devinfo.target_mask) != 0)
|
|
|
|
|
spi->flags |= CTS_SPI_FLAGS_DISC_ENB;
|
|
|
|
|
|
|
|
|
|
if ((ahc->user_tagenable & devinfo.target_mask) != 0)
|
|
|
|
|
scsi->flags |= CTS_SCSI_FLAGS_TAG_ENB;
|
|
|
|
|
} else {
|
|
|
|
|
if ((tstate->discenable & devinfo.target_mask) != 0)
|
|
|
|
|
spi->flags |= CTS_SPI_FLAGS_DISC_ENB;
|
|
|
|
|
|
|
|
|
|
if ((tstate->tagenable & devinfo.target_mask) != 0)
|
|
|
|
|
scsi->flags |= CTS_SCSI_FLAGS_TAG_ENB;
|
|
|
|
|
}
|
|
|
|
|
cts->protocol_version = tinfo->protocol_version;
|
|
|
|
|
cts->transport_version = tinfo->transport_version;
|
|
|
|
|
|
|
|
|
|
spi->sync_period = tinfo->period;
|
|
|
|
|
spi->sync_offset = tinfo->offset;
|
|
|
|
|
spi->bus_width = tinfo->width;
|
|
|
|
|
spi->ppr_options = tinfo->ppr_options;
|
|
|
|
|
|
|
|
|
|
cts->protocol = PROTO_SCSI;
|
|
|
|
|
cts->transport = XPORT_SPI;
|
|
|
|
|
spi->valid = CTS_SPI_VALID_SYNC_RATE
|
|
|
|
|
| CTS_SPI_VALID_SYNC_OFFSET
|
|
|
|
|
| CTS_SPI_VALID_BUS_WIDTH
|
|
|
|
|
| CTS_SPI_VALID_PPR_OPTIONS;
|
|
|
|
|
|
2000-10-06 04:01:06 +00:00
|
|
|
if (cts->ccb_h.target_lun != CAM_LUN_WILDCARD) {
|
|
|
|
|
scsi->valid = CTS_SCSI_VALID_TQ;
|
|
|
|
|
spi->valid |= CTS_SPI_VALID_DISC;
|
|
|
|
|
} else {
|
|
|
|
|
scsi->valid = 0;
|
|
|
|
|
}
|
|
|
|
|
|
2000-09-16 20:02:28 +00:00
|
|
|
cts->ccb_h.status = CAM_REQ_CMP;
|
|
|
|
|
#else
|
|
|
|
|
struct ahc_devinfo devinfo;
|
|
|
|
|
struct ahc_initiator_tinfo *targ_info;
|
2001-03-29 00:36:35 +00:00
|
|
|
struct ahc_tmode_tstate *tstate;
|
2000-09-16 20:02:28 +00:00
|
|
|
struct ahc_transinfo *tinfo;
|
|
|
|
|
|
|
|
|
|
ahc_compile_devinfo(&devinfo, our_id,
|
|
|
|
|
cts->ccb_h.target_id,
|
|
|
|
|
cts->ccb_h.target_lun,
|
|
|
|
|
channel, ROLE_UNKNOWN);
|
|
|
|
|
targ_info = ahc_fetch_transinfo(ahc, devinfo.channel,
|
|
|
|
|
devinfo.our_scsiid,
|
|
|
|
|
devinfo.target, &tstate);
|
|
|
|
|
|
|
|
|
|
if ((cts->flags & CCB_TRANS_CURRENT_SETTINGS) != 0)
|
2001-05-15 19:41:12 +00:00
|
|
|
tinfo = &targ_info->curr;
|
2000-09-16 20:02:28 +00:00
|
|
|
else
|
|
|
|
|
tinfo = &targ_info->user;
|
|
|
|
|
|
|
|
|
|
cts->flags &= ~(CCB_TRANS_DISC_ENB|CCB_TRANS_TAG_ENB);
|
2000-10-06 19:34:40 +00:00
|
|
|
if ((cts->flags & CCB_TRANS_CURRENT_SETTINGS) == 0) {
|
2000-09-16 20:02:28 +00:00
|
|
|
if ((ahc->user_discenable & devinfo.target_mask) != 0)
|
|
|
|
|
cts->flags |= CCB_TRANS_DISC_ENB;
|
|
|
|
|
|
|
|
|
|
if ((ahc->user_tagenable & devinfo.target_mask) != 0)
|
|
|
|
|
cts->flags |= CCB_TRANS_TAG_ENB;
|
|
|
|
|
} else {
|
|
|
|
|
if ((tstate->discenable & devinfo.target_mask) != 0)
|
|
|
|
|
cts->flags |= CCB_TRANS_DISC_ENB;
|
|
|
|
|
|
|
|
|
|
if ((tstate->tagenable & devinfo.target_mask) != 0)
|
|
|
|
|
cts->flags |= CCB_TRANS_TAG_ENB;
|
|
|
|
|
}
|
|
|
|
|
cts->sync_period = tinfo->period;
|
|
|
|
|
cts->sync_offset = tinfo->offset;
|
|
|
|
|
cts->bus_width = tinfo->width;
|
|
|
|
|
|
|
|
|
|
cts->valid = CCB_TRANS_SYNC_RATE_VALID
|
|
|
|
|
| CCB_TRANS_SYNC_OFFSET_VALID
|
2000-10-06 04:01:06 +00:00
|
|
|
| CCB_TRANS_BUS_WIDTH_VALID;
|
|
|
|
|
|
|
|
|
|
if (cts->ccb_h.target_lun != CAM_LUN_WILDCARD)
|
|
|
|
|
cts->valid |= CCB_TRANS_DISC_VALID|CCB_TRANS_TQ_VALID;
|
2000-09-16 20:02:28 +00:00
|
|
|
|
|
|
|
|
cts->ccb_h.status = CAM_REQ_CMP;
|
|
|
|
|
#endif
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static void
|
|
|
|
|
ahc_async(void *callback_arg, uint32_t code, struct cam_path *path, void *arg)
|
|
|
|
|
{
|
|
|
|
|
struct ahc_softc *ahc;
|
|
|
|
|
struct cam_sim *sim;
|
|
|
|
|
|
|
|
|
|
sim = (struct cam_sim *)callback_arg;
|
|
|
|
|
ahc = (struct ahc_softc *)cam_sim_softc(sim);
|
|
|
|
|
switch (code) {
|
|
|
|
|
case AC_LOST_DEVICE:
|
|
|
|
|
{
|
|
|
|
|
struct ahc_devinfo devinfo;
|
|
|
|
|
long s;
|
|
|
|
|
|
|
|
|
|
ahc_compile_devinfo(&devinfo, SIM_SCSI_ID(ahc, sim),
|
|
|
|
|
xpt_path_target_id(path),
|
|
|
|
|
xpt_path_lun_id(path),
|
|
|
|
|
SIM_CHANNEL(ahc, sim),
|
|
|
|
|
ROLE_UNKNOWN);
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Revert to async/narrow transfers
|
|
|
|
|
* for the next device.
|
|
|
|
|
*/
|
|
|
|
|
ahc_lock(ahc, &s);
|
|
|
|
|
ahc_set_width(ahc, &devinfo, MSG_EXT_WDTR_BUS_8_BIT,
|
|
|
|
|
AHC_TRANS_GOAL|AHC_TRANS_CUR, /*paused*/FALSE);
|
|
|
|
|
ahc_set_syncrate(ahc, &devinfo, /*syncrate*/NULL,
|
|
|
|
|
/*period*/0, /*offset*/0, /*ppr_options*/0,
|
|
|
|
|
AHC_TRANS_GOAL|AHC_TRANS_CUR,
|
|
|
|
|
/*paused*/FALSE);
|
|
|
|
|
ahc_unlock(ahc, &s);
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
default:
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static void
|
|
|
|
|
ahc_execute_scb(void *arg, bus_dma_segment_t *dm_segs, int nsegments,
|
|
|
|
|
int error)
|
|
|
|
|
{
|
2000-10-06 04:01:06 +00:00
|
|
|
struct scb *scb;
|
|
|
|
|
union ccb *ccb;
|
|
|
|
|
struct ahc_softc *ahc;
|
|
|
|
|
struct ahc_initiator_tinfo *tinfo;
|
2001-03-29 00:36:35 +00:00
|
|
|
struct ahc_tmode_tstate *tstate;
|
2000-10-06 04:01:06 +00:00
|
|
|
u_int mask;
|
|
|
|
|
long s;
|
2000-09-16 20:02:28 +00:00
|
|
|
|
|
|
|
|
scb = (struct scb *)arg;
|
|
|
|
|
ccb = scb->io_ctx;
|
2000-10-11 23:46:34 +00:00
|
|
|
ahc = scb->ahc_softc;
|
2000-09-16 20:02:28 +00:00
|
|
|
|
|
|
|
|
if (error != 0) {
|
|
|
|
|
if (error == EFBIG)
|
|
|
|
|
ahc_set_transaction_status(scb, CAM_REQ_TOO_BIG);
|
|
|
|
|
else
|
|
|
|
|
ahc_set_transaction_status(scb, CAM_REQ_CMP_ERR);
|
|
|
|
|
if (nsegments != 0)
|
|
|
|
|
bus_dmamap_unload(ahc->buffer_dmat, scb->dmamap);
|
2000-10-05 04:24:14 +00:00
|
|
|
ahc_lock(ahc, &s);
|
2000-09-16 20:02:28 +00:00
|
|
|
ahc_free_scb(ahc, scb);
|
2000-10-05 04:24:14 +00:00
|
|
|
ahc_unlock(ahc, &s);
|
2000-09-16 20:02:28 +00:00
|
|
|
xpt_done(ccb);
|
|
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
if (nsegments != 0) {
|
|
|
|
|
struct ahc_dma_seg *sg;
|
|
|
|
|
bus_dma_segment_t *end_seg;
|
|
|
|
|
bus_dmasync_op_t op;
|
|
|
|
|
|
|
|
|
|
end_seg = dm_segs + nsegments;
|
|
|
|
|
|
|
|
|
|
/* Copy the segments into our SG list */
|
|
|
|
|
sg = scb->sg_list;
|
|
|
|
|
while (dm_segs < end_seg) {
|
2001-07-18 21:39:48 +00:00
|
|
|
uint32_t len;
|
|
|
|
|
|
|
|
|
|
sg->addr = ahc_htole32(dm_segs->ds_addr);
|
|
|
|
|
len = dm_segs->ds_len
|
|
|
|
|
| ((dm_segs->ds_addr >> 8) & 0x7F000000);
|
|
|
|
|
sg->len = ahc_htole32(len);
|
2000-09-16 20:02:28 +00:00
|
|
|
sg++;
|
|
|
|
|
dm_segs++;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Note where to find the SG entries in bus space.
|
|
|
|
|
* We also set the full residual flag which the
|
|
|
|
|
* sequencer will clear as soon as a data transfer
|
|
|
|
|
* occurs.
|
|
|
|
|
*/
|
2001-07-18 21:39:48 +00:00
|
|
|
scb->hscb->sgptr = ahc_htole32(scb->sg_list_phys|SG_FULL_RESID);
|
2000-09-16 20:02:28 +00:00
|
|
|
|
|
|
|
|
if ((ccb->ccb_h.flags & CAM_DIR_MASK) == CAM_DIR_IN)
|
|
|
|
|
op = BUS_DMASYNC_PREREAD;
|
|
|
|
|
else
|
|
|
|
|
op = BUS_DMASYNC_PREWRITE;
|
|
|
|
|
|
|
|
|
|
bus_dmamap_sync(ahc->buffer_dmat, scb->dmamap, op);
|
|
|
|
|
|
|
|
|
|
if (ccb->ccb_h.func_code == XPT_CONT_TARGET_IO) {
|
|
|
|
|
struct target_data *tdata;
|
|
|
|
|
|
|
|
|
|
tdata = &scb->hscb->shared_data.tdata;
|
|
|
|
|
tdata->target_phases |= DPHASE_PENDING;
|
|
|
|
|
if ((ccb->ccb_h.flags & CAM_DIR_MASK) == CAM_DIR_OUT)
|
|
|
|
|
tdata->data_phase = P_DATAOUT;
|
|
|
|
|
else
|
|
|
|
|
tdata->data_phase = P_DATAIN;
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* If the transfer is of an odd length and in the
|
|
|
|
|
* "in" direction (scsi->HostBus), then it may
|
|
|
|
|
* trigger a bug in the 'WideODD' feature of
|
|
|
|
|
* non-Ultra2 chips. Force the total data-length
|
|
|
|
|
* to be even by adding an extra, 1 byte, SG,
|
|
|
|
|
* element. We do this even if we are not currently
|
|
|
|
|
* negotiated wide as negotiation could occur before
|
|
|
|
|
* this command is executed.
|
|
|
|
|
*/
|
|
|
|
|
if ((ahc->bugs & AHC_TMODE_WIDEODD_BUG) != 0
|
|
|
|
|
&& (ccb->csio.dxfer_len & 0x1) != 0
|
|
|
|
|
&& (ccb->ccb_h.flags & CAM_DIR_MASK) == CAM_DIR_IN) {
|
|
|
|
|
|
|
|
|
|
nsegments++;
|
|
|
|
|
if (nsegments > AHC_NSEG) {
|
|
|
|
|
|
|
|
|
|
ahc_set_transaction_status(scb,
|
|
|
|
|
CAM_REQ_TOO_BIG);
|
|
|
|
|
bus_dmamap_unload(ahc->buffer_dmat,
|
|
|
|
|
scb->dmamap);
|
2000-10-05 04:24:14 +00:00
|
|
|
ahc_lock(ahc, &s);
|
2000-09-16 20:02:28 +00:00
|
|
|
ahc_free_scb(ahc, scb);
|
2000-10-05 04:24:14 +00:00
|
|
|
ahc_unlock(ahc, &s);
|
2000-09-16 20:02:28 +00:00
|
|
|
xpt_done(ccb);
|
|
|
|
|
return;
|
|
|
|
|
}
|
2001-07-18 21:39:48 +00:00
|
|
|
sg->addr = ahc_htole32(ahc->dma_bug_buf);
|
|
|
|
|
sg->len = ahc_htole32(1);
|
2000-09-16 20:02:28 +00:00
|
|
|
sg++;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
sg--;
|
2001-07-18 21:39:48 +00:00
|
|
|
sg->len |= ahc_htole32(AHC_DMA_LAST_SEG);
|
2000-09-16 20:02:28 +00:00
|
|
|
|
|
|
|
|
/* Copy the first SG into the "current" data pointer area */
|
|
|
|
|
scb->hscb->dataptr = scb->sg_list->addr;
|
|
|
|
|
scb->hscb->datacnt = scb->sg_list->len;
|
|
|
|
|
} else {
|
2002-04-24 16:58:51 +00:00
|
|
|
scb->hscb->sgptr = ahc_htole32(SG_LIST_NULL);
|
2000-09-16 20:02:28 +00:00
|
|
|
scb->hscb->dataptr = 0;
|
|
|
|
|
scb->hscb->datacnt = 0;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
scb->sg_count = nsegments;
|
|
|
|
|
|
|
|
|
|
ahc_lock(ahc, &s);
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Last time we need to check if this SCB needs to
|
|
|
|
|
* be aborted.
|
|
|
|
|
*/
|
|
|
|
|
if (ahc_get_transaction_status(scb) != CAM_REQ_INPROG) {
|
|
|
|
|
if (nsegments != 0)
|
2002-04-24 16:58:51 +00:00
|
|
|
bus_dmamap_unload(ahc->buffer_dmat, scb->dmamap);
|
2000-09-16 20:02:28 +00:00
|
|
|
ahc_free_scb(ahc, scb);
|
|
|
|
|
ahc_unlock(ahc, &s);
|
2000-10-05 04:24:14 +00:00
|
|
|
xpt_done(ccb);
|
2000-09-16 20:02:28 +00:00
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
|
2000-10-06 04:01:06 +00:00
|
|
|
tinfo = ahc_fetch_transinfo(ahc, SCSIID_CHANNEL(ahc, scb->hscb->scsiid),
|
|
|
|
|
SCSIID_OUR_ID(scb->hscb->scsiid),
|
aic7xxx.c:
Filter incoming transfer negotiation requests to ensure they
never exceed the settings specified by the user.
In restart sequencer attempt to deal with a bug in the aic7895.
If a third party reset occurs at just the right time, the
stack register can lock up. When restarting the sequencer
after handling the SCSI reset, poke SEQADDR1 before resting
the sequencers program counter.
When something strange happens, dump the card's transaction
state via ahc_dump_card_state(). This should aid in debugging.
Handle request sense transactions via the QINFIFO instead of
attaching them to the waiting queue directly. The waiting
queue consumes card SCB resources and, in the pathological case
of every target on the bus beating our selection attemps and
issuing a check condition, could have caused us to run out
of SCBs. I have never seen this happen, and only early
cards with 3 or 4 SCBs had any real chance of ever getting
into this state.
Add additional sequencer interrupt codes to support firmware
diagnostics. The diagnostic code is enabled with the
AHC_DEBUG_SEQUENCER kernel option.
Make it possible to switch into and out of target mode on
the fly. The card comes up by default as an initiator but
will switch into target mode as soon as an enable lun operation
is performed. As always, target mode behavior is gated
by the AHC_TMODE_ENABLE kernel option so most users will
not be affected by this change.
In ahc_update_target_msg_request(), also issue a new
request if the ppr_options have changed.
Never issue a PPR as a target. It is forbidden by the spec.
Correct a bug in ahc_parse_msg() that prevented us from
responding to PPR messages as a target.
Mark SCBs that are on the untagged queue with a flag instead
of checking several fields in the SCB to see if the SCB should
be on the queue. This makes it easier for things like automatic
request sense requests to be queued without touching the
untagged queues even though they are untagged requests.
When dealing with ignore wide residue messages that occur
in the middle of a transfer, reset HADDR, not SHADDR for
non-ultra2 chips. Although SHADDR is where the firmware
fetches the ending transfer address for a save data pointers
request, it is readonly. Setting HADDR has the side effect
of also updating SHADDR.
Cleanup the output of ahc_dump_card_state() by nulling out the
free scb list in the non-paging case. The free list is only
used if we must page SCBs.
Correct the transmission of cdbs > 12 bytes in length. When
swapping HSCBs prior to notifing the sequencer of the new
transaction, the bus address pointer for the cdb must also
be recalculated to reflect its new location. We now defer
the calculation of the cdb address until just before queing
it to the card.
When pulling transfer negotiation settings out of scratch
ram, convert 5MHz/clock doubled settings to 10MHz.
Add a new function ahc_qinfifo_requeue_tail() for use by
error recovery actions and auto-request sense operations.
These operations always occur when the sequencer is paused,
so we can avoid the extra expense incurred in the normal
SCB queue method.
Use the BMOV instruction for all single byte moves on
controllers that support it. The bmov instruction is
twice as fast as an AND with an immediate of 0xFF as
is used on older controllers.
Correct a few bugs in ahc_dump_card_state(). If we have
hardware assisted queue registers, use them to get the
sequencer's idea of the head of the queue. When enumerating
the untagged queue, it helps to use the correct index for
the queue.
aic7xxx.h:
Indicate via a feature flag, which controllers can take
on both the target and the initiator role at the same time.
Add the AHC_SEQUENCER_DEBUG flag.
Add the SCB_CDB32_PTR flag used for dealing with cdbs
with lengths between 13 and 32 bytes.
Add new prototypes.
aic7xxx.reg:
Allow the SCSIBUSL register to be written to. This is
required to fix a selection timeout problem on the 7892/99.
Cleanup the sequencer interrupt codes so that all debugging
codes are grouped at the end of the list.
Correct the definition of the ULTRA_ENB and DISC_DSB locations
in scratch ram. This prevented the driver from properly honoring
these settings when no serial eeprom was available.
Remove an unused sequencer flag.
aic7xxx.seq:
Just before a potential select-out, clear the SCSIBUSL
register. Occasionally, during a selection timeout, the
contents of the register may be presented on the bus,
causing much confusion.
Add sequencer diagnostic code to detect software and or
hardware bugs. The code attempts to verify most list
operations so any corruption is caught before it occurs.
We also track information about why a particular reconnection
request was rejected.
Don't clobber the digital REQ/ACK filter setting in SXFRCTL0
when clearing the channel.
Fix a target mode bug that would cause us to return busy
status instead of queue full in respnse to a tagged transaction.
Cleanup the overrun case. It turns out that by simply
butting the chip in bitbucket mode, it will ack any
bytes until the phase changes. This drasticaly simplifies
things.
Prior to leaving the data phase, make sure that the S/G
preload queue is empty.
Remove code to place a request sense request on the waiting
queue. This is all handled by the kernel now.
Change the semantics of "findSCB". In the past, findSCB
ensured that a freshly paged in SCB appeared on the disconnected
list. The problem with this is that there is no guarantee that
the paged in SCB is for a disconnected transation. We now
defer any list manipulation to the caller who usually discards
the SCB via the free list.
Inline some busy target table operations.
Add a critical section to protect adding an SCB to
the disconnected list.
aic7xxx_freebsd.c:
Handle changes in the transfer negotiation setting API
to filter incoming requests. No filtering is necessary
for "goal" requests from the XPT.
Set the SCB_CDB32_PTR flag when queing a transaction with
a large cdb.
In ahc_timeout, only take action if the active SCB is
the timedout SCB. This deals with the case of two
transactions to the same device with different timeout
values.
Use ahc_qinfifo_requeu_tail() instead of home grown
version.
aic7xxx_inline.h:
Honor SCB_CDB32_PTR when queuing a new request.
aic7xxx_pci.c:
Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
|
|
|
SCSIID_TARGET(ahc, scb->hscb->scsiid),
|
|
|
|
|
&tstate);
|
2000-10-06 04:01:06 +00:00
|
|
|
|
|
|
|
|
mask = SCB_GET_TARGET_MASK(ahc, scb);
|
|
|
|
|
scb->hscb->scsirate = tinfo->scsirate;
|
2001-05-15 19:41:12 +00:00
|
|
|
scb->hscb->scsioffset = tinfo->curr.offset;
|
2000-10-06 04:01:06 +00:00
|
|
|
if ((tstate->ultraenb & mask) != 0)
|
|
|
|
|
scb->hscb->control |= ULTRAENB;
|
2001-03-29 00:36:35 +00:00
|
|
|
|
2000-10-06 04:01:06 +00:00
|
|
|
if ((tstate->discenable & mask) != 0
|
|
|
|
|
&& (ccb->ccb_h.flags & CAM_DIS_DISCONNECT) == 0)
|
|
|
|
|
scb->hscb->control |= DISCENB;
|
|
|
|
|
|
|
|
|
|
if ((ccb->ccb_h.flags & CAM_NEGOTIATE) != 0
|
ahc_eisa.c:
ahc_pci.c:
Add detach support.
Make use of soft allocated on our behalf by newbus.
For PCI devices, disable the mapping type we aren't
using for extra protection from rogue code.
aic7xxx_93cx6.c:
aic7xxx_93cx6.h:
Sync perforce IDs.
aic7xxx_freebsd.c:
Capture the eventhandle returned by EVENTHANDER_REGISTER
so we can kill the handler off during detach.
Use AHC_* constants instead of hard coded numbers in a
few more places.
Test PPR option state when deciding to "really" negotiate
when the CAM_NEGOTIATE flag is passed in a CCB.
Make use of core "ahc_pause_and_flushwork" routine in our
timeout handler rather than re-inventing this code.
Cleanup all of our resources (really!) in ahc_platform_free().
We should be all set to become a module now.
Implement the core ahc_detach() routine shared by all of
the FreeBSD front-ends.
aic7xxx_freebsd.h:
Softc storage for our event handler.
Null implementation for the ahc_platform_flushwork() OSM
callback. FreeBSD doesn't need this as XPT callbacks are
safe from all contexts and are done directly in ahc_done().
aic7xxx_inline.h:
Implement new lazy interrupt scheme. To avoid an extra
PCI bus read, we first check our completion queues to
see if any work has completed. If work is available, we
assume that this is the source of the interrupt and skip
reading INTSTAT. Any remaining interrupt status will be
cleared by a second call to the interrupt handler should
the interrupt line still be asserted. This drops the
interrupt handler down to a single PCI bus read in the
common case of I/O completion. This is the same overhead
as in the not so distant past, but the extra sanity of
perforning a PCI read after clearing the command complete
interrupt and before running the completion queue to avoid
missing command complete interrupts added a cycle.
aic7xxx.c:
During initialization, be sure to initialize all scratch
ram locations before they are read to avoid parity errors.
In this case, we use a new function, ahc_unbusy_tcl() to
initialize the scratch ram busy target table.
Replace instances of ahc_index_busy_tcl() used to unbusy
a tcl without looking at the old value with ahc_unbusy_tcl().
Modify ahc_sent_msg so that it can find single byte messages.
ahc_sent_msg is now used to determine if a transfer negotiation
attempt resulted in a bus free.
Be more careful in filtering out only the SCSI interrupts
of interest in ahc_handle_scsiint.
Rearrange interrupt clearing code to ensure that at least
one PCI transaction occurrs after hitting CLRSINT1 and
writting to CLRINT. CLRSINT1 writes take a bit to
take effect, and the re-arrangement provides sufficient
delay to ensure the write to CLRINT is effective. The
old code might report a spurious interrupt on some "fast"
chipsets.
export ahc-update_target_msg_request for use by OSM code.
If a target does not respond to our ATN request, clear
it once we move to a non-message phase. This avoids
sending a MSG_NOOP in some later message out phase.
Use max lun and max target constants instead of
hard-coded values.
Use softc storage built into our device_t under FreeBSD.
Fix a bug in ahc_free() that caused us to delete
resources that were not allocated.
Clean up any tstate/lstate info in ahc_free().
Clear the powerdown state in ahc_reset() so that
registers can be accessed.
Add a preliminary function for pausing the chip and
processing any posted work.
Add a preliminary suspend and resume functions.
aic7xxx.h:
Limit the number of supported luns to 64. We don't
support information unit transfers, so this is the
maximum that makes sense for these chips.
Add a new flag AHC_ALL_INTERRUPTS that forces the
processing of all interrupt state in a single invokation
of ahc_intr(). When the flag is not set, we use the
lazy interrupt handling scheme.
Add data structures to store controller state while
we are suspended.
Use constants instead of hard coded values where appropriate.
Correct some harmless "unsigned/signed" conflicts.
aic7xxx.seq:
Only perform the SCSIBUSL fix on ULTRA2 or newer controllers.
Older controllers seem to be confused by this.
In target mode, ignore PHASEMIS during data phases.
This bit seems to be flakey on U160 controllers acting
in target mode.
aic7xxx_pci.c:
Add support for the 29160C CPCI adapter.
Add definitions for subvendor ID information
available for devices with the "9005" vendor id.
We currently use this information to determine
if a multi-function device doesn't have the second
channel hooked up on a board.
Add rudimentary power mode code so we can put the
controller into the D0 state. In the future this
will be an OSM callback so that in FreeBSD we don't
duplicate functionality provided by the PCI code.
The powerstate code was added after I'd completed
my regression tests on this code.
Only capture "left over BIOS state" if the POWRDN
setting is not set in HCNTRL.
In target mode, don't bother sending incremental
CRC data.
2000-12-20 01:11:37 +00:00
|
|
|
&& (tinfo->goal.width != 0
|
2002-08-31 06:43:55 +00:00
|
|
|
|| tinfo->goal.offset != 0
|
ahc_eisa.c:
ahc_pci.c:
Add detach support.
Make use of soft allocated on our behalf by newbus.
For PCI devices, disable the mapping type we aren't
using for extra protection from rogue code.
aic7xxx_93cx6.c:
aic7xxx_93cx6.h:
Sync perforce IDs.
aic7xxx_freebsd.c:
Capture the eventhandle returned by EVENTHANDER_REGISTER
so we can kill the handler off during detach.
Use AHC_* constants instead of hard coded numbers in a
few more places.
Test PPR option state when deciding to "really" negotiate
when the CAM_NEGOTIATE flag is passed in a CCB.
Make use of core "ahc_pause_and_flushwork" routine in our
timeout handler rather than re-inventing this code.
Cleanup all of our resources (really!) in ahc_platform_free().
We should be all set to become a module now.
Implement the core ahc_detach() routine shared by all of
the FreeBSD front-ends.
aic7xxx_freebsd.h:
Softc storage for our event handler.
Null implementation for the ahc_platform_flushwork() OSM
callback. FreeBSD doesn't need this as XPT callbacks are
safe from all contexts and are done directly in ahc_done().
aic7xxx_inline.h:
Implement new lazy interrupt scheme. To avoid an extra
PCI bus read, we first check our completion queues to
see if any work has completed. If work is available, we
assume that this is the source of the interrupt and skip
reading INTSTAT. Any remaining interrupt status will be
cleared by a second call to the interrupt handler should
the interrupt line still be asserted. This drops the
interrupt handler down to a single PCI bus read in the
common case of I/O completion. This is the same overhead
as in the not so distant past, but the extra sanity of
perforning a PCI read after clearing the command complete
interrupt and before running the completion queue to avoid
missing command complete interrupts added a cycle.
aic7xxx.c:
During initialization, be sure to initialize all scratch
ram locations before they are read to avoid parity errors.
In this case, we use a new function, ahc_unbusy_tcl() to
initialize the scratch ram busy target table.
Replace instances of ahc_index_busy_tcl() used to unbusy
a tcl without looking at the old value with ahc_unbusy_tcl().
Modify ahc_sent_msg so that it can find single byte messages.
ahc_sent_msg is now used to determine if a transfer negotiation
attempt resulted in a bus free.
Be more careful in filtering out only the SCSI interrupts
of interest in ahc_handle_scsiint.
Rearrange interrupt clearing code to ensure that at least
one PCI transaction occurrs after hitting CLRSINT1 and
writting to CLRINT. CLRSINT1 writes take a bit to
take effect, and the re-arrangement provides sufficient
delay to ensure the write to CLRINT is effective. The
old code might report a spurious interrupt on some "fast"
chipsets.
export ahc-update_target_msg_request for use by OSM code.
If a target does not respond to our ATN request, clear
it once we move to a non-message phase. This avoids
sending a MSG_NOOP in some later message out phase.
Use max lun and max target constants instead of
hard-coded values.
Use softc storage built into our device_t under FreeBSD.
Fix a bug in ahc_free() that caused us to delete
resources that were not allocated.
Clean up any tstate/lstate info in ahc_free().
Clear the powerdown state in ahc_reset() so that
registers can be accessed.
Add a preliminary function for pausing the chip and
processing any posted work.
Add a preliminary suspend and resume functions.
aic7xxx.h:
Limit the number of supported luns to 64. We don't
support information unit transfers, so this is the
maximum that makes sense for these chips.
Add a new flag AHC_ALL_INTERRUPTS that forces the
processing of all interrupt state in a single invokation
of ahc_intr(). When the flag is not set, we use the
lazy interrupt handling scheme.
Add data structures to store controller state while
we are suspended.
Use constants instead of hard coded values where appropriate.
Correct some harmless "unsigned/signed" conflicts.
aic7xxx.seq:
Only perform the SCSIBUSL fix on ULTRA2 or newer controllers.
Older controllers seem to be confused by this.
In target mode, ignore PHASEMIS during data phases.
This bit seems to be flakey on U160 controllers acting
in target mode.
aic7xxx_pci.c:
Add support for the 29160C CPCI adapter.
Add definitions for subvendor ID information
available for devices with the "9005" vendor id.
We currently use this information to determine
if a multi-function device doesn't have the second
channel hooked up on a board.
Add rudimentary power mode code so we can put the
controller into the D0 state. In the future this
will be an OSM callback so that in FreeBSD we don't
duplicate functionality provided by the PCI code.
The powerstate code was added after I'd completed
my regression tests on this code.
Only capture "left over BIOS state" if the POWRDN
setting is not set in HCNTRL.
In target mode, don't bother sending incremental
CRC data.
2000-12-20 01:11:37 +00:00
|
|
|
|| tinfo->goal.ppr_options != 0)) {
|
2000-10-06 04:01:06 +00:00
|
|
|
scb->flags |= SCB_NEGOTIATE;
|
|
|
|
|
scb->hscb->control |= MK_MESSAGE;
|
2001-03-29 00:36:35 +00:00
|
|
|
} else if ((tstate->auto_negotiate & mask) != 0) {
|
|
|
|
|
scb->flags |= SCB_AUTO_NEGOTIATE;
|
|
|
|
|
scb->hscb->control |= MK_MESSAGE;
|
2000-10-06 04:01:06 +00:00
|
|
|
}
|
|
|
|
|
|
2000-09-16 20:02:28 +00:00
|
|
|
LIST_INSERT_HEAD(&ahc->pending_scbs, scb, pending_links);
|
|
|
|
|
|
|
|
|
|
ccb->ccb_h.status |= CAM_SIM_QUEUED;
|
|
|
|
|
|
|
|
|
|
if (ccb->ccb_h.timeout != CAM_TIME_INFINITY) {
|
2001-01-22 21:03:48 +00:00
|
|
|
uint64_t time;
|
|
|
|
|
|
2000-09-16 20:02:28 +00:00
|
|
|
if (ccb->ccb_h.timeout == CAM_TIME_DEFAULT)
|
|
|
|
|
ccb->ccb_h.timeout = 5 * 1000;
|
2001-01-22 21:03:48 +00:00
|
|
|
|
|
|
|
|
time = ccb->ccb_h.timeout;
|
|
|
|
|
time *= hz;
|
|
|
|
|
time /= 1000;
|
2000-09-16 20:02:28 +00:00
|
|
|
ccb->ccb_h.timeout_ch =
|
2001-01-22 21:03:48 +00:00
|
|
|
timeout(ahc_timeout, (caddr_t)scb, time);
|
2000-09-16 20:02:28 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* We only allow one untagged transaction
|
|
|
|
|
* per target in the initiator role unless
|
|
|
|
|
* we are storing a full busy target *lun*
|
|
|
|
|
* table in SCB space.
|
|
|
|
|
*/
|
|
|
|
|
if ((scb->hscb->control & (TARGET_SCB|TAG_ENB)) == 0
|
2001-01-22 21:03:48 +00:00
|
|
|
&& (ahc->flags & AHC_SCB_BTT) == 0) {
|
2000-09-16 20:02:28 +00:00
|
|
|
struct scb_tailq *untagged_q;
|
2001-02-21 20:50:36 +00:00
|
|
|
int target_offset;
|
2000-09-16 20:02:28 +00:00
|
|
|
|
2001-02-21 20:50:36 +00:00
|
|
|
target_offset = SCB_GET_TARGET_OFFSET(ahc, scb);
|
|
|
|
|
untagged_q = &(ahc->untagged_queues[target_offset]);
|
2000-09-16 20:02:28 +00:00
|
|
|
TAILQ_INSERT_TAIL(untagged_q, scb, links.tqe);
|
2000-10-11 23:46:34 +00:00
|
|
|
scb->flags |= SCB_UNTAGGEDQ;
|
2000-09-16 20:02:28 +00:00
|
|
|
if (TAILQ_FIRST(untagged_q) != scb) {
|
|
|
|
|
ahc_unlock(ahc, &s);
|
|
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
scb->flags |= SCB_ACTIVE;
|
|
|
|
|
|
|
|
|
|
if ((scb->flags & SCB_TARGET_IMMEDIATE) != 0) {
|
2001-03-29 00:36:35 +00:00
|
|
|
/* Define a mapping from our tag to the SCB. */
|
|
|
|
|
ahc->scb_data->scbindex[scb->hscb->tag] = scb;
|
This is an MFC candidate.
ahc_eisa.c:
Change aic7770_map_int to take an additional irq parameter.
Although we can get the irq from the eisa dev under FreeBSD,
we can't do this under linux, so the OSM interface must supply
this.
ahc_pci.c:
Move ahc_power_state_change() to the OSM. This allows us to
use a platform supplied function that does the same thing.
-current will move to the FreeBSD native API in the near
future.
aic7770.c:
Sync up with core changes to support Linux EISA.
We now store a 2 bit primary channel number rather
than a bit flag that only allows b to be the primary
channel. Adjust for this change.
aic7xxx.c:
Namespace and staticization cleanup. All exported symbols
use an "ahc_" prefix to avoid collisions with other modules.
Correct a logic bug that prevented us from dropping
ATN during some exceptional conditions during message
processing.
Take advantage of a new flag managed by the sequencer
that indicates if an SCB fetch is in progress. If so,
the currently selected SCB needs to be returned to the
free list to prevent an SCB leak. This leak is a rarity
and would only occur if a bus reset or timeout resulting
in a bus reset occurred in the middle of an SCB fetch.
Don't attempt to perform ULTRA transfers on ultra capable
adapters missing the external precision resistor required
for ultra speeds. I've never encountered an adapter
configured this way, but better safe than sorry.
Handle the case of 5MHz user sync rate set as "0" instead of 0x1c
in scratch ram.
If we lookup a period of 0 in our table (async), clear the scsi offset.
aic7xxx.h:
Adjust for the primary channel being represented as
a 2 bit integer in the flags member of the ahc softc.
Cleanup the flags definitions so that comment blocks are
not cramped.
Update seeprom definitions to correctly reflect the fact
that the primary channel is represented as a 2 bit integer.
Add AHC_ULTRA_DIASABLED softc flag to denote controllers
missing the external precision resistor.
aic7xxx.reg:
Add DFCACHETH to the definition of DFSTATUS for completness sake.
Add SEQ_FLAGS2 which currently only contains the SCB_DMA
(SCB DMA in progress) flag.
aic7xxx.seq:
Correct a problem when one lun has a disconnected untagged
transaction and another lun has disconnected tagged transactions.
Just because an entry is found in the untagged table doesn't
mean that it will match. If the match on the lun fails, cleanup
the SCB (return it to the disconnected list or free it), and snoop
for a tag message. Before this change, we reported an unsolicited
reselection. This bug was introduced about a month ago during an
overly aggressive optimization pass on the reselection code.
When cleaning up an SCB, we can't just blindly free the SCB. In
the paging case, if the SCB came off of the disconnected list, its
state may never have been updated in host memory. So, check the
disconnected bit in SCB_CONTROL and return the SCB to the disconnected
list if appropriate.
Manage the SCB_DMA flag of SEQ_FLAGS2.
More carefully shutdown the S/G dma engine in all cases by using
a subroutine. Supposedly not doing this can cause an arbiter hang
on some ULTRA2 chips.
Formatting cleanup.
On some chips, at least the aic7856, the transition from
MREQPEND to HDONE can take a full 4 clock cycles. Test
HDONE one more time to avoid this race. We only want our
FIFO hung recovery code to execute when the engine is
really hung.
aic7xxx_93cx6.c:
Sync perforce ids.
aic7xxx_freebsd.c:
Adjust for the primary channel being a 2 bit integer
rather than a flag for 'B' channel being the primary.
Namespace cleanup.
Unpause the sequencer in one error recovery path that
neglected to do so. This could have caused us to perform
a bus reset when a recovery message might have otherwise been
successful.
aic7xxx_freebsd.h:
Use AHC_PCI_CONFIG for controlling compilation of PCI
support consistently throughout the driver.
Move ahc_power_state_change() to OSM.
aic7xxx_inline.h
Namespace cleanup.
Adjust our interrupt handler so it will work in the edge
interrupt case. We must process all interrupt sources
when the interrupt fires or risk not ever getting an
interrupt again. This involves marking the fact
that we are relying on an edge interrupt in ahc->flags
and checking for this condition in addition to the
AHC_ALL_INTERRUPTS flag. This fixes hangs on the
284X and any other aic7770 installation where level
interrupts are not available.
aic7xxx_pci.c:
Move the powerstate manipulation code into the OSM. Several
OSes now provide this functionality natively.
Take another shot at using the data stored in scratch ram
if the SCB2 signature is correct and no SEEPROM data is
available. In the past this failed if external SCB ram
was configured because the memory port was locked. We
now release the memory port prior to testing the values
in SCB2 and re-acquire it prior to doing termination control.
Adjust for new 2 bit primary channel setting.
Trust the STPWLEVEL setting on v 3.X BIOSes too.
Configure any 785X ID in the same fashion and assume
that any device with a rev id of 1 or higher has the
PCI 2.1 retry bug.
2001-03-11 06:34:17 +00:00
|
|
|
ahc_pause(ahc);
|
2000-09-16 20:02:28 +00:00
|
|
|
if ((ahc->flags & AHC_PAGESCBS) == 0)
|
|
|
|
|
ahc_outb(ahc, SCBPTR, scb->hscb->tag);
|
2002-08-31 06:43:55 +00:00
|
|
|
ahc_outb(ahc, TARG_IMMEDIATE_SCB, scb->hscb->tag);
|
This is an MFC candidate.
ahc_eisa.c:
Change aic7770_map_int to take an additional irq parameter.
Although we can get the irq from the eisa dev under FreeBSD,
we can't do this under linux, so the OSM interface must supply
this.
ahc_pci.c:
Move ahc_power_state_change() to the OSM. This allows us to
use a platform supplied function that does the same thing.
-current will move to the FreeBSD native API in the near
future.
aic7770.c:
Sync up with core changes to support Linux EISA.
We now store a 2 bit primary channel number rather
than a bit flag that only allows b to be the primary
channel. Adjust for this change.
aic7xxx.c:
Namespace and staticization cleanup. All exported symbols
use an "ahc_" prefix to avoid collisions with other modules.
Correct a logic bug that prevented us from dropping
ATN during some exceptional conditions during message
processing.
Take advantage of a new flag managed by the sequencer
that indicates if an SCB fetch is in progress. If so,
the currently selected SCB needs to be returned to the
free list to prevent an SCB leak. This leak is a rarity
and would only occur if a bus reset or timeout resulting
in a bus reset occurred in the middle of an SCB fetch.
Don't attempt to perform ULTRA transfers on ultra capable
adapters missing the external precision resistor required
for ultra speeds. I've never encountered an adapter
configured this way, but better safe than sorry.
Handle the case of 5MHz user sync rate set as "0" instead of 0x1c
in scratch ram.
If we lookup a period of 0 in our table (async), clear the scsi offset.
aic7xxx.h:
Adjust for the primary channel being represented as
a 2 bit integer in the flags member of the ahc softc.
Cleanup the flags definitions so that comment blocks are
not cramped.
Update seeprom definitions to correctly reflect the fact
that the primary channel is represented as a 2 bit integer.
Add AHC_ULTRA_DIASABLED softc flag to denote controllers
missing the external precision resistor.
aic7xxx.reg:
Add DFCACHETH to the definition of DFSTATUS for completness sake.
Add SEQ_FLAGS2 which currently only contains the SCB_DMA
(SCB DMA in progress) flag.
aic7xxx.seq:
Correct a problem when one lun has a disconnected untagged
transaction and another lun has disconnected tagged transactions.
Just because an entry is found in the untagged table doesn't
mean that it will match. If the match on the lun fails, cleanup
the SCB (return it to the disconnected list or free it), and snoop
for a tag message. Before this change, we reported an unsolicited
reselection. This bug was introduced about a month ago during an
overly aggressive optimization pass on the reselection code.
When cleaning up an SCB, we can't just blindly free the SCB. In
the paging case, if the SCB came off of the disconnected list, its
state may never have been updated in host memory. So, check the
disconnected bit in SCB_CONTROL and return the SCB to the disconnected
list if appropriate.
Manage the SCB_DMA flag of SEQ_FLAGS2.
More carefully shutdown the S/G dma engine in all cases by using
a subroutine. Supposedly not doing this can cause an arbiter hang
on some ULTRA2 chips.
Formatting cleanup.
On some chips, at least the aic7856, the transition from
MREQPEND to HDONE can take a full 4 clock cycles. Test
HDONE one more time to avoid this race. We only want our
FIFO hung recovery code to execute when the engine is
really hung.
aic7xxx_93cx6.c:
Sync perforce ids.
aic7xxx_freebsd.c:
Adjust for the primary channel being a 2 bit integer
rather than a flag for 'B' channel being the primary.
Namespace cleanup.
Unpause the sequencer in one error recovery path that
neglected to do so. This could have caused us to perform
a bus reset when a recovery message might have otherwise been
successful.
aic7xxx_freebsd.h:
Use AHC_PCI_CONFIG for controlling compilation of PCI
support consistently throughout the driver.
Move ahc_power_state_change() to OSM.
aic7xxx_inline.h
Namespace cleanup.
Adjust our interrupt handler so it will work in the edge
interrupt case. We must process all interrupt sources
when the interrupt fires or risk not ever getting an
interrupt again. This involves marking the fact
that we are relying on an edge interrupt in ahc->flags
and checking for this condition in addition to the
AHC_ALL_INTERRUPTS flag. This fixes hangs on the
284X and any other aic7770 installation where level
interrupts are not available.
aic7xxx_pci.c:
Move the powerstate manipulation code into the OSM. Several
OSes now provide this functionality natively.
Take another shot at using the data stored in scratch ram
if the SCB2 signature is correct and no SEEPROM data is
available. In the past this failed if external SCB ram
was configured because the memory port was locked. We
now release the memory port prior to testing the values
in SCB2 and re-acquire it prior to doing termination control.
Adjust for new 2 bit primary channel setting.
Trust the STPWLEVEL setting on v 3.X BIOSes too.
Configure any 785X ID in the same fashion and assume
that any device with a rev id of 1 or higher has the
PCI 2.1 retry bug.
2001-03-11 06:34:17 +00:00
|
|
|
ahc_unpause(ahc);
|
2000-09-16 20:02:28 +00:00
|
|
|
} else {
|
|
|
|
|
ahc_queue_scb(ahc, scb);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
ahc_unlock(ahc, &s);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static void
|
|
|
|
|
ahc_poll(struct cam_sim *sim)
|
|
|
|
|
{
|
2002-08-31 06:43:55 +00:00
|
|
|
struct ahc_softc *ahc;
|
|
|
|
|
|
|
|
|
|
ahc = (struct ahc_softc *)cam_sim_softc(sim);
|
|
|
|
|
ahc_intr(ahc);
|
2000-09-16 20:02:28 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static void
|
|
|
|
|
ahc_setup_data(struct ahc_softc *ahc, struct cam_sim *sim,
|
|
|
|
|
struct ccb_scsiio *csio, struct scb *scb)
|
|
|
|
|
{
|
|
|
|
|
struct hardware_scb *hscb;
|
|
|
|
|
struct ccb_hdr *ccb_h;
|
|
|
|
|
|
|
|
|
|
hscb = scb->hscb;
|
|
|
|
|
ccb_h = &csio->ccb_h;
|
|
|
|
|
|
2001-07-18 21:39:48 +00:00
|
|
|
csio->resid = 0;
|
|
|
|
|
csio->sense_resid = 0;
|
2000-09-16 20:02:28 +00:00
|
|
|
if (ccb_h->func_code == XPT_SCSI_IO) {
|
|
|
|
|
hscb->cdb_len = csio->cdb_len;
|
|
|
|
|
if ((ccb_h->flags & CAM_CDB_POINTER) != 0) {
|
|
|
|
|
|
|
|
|
|
if (hscb->cdb_len > sizeof(hscb->cdb32)
|
|
|
|
|
|| (ccb_h->flags & CAM_CDB_PHYS) != 0) {
|
2000-10-05 04:24:14 +00:00
|
|
|
u_long s;
|
|
|
|
|
|
2000-09-16 20:02:28 +00:00
|
|
|
ahc_set_transaction_status(scb,
|
|
|
|
|
CAM_REQ_INVALID);
|
2000-10-05 04:24:14 +00:00
|
|
|
ahc_lock(ahc, &s);
|
2000-09-16 20:02:28 +00:00
|
|
|
ahc_free_scb(ahc, scb);
|
2000-10-05 04:24:14 +00:00
|
|
|
ahc_unlock(ahc, &s);
|
|
|
|
|
xpt_done((union ccb *)csio);
|
2000-09-16 20:02:28 +00:00
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
if (hscb->cdb_len > 12) {
|
|
|
|
|
memcpy(hscb->cdb32,
|
|
|
|
|
csio->cdb_io.cdb_ptr,
|
|
|
|
|
hscb->cdb_len);
|
aic7xxx.c:
Filter incoming transfer negotiation requests to ensure they
never exceed the settings specified by the user.
In restart sequencer attempt to deal with a bug in the aic7895.
If a third party reset occurs at just the right time, the
stack register can lock up. When restarting the sequencer
after handling the SCSI reset, poke SEQADDR1 before resting
the sequencers program counter.
When something strange happens, dump the card's transaction
state via ahc_dump_card_state(). This should aid in debugging.
Handle request sense transactions via the QINFIFO instead of
attaching them to the waiting queue directly. The waiting
queue consumes card SCB resources and, in the pathological case
of every target on the bus beating our selection attemps and
issuing a check condition, could have caused us to run out
of SCBs. I have never seen this happen, and only early
cards with 3 or 4 SCBs had any real chance of ever getting
into this state.
Add additional sequencer interrupt codes to support firmware
diagnostics. The diagnostic code is enabled with the
AHC_DEBUG_SEQUENCER kernel option.
Make it possible to switch into and out of target mode on
the fly. The card comes up by default as an initiator but
will switch into target mode as soon as an enable lun operation
is performed. As always, target mode behavior is gated
by the AHC_TMODE_ENABLE kernel option so most users will
not be affected by this change.
In ahc_update_target_msg_request(), also issue a new
request if the ppr_options have changed.
Never issue a PPR as a target. It is forbidden by the spec.
Correct a bug in ahc_parse_msg() that prevented us from
responding to PPR messages as a target.
Mark SCBs that are on the untagged queue with a flag instead
of checking several fields in the SCB to see if the SCB should
be on the queue. This makes it easier for things like automatic
request sense requests to be queued without touching the
untagged queues even though they are untagged requests.
When dealing with ignore wide residue messages that occur
in the middle of a transfer, reset HADDR, not SHADDR for
non-ultra2 chips. Although SHADDR is where the firmware
fetches the ending transfer address for a save data pointers
request, it is readonly. Setting HADDR has the side effect
of also updating SHADDR.
Cleanup the output of ahc_dump_card_state() by nulling out the
free scb list in the non-paging case. The free list is only
used if we must page SCBs.
Correct the transmission of cdbs > 12 bytes in length. When
swapping HSCBs prior to notifing the sequencer of the new
transaction, the bus address pointer for the cdb must also
be recalculated to reflect its new location. We now defer
the calculation of the cdb address until just before queing
it to the card.
When pulling transfer negotiation settings out of scratch
ram, convert 5MHz/clock doubled settings to 10MHz.
Add a new function ahc_qinfifo_requeue_tail() for use by
error recovery actions and auto-request sense operations.
These operations always occur when the sequencer is paused,
so we can avoid the extra expense incurred in the normal
SCB queue method.
Use the BMOV instruction for all single byte moves on
controllers that support it. The bmov instruction is
twice as fast as an AND with an immediate of 0xFF as
is used on older controllers.
Correct a few bugs in ahc_dump_card_state(). If we have
hardware assisted queue registers, use them to get the
sequencer's idea of the head of the queue. When enumerating
the untagged queue, it helps to use the correct index for
the queue.
aic7xxx.h:
Indicate via a feature flag, which controllers can take
on both the target and the initiator role at the same time.
Add the AHC_SEQUENCER_DEBUG flag.
Add the SCB_CDB32_PTR flag used for dealing with cdbs
with lengths between 13 and 32 bytes.
Add new prototypes.
aic7xxx.reg:
Allow the SCSIBUSL register to be written to. This is
required to fix a selection timeout problem on the 7892/99.
Cleanup the sequencer interrupt codes so that all debugging
codes are grouped at the end of the list.
Correct the definition of the ULTRA_ENB and DISC_DSB locations
in scratch ram. This prevented the driver from properly honoring
these settings when no serial eeprom was available.
Remove an unused sequencer flag.
aic7xxx.seq:
Just before a potential select-out, clear the SCSIBUSL
register. Occasionally, during a selection timeout, the
contents of the register may be presented on the bus,
causing much confusion.
Add sequencer diagnostic code to detect software and or
hardware bugs. The code attempts to verify most list
operations so any corruption is caught before it occurs.
We also track information about why a particular reconnection
request was rejected.
Don't clobber the digital REQ/ACK filter setting in SXFRCTL0
when clearing the channel.
Fix a target mode bug that would cause us to return busy
status instead of queue full in respnse to a tagged transaction.
Cleanup the overrun case. It turns out that by simply
butting the chip in bitbucket mode, it will ack any
bytes until the phase changes. This drasticaly simplifies
things.
Prior to leaving the data phase, make sure that the S/G
preload queue is empty.
Remove code to place a request sense request on the waiting
queue. This is all handled by the kernel now.
Change the semantics of "findSCB". In the past, findSCB
ensured that a freshly paged in SCB appeared on the disconnected
list. The problem with this is that there is no guarantee that
the paged in SCB is for a disconnected transation. We now
defer any list manipulation to the caller who usually discards
the SCB via the free list.
Inline some busy target table operations.
Add a critical section to protect adding an SCB to
the disconnected list.
aic7xxx_freebsd.c:
Handle changes in the transfer negotiation setting API
to filter incoming requests. No filtering is necessary
for "goal" requests from the XPT.
Set the SCB_CDB32_PTR flag when queing a transaction with
a large cdb.
In ahc_timeout, only take action if the active SCB is
the timedout SCB. This deals with the case of two
transactions to the same device with different timeout
values.
Use ahc_qinfifo_requeu_tail() instead of home grown
version.
aic7xxx_inline.h:
Honor SCB_CDB32_PTR when queuing a new request.
aic7xxx_pci.c:
Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
|
|
|
scb->flags |= SCB_CDB32_PTR;
|
2000-09-16 20:02:28 +00:00
|
|
|
} else {
|
|
|
|
|
memcpy(hscb->shared_data.cdb,
|
|
|
|
|
csio->cdb_io.cdb_ptr,
|
|
|
|
|
hscb->cdb_len);
|
|
|
|
|
}
|
|
|
|
|
} else {
|
|
|
|
|
if (hscb->cdb_len > 12) {
|
|
|
|
|
memcpy(hscb->cdb32, csio->cdb_io.cdb_bytes,
|
|
|
|
|
hscb->cdb_len);
|
aic7xxx.c:
Filter incoming transfer negotiation requests to ensure they
never exceed the settings specified by the user.
In restart sequencer attempt to deal with a bug in the aic7895.
If a third party reset occurs at just the right time, the
stack register can lock up. When restarting the sequencer
after handling the SCSI reset, poke SEQADDR1 before resting
the sequencers program counter.
When something strange happens, dump the card's transaction
state via ahc_dump_card_state(). This should aid in debugging.
Handle request sense transactions via the QINFIFO instead of
attaching them to the waiting queue directly. The waiting
queue consumes card SCB resources and, in the pathological case
of every target on the bus beating our selection attemps and
issuing a check condition, could have caused us to run out
of SCBs. I have never seen this happen, and only early
cards with 3 or 4 SCBs had any real chance of ever getting
into this state.
Add additional sequencer interrupt codes to support firmware
diagnostics. The diagnostic code is enabled with the
AHC_DEBUG_SEQUENCER kernel option.
Make it possible to switch into and out of target mode on
the fly. The card comes up by default as an initiator but
will switch into target mode as soon as an enable lun operation
is performed. As always, target mode behavior is gated
by the AHC_TMODE_ENABLE kernel option so most users will
not be affected by this change.
In ahc_update_target_msg_request(), also issue a new
request if the ppr_options have changed.
Never issue a PPR as a target. It is forbidden by the spec.
Correct a bug in ahc_parse_msg() that prevented us from
responding to PPR messages as a target.
Mark SCBs that are on the untagged queue with a flag instead
of checking several fields in the SCB to see if the SCB should
be on the queue. This makes it easier for things like automatic
request sense requests to be queued without touching the
untagged queues even though they are untagged requests.
When dealing with ignore wide residue messages that occur
in the middle of a transfer, reset HADDR, not SHADDR for
non-ultra2 chips. Although SHADDR is where the firmware
fetches the ending transfer address for a save data pointers
request, it is readonly. Setting HADDR has the side effect
of also updating SHADDR.
Cleanup the output of ahc_dump_card_state() by nulling out the
free scb list in the non-paging case. The free list is only
used if we must page SCBs.
Correct the transmission of cdbs > 12 bytes in length. When
swapping HSCBs prior to notifing the sequencer of the new
transaction, the bus address pointer for the cdb must also
be recalculated to reflect its new location. We now defer
the calculation of the cdb address until just before queing
it to the card.
When pulling transfer negotiation settings out of scratch
ram, convert 5MHz/clock doubled settings to 10MHz.
Add a new function ahc_qinfifo_requeue_tail() for use by
error recovery actions and auto-request sense operations.
These operations always occur when the sequencer is paused,
so we can avoid the extra expense incurred in the normal
SCB queue method.
Use the BMOV instruction for all single byte moves on
controllers that support it. The bmov instruction is
twice as fast as an AND with an immediate of 0xFF as
is used on older controllers.
Correct a few bugs in ahc_dump_card_state(). If we have
hardware assisted queue registers, use them to get the
sequencer's idea of the head of the queue. When enumerating
the untagged queue, it helps to use the correct index for
the queue.
aic7xxx.h:
Indicate via a feature flag, which controllers can take
on both the target and the initiator role at the same time.
Add the AHC_SEQUENCER_DEBUG flag.
Add the SCB_CDB32_PTR flag used for dealing with cdbs
with lengths between 13 and 32 bytes.
Add new prototypes.
aic7xxx.reg:
Allow the SCSIBUSL register to be written to. This is
required to fix a selection timeout problem on the 7892/99.
Cleanup the sequencer interrupt codes so that all debugging
codes are grouped at the end of the list.
Correct the definition of the ULTRA_ENB and DISC_DSB locations
in scratch ram. This prevented the driver from properly honoring
these settings when no serial eeprom was available.
Remove an unused sequencer flag.
aic7xxx.seq:
Just before a potential select-out, clear the SCSIBUSL
register. Occasionally, during a selection timeout, the
contents of the register may be presented on the bus,
causing much confusion.
Add sequencer diagnostic code to detect software and or
hardware bugs. The code attempts to verify most list
operations so any corruption is caught before it occurs.
We also track information about why a particular reconnection
request was rejected.
Don't clobber the digital REQ/ACK filter setting in SXFRCTL0
when clearing the channel.
Fix a target mode bug that would cause us to return busy
status instead of queue full in respnse to a tagged transaction.
Cleanup the overrun case. It turns out that by simply
butting the chip in bitbucket mode, it will ack any
bytes until the phase changes. This drasticaly simplifies
things.
Prior to leaving the data phase, make sure that the S/G
preload queue is empty.
Remove code to place a request sense request on the waiting
queue. This is all handled by the kernel now.
Change the semantics of "findSCB". In the past, findSCB
ensured that a freshly paged in SCB appeared on the disconnected
list. The problem with this is that there is no guarantee that
the paged in SCB is for a disconnected transation. We now
defer any list manipulation to the caller who usually discards
the SCB via the free list.
Inline some busy target table operations.
Add a critical section to protect adding an SCB to
the disconnected list.
aic7xxx_freebsd.c:
Handle changes in the transfer negotiation setting API
to filter incoming requests. No filtering is necessary
for "goal" requests from the XPT.
Set the SCB_CDB32_PTR flag when queing a transaction with
a large cdb.
In ahc_timeout, only take action if the active SCB is
the timedout SCB. This deals with the case of two
transactions to the same device with different timeout
values.
Use ahc_qinfifo_requeu_tail() instead of home grown
version.
aic7xxx_inline.h:
Honor SCB_CDB32_PTR when queuing a new request.
aic7xxx_pci.c:
Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
|
|
|
scb->flags |= SCB_CDB32_PTR;
|
2000-09-16 20:02:28 +00:00
|
|
|
} else {
|
|
|
|
|
memcpy(hscb->shared_data.cdb,
|
|
|
|
|
csio->cdb_io.cdb_bytes,
|
|
|
|
|
hscb->cdb_len);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* 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(ahc->buffer_dmat,
|
|
|
|
|
scb->dmamap,
|
|
|
|
|
csio->data_ptr,
|
|
|
|
|
csio->dxfer_len,
|
|
|
|
|
ahc_execute_scb,
|
|
|
|
|
scb, /*flags*/0);
|
|
|
|
|
if (error == EINPROGRESS) {
|
|
|
|
|
/*
|
|
|
|
|
* So as to maintain ordering,
|
|
|
|
|
* freeze the controller queue
|
|
|
|
|
* until our mapping is
|
|
|
|
|
* returned.
|
|
|
|
|
*/
|
|
|
|
|
xpt_freeze_simq(sim,
|
|
|
|
|
/*count*/1);
|
|
|
|
|
scb->io_ctx->ccb_h.status |=
|
|
|
|
|
CAM_RELEASE_SIMQ;
|
|
|
|
|
}
|
|
|
|
|
splx(s);
|
|
|
|
|
} else {
|
|
|
|
|
struct bus_dma_segment seg;
|
|
|
|
|
|
|
|
|
|
/* Pointer to physical buffer */
|
|
|
|
|
if (csio->dxfer_len > AHC_MAXTRANSFER_SIZE)
|
|
|
|
|
panic("ahc_setup_data - Transfer size "
|
|
|
|
|
"larger than can device max");
|
|
|
|
|
|
|
|
|
|
seg.ds_addr = (bus_addr_t)csio->data_ptr;
|
|
|
|
|
seg.ds_len = csio->dxfer_len;
|
|
|
|
|
ahc_execute_scb(scb, &seg, 1, 0);
|
|
|
|
|
}
|
|
|
|
|
} else {
|
|
|
|
|
struct bus_dma_segment *segs;
|
|
|
|
|
|
|
|
|
|
if ((ccb_h->flags & CAM_DATA_PHYS) != 0)
|
|
|
|
|
panic("ahc_setup_data - Physical segment "
|
|
|
|
|
"pointers unsupported");
|
|
|
|
|
|
|
|
|
|
if ((ccb_h->flags & CAM_SG_LIST_PHYS) == 0)
|
|
|
|
|
panic("ahc_setup_data - Virtual segment "
|
|
|
|
|
"addresses unsupported");
|
|
|
|
|
|
|
|
|
|
/* Just use the segments provided */
|
|
|
|
|
segs = (struct bus_dma_segment *)csio->data_ptr;
|
|
|
|
|
ahc_execute_scb(scb, segs, csio->sglist_cnt, 0);
|
|
|
|
|
}
|
|
|
|
|
} else {
|
|
|
|
|
ahc_execute_scb(scb, NULL, 0, 0);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static void
|
|
|
|
|
ahc_set_recoveryscb(struct ahc_softc *ahc, struct scb *scb) {
|
|
|
|
|
|
|
|
|
|
if ((scb->flags & SCB_RECOVERY_SCB) == 0) {
|
|
|
|
|
struct scb *list_scb;
|
|
|
|
|
|
|
|
|
|
scb->flags |= SCB_RECOVERY_SCB;
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Take all queued, but not sent SCBs out of the equation.
|
|
|
|
|
* Also ensure that no new CCBs are queued to us while we
|
|
|
|
|
* try to fix this problem.
|
|
|
|
|
*/
|
|
|
|
|
if ((scb->io_ctx->ccb_h.status & CAM_RELEASE_SIMQ) == 0) {
|
|
|
|
|
xpt_freeze_simq(SCB_GET_SIM(ahc, scb), /*count*/1);
|
|
|
|
|
scb->io_ctx->ccb_h.status |= CAM_RELEASE_SIMQ;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Go through all of our pending SCBs and remove
|
|
|
|
|
* any scheduled timeouts for them. We will reschedule
|
|
|
|
|
* them after we've successfully fixed this problem.
|
|
|
|
|
*/
|
|
|
|
|
LIST_FOREACH(list_scb, &ahc->pending_scbs, pending_links) {
|
|
|
|
|
union ccb *ccb;
|
|
|
|
|
|
|
|
|
|
ccb = list_scb->io_ctx;
|
|
|
|
|
untimeout(ahc_timeout, list_scb, ccb->ccb_h.timeout_ch);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
void
|
|
|
|
|
ahc_timeout(void *arg)
|
|
|
|
|
{
|
|
|
|
|
struct scb *scb;
|
|
|
|
|
struct ahc_softc *ahc;
|
|
|
|
|
long s;
|
|
|
|
|
int found;
|
|
|
|
|
u_int last_phase;
|
|
|
|
|
int target;
|
|
|
|
|
int lun;
|
|
|
|
|
int i;
|
|
|
|
|
char channel;
|
|
|
|
|
|
|
|
|
|
scb = (struct scb *)arg;
|
2000-10-11 23:46:34 +00:00
|
|
|
ahc = (struct ahc_softc *)scb->ahc_softc;
|
2000-09-16 20:02:28 +00:00
|
|
|
|
|
|
|
|
ahc_lock(ahc, &s);
|
|
|
|
|
|
ahc_eisa.c:
ahc_pci.c:
Add detach support.
Make use of soft allocated on our behalf by newbus.
For PCI devices, disable the mapping type we aren't
using for extra protection from rogue code.
aic7xxx_93cx6.c:
aic7xxx_93cx6.h:
Sync perforce IDs.
aic7xxx_freebsd.c:
Capture the eventhandle returned by EVENTHANDER_REGISTER
so we can kill the handler off during detach.
Use AHC_* constants instead of hard coded numbers in a
few more places.
Test PPR option state when deciding to "really" negotiate
when the CAM_NEGOTIATE flag is passed in a CCB.
Make use of core "ahc_pause_and_flushwork" routine in our
timeout handler rather than re-inventing this code.
Cleanup all of our resources (really!) in ahc_platform_free().
We should be all set to become a module now.
Implement the core ahc_detach() routine shared by all of
the FreeBSD front-ends.
aic7xxx_freebsd.h:
Softc storage for our event handler.
Null implementation for the ahc_platform_flushwork() OSM
callback. FreeBSD doesn't need this as XPT callbacks are
safe from all contexts and are done directly in ahc_done().
aic7xxx_inline.h:
Implement new lazy interrupt scheme. To avoid an extra
PCI bus read, we first check our completion queues to
see if any work has completed. If work is available, we
assume that this is the source of the interrupt and skip
reading INTSTAT. Any remaining interrupt status will be
cleared by a second call to the interrupt handler should
the interrupt line still be asserted. This drops the
interrupt handler down to a single PCI bus read in the
common case of I/O completion. This is the same overhead
as in the not so distant past, but the extra sanity of
perforning a PCI read after clearing the command complete
interrupt and before running the completion queue to avoid
missing command complete interrupts added a cycle.
aic7xxx.c:
During initialization, be sure to initialize all scratch
ram locations before they are read to avoid parity errors.
In this case, we use a new function, ahc_unbusy_tcl() to
initialize the scratch ram busy target table.
Replace instances of ahc_index_busy_tcl() used to unbusy
a tcl without looking at the old value with ahc_unbusy_tcl().
Modify ahc_sent_msg so that it can find single byte messages.
ahc_sent_msg is now used to determine if a transfer negotiation
attempt resulted in a bus free.
Be more careful in filtering out only the SCSI interrupts
of interest in ahc_handle_scsiint.
Rearrange interrupt clearing code to ensure that at least
one PCI transaction occurrs after hitting CLRSINT1 and
writting to CLRINT. CLRSINT1 writes take a bit to
take effect, and the re-arrangement provides sufficient
delay to ensure the write to CLRINT is effective. The
old code might report a spurious interrupt on some "fast"
chipsets.
export ahc-update_target_msg_request for use by OSM code.
If a target does not respond to our ATN request, clear
it once we move to a non-message phase. This avoids
sending a MSG_NOOP in some later message out phase.
Use max lun and max target constants instead of
hard-coded values.
Use softc storage built into our device_t under FreeBSD.
Fix a bug in ahc_free() that caused us to delete
resources that were not allocated.
Clean up any tstate/lstate info in ahc_free().
Clear the powerdown state in ahc_reset() so that
registers can be accessed.
Add a preliminary function for pausing the chip and
processing any posted work.
Add a preliminary suspend and resume functions.
aic7xxx.h:
Limit the number of supported luns to 64. We don't
support information unit transfers, so this is the
maximum that makes sense for these chips.
Add a new flag AHC_ALL_INTERRUPTS that forces the
processing of all interrupt state in a single invokation
of ahc_intr(). When the flag is not set, we use the
lazy interrupt handling scheme.
Add data structures to store controller state while
we are suspended.
Use constants instead of hard coded values where appropriate.
Correct some harmless "unsigned/signed" conflicts.
aic7xxx.seq:
Only perform the SCSIBUSL fix on ULTRA2 or newer controllers.
Older controllers seem to be confused by this.
In target mode, ignore PHASEMIS during data phases.
This bit seems to be flakey on U160 controllers acting
in target mode.
aic7xxx_pci.c:
Add support for the 29160C CPCI adapter.
Add definitions for subvendor ID information
available for devices with the "9005" vendor id.
We currently use this information to determine
if a multi-function device doesn't have the second
channel hooked up on a board.
Add rudimentary power mode code so we can put the
controller into the D0 state. In the future this
will be an OSM callback so that in FreeBSD we don't
duplicate functionality provided by the PCI code.
The powerstate code was added after I'd completed
my regression tests on this code.
Only capture "left over BIOS state" if the POWRDN
setting is not set in HCNTRL.
In target mode, don't bother sending incremental
CRC data.
2000-12-20 01:11:37 +00:00
|
|
|
ahc_pause_and_flushwork(ahc);
|
2000-10-08 03:37:52 +00:00
|
|
|
|
2000-09-16 20:02:28 +00:00
|
|
|
if ((scb->flags & SCB_ACTIVE) == 0) {
|
|
|
|
|
/* Previous timeout took care of me already */
|
2000-10-11 23:46:34 +00:00
|
|
|
printf("%s: Timedout SCB already complete. "
|
|
|
|
|
"Interrupts may not be functioning.\n", ahc_name(ahc));
|
This is an MFC candidate.
ahc_eisa.c:
Change aic7770_map_int to take an additional irq parameter.
Although we can get the irq from the eisa dev under FreeBSD,
we can't do this under linux, so the OSM interface must supply
this.
ahc_pci.c:
Move ahc_power_state_change() to the OSM. This allows us to
use a platform supplied function that does the same thing.
-current will move to the FreeBSD native API in the near
future.
aic7770.c:
Sync up with core changes to support Linux EISA.
We now store a 2 bit primary channel number rather
than a bit flag that only allows b to be the primary
channel. Adjust for this change.
aic7xxx.c:
Namespace and staticization cleanup. All exported symbols
use an "ahc_" prefix to avoid collisions with other modules.
Correct a logic bug that prevented us from dropping
ATN during some exceptional conditions during message
processing.
Take advantage of a new flag managed by the sequencer
that indicates if an SCB fetch is in progress. If so,
the currently selected SCB needs to be returned to the
free list to prevent an SCB leak. This leak is a rarity
and would only occur if a bus reset or timeout resulting
in a bus reset occurred in the middle of an SCB fetch.
Don't attempt to perform ULTRA transfers on ultra capable
adapters missing the external precision resistor required
for ultra speeds. I've never encountered an adapter
configured this way, but better safe than sorry.
Handle the case of 5MHz user sync rate set as "0" instead of 0x1c
in scratch ram.
If we lookup a period of 0 in our table (async), clear the scsi offset.
aic7xxx.h:
Adjust for the primary channel being represented as
a 2 bit integer in the flags member of the ahc softc.
Cleanup the flags definitions so that comment blocks are
not cramped.
Update seeprom definitions to correctly reflect the fact
that the primary channel is represented as a 2 bit integer.
Add AHC_ULTRA_DIASABLED softc flag to denote controllers
missing the external precision resistor.
aic7xxx.reg:
Add DFCACHETH to the definition of DFSTATUS for completness sake.
Add SEQ_FLAGS2 which currently only contains the SCB_DMA
(SCB DMA in progress) flag.
aic7xxx.seq:
Correct a problem when one lun has a disconnected untagged
transaction and another lun has disconnected tagged transactions.
Just because an entry is found in the untagged table doesn't
mean that it will match. If the match on the lun fails, cleanup
the SCB (return it to the disconnected list or free it), and snoop
for a tag message. Before this change, we reported an unsolicited
reselection. This bug was introduced about a month ago during an
overly aggressive optimization pass on the reselection code.
When cleaning up an SCB, we can't just blindly free the SCB. In
the paging case, if the SCB came off of the disconnected list, its
state may never have been updated in host memory. So, check the
disconnected bit in SCB_CONTROL and return the SCB to the disconnected
list if appropriate.
Manage the SCB_DMA flag of SEQ_FLAGS2.
More carefully shutdown the S/G dma engine in all cases by using
a subroutine. Supposedly not doing this can cause an arbiter hang
on some ULTRA2 chips.
Formatting cleanup.
On some chips, at least the aic7856, the transition from
MREQPEND to HDONE can take a full 4 clock cycles. Test
HDONE one more time to avoid this race. We only want our
FIFO hung recovery code to execute when the engine is
really hung.
aic7xxx_93cx6.c:
Sync perforce ids.
aic7xxx_freebsd.c:
Adjust for the primary channel being a 2 bit integer
rather than a flag for 'B' channel being the primary.
Namespace cleanup.
Unpause the sequencer in one error recovery path that
neglected to do so. This could have caused us to perform
a bus reset when a recovery message might have otherwise been
successful.
aic7xxx_freebsd.h:
Use AHC_PCI_CONFIG for controlling compilation of PCI
support consistently throughout the driver.
Move ahc_power_state_change() to OSM.
aic7xxx_inline.h
Namespace cleanup.
Adjust our interrupt handler so it will work in the edge
interrupt case. We must process all interrupt sources
when the interrupt fires or risk not ever getting an
interrupt again. This involves marking the fact
that we are relying on an edge interrupt in ahc->flags
and checking for this condition in addition to the
AHC_ALL_INTERRUPTS flag. This fixes hangs on the
284X and any other aic7770 installation where level
interrupts are not available.
aic7xxx_pci.c:
Move the powerstate manipulation code into the OSM. Several
OSes now provide this functionality natively.
Take another shot at using the data stored in scratch ram
if the SCB2 signature is correct and no SEEPROM data is
available. In the past this failed if external SCB ram
was configured because the memory port was locked. We
now release the memory port prior to testing the values
in SCB2 and re-acquire it prior to doing termination control.
Adjust for new 2 bit primary channel setting.
Trust the STPWLEVEL setting on v 3.X BIOSes too.
Configure any 785X ID in the same fashion and assume
that any device with a rev id of 1 or higher has the
PCI 2.1 retry bug.
2001-03-11 06:34:17 +00:00
|
|
|
ahc_unpause(ahc);
|
2000-09-16 20:02:28 +00:00
|
|
|
ahc_unlock(ahc, &s);
|
|
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
target = SCB_GET_TARGET(ahc, scb);
|
|
|
|
|
channel = SCB_GET_CHANNEL(ahc, scb);
|
|
|
|
|
lun = SCB_GET_LUN(scb);
|
|
|
|
|
|
aic7xxx.c:
Filter incoming transfer negotiation requests to ensure they
never exceed the settings specified by the user.
In restart sequencer attempt to deal with a bug in the aic7895.
If a third party reset occurs at just the right time, the
stack register can lock up. When restarting the sequencer
after handling the SCSI reset, poke SEQADDR1 before resting
the sequencers program counter.
When something strange happens, dump the card's transaction
state via ahc_dump_card_state(). This should aid in debugging.
Handle request sense transactions via the QINFIFO instead of
attaching them to the waiting queue directly. The waiting
queue consumes card SCB resources and, in the pathological case
of every target on the bus beating our selection attemps and
issuing a check condition, could have caused us to run out
of SCBs. I have never seen this happen, and only early
cards with 3 or 4 SCBs had any real chance of ever getting
into this state.
Add additional sequencer interrupt codes to support firmware
diagnostics. The diagnostic code is enabled with the
AHC_DEBUG_SEQUENCER kernel option.
Make it possible to switch into and out of target mode on
the fly. The card comes up by default as an initiator but
will switch into target mode as soon as an enable lun operation
is performed. As always, target mode behavior is gated
by the AHC_TMODE_ENABLE kernel option so most users will
not be affected by this change.
In ahc_update_target_msg_request(), also issue a new
request if the ppr_options have changed.
Never issue a PPR as a target. It is forbidden by the spec.
Correct a bug in ahc_parse_msg() that prevented us from
responding to PPR messages as a target.
Mark SCBs that are on the untagged queue with a flag instead
of checking several fields in the SCB to see if the SCB should
be on the queue. This makes it easier for things like automatic
request sense requests to be queued without touching the
untagged queues even though they are untagged requests.
When dealing with ignore wide residue messages that occur
in the middle of a transfer, reset HADDR, not SHADDR for
non-ultra2 chips. Although SHADDR is where the firmware
fetches the ending transfer address for a save data pointers
request, it is readonly. Setting HADDR has the side effect
of also updating SHADDR.
Cleanup the output of ahc_dump_card_state() by nulling out the
free scb list in the non-paging case. The free list is only
used if we must page SCBs.
Correct the transmission of cdbs > 12 bytes in length. When
swapping HSCBs prior to notifing the sequencer of the new
transaction, the bus address pointer for the cdb must also
be recalculated to reflect its new location. We now defer
the calculation of the cdb address until just before queing
it to the card.
When pulling transfer negotiation settings out of scratch
ram, convert 5MHz/clock doubled settings to 10MHz.
Add a new function ahc_qinfifo_requeue_tail() for use by
error recovery actions and auto-request sense operations.
These operations always occur when the sequencer is paused,
so we can avoid the extra expense incurred in the normal
SCB queue method.
Use the BMOV instruction for all single byte moves on
controllers that support it. The bmov instruction is
twice as fast as an AND with an immediate of 0xFF as
is used on older controllers.
Correct a few bugs in ahc_dump_card_state(). If we have
hardware assisted queue registers, use them to get the
sequencer's idea of the head of the queue. When enumerating
the untagged queue, it helps to use the correct index for
the queue.
aic7xxx.h:
Indicate via a feature flag, which controllers can take
on both the target and the initiator role at the same time.
Add the AHC_SEQUENCER_DEBUG flag.
Add the SCB_CDB32_PTR flag used for dealing with cdbs
with lengths between 13 and 32 bytes.
Add new prototypes.
aic7xxx.reg:
Allow the SCSIBUSL register to be written to. This is
required to fix a selection timeout problem on the 7892/99.
Cleanup the sequencer interrupt codes so that all debugging
codes are grouped at the end of the list.
Correct the definition of the ULTRA_ENB and DISC_DSB locations
in scratch ram. This prevented the driver from properly honoring
these settings when no serial eeprom was available.
Remove an unused sequencer flag.
aic7xxx.seq:
Just before a potential select-out, clear the SCSIBUSL
register. Occasionally, during a selection timeout, the
contents of the register may be presented on the bus,
causing much confusion.
Add sequencer diagnostic code to detect software and or
hardware bugs. The code attempts to verify most list
operations so any corruption is caught before it occurs.
We also track information about why a particular reconnection
request was rejected.
Don't clobber the digital REQ/ACK filter setting in SXFRCTL0
when clearing the channel.
Fix a target mode bug that would cause us to return busy
status instead of queue full in respnse to a tagged transaction.
Cleanup the overrun case. It turns out that by simply
butting the chip in bitbucket mode, it will ack any
bytes until the phase changes. This drasticaly simplifies
things.
Prior to leaving the data phase, make sure that the S/G
preload queue is empty.
Remove code to place a request sense request on the waiting
queue. This is all handled by the kernel now.
Change the semantics of "findSCB". In the past, findSCB
ensured that a freshly paged in SCB appeared on the disconnected
list. The problem with this is that there is no guarantee that
the paged in SCB is for a disconnected transation. We now
defer any list manipulation to the caller who usually discards
the SCB via the free list.
Inline some busy target table operations.
Add a critical section to protect adding an SCB to
the disconnected list.
aic7xxx_freebsd.c:
Handle changes in the transfer negotiation setting API
to filter incoming requests. No filtering is necessary
for "goal" requests from the XPT.
Set the SCB_CDB32_PTR flag when queing a transaction with
a large cdb.
In ahc_timeout, only take action if the active SCB is
the timedout SCB. This deals with the case of two
transactions to the same device with different timeout
values.
Use ahc_qinfifo_requeu_tail() instead of home grown
version.
aic7xxx_inline.h:
Honor SCB_CDB32_PTR when queuing a new request.
aic7xxx_pci.c:
Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
|
|
|
ahc_print_path(ahc, scb);
|
2001-03-29 00:36:35 +00:00
|
|
|
printf("SCB 0x%x - timed out\n", scb->hscb->tag);
|
aic7xxx.c:
Filter incoming transfer negotiation requests to ensure they
never exceed the settings specified by the user.
In restart sequencer attempt to deal with a bug in the aic7895.
If a third party reset occurs at just the right time, the
stack register can lock up. When restarting the sequencer
after handling the SCSI reset, poke SEQADDR1 before resting
the sequencers program counter.
When something strange happens, dump the card's transaction
state via ahc_dump_card_state(). This should aid in debugging.
Handle request sense transactions via the QINFIFO instead of
attaching them to the waiting queue directly. The waiting
queue consumes card SCB resources and, in the pathological case
of every target on the bus beating our selection attemps and
issuing a check condition, could have caused us to run out
of SCBs. I have never seen this happen, and only early
cards with 3 or 4 SCBs had any real chance of ever getting
into this state.
Add additional sequencer interrupt codes to support firmware
diagnostics. The diagnostic code is enabled with the
AHC_DEBUG_SEQUENCER kernel option.
Make it possible to switch into and out of target mode on
the fly. The card comes up by default as an initiator but
will switch into target mode as soon as an enable lun operation
is performed. As always, target mode behavior is gated
by the AHC_TMODE_ENABLE kernel option so most users will
not be affected by this change.
In ahc_update_target_msg_request(), also issue a new
request if the ppr_options have changed.
Never issue a PPR as a target. It is forbidden by the spec.
Correct a bug in ahc_parse_msg() that prevented us from
responding to PPR messages as a target.
Mark SCBs that are on the untagged queue with a flag instead
of checking several fields in the SCB to see if the SCB should
be on the queue. This makes it easier for things like automatic
request sense requests to be queued without touching the
untagged queues even though they are untagged requests.
When dealing with ignore wide residue messages that occur
in the middle of a transfer, reset HADDR, not SHADDR for
non-ultra2 chips. Although SHADDR is where the firmware
fetches the ending transfer address for a save data pointers
request, it is readonly. Setting HADDR has the side effect
of also updating SHADDR.
Cleanup the output of ahc_dump_card_state() by nulling out the
free scb list in the non-paging case. The free list is only
used if we must page SCBs.
Correct the transmission of cdbs > 12 bytes in length. When
swapping HSCBs prior to notifing the sequencer of the new
transaction, the bus address pointer for the cdb must also
be recalculated to reflect its new location. We now defer
the calculation of the cdb address until just before queing
it to the card.
When pulling transfer negotiation settings out of scratch
ram, convert 5MHz/clock doubled settings to 10MHz.
Add a new function ahc_qinfifo_requeue_tail() for use by
error recovery actions and auto-request sense operations.
These operations always occur when the sequencer is paused,
so we can avoid the extra expense incurred in the normal
SCB queue method.
Use the BMOV instruction for all single byte moves on
controllers that support it. The bmov instruction is
twice as fast as an AND with an immediate of 0xFF as
is used on older controllers.
Correct a few bugs in ahc_dump_card_state(). If we have
hardware assisted queue registers, use them to get the
sequencer's idea of the head of the queue. When enumerating
the untagged queue, it helps to use the correct index for
the queue.
aic7xxx.h:
Indicate via a feature flag, which controllers can take
on both the target and the initiator role at the same time.
Add the AHC_SEQUENCER_DEBUG flag.
Add the SCB_CDB32_PTR flag used for dealing with cdbs
with lengths between 13 and 32 bytes.
Add new prototypes.
aic7xxx.reg:
Allow the SCSIBUSL register to be written to. This is
required to fix a selection timeout problem on the 7892/99.
Cleanup the sequencer interrupt codes so that all debugging
codes are grouped at the end of the list.
Correct the definition of the ULTRA_ENB and DISC_DSB locations
in scratch ram. This prevented the driver from properly honoring
these settings when no serial eeprom was available.
Remove an unused sequencer flag.
aic7xxx.seq:
Just before a potential select-out, clear the SCSIBUSL
register. Occasionally, during a selection timeout, the
contents of the register may be presented on the bus,
causing much confusion.
Add sequencer diagnostic code to detect software and or
hardware bugs. The code attempts to verify most list
operations so any corruption is caught before it occurs.
We also track information about why a particular reconnection
request was rejected.
Don't clobber the digital REQ/ACK filter setting in SXFRCTL0
when clearing the channel.
Fix a target mode bug that would cause us to return busy
status instead of queue full in respnse to a tagged transaction.
Cleanup the overrun case. It turns out that by simply
butting the chip in bitbucket mode, it will ack any
bytes until the phase changes. This drasticaly simplifies
things.
Prior to leaving the data phase, make sure that the S/G
preload queue is empty.
Remove code to place a request sense request on the waiting
queue. This is all handled by the kernel now.
Change the semantics of "findSCB". In the past, findSCB
ensured that a freshly paged in SCB appeared on the disconnected
list. The problem with this is that there is no guarantee that
the paged in SCB is for a disconnected transation. We now
defer any list manipulation to the caller who usually discards
the SCB via the free list.
Inline some busy target table operations.
Add a critical section to protect adding an SCB to
the disconnected list.
aic7xxx_freebsd.c:
Handle changes in the transfer negotiation setting API
to filter incoming requests. No filtering is necessary
for "goal" requests from the XPT.
Set the SCB_CDB32_PTR flag when queing a transaction with
a large cdb.
In ahc_timeout, only take action if the active SCB is
the timedout SCB. This deals with the case of two
transactions to the same device with different timeout
values.
Use ahc_qinfifo_requeu_tail() instead of home grown
version.
aic7xxx_inline.h:
Honor SCB_CDB32_PTR when queuing a new request.
aic7xxx_pci.c:
Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
|
|
|
ahc_dump_card_state(ahc);
|
2001-03-29 00:36:35 +00:00
|
|
|
last_phase = ahc_inb(ahc, LASTPHASE);
|
2000-09-16 20:02:28 +00:00
|
|
|
if (scb->sg_count > 0) {
|
|
|
|
|
for (i = 0; i < scb->sg_count; i++) {
|
|
|
|
|
printf("sg[%d] - Addr 0x%x : Length %d\n",
|
|
|
|
|
i,
|
|
|
|
|
scb->sg_list[i].addr,
|
2000-10-08 03:37:52 +00:00
|
|
|
scb->sg_list[i].len & AHC_SG_LEN_MASK);
|
2000-09-16 20:02:28 +00:00
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
if (scb->flags & (SCB_DEVICE_RESET|SCB_ABORT)) {
|
|
|
|
|
/*
|
|
|
|
|
* Been down this road before.
|
|
|
|
|
* Do a full bus reset.
|
|
|
|
|
*/
|
|
|
|
|
bus_reset:
|
|
|
|
|
ahc_set_transaction_status(scb, CAM_CMD_TIMEOUT);
|
|
|
|
|
found = ahc_reset_channel(ahc, channel, /*Initiate Reset*/TRUE);
|
|
|
|
|
printf("%s: Issued Channel %c Bus Reset. "
|
|
|
|
|
"%d SCBs aborted\n", ahc_name(ahc), channel, found);
|
|
|
|
|
} else {
|
|
|
|
|
/*
|
|
|
|
|
* If we are a target, transition to bus free and report
|
|
|
|
|
* the timeout.
|
|
|
|
|
*
|
|
|
|
|
* The target/initiator that is holding up the bus may not
|
|
|
|
|
* be the same as the one that triggered this timeout
|
|
|
|
|
* (different commands have different timeout lengths).
|
|
|
|
|
* If the bus is idle and we are actiing as the initiator
|
|
|
|
|
* for this request, queue a BDR message to the timed out
|
|
|
|
|
* target. Otherwise, if the timed out transaction is
|
|
|
|
|
* active:
|
|
|
|
|
* Initiator transaction:
|
|
|
|
|
* Stuff the message buffer with a BDR message and assert
|
|
|
|
|
* ATN in the hopes that the target will let go of the bus
|
|
|
|
|
* and go to the mesgout phase. If this fails, we'll
|
|
|
|
|
* get another timeout 2 seconds later which will attempt
|
|
|
|
|
* a bus reset.
|
|
|
|
|
*
|
|
|
|
|
* Target transaction:
|
|
|
|
|
* Transition to BUS FREE and report the error.
|
|
|
|
|
* It's good to be the target!
|
|
|
|
|
*/
|
|
|
|
|
u_int active_scb_index;
|
aic7xxx.c:
Filter incoming transfer negotiation requests to ensure they
never exceed the settings specified by the user.
In restart sequencer attempt to deal with a bug in the aic7895.
If a third party reset occurs at just the right time, the
stack register can lock up. When restarting the sequencer
after handling the SCSI reset, poke SEQADDR1 before resting
the sequencers program counter.
When something strange happens, dump the card's transaction
state via ahc_dump_card_state(). This should aid in debugging.
Handle request sense transactions via the QINFIFO instead of
attaching them to the waiting queue directly. The waiting
queue consumes card SCB resources and, in the pathological case
of every target on the bus beating our selection attemps and
issuing a check condition, could have caused us to run out
of SCBs. I have never seen this happen, and only early
cards with 3 or 4 SCBs had any real chance of ever getting
into this state.
Add additional sequencer interrupt codes to support firmware
diagnostics. The diagnostic code is enabled with the
AHC_DEBUG_SEQUENCER kernel option.
Make it possible to switch into and out of target mode on
the fly. The card comes up by default as an initiator but
will switch into target mode as soon as an enable lun operation
is performed. As always, target mode behavior is gated
by the AHC_TMODE_ENABLE kernel option so most users will
not be affected by this change.
In ahc_update_target_msg_request(), also issue a new
request if the ppr_options have changed.
Never issue a PPR as a target. It is forbidden by the spec.
Correct a bug in ahc_parse_msg() that prevented us from
responding to PPR messages as a target.
Mark SCBs that are on the untagged queue with a flag instead
of checking several fields in the SCB to see if the SCB should
be on the queue. This makes it easier for things like automatic
request sense requests to be queued without touching the
untagged queues even though they are untagged requests.
When dealing with ignore wide residue messages that occur
in the middle of a transfer, reset HADDR, not SHADDR for
non-ultra2 chips. Although SHADDR is where the firmware
fetches the ending transfer address for a save data pointers
request, it is readonly. Setting HADDR has the side effect
of also updating SHADDR.
Cleanup the output of ahc_dump_card_state() by nulling out the
free scb list in the non-paging case. The free list is only
used if we must page SCBs.
Correct the transmission of cdbs > 12 bytes in length. When
swapping HSCBs prior to notifing the sequencer of the new
transaction, the bus address pointer for the cdb must also
be recalculated to reflect its new location. We now defer
the calculation of the cdb address until just before queing
it to the card.
When pulling transfer negotiation settings out of scratch
ram, convert 5MHz/clock doubled settings to 10MHz.
Add a new function ahc_qinfifo_requeue_tail() for use by
error recovery actions and auto-request sense operations.
These operations always occur when the sequencer is paused,
so we can avoid the extra expense incurred in the normal
SCB queue method.
Use the BMOV instruction for all single byte moves on
controllers that support it. The bmov instruction is
twice as fast as an AND with an immediate of 0xFF as
is used on older controllers.
Correct a few bugs in ahc_dump_card_state(). If we have
hardware assisted queue registers, use them to get the
sequencer's idea of the head of the queue. When enumerating
the untagged queue, it helps to use the correct index for
the queue.
aic7xxx.h:
Indicate via a feature flag, which controllers can take
on both the target and the initiator role at the same time.
Add the AHC_SEQUENCER_DEBUG flag.
Add the SCB_CDB32_PTR flag used for dealing with cdbs
with lengths between 13 and 32 bytes.
Add new prototypes.
aic7xxx.reg:
Allow the SCSIBUSL register to be written to. This is
required to fix a selection timeout problem on the 7892/99.
Cleanup the sequencer interrupt codes so that all debugging
codes are grouped at the end of the list.
Correct the definition of the ULTRA_ENB and DISC_DSB locations
in scratch ram. This prevented the driver from properly honoring
these settings when no serial eeprom was available.
Remove an unused sequencer flag.
aic7xxx.seq:
Just before a potential select-out, clear the SCSIBUSL
register. Occasionally, during a selection timeout, the
contents of the register may be presented on the bus,
causing much confusion.
Add sequencer diagnostic code to detect software and or
hardware bugs. The code attempts to verify most list
operations so any corruption is caught before it occurs.
We also track information about why a particular reconnection
request was rejected.
Don't clobber the digital REQ/ACK filter setting in SXFRCTL0
when clearing the channel.
Fix a target mode bug that would cause us to return busy
status instead of queue full in respnse to a tagged transaction.
Cleanup the overrun case. It turns out that by simply
butting the chip in bitbucket mode, it will ack any
bytes until the phase changes. This drasticaly simplifies
things.
Prior to leaving the data phase, make sure that the S/G
preload queue is empty.
Remove code to place a request sense request on the waiting
queue. This is all handled by the kernel now.
Change the semantics of "findSCB". In the past, findSCB
ensured that a freshly paged in SCB appeared on the disconnected
list. The problem with this is that there is no guarantee that
the paged in SCB is for a disconnected transation. We now
defer any list manipulation to the caller who usually discards
the SCB via the free list.
Inline some busy target table operations.
Add a critical section to protect adding an SCB to
the disconnected list.
aic7xxx_freebsd.c:
Handle changes in the transfer negotiation setting API
to filter incoming requests. No filtering is necessary
for "goal" requests from the XPT.
Set the SCB_CDB32_PTR flag when queing a transaction with
a large cdb.
In ahc_timeout, only take action if the active SCB is
the timedout SCB. This deals with the case of two
transactions to the same device with different timeout
values.
Use ahc_qinfifo_requeu_tail() instead of home grown
version.
aic7xxx_inline.h:
Honor SCB_CDB32_PTR when queuing a new request.
aic7xxx_pci.c:
Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
|
|
|
u_int saved_scbptr;
|
2000-09-16 20:02:28 +00:00
|
|
|
|
aic7xxx.c:
Filter incoming transfer negotiation requests to ensure they
never exceed the settings specified by the user.
In restart sequencer attempt to deal with a bug in the aic7895.
If a third party reset occurs at just the right time, the
stack register can lock up. When restarting the sequencer
after handling the SCSI reset, poke SEQADDR1 before resting
the sequencers program counter.
When something strange happens, dump the card's transaction
state via ahc_dump_card_state(). This should aid in debugging.
Handle request sense transactions via the QINFIFO instead of
attaching them to the waiting queue directly. The waiting
queue consumes card SCB resources and, in the pathological case
of every target on the bus beating our selection attemps and
issuing a check condition, could have caused us to run out
of SCBs. I have never seen this happen, and only early
cards with 3 or 4 SCBs had any real chance of ever getting
into this state.
Add additional sequencer interrupt codes to support firmware
diagnostics. The diagnostic code is enabled with the
AHC_DEBUG_SEQUENCER kernel option.
Make it possible to switch into and out of target mode on
the fly. The card comes up by default as an initiator but
will switch into target mode as soon as an enable lun operation
is performed. As always, target mode behavior is gated
by the AHC_TMODE_ENABLE kernel option so most users will
not be affected by this change.
In ahc_update_target_msg_request(), also issue a new
request if the ppr_options have changed.
Never issue a PPR as a target. It is forbidden by the spec.
Correct a bug in ahc_parse_msg() that prevented us from
responding to PPR messages as a target.
Mark SCBs that are on the untagged queue with a flag instead
of checking several fields in the SCB to see if the SCB should
be on the queue. This makes it easier for things like automatic
request sense requests to be queued without touching the
untagged queues even though they are untagged requests.
When dealing with ignore wide residue messages that occur
in the middle of a transfer, reset HADDR, not SHADDR for
non-ultra2 chips. Although SHADDR is where the firmware
fetches the ending transfer address for a save data pointers
request, it is readonly. Setting HADDR has the side effect
of also updating SHADDR.
Cleanup the output of ahc_dump_card_state() by nulling out the
free scb list in the non-paging case. The free list is only
used if we must page SCBs.
Correct the transmission of cdbs > 12 bytes in length. When
swapping HSCBs prior to notifing the sequencer of the new
transaction, the bus address pointer for the cdb must also
be recalculated to reflect its new location. We now defer
the calculation of the cdb address until just before queing
it to the card.
When pulling transfer negotiation settings out of scratch
ram, convert 5MHz/clock doubled settings to 10MHz.
Add a new function ahc_qinfifo_requeue_tail() for use by
error recovery actions and auto-request sense operations.
These operations always occur when the sequencer is paused,
so we can avoid the extra expense incurred in the normal
SCB queue method.
Use the BMOV instruction for all single byte moves on
controllers that support it. The bmov instruction is
twice as fast as an AND with an immediate of 0xFF as
is used on older controllers.
Correct a few bugs in ahc_dump_card_state(). If we have
hardware assisted queue registers, use them to get the
sequencer's idea of the head of the queue. When enumerating
the untagged queue, it helps to use the correct index for
the queue.
aic7xxx.h:
Indicate via a feature flag, which controllers can take
on both the target and the initiator role at the same time.
Add the AHC_SEQUENCER_DEBUG flag.
Add the SCB_CDB32_PTR flag used for dealing with cdbs
with lengths between 13 and 32 bytes.
Add new prototypes.
aic7xxx.reg:
Allow the SCSIBUSL register to be written to. This is
required to fix a selection timeout problem on the 7892/99.
Cleanup the sequencer interrupt codes so that all debugging
codes are grouped at the end of the list.
Correct the definition of the ULTRA_ENB and DISC_DSB locations
in scratch ram. This prevented the driver from properly honoring
these settings when no serial eeprom was available.
Remove an unused sequencer flag.
aic7xxx.seq:
Just before a potential select-out, clear the SCSIBUSL
register. Occasionally, during a selection timeout, the
contents of the register may be presented on the bus,
causing much confusion.
Add sequencer diagnostic code to detect software and or
hardware bugs. The code attempts to verify most list
operations so any corruption is caught before it occurs.
We also track information about why a particular reconnection
request was rejected.
Don't clobber the digital REQ/ACK filter setting in SXFRCTL0
when clearing the channel.
Fix a target mode bug that would cause us to return busy
status instead of queue full in respnse to a tagged transaction.
Cleanup the overrun case. It turns out that by simply
butting the chip in bitbucket mode, it will ack any
bytes until the phase changes. This drasticaly simplifies
things.
Prior to leaving the data phase, make sure that the S/G
preload queue is empty.
Remove code to place a request sense request on the waiting
queue. This is all handled by the kernel now.
Change the semantics of "findSCB". In the past, findSCB
ensured that a freshly paged in SCB appeared on the disconnected
list. The problem with this is that there is no guarantee that
the paged in SCB is for a disconnected transation. We now
defer any list manipulation to the caller who usually discards
the SCB via the free list.
Inline some busy target table operations.
Add a critical section to protect adding an SCB to
the disconnected list.
aic7xxx_freebsd.c:
Handle changes in the transfer negotiation setting API
to filter incoming requests. No filtering is necessary
for "goal" requests from the XPT.
Set the SCB_CDB32_PTR flag when queing a transaction with
a large cdb.
In ahc_timeout, only take action if the active SCB is
the timedout SCB. This deals with the case of two
transactions to the same device with different timeout
values.
Use ahc_qinfifo_requeu_tail() instead of home grown
version.
aic7xxx_inline.h:
Honor SCB_CDB32_PTR when queuing a new request.
aic7xxx_pci.c:
Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
|
|
|
saved_scbptr = ahc_inb(ahc, SCBPTR);
|
2000-09-16 20:02:28 +00:00
|
|
|
active_scb_index = ahc_inb(ahc, SCB_TAG);
|
|
|
|
|
|
2002-08-31 06:43:55 +00:00
|
|
|
if ((ahc_inb(ahc, SEQ_FLAGS) & IDENTIFY_SEEN) != 0
|
2000-09-16 20:02:28 +00:00
|
|
|
&& (active_scb_index < ahc->scb_data->numscbs)) {
|
|
|
|
|
struct scb *active_scb;
|
|
|
|
|
|
|
|
|
|
/*
|
aic7xxx.c:
Filter incoming transfer negotiation requests to ensure they
never exceed the settings specified by the user.
In restart sequencer attempt to deal with a bug in the aic7895.
If a third party reset occurs at just the right time, the
stack register can lock up. When restarting the sequencer
after handling the SCSI reset, poke SEQADDR1 before resting
the sequencers program counter.
When something strange happens, dump the card's transaction
state via ahc_dump_card_state(). This should aid in debugging.
Handle request sense transactions via the QINFIFO instead of
attaching them to the waiting queue directly. The waiting
queue consumes card SCB resources and, in the pathological case
of every target on the bus beating our selection attemps and
issuing a check condition, could have caused us to run out
of SCBs. I have never seen this happen, and only early
cards with 3 or 4 SCBs had any real chance of ever getting
into this state.
Add additional sequencer interrupt codes to support firmware
diagnostics. The diagnostic code is enabled with the
AHC_DEBUG_SEQUENCER kernel option.
Make it possible to switch into and out of target mode on
the fly. The card comes up by default as an initiator but
will switch into target mode as soon as an enable lun operation
is performed. As always, target mode behavior is gated
by the AHC_TMODE_ENABLE kernel option so most users will
not be affected by this change.
In ahc_update_target_msg_request(), also issue a new
request if the ppr_options have changed.
Never issue a PPR as a target. It is forbidden by the spec.
Correct a bug in ahc_parse_msg() that prevented us from
responding to PPR messages as a target.
Mark SCBs that are on the untagged queue with a flag instead
of checking several fields in the SCB to see if the SCB should
be on the queue. This makes it easier for things like automatic
request sense requests to be queued without touching the
untagged queues even though they are untagged requests.
When dealing with ignore wide residue messages that occur
in the middle of a transfer, reset HADDR, not SHADDR for
non-ultra2 chips. Although SHADDR is where the firmware
fetches the ending transfer address for a save data pointers
request, it is readonly. Setting HADDR has the side effect
of also updating SHADDR.
Cleanup the output of ahc_dump_card_state() by nulling out the
free scb list in the non-paging case. The free list is only
used if we must page SCBs.
Correct the transmission of cdbs > 12 bytes in length. When
swapping HSCBs prior to notifing the sequencer of the new
transaction, the bus address pointer for the cdb must also
be recalculated to reflect its new location. We now defer
the calculation of the cdb address until just before queing
it to the card.
When pulling transfer negotiation settings out of scratch
ram, convert 5MHz/clock doubled settings to 10MHz.
Add a new function ahc_qinfifo_requeue_tail() for use by
error recovery actions and auto-request sense operations.
These operations always occur when the sequencer is paused,
so we can avoid the extra expense incurred in the normal
SCB queue method.
Use the BMOV instruction for all single byte moves on
controllers that support it. The bmov instruction is
twice as fast as an AND with an immediate of 0xFF as
is used on older controllers.
Correct a few bugs in ahc_dump_card_state(). If we have
hardware assisted queue registers, use them to get the
sequencer's idea of the head of the queue. When enumerating
the untagged queue, it helps to use the correct index for
the queue.
aic7xxx.h:
Indicate via a feature flag, which controllers can take
on both the target and the initiator role at the same time.
Add the AHC_SEQUENCER_DEBUG flag.
Add the SCB_CDB32_PTR flag used for dealing with cdbs
with lengths between 13 and 32 bytes.
Add new prototypes.
aic7xxx.reg:
Allow the SCSIBUSL register to be written to. This is
required to fix a selection timeout problem on the 7892/99.
Cleanup the sequencer interrupt codes so that all debugging
codes are grouped at the end of the list.
Correct the definition of the ULTRA_ENB and DISC_DSB locations
in scratch ram. This prevented the driver from properly honoring
these settings when no serial eeprom was available.
Remove an unused sequencer flag.
aic7xxx.seq:
Just before a potential select-out, clear the SCSIBUSL
register. Occasionally, during a selection timeout, the
contents of the register may be presented on the bus,
causing much confusion.
Add sequencer diagnostic code to detect software and or
hardware bugs. The code attempts to verify most list
operations so any corruption is caught before it occurs.
We also track information about why a particular reconnection
request was rejected.
Don't clobber the digital REQ/ACK filter setting in SXFRCTL0
when clearing the channel.
Fix a target mode bug that would cause us to return busy
status instead of queue full in respnse to a tagged transaction.
Cleanup the overrun case. It turns out that by simply
butting the chip in bitbucket mode, it will ack any
bytes until the phase changes. This drasticaly simplifies
things.
Prior to leaving the data phase, make sure that the S/G
preload queue is empty.
Remove code to place a request sense request on the waiting
queue. This is all handled by the kernel now.
Change the semantics of "findSCB". In the past, findSCB
ensured that a freshly paged in SCB appeared on the disconnected
list. The problem with this is that there is no guarantee that
the paged in SCB is for a disconnected transation. We now
defer any list manipulation to the caller who usually discards
the SCB via the free list.
Inline some busy target table operations.
Add a critical section to protect adding an SCB to
the disconnected list.
aic7xxx_freebsd.c:
Handle changes in the transfer negotiation setting API
to filter incoming requests. No filtering is necessary
for "goal" requests from the XPT.
Set the SCB_CDB32_PTR flag when queing a transaction with
a large cdb.
In ahc_timeout, only take action if the active SCB is
the timedout SCB. This deals with the case of two
transactions to the same device with different timeout
values.
Use ahc_qinfifo_requeu_tail() instead of home grown
version.
aic7xxx_inline.h:
Honor SCB_CDB32_PTR when queuing a new request.
aic7xxx_pci.c:
Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
|
|
|
* If the active SCB is not us, assume that
|
|
|
|
|
* the active SCB has a longer timeout than
|
|
|
|
|
* the timedout SCB, and wait for the active
|
|
|
|
|
* SCB to timeout.
|
2000-09-16 20:02:28 +00:00
|
|
|
*/
|
2000-10-08 03:37:52 +00:00
|
|
|
active_scb = ahc_lookup_scb(ahc, active_scb_index);
|
aic7xxx.c:
Filter incoming transfer negotiation requests to ensure they
never exceed the settings specified by the user.
In restart sequencer attempt to deal with a bug in the aic7895.
If a third party reset occurs at just the right time, the
stack register can lock up. When restarting the sequencer
after handling the SCSI reset, poke SEQADDR1 before resting
the sequencers program counter.
When something strange happens, dump the card's transaction
state via ahc_dump_card_state(). This should aid in debugging.
Handle request sense transactions via the QINFIFO instead of
attaching them to the waiting queue directly. The waiting
queue consumes card SCB resources and, in the pathological case
of every target on the bus beating our selection attemps and
issuing a check condition, could have caused us to run out
of SCBs. I have never seen this happen, and only early
cards with 3 or 4 SCBs had any real chance of ever getting
into this state.
Add additional sequencer interrupt codes to support firmware
diagnostics. The diagnostic code is enabled with the
AHC_DEBUG_SEQUENCER kernel option.
Make it possible to switch into and out of target mode on
the fly. The card comes up by default as an initiator but
will switch into target mode as soon as an enable lun operation
is performed. As always, target mode behavior is gated
by the AHC_TMODE_ENABLE kernel option so most users will
not be affected by this change.
In ahc_update_target_msg_request(), also issue a new
request if the ppr_options have changed.
Never issue a PPR as a target. It is forbidden by the spec.
Correct a bug in ahc_parse_msg() that prevented us from
responding to PPR messages as a target.
Mark SCBs that are on the untagged queue with a flag instead
of checking several fields in the SCB to see if the SCB should
be on the queue. This makes it easier for things like automatic
request sense requests to be queued without touching the
untagged queues even though they are untagged requests.
When dealing with ignore wide residue messages that occur
in the middle of a transfer, reset HADDR, not SHADDR for
non-ultra2 chips. Although SHADDR is where the firmware
fetches the ending transfer address for a save data pointers
request, it is readonly. Setting HADDR has the side effect
of also updating SHADDR.
Cleanup the output of ahc_dump_card_state() by nulling out the
free scb list in the non-paging case. The free list is only
used if we must page SCBs.
Correct the transmission of cdbs > 12 bytes in length. When
swapping HSCBs prior to notifing the sequencer of the new
transaction, the bus address pointer for the cdb must also
be recalculated to reflect its new location. We now defer
the calculation of the cdb address until just before queing
it to the card.
When pulling transfer negotiation settings out of scratch
ram, convert 5MHz/clock doubled settings to 10MHz.
Add a new function ahc_qinfifo_requeue_tail() for use by
error recovery actions and auto-request sense operations.
These operations always occur when the sequencer is paused,
so we can avoid the extra expense incurred in the normal
SCB queue method.
Use the BMOV instruction for all single byte moves on
controllers that support it. The bmov instruction is
twice as fast as an AND with an immediate of 0xFF as
is used on older controllers.
Correct a few bugs in ahc_dump_card_state(). If we have
hardware assisted queue registers, use them to get the
sequencer's idea of the head of the queue. When enumerating
the untagged queue, it helps to use the correct index for
the queue.
aic7xxx.h:
Indicate via a feature flag, which controllers can take
on both the target and the initiator role at the same time.
Add the AHC_SEQUENCER_DEBUG flag.
Add the SCB_CDB32_PTR flag used for dealing with cdbs
with lengths between 13 and 32 bytes.
Add new prototypes.
aic7xxx.reg:
Allow the SCSIBUSL register to be written to. This is
required to fix a selection timeout problem on the 7892/99.
Cleanup the sequencer interrupt codes so that all debugging
codes are grouped at the end of the list.
Correct the definition of the ULTRA_ENB and DISC_DSB locations
in scratch ram. This prevented the driver from properly honoring
these settings when no serial eeprom was available.
Remove an unused sequencer flag.
aic7xxx.seq:
Just before a potential select-out, clear the SCSIBUSL
register. Occasionally, during a selection timeout, the
contents of the register may be presented on the bus,
causing much confusion.
Add sequencer diagnostic code to detect software and or
hardware bugs. The code attempts to verify most list
operations so any corruption is caught before it occurs.
We also track information about why a particular reconnection
request was rejected.
Don't clobber the digital REQ/ACK filter setting in SXFRCTL0
when clearing the channel.
Fix a target mode bug that would cause us to return busy
status instead of queue full in respnse to a tagged transaction.
Cleanup the overrun case. It turns out that by simply
butting the chip in bitbucket mode, it will ack any
bytes until the phase changes. This drasticaly simplifies
things.
Prior to leaving the data phase, make sure that the S/G
preload queue is empty.
Remove code to place a request sense request on the waiting
queue. This is all handled by the kernel now.
Change the semantics of "findSCB". In the past, findSCB
ensured that a freshly paged in SCB appeared on the disconnected
list. The problem with this is that there is no guarantee that
the paged in SCB is for a disconnected transation. We now
defer any list manipulation to the caller who usually discards
the SCB via the free list.
Inline some busy target table operations.
Add a critical section to protect adding an SCB to
the disconnected list.
aic7xxx_freebsd.c:
Handle changes in the transfer negotiation setting API
to filter incoming requests. No filtering is necessary
for "goal" requests from the XPT.
Set the SCB_CDB32_PTR flag when queing a transaction with
a large cdb.
In ahc_timeout, only take action if the active SCB is
the timedout SCB. This deals with the case of two
transactions to the same device with different timeout
values.
Use ahc_qinfifo_requeu_tail() instead of home grown
version.
aic7xxx_inline.h:
Honor SCB_CDB32_PTR when queuing a new request.
aic7xxx_pci.c:
Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
|
|
|
if (active_scb != scb) {
|
2001-01-22 21:03:48 +00:00
|
|
|
struct ccb_hdr *ccbh;
|
|
|
|
|
uint64_t newtimeout;
|
2000-09-16 20:02:28 +00:00
|
|
|
|
|
|
|
|
ahc_print_path(ahc, scb);
|
|
|
|
|
printf("Other SCB Timeout%s",
|
|
|
|
|
(scb->flags & SCB_OTHERTCL_TIMEOUT) != 0
|
|
|
|
|
? " again\n" : "\n");
|
|
|
|
|
scb->flags |= SCB_OTHERTCL_TIMEOUT;
|
|
|
|
|
newtimeout =
|
|
|
|
|
MAX(active_scb->io_ctx->ccb_h.timeout,
|
|
|
|
|
scb->io_ctx->ccb_h.timeout);
|
2001-01-22 21:03:48 +00:00
|
|
|
newtimeout *= hz;
|
|
|
|
|
newtimeout /= 1000;
|
2000-09-16 20:02:28 +00:00
|
|
|
ccbh = &scb->io_ctx->ccb_h;
|
|
|
|
|
scb->io_ctx->ccb_h.timeout_ch =
|
2001-01-22 21:03:48 +00:00
|
|
|
timeout(ahc_timeout, scb, newtimeout);
|
This is an MFC candidate.
ahc_eisa.c:
Change aic7770_map_int to take an additional irq parameter.
Although we can get the irq from the eisa dev under FreeBSD,
we can't do this under linux, so the OSM interface must supply
this.
ahc_pci.c:
Move ahc_power_state_change() to the OSM. This allows us to
use a platform supplied function that does the same thing.
-current will move to the FreeBSD native API in the near
future.
aic7770.c:
Sync up with core changes to support Linux EISA.
We now store a 2 bit primary channel number rather
than a bit flag that only allows b to be the primary
channel. Adjust for this change.
aic7xxx.c:
Namespace and staticization cleanup. All exported symbols
use an "ahc_" prefix to avoid collisions with other modules.
Correct a logic bug that prevented us from dropping
ATN during some exceptional conditions during message
processing.
Take advantage of a new flag managed by the sequencer
that indicates if an SCB fetch is in progress. If so,
the currently selected SCB needs to be returned to the
free list to prevent an SCB leak. This leak is a rarity
and would only occur if a bus reset or timeout resulting
in a bus reset occurred in the middle of an SCB fetch.
Don't attempt to perform ULTRA transfers on ultra capable
adapters missing the external precision resistor required
for ultra speeds. I've never encountered an adapter
configured this way, but better safe than sorry.
Handle the case of 5MHz user sync rate set as "0" instead of 0x1c
in scratch ram.
If we lookup a period of 0 in our table (async), clear the scsi offset.
aic7xxx.h:
Adjust for the primary channel being represented as
a 2 bit integer in the flags member of the ahc softc.
Cleanup the flags definitions so that comment blocks are
not cramped.
Update seeprom definitions to correctly reflect the fact
that the primary channel is represented as a 2 bit integer.
Add AHC_ULTRA_DIASABLED softc flag to denote controllers
missing the external precision resistor.
aic7xxx.reg:
Add DFCACHETH to the definition of DFSTATUS for completness sake.
Add SEQ_FLAGS2 which currently only contains the SCB_DMA
(SCB DMA in progress) flag.
aic7xxx.seq:
Correct a problem when one lun has a disconnected untagged
transaction and another lun has disconnected tagged transactions.
Just because an entry is found in the untagged table doesn't
mean that it will match. If the match on the lun fails, cleanup
the SCB (return it to the disconnected list or free it), and snoop
for a tag message. Before this change, we reported an unsolicited
reselection. This bug was introduced about a month ago during an
overly aggressive optimization pass on the reselection code.
When cleaning up an SCB, we can't just blindly free the SCB. In
the paging case, if the SCB came off of the disconnected list, its
state may never have been updated in host memory. So, check the
disconnected bit in SCB_CONTROL and return the SCB to the disconnected
list if appropriate.
Manage the SCB_DMA flag of SEQ_FLAGS2.
More carefully shutdown the S/G dma engine in all cases by using
a subroutine. Supposedly not doing this can cause an arbiter hang
on some ULTRA2 chips.
Formatting cleanup.
On some chips, at least the aic7856, the transition from
MREQPEND to HDONE can take a full 4 clock cycles. Test
HDONE one more time to avoid this race. We only want our
FIFO hung recovery code to execute when the engine is
really hung.
aic7xxx_93cx6.c:
Sync perforce ids.
aic7xxx_freebsd.c:
Adjust for the primary channel being a 2 bit integer
rather than a flag for 'B' channel being the primary.
Namespace cleanup.
Unpause the sequencer in one error recovery path that
neglected to do so. This could have caused us to perform
a bus reset when a recovery message might have otherwise been
successful.
aic7xxx_freebsd.h:
Use AHC_PCI_CONFIG for controlling compilation of PCI
support consistently throughout the driver.
Move ahc_power_state_change() to OSM.
aic7xxx_inline.h
Namespace cleanup.
Adjust our interrupt handler so it will work in the edge
interrupt case. We must process all interrupt sources
when the interrupt fires or risk not ever getting an
interrupt again. This involves marking the fact
that we are relying on an edge interrupt in ahc->flags
and checking for this condition in addition to the
AHC_ALL_INTERRUPTS flag. This fixes hangs on the
284X and any other aic7770 installation where level
interrupts are not available.
aic7xxx_pci.c:
Move the powerstate manipulation code into the OSM. Several
OSes now provide this functionality natively.
Take another shot at using the data stored in scratch ram
if the SCB2 signature is correct and no SEEPROM data is
available. In the past this failed if external SCB ram
was configured because the memory port was locked. We
now release the memory port prior to testing the values
in SCB2 and re-acquire it prior to doing termination control.
Adjust for new 2 bit primary channel setting.
Trust the STPWLEVEL setting on v 3.X BIOSes too.
Configure any 785X ID in the same fashion and assume
that any device with a rev id of 1 or higher has the
PCI 2.1 retry bug.
2001-03-11 06:34:17 +00:00
|
|
|
ahc_unpause(ahc);
|
2000-09-16 20:02:28 +00:00
|
|
|
ahc_unlock(ahc, &s);
|
|
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* It's us */
|
|
|
|
|
if ((scb->hscb->control & TARGET_SCB) != 0) {
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Send back any queued up transactions
|
|
|
|
|
* and properly record the error condition.
|
|
|
|
|
*/
|
2002-04-24 16:58:51 +00:00
|
|
|
ahc_abort_scbs(ahc, SCB_GET_TARGET(ahc, scb),
|
|
|
|
|
SCB_GET_CHANNEL(ahc, scb),
|
|
|
|
|
SCB_GET_LUN(scb),
|
|
|
|
|
scb->hscb->tag,
|
|
|
|
|
ROLE_TARGET,
|
|
|
|
|
CAM_CMD_TIMEOUT);
|
2000-09-16 20:02:28 +00:00
|
|
|
|
|
|
|
|
/* Will clear us from the bus */
|
This is an MFC candidate.
ahc_eisa.c:
Change aic7770_map_int to take an additional irq parameter.
Although we can get the irq from the eisa dev under FreeBSD,
we can't do this under linux, so the OSM interface must supply
this.
ahc_pci.c:
Move ahc_power_state_change() to the OSM. This allows us to
use a platform supplied function that does the same thing.
-current will move to the FreeBSD native API in the near
future.
aic7770.c:
Sync up with core changes to support Linux EISA.
We now store a 2 bit primary channel number rather
than a bit flag that only allows b to be the primary
channel. Adjust for this change.
aic7xxx.c:
Namespace and staticization cleanup. All exported symbols
use an "ahc_" prefix to avoid collisions with other modules.
Correct a logic bug that prevented us from dropping
ATN during some exceptional conditions during message
processing.
Take advantage of a new flag managed by the sequencer
that indicates if an SCB fetch is in progress. If so,
the currently selected SCB needs to be returned to the
free list to prevent an SCB leak. This leak is a rarity
and would only occur if a bus reset or timeout resulting
in a bus reset occurred in the middle of an SCB fetch.
Don't attempt to perform ULTRA transfers on ultra capable
adapters missing the external precision resistor required
for ultra speeds. I've never encountered an adapter
configured this way, but better safe than sorry.
Handle the case of 5MHz user sync rate set as "0" instead of 0x1c
in scratch ram.
If we lookup a period of 0 in our table (async), clear the scsi offset.
aic7xxx.h:
Adjust for the primary channel being represented as
a 2 bit integer in the flags member of the ahc softc.
Cleanup the flags definitions so that comment blocks are
not cramped.
Update seeprom definitions to correctly reflect the fact
that the primary channel is represented as a 2 bit integer.
Add AHC_ULTRA_DIASABLED softc flag to denote controllers
missing the external precision resistor.
aic7xxx.reg:
Add DFCACHETH to the definition of DFSTATUS for completness sake.
Add SEQ_FLAGS2 which currently only contains the SCB_DMA
(SCB DMA in progress) flag.
aic7xxx.seq:
Correct a problem when one lun has a disconnected untagged
transaction and another lun has disconnected tagged transactions.
Just because an entry is found in the untagged table doesn't
mean that it will match. If the match on the lun fails, cleanup
the SCB (return it to the disconnected list or free it), and snoop
for a tag message. Before this change, we reported an unsolicited
reselection. This bug was introduced about a month ago during an
overly aggressive optimization pass on the reselection code.
When cleaning up an SCB, we can't just blindly free the SCB. In
the paging case, if the SCB came off of the disconnected list, its
state may never have been updated in host memory. So, check the
disconnected bit in SCB_CONTROL and return the SCB to the disconnected
list if appropriate.
Manage the SCB_DMA flag of SEQ_FLAGS2.
More carefully shutdown the S/G dma engine in all cases by using
a subroutine. Supposedly not doing this can cause an arbiter hang
on some ULTRA2 chips.
Formatting cleanup.
On some chips, at least the aic7856, the transition from
MREQPEND to HDONE can take a full 4 clock cycles. Test
HDONE one more time to avoid this race. We only want our
FIFO hung recovery code to execute when the engine is
really hung.
aic7xxx_93cx6.c:
Sync perforce ids.
aic7xxx_freebsd.c:
Adjust for the primary channel being a 2 bit integer
rather than a flag for 'B' channel being the primary.
Namespace cleanup.
Unpause the sequencer in one error recovery path that
neglected to do so. This could have caused us to perform
a bus reset when a recovery message might have otherwise been
successful.
aic7xxx_freebsd.h:
Use AHC_PCI_CONFIG for controlling compilation of PCI
support consistently throughout the driver.
Move ahc_power_state_change() to OSM.
aic7xxx_inline.h
Namespace cleanup.
Adjust our interrupt handler so it will work in the edge
interrupt case. We must process all interrupt sources
when the interrupt fires or risk not ever getting an
interrupt again. This involves marking the fact
that we are relying on an edge interrupt in ahc->flags
and checking for this condition in addition to the
AHC_ALL_INTERRUPTS flag. This fixes hangs on the
284X and any other aic7770 installation where level
interrupts are not available.
aic7xxx_pci.c:
Move the powerstate manipulation code into the OSM. Several
OSes now provide this functionality natively.
Take another shot at using the data stored in scratch ram
if the SCB2 signature is correct and no SEEPROM data is
available. In the past this failed if external SCB ram
was configured because the memory port was locked. We
now release the memory port prior to testing the values
in SCB2 and re-acquire it prior to doing termination control.
Adjust for new 2 bit primary channel setting.
Trust the STPWLEVEL setting on v 3.X BIOSes too.
Configure any 785X ID in the same fashion and assume
that any device with a rev id of 1 or higher has the
PCI 2.1 retry bug.
2001-03-11 06:34:17 +00:00
|
|
|
ahc_restart(ahc);
|
2000-10-06 04:01:06 +00:00
|
|
|
ahc_unlock(ahc, &s);
|
2000-09-16 20:02:28 +00:00
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
ahc_set_recoveryscb(ahc, active_scb);
|
Clean up error recovery code:
aic7xxx.c:
In target mode, reset the TQINPOS on every restart of the sequencer.
In the past we did this only during a bus reset, but there are other
reasons the sequencer might be reset.
In ahc_clear_critical_section(), disable pausing chip interrupts while
we step the sequencer out of a critical section. This avoids the
possibility of getting a pausing interrupt (unexpected bus free,
bus reset, etc.) that would prevent the sequencer from stepping.
Send the correct async notifications in the case of a BDR or bus reset.
In ahc_loadseq(), correct the calculation of our critical sections.
In some cases, the sections would be larger than needed.
aic7xxx.h:
Remove an unused SCB flag.
aic7xxx.seq:
MK_MESSAGE is cleared by the kernel, there is no need to waste
a sequencer instruction clearing it.
aic7xxx_freebsd.c:
Go through the host message loop instead of issuing a single
byte message directly in the ahc_timeout() case where we
are currently on the bus to the device. The effect is the same,
but this way we get a nice printf saying that an expected BDR
was delivered instead of an unexpected bus free.
If we are requeuing an SCB for an error recovery action, be sure
to set the DISCONNECTED flag in the in-core version of the SCB.
This ensures that, in the SCB-paging case, the sequencer will
still recognize the reselection as valid even if the version
of the SCB with this flag set was never previously paged out
to system memory. In the non-paging case, set the MK_MESSAGE
flag in SCB_CONTROL directly.
aic7xxx_pci.c:
Enable the Memeory Write and Invalidate bug workaround for
all aic7880 chips with revs < 1. This bug is rarely triggered
in FreeBSD as most transfers end on cache-aligned boundaries,
but a recheck of my references indicates that these chips
are affected.
2000-10-09 01:46:01 +00:00
|
|
|
ahc_outb(ahc, MSG_OUT, HOST_MSG);
|
2000-09-16 20:02:28 +00:00
|
|
|
ahc_outb(ahc, SCSISIGO, last_phase|ATNO);
|
|
|
|
|
ahc_print_path(ahc, active_scb);
|
|
|
|
|
printf("BDR message in message buffer\n");
|
aic7xxx.c:
Filter incoming transfer negotiation requests to ensure they
never exceed the settings specified by the user.
In restart sequencer attempt to deal with a bug in the aic7895.
If a third party reset occurs at just the right time, the
stack register can lock up. When restarting the sequencer
after handling the SCSI reset, poke SEQADDR1 before resting
the sequencers program counter.
When something strange happens, dump the card's transaction
state via ahc_dump_card_state(). This should aid in debugging.
Handle request sense transactions via the QINFIFO instead of
attaching them to the waiting queue directly. The waiting
queue consumes card SCB resources and, in the pathological case
of every target on the bus beating our selection attemps and
issuing a check condition, could have caused us to run out
of SCBs. I have never seen this happen, and only early
cards with 3 or 4 SCBs had any real chance of ever getting
into this state.
Add additional sequencer interrupt codes to support firmware
diagnostics. The diagnostic code is enabled with the
AHC_DEBUG_SEQUENCER kernel option.
Make it possible to switch into and out of target mode on
the fly. The card comes up by default as an initiator but
will switch into target mode as soon as an enable lun operation
is performed. As always, target mode behavior is gated
by the AHC_TMODE_ENABLE kernel option so most users will
not be affected by this change.
In ahc_update_target_msg_request(), also issue a new
request if the ppr_options have changed.
Never issue a PPR as a target. It is forbidden by the spec.
Correct a bug in ahc_parse_msg() that prevented us from
responding to PPR messages as a target.
Mark SCBs that are on the untagged queue with a flag instead
of checking several fields in the SCB to see if the SCB should
be on the queue. This makes it easier for things like automatic
request sense requests to be queued without touching the
untagged queues even though they are untagged requests.
When dealing with ignore wide residue messages that occur
in the middle of a transfer, reset HADDR, not SHADDR for
non-ultra2 chips. Although SHADDR is where the firmware
fetches the ending transfer address for a save data pointers
request, it is readonly. Setting HADDR has the side effect
of also updating SHADDR.
Cleanup the output of ahc_dump_card_state() by nulling out the
free scb list in the non-paging case. The free list is only
used if we must page SCBs.
Correct the transmission of cdbs > 12 bytes in length. When
swapping HSCBs prior to notifing the sequencer of the new
transaction, the bus address pointer for the cdb must also
be recalculated to reflect its new location. We now defer
the calculation of the cdb address until just before queing
it to the card.
When pulling transfer negotiation settings out of scratch
ram, convert 5MHz/clock doubled settings to 10MHz.
Add a new function ahc_qinfifo_requeue_tail() for use by
error recovery actions and auto-request sense operations.
These operations always occur when the sequencer is paused,
so we can avoid the extra expense incurred in the normal
SCB queue method.
Use the BMOV instruction for all single byte moves on
controllers that support it. The bmov instruction is
twice as fast as an AND with an immediate of 0xFF as
is used on older controllers.
Correct a few bugs in ahc_dump_card_state(). If we have
hardware assisted queue registers, use them to get the
sequencer's idea of the head of the queue. When enumerating
the untagged queue, it helps to use the correct index for
the queue.
aic7xxx.h:
Indicate via a feature flag, which controllers can take
on both the target and the initiator role at the same time.
Add the AHC_SEQUENCER_DEBUG flag.
Add the SCB_CDB32_PTR flag used for dealing with cdbs
with lengths between 13 and 32 bytes.
Add new prototypes.
aic7xxx.reg:
Allow the SCSIBUSL register to be written to. This is
required to fix a selection timeout problem on the 7892/99.
Cleanup the sequencer interrupt codes so that all debugging
codes are grouped at the end of the list.
Correct the definition of the ULTRA_ENB and DISC_DSB locations
in scratch ram. This prevented the driver from properly honoring
these settings when no serial eeprom was available.
Remove an unused sequencer flag.
aic7xxx.seq:
Just before a potential select-out, clear the SCSIBUSL
register. Occasionally, during a selection timeout, the
contents of the register may be presented on the bus,
causing much confusion.
Add sequencer diagnostic code to detect software and or
hardware bugs. The code attempts to verify most list
operations so any corruption is caught before it occurs.
We also track information about why a particular reconnection
request was rejected.
Don't clobber the digital REQ/ACK filter setting in SXFRCTL0
when clearing the channel.
Fix a target mode bug that would cause us to return busy
status instead of queue full in respnse to a tagged transaction.
Cleanup the overrun case. It turns out that by simply
butting the chip in bitbucket mode, it will ack any
bytes until the phase changes. This drasticaly simplifies
things.
Prior to leaving the data phase, make sure that the S/G
preload queue is empty.
Remove code to place a request sense request on the waiting
queue. This is all handled by the kernel now.
Change the semantics of "findSCB". In the past, findSCB
ensured that a freshly paged in SCB appeared on the disconnected
list. The problem with this is that there is no guarantee that
the paged in SCB is for a disconnected transation. We now
defer any list manipulation to the caller who usually discards
the SCB via the free list.
Inline some busy target table operations.
Add a critical section to protect adding an SCB to
the disconnected list.
aic7xxx_freebsd.c:
Handle changes in the transfer negotiation setting API
to filter incoming requests. No filtering is necessary
for "goal" requests from the XPT.
Set the SCB_CDB32_PTR flag when queing a transaction with
a large cdb.
In ahc_timeout, only take action if the active SCB is
the timedout SCB. This deals with the case of two
transactions to the same device with different timeout
values.
Use ahc_qinfifo_requeu_tail() instead of home grown
version.
aic7xxx_inline.h:
Honor SCB_CDB32_PTR when queuing a new request.
aic7xxx_pci.c:
Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
|
|
|
active_scb->flags |= SCB_DEVICE_RESET;
|
2000-09-16 20:02:28 +00:00
|
|
|
active_scb->io_ctx->ccb_h.timeout_ch =
|
|
|
|
|
timeout(ahc_timeout, (caddr_t)active_scb, 2 * hz);
|
This is an MFC candidate.
ahc_eisa.c:
Change aic7770_map_int to take an additional irq parameter.
Although we can get the irq from the eisa dev under FreeBSD,
we can't do this under linux, so the OSM interface must supply
this.
ahc_pci.c:
Move ahc_power_state_change() to the OSM. This allows us to
use a platform supplied function that does the same thing.
-current will move to the FreeBSD native API in the near
future.
aic7770.c:
Sync up with core changes to support Linux EISA.
We now store a 2 bit primary channel number rather
than a bit flag that only allows b to be the primary
channel. Adjust for this change.
aic7xxx.c:
Namespace and staticization cleanup. All exported symbols
use an "ahc_" prefix to avoid collisions with other modules.
Correct a logic bug that prevented us from dropping
ATN during some exceptional conditions during message
processing.
Take advantage of a new flag managed by the sequencer
that indicates if an SCB fetch is in progress. If so,
the currently selected SCB needs to be returned to the
free list to prevent an SCB leak. This leak is a rarity
and would only occur if a bus reset or timeout resulting
in a bus reset occurred in the middle of an SCB fetch.
Don't attempt to perform ULTRA transfers on ultra capable
adapters missing the external precision resistor required
for ultra speeds. I've never encountered an adapter
configured this way, but better safe than sorry.
Handle the case of 5MHz user sync rate set as "0" instead of 0x1c
in scratch ram.
If we lookup a period of 0 in our table (async), clear the scsi offset.
aic7xxx.h:
Adjust for the primary channel being represented as
a 2 bit integer in the flags member of the ahc softc.
Cleanup the flags definitions so that comment blocks are
not cramped.
Update seeprom definitions to correctly reflect the fact
that the primary channel is represented as a 2 bit integer.
Add AHC_ULTRA_DIASABLED softc flag to denote controllers
missing the external precision resistor.
aic7xxx.reg:
Add DFCACHETH to the definition of DFSTATUS for completness sake.
Add SEQ_FLAGS2 which currently only contains the SCB_DMA
(SCB DMA in progress) flag.
aic7xxx.seq:
Correct a problem when one lun has a disconnected untagged
transaction and another lun has disconnected tagged transactions.
Just because an entry is found in the untagged table doesn't
mean that it will match. If the match on the lun fails, cleanup
the SCB (return it to the disconnected list or free it), and snoop
for a tag message. Before this change, we reported an unsolicited
reselection. This bug was introduced about a month ago during an
overly aggressive optimization pass on the reselection code.
When cleaning up an SCB, we can't just blindly free the SCB. In
the paging case, if the SCB came off of the disconnected list, its
state may never have been updated in host memory. So, check the
disconnected bit in SCB_CONTROL and return the SCB to the disconnected
list if appropriate.
Manage the SCB_DMA flag of SEQ_FLAGS2.
More carefully shutdown the S/G dma engine in all cases by using
a subroutine. Supposedly not doing this can cause an arbiter hang
on some ULTRA2 chips.
Formatting cleanup.
On some chips, at least the aic7856, the transition from
MREQPEND to HDONE can take a full 4 clock cycles. Test
HDONE one more time to avoid this race. We only want our
FIFO hung recovery code to execute when the engine is
really hung.
aic7xxx_93cx6.c:
Sync perforce ids.
aic7xxx_freebsd.c:
Adjust for the primary channel being a 2 bit integer
rather than a flag for 'B' channel being the primary.
Namespace cleanup.
Unpause the sequencer in one error recovery path that
neglected to do so. This could have caused us to perform
a bus reset when a recovery message might have otherwise been
successful.
aic7xxx_freebsd.h:
Use AHC_PCI_CONFIG for controlling compilation of PCI
support consistently throughout the driver.
Move ahc_power_state_change() to OSM.
aic7xxx_inline.h
Namespace cleanup.
Adjust our interrupt handler so it will work in the edge
interrupt case. We must process all interrupt sources
when the interrupt fires or risk not ever getting an
interrupt again. This involves marking the fact
that we are relying on an edge interrupt in ahc->flags
and checking for this condition in addition to the
AHC_ALL_INTERRUPTS flag. This fixes hangs on the
284X and any other aic7770 installation where level
interrupts are not available.
aic7xxx_pci.c:
Move the powerstate manipulation code into the OSM. Several
OSes now provide this functionality natively.
Take another shot at using the data stored in scratch ram
if the SCB2 signature is correct and no SEEPROM data is
available. In the past this failed if external SCB ram
was configured because the memory port was locked. We
now release the memory port prior to testing the values
in SCB2 and re-acquire it prior to doing termination control.
Adjust for new 2 bit primary channel setting.
Trust the STPWLEVEL setting on v 3.X BIOSes too.
Configure any 785X ID in the same fashion and assume
that any device with a rev id of 1 or higher has the
PCI 2.1 retry bug.
2001-03-11 06:34:17 +00:00
|
|
|
ahc_unpause(ahc);
|
2000-09-16 20:02:28 +00:00
|
|
|
} else {
|
|
|
|
|
int disconnected;
|
|
|
|
|
|
2002-04-24 16:58:51 +00:00
|
|
|
/* XXX Shouldn't panic. Just punt instead? */
|
2000-09-16 20:02:28 +00:00
|
|
|
if ((scb->hscb->control & TARGET_SCB) != 0)
|
|
|
|
|
panic("Timed-out target SCB but bus idle");
|
|
|
|
|
|
|
|
|
|
if (last_phase != P_BUSFREE
|
|
|
|
|
&& (ahc_inb(ahc, SSTAT0) & TARGET) != 0) {
|
|
|
|
|
/* XXX What happened to the SCB? */
|
|
|
|
|
/* Hung target selection. Goto busfree */
|
|
|
|
|
printf("%s: Hung target selection\n",
|
|
|
|
|
ahc_name(ahc));
|
This is an MFC candidate.
ahc_eisa.c:
Change aic7770_map_int to take an additional irq parameter.
Although we can get the irq from the eisa dev under FreeBSD,
we can't do this under linux, so the OSM interface must supply
this.
ahc_pci.c:
Move ahc_power_state_change() to the OSM. This allows us to
use a platform supplied function that does the same thing.
-current will move to the FreeBSD native API in the near
future.
aic7770.c:
Sync up with core changes to support Linux EISA.
We now store a 2 bit primary channel number rather
than a bit flag that only allows b to be the primary
channel. Adjust for this change.
aic7xxx.c:
Namespace and staticization cleanup. All exported symbols
use an "ahc_" prefix to avoid collisions with other modules.
Correct a logic bug that prevented us from dropping
ATN during some exceptional conditions during message
processing.
Take advantage of a new flag managed by the sequencer
that indicates if an SCB fetch is in progress. If so,
the currently selected SCB needs to be returned to the
free list to prevent an SCB leak. This leak is a rarity
and would only occur if a bus reset or timeout resulting
in a bus reset occurred in the middle of an SCB fetch.
Don't attempt to perform ULTRA transfers on ultra capable
adapters missing the external precision resistor required
for ultra speeds. I've never encountered an adapter
configured this way, but better safe than sorry.
Handle the case of 5MHz user sync rate set as "0" instead of 0x1c
in scratch ram.
If we lookup a period of 0 in our table (async), clear the scsi offset.
aic7xxx.h:
Adjust for the primary channel being represented as
a 2 bit integer in the flags member of the ahc softc.
Cleanup the flags definitions so that comment blocks are
not cramped.
Update seeprom definitions to correctly reflect the fact
that the primary channel is represented as a 2 bit integer.
Add AHC_ULTRA_DIASABLED softc flag to denote controllers
missing the external precision resistor.
aic7xxx.reg:
Add DFCACHETH to the definition of DFSTATUS for completness sake.
Add SEQ_FLAGS2 which currently only contains the SCB_DMA
(SCB DMA in progress) flag.
aic7xxx.seq:
Correct a problem when one lun has a disconnected untagged
transaction and another lun has disconnected tagged transactions.
Just because an entry is found in the untagged table doesn't
mean that it will match. If the match on the lun fails, cleanup
the SCB (return it to the disconnected list or free it), and snoop
for a tag message. Before this change, we reported an unsolicited
reselection. This bug was introduced about a month ago during an
overly aggressive optimization pass on the reselection code.
When cleaning up an SCB, we can't just blindly free the SCB. In
the paging case, if the SCB came off of the disconnected list, its
state may never have been updated in host memory. So, check the
disconnected bit in SCB_CONTROL and return the SCB to the disconnected
list if appropriate.
Manage the SCB_DMA flag of SEQ_FLAGS2.
More carefully shutdown the S/G dma engine in all cases by using
a subroutine. Supposedly not doing this can cause an arbiter hang
on some ULTRA2 chips.
Formatting cleanup.
On some chips, at least the aic7856, the transition from
MREQPEND to HDONE can take a full 4 clock cycles. Test
HDONE one more time to avoid this race. We only want our
FIFO hung recovery code to execute when the engine is
really hung.
aic7xxx_93cx6.c:
Sync perforce ids.
aic7xxx_freebsd.c:
Adjust for the primary channel being a 2 bit integer
rather than a flag for 'B' channel being the primary.
Namespace cleanup.
Unpause the sequencer in one error recovery path that
neglected to do so. This could have caused us to perform
a bus reset when a recovery message might have otherwise been
successful.
aic7xxx_freebsd.h:
Use AHC_PCI_CONFIG for controlling compilation of PCI
support consistently throughout the driver.
Move ahc_power_state_change() to OSM.
aic7xxx_inline.h
Namespace cleanup.
Adjust our interrupt handler so it will work in the edge
interrupt case. We must process all interrupt sources
when the interrupt fires or risk not ever getting an
interrupt again. This involves marking the fact
that we are relying on an edge interrupt in ahc->flags
and checking for this condition in addition to the
AHC_ALL_INTERRUPTS flag. This fixes hangs on the
284X and any other aic7770 installation where level
interrupts are not available.
aic7xxx_pci.c:
Move the powerstate manipulation code into the OSM. Several
OSes now provide this functionality natively.
Take another shot at using the data stored in scratch ram
if the SCB2 signature is correct and no SEEPROM data is
available. In the past this failed if external SCB ram
was configured because the memory port was locked. We
now release the memory port prior to testing the values
in SCB2 and re-acquire it prior to doing termination control.
Adjust for new 2 bit primary channel setting.
Trust the STPWLEVEL setting on v 3.X BIOSes too.
Configure any 785X ID in the same fashion and assume
that any device with a rev id of 1 or higher has the
PCI 2.1 retry bug.
2001-03-11 06:34:17 +00:00
|
|
|
ahc_restart(ahc);
|
2000-10-06 04:01:06 +00:00
|
|
|
ahc_unlock(ahc, &s);
|
2000-09-16 20:02:28 +00:00
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (ahc_search_qinfifo(ahc, target, channel, lun,
|
|
|
|
|
scb->hscb->tag, ROLE_INITIATOR,
|
|
|
|
|
/*status*/0, SEARCH_COUNT) > 0) {
|
|
|
|
|
disconnected = FALSE;
|
|
|
|
|
} else {
|
|
|
|
|
disconnected = TRUE;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (disconnected) {
|
|
|
|
|
|
|
|
|
|
ahc_set_recoveryscb(ahc, scb);
|
|
|
|
|
/*
|
2000-10-08 03:37:52 +00:00
|
|
|
* Actually re-queue this SCB in an attempt
|
|
|
|
|
* to select the device before it reconnects.
|
|
|
|
|
* In either case (selection or reselection),
|
|
|
|
|
* we will now issue a target reset to the
|
|
|
|
|
* timed-out device.
|
|
|
|
|
*
|
Clean up error recovery code:
aic7xxx.c:
In target mode, reset the TQINPOS on every restart of the sequencer.
In the past we did this only during a bus reset, but there are other
reasons the sequencer might be reset.
In ahc_clear_critical_section(), disable pausing chip interrupts while
we step the sequencer out of a critical section. This avoids the
possibility of getting a pausing interrupt (unexpected bus free,
bus reset, etc.) that would prevent the sequencer from stepping.
Send the correct async notifications in the case of a BDR or bus reset.
In ahc_loadseq(), correct the calculation of our critical sections.
In some cases, the sections would be larger than needed.
aic7xxx.h:
Remove an unused SCB flag.
aic7xxx.seq:
MK_MESSAGE is cleared by the kernel, there is no need to waste
a sequencer instruction clearing it.
aic7xxx_freebsd.c:
Go through the host message loop instead of issuing a single
byte message directly in the ahc_timeout() case where we
are currently on the bus to the device. The effect is the same,
but this way we get a nice printf saying that an expected BDR
was delivered instead of an unexpected bus free.
If we are requeuing an SCB for an error recovery action, be sure
to set the DISCONNECTED flag in the in-core version of the SCB.
This ensures that, in the SCB-paging case, the sequencer will
still recognize the reselection as valid even if the version
of the SCB with this flag set was never previously paged out
to system memory. In the non-paging case, set the MK_MESSAGE
flag in SCB_CONTROL directly.
aic7xxx_pci.c:
Enable the Memeory Write and Invalidate bug workaround for
all aic7880 chips with revs < 1. This bug is rarely triggered
in FreeBSD as most transfers end on cache-aligned boundaries,
but a recheck of my references indicates that these chips
are affected.
2000-10-09 01:46:01 +00:00
|
|
|
* Set the MK_MESSAGE control bit indicating
|
|
|
|
|
* that we desire to send a message. We
|
|
|
|
|
* also set the disconnected flag since
|
|
|
|
|
* in the paging case there is no guarantee
|
|
|
|
|
* that our SCB control byte matches the
|
|
|
|
|
* version on the card. We don't want the
|
|
|
|
|
* sequencer to abort the command thinking
|
|
|
|
|
* an unsolicited reselection occurred.
|
|
|
|
|
*/
|
|
|
|
|
scb->hscb->control |= MK_MESSAGE|DISCONNECTED;
|
|
|
|
|
scb->flags |= SCB_DEVICE_RESET;
|
|
|
|
|
|
|
|
|
|
/*
|
2000-10-08 03:37:52 +00:00
|
|
|
* Remove any cached copy of this SCB in the
|
|
|
|
|
* disconnected list in preparation for the
|
|
|
|
|
* queuing of our abort SCB. We use the
|
|
|
|
|
* same element in the SCB, SCB_NEXT, for
|
|
|
|
|
* both the qinfifo and the disconnected list.
|
2000-09-16 20:02:28 +00:00
|
|
|
*/
|
2000-10-08 03:37:52 +00:00
|
|
|
ahc_search_disc_list(ahc, target, channel,
|
|
|
|
|
lun, scb->hscb->tag,
|
|
|
|
|
/*stop_on_first*/TRUE,
|
|
|
|
|
/*remove*/TRUE,
|
Clean up error recovery code:
aic7xxx.c:
In target mode, reset the TQINPOS on every restart of the sequencer.
In the past we did this only during a bus reset, but there are other
reasons the sequencer might be reset.
In ahc_clear_critical_section(), disable pausing chip interrupts while
we step the sequencer out of a critical section. This avoids the
possibility of getting a pausing interrupt (unexpected bus free,
bus reset, etc.) that would prevent the sequencer from stepping.
Send the correct async notifications in the case of a BDR or bus reset.
In ahc_loadseq(), correct the calculation of our critical sections.
In some cases, the sections would be larger than needed.
aic7xxx.h:
Remove an unused SCB flag.
aic7xxx.seq:
MK_MESSAGE is cleared by the kernel, there is no need to waste
a sequencer instruction clearing it.
aic7xxx_freebsd.c:
Go through the host message loop instead of issuing a single
byte message directly in the ahc_timeout() case where we
are currently on the bus to the device. The effect is the same,
but this way we get a nice printf saying that an expected BDR
was delivered instead of an unexpected bus free.
If we are requeuing an SCB for an error recovery action, be sure
to set the DISCONNECTED flag in the in-core version of the SCB.
This ensures that, in the SCB-paging case, the sequencer will
still recognize the reselection as valid even if the version
of the SCB with this flag set was never previously paged out
to system memory. In the non-paging case, set the MK_MESSAGE
flag in SCB_CONTROL directly.
aic7xxx_pci.c:
Enable the Memeory Write and Invalidate bug workaround for
all aic7880 chips with revs < 1. This bug is rarely triggered
in FreeBSD as most transfers end on cache-aligned boundaries,
but a recheck of my references indicates that these chips
are affected.
2000-10-09 01:46:01 +00:00
|
|
|
/*save_state*/FALSE);
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* In the non-paging case, the sequencer will
|
|
|
|
|
* never re-reference the in-core SCB.
|
|
|
|
|
* To make sure we are notified during
|
|
|
|
|
* reslection, set the MK_MESSAGE flag in
|
|
|
|
|
* the card's copy of the SCB.
|
|
|
|
|
*/
|
Sync Perforce IDs, add tranceiver state change support, and correct
numerous error recovery buglets.
Many thanks to Tor Egge for his assistance in diagnosing problems with
the error recovery code.
aic7xxx.c:
Report missed bus free events using their own sequencer interrupt
code to avoid confusion with other "bad phase" interrupts.
Remove a delay used in debugging. This delay could only be hit
in certain, very extreme, error recovery scenarios.
Handle transceiver state changes correctly. You can now
plug an SE device into a hot-plug LVD bus without hanging
the controller.
When stepping through a critical section, panic if we step
more than a reasonable number of times.
After a bus reset, disable bus reset interupts until we either
our first attempt to (re)select another device, or another device
attemps to select us. This removes the need to busy wait in
kernel for the scsi reset line to fall yet still ensures we
see any reset events that impact the state of either our initiator
or target roles. Before this change, we had the potential of
servicing a "storm" of reset interrupts if the reset line was
held for a significant amount of time.
Indicate the current sequencer address whenever we dump the
card's state.
aic7xxx.reg:
Transceiver state change register definitions.
Add the missed bussfree sequencer interrupt code.
Re-enable the scsi reset interrupt if it has been
disabled before every attempt to (re)select a device
and when we have been selected as a target.
When being (re)selected, check to see if the selection
dissappeared just after we enabled our bus free interrupt.
If the bus has gone free again, go back to the idle loop
and wait for another selection.
Note two locations where we should change our behavior
if ATN is still raised. If ATN is raised during the
presentation of a command complete or disconnect message,
we should ignore the message and expect the target to put
us in msgout phase. We don't currently do this as it
requires some code re-arrangement so that critical sections
can be properly placed around our handling of these two
events. Otherwise, we cannot guarantee that the check of
ATN is atomic relative to our acking of the message in
byte (the kernel could assert ATN).
Only set the IDENTIFY_SEEN flag after we have settled
on the SCB for this transaction. The kernel looks at
this flag before assuming that SCB_TAG is valid. This
avoids confusion during certain types of error recovery.
Add a critical section around findSCB. We cannot allow
the kernel to remove an entry from the disconnected
list while we are traversing it. Ditto for get_free_or_disc_scb.
aic7xxx_freebsd.c:
Only assume that SCB_TAG is accurate if IDENTIFY_SEEN is
set in SEQ_FLAGS.
Fix a typo that caused us to execute some code for the
non-SCB paging case when paging SCBs. This only occurred
during error recovery.
2000-11-10 20:13:41 +00:00
|
|
|
if ((ahc->flags & AHC_PAGESCBS) == 0) {
|
Clean up error recovery code:
aic7xxx.c:
In target mode, reset the TQINPOS on every restart of the sequencer.
In the past we did this only during a bus reset, but there are other
reasons the sequencer might be reset.
In ahc_clear_critical_section(), disable pausing chip interrupts while
we step the sequencer out of a critical section. This avoids the
possibility of getting a pausing interrupt (unexpected bus free,
bus reset, etc.) that would prevent the sequencer from stepping.
Send the correct async notifications in the case of a BDR or bus reset.
In ahc_loadseq(), correct the calculation of our critical sections.
In some cases, the sections would be larger than needed.
aic7xxx.h:
Remove an unused SCB flag.
aic7xxx.seq:
MK_MESSAGE is cleared by the kernel, there is no need to waste
a sequencer instruction clearing it.
aic7xxx_freebsd.c:
Go through the host message loop instead of issuing a single
byte message directly in the ahc_timeout() case where we
are currently on the bus to the device. The effect is the same,
but this way we get a nice printf saying that an expected BDR
was delivered instead of an unexpected bus free.
If we are requeuing an SCB for an error recovery action, be sure
to set the DISCONNECTED flag in the in-core version of the SCB.
This ensures that, in the SCB-paging case, the sequencer will
still recognize the reselection as valid even if the version
of the SCB with this flag set was never previously paged out
to system memory. In the non-paging case, set the MK_MESSAGE
flag in SCB_CONTROL directly.
aic7xxx_pci.c:
Enable the Memeory Write and Invalidate bug workaround for
all aic7880 chips with revs < 1. This bug is rarely triggered
in FreeBSD as most transfers end on cache-aligned boundaries,
but a recheck of my references indicates that these chips
are affected.
2000-10-09 01:46:01 +00:00
|
|
|
ahc_outb(ahc, SCBPTR, scb->hscb->tag);
|
|
|
|
|
ahc_outb(ahc, SCB_CONTROL,
|
|
|
|
|
ahc_inb(ahc, SCB_CONTROL)
|
|
|
|
|
| MK_MESSAGE);
|
|
|
|
|
}
|
2000-09-16 20:02:28 +00:00
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Clear out any entries in the QINFIFO first
|
|
|
|
|
* so we are the next SCB for this target
|
|
|
|
|
* to run.
|
|
|
|
|
*/
|
|
|
|
|
ahc_search_qinfifo(ahc,
|
|
|
|
|
SCB_GET_TARGET(ahc, scb),
|
|
|
|
|
channel, SCB_GET_LUN(scb),
|
|
|
|
|
SCB_LIST_NULL,
|
|
|
|
|
ROLE_INITIATOR,
|
|
|
|
|
CAM_REQUEUE_REQ,
|
|
|
|
|
SEARCH_COMPLETE);
|
|
|
|
|
ahc_print_path(ahc, scb);
|
|
|
|
|
printf("Queuing a BDR SCB\n");
|
aic7xxx.c:
Filter incoming transfer negotiation requests to ensure they
never exceed the settings specified by the user.
In restart sequencer attempt to deal with a bug in the aic7895.
If a third party reset occurs at just the right time, the
stack register can lock up. When restarting the sequencer
after handling the SCSI reset, poke SEQADDR1 before resting
the sequencers program counter.
When something strange happens, dump the card's transaction
state via ahc_dump_card_state(). This should aid in debugging.
Handle request sense transactions via the QINFIFO instead of
attaching them to the waiting queue directly. The waiting
queue consumes card SCB resources and, in the pathological case
of every target on the bus beating our selection attemps and
issuing a check condition, could have caused us to run out
of SCBs. I have never seen this happen, and only early
cards with 3 or 4 SCBs had any real chance of ever getting
into this state.
Add additional sequencer interrupt codes to support firmware
diagnostics. The diagnostic code is enabled with the
AHC_DEBUG_SEQUENCER kernel option.
Make it possible to switch into and out of target mode on
the fly. The card comes up by default as an initiator but
will switch into target mode as soon as an enable lun operation
is performed. As always, target mode behavior is gated
by the AHC_TMODE_ENABLE kernel option so most users will
not be affected by this change.
In ahc_update_target_msg_request(), also issue a new
request if the ppr_options have changed.
Never issue a PPR as a target. It is forbidden by the spec.
Correct a bug in ahc_parse_msg() that prevented us from
responding to PPR messages as a target.
Mark SCBs that are on the untagged queue with a flag instead
of checking several fields in the SCB to see if the SCB should
be on the queue. This makes it easier for things like automatic
request sense requests to be queued without touching the
untagged queues even though they are untagged requests.
When dealing with ignore wide residue messages that occur
in the middle of a transfer, reset HADDR, not SHADDR for
non-ultra2 chips. Although SHADDR is where the firmware
fetches the ending transfer address for a save data pointers
request, it is readonly. Setting HADDR has the side effect
of also updating SHADDR.
Cleanup the output of ahc_dump_card_state() by nulling out the
free scb list in the non-paging case. The free list is only
used if we must page SCBs.
Correct the transmission of cdbs > 12 bytes in length. When
swapping HSCBs prior to notifing the sequencer of the new
transaction, the bus address pointer for the cdb must also
be recalculated to reflect its new location. We now defer
the calculation of the cdb address until just before queing
it to the card.
When pulling transfer negotiation settings out of scratch
ram, convert 5MHz/clock doubled settings to 10MHz.
Add a new function ahc_qinfifo_requeue_tail() for use by
error recovery actions and auto-request sense operations.
These operations always occur when the sequencer is paused,
so we can avoid the extra expense incurred in the normal
SCB queue method.
Use the BMOV instruction for all single byte moves on
controllers that support it. The bmov instruction is
twice as fast as an AND with an immediate of 0xFF as
is used on older controllers.
Correct a few bugs in ahc_dump_card_state(). If we have
hardware assisted queue registers, use them to get the
sequencer's idea of the head of the queue. When enumerating
the untagged queue, it helps to use the correct index for
the queue.
aic7xxx.h:
Indicate via a feature flag, which controllers can take
on both the target and the initiator role at the same time.
Add the AHC_SEQUENCER_DEBUG flag.
Add the SCB_CDB32_PTR flag used for dealing with cdbs
with lengths between 13 and 32 bytes.
Add new prototypes.
aic7xxx.reg:
Allow the SCSIBUSL register to be written to. This is
required to fix a selection timeout problem on the 7892/99.
Cleanup the sequencer interrupt codes so that all debugging
codes are grouped at the end of the list.
Correct the definition of the ULTRA_ENB and DISC_DSB locations
in scratch ram. This prevented the driver from properly honoring
these settings when no serial eeprom was available.
Remove an unused sequencer flag.
aic7xxx.seq:
Just before a potential select-out, clear the SCSIBUSL
register. Occasionally, during a selection timeout, the
contents of the register may be presented on the bus,
causing much confusion.
Add sequencer diagnostic code to detect software and or
hardware bugs. The code attempts to verify most list
operations so any corruption is caught before it occurs.
We also track information about why a particular reconnection
request was rejected.
Don't clobber the digital REQ/ACK filter setting in SXFRCTL0
when clearing the channel.
Fix a target mode bug that would cause us to return busy
status instead of queue full in respnse to a tagged transaction.
Cleanup the overrun case. It turns out that by simply
butting the chip in bitbucket mode, it will ack any
bytes until the phase changes. This drasticaly simplifies
things.
Prior to leaving the data phase, make sure that the S/G
preload queue is empty.
Remove code to place a request sense request on the waiting
queue. This is all handled by the kernel now.
Change the semantics of "findSCB". In the past, findSCB
ensured that a freshly paged in SCB appeared on the disconnected
list. The problem with this is that there is no guarantee that
the paged in SCB is for a disconnected transation. We now
defer any list manipulation to the caller who usually discards
the SCB via the free list.
Inline some busy target table operations.
Add a critical section to protect adding an SCB to
the disconnected list.
aic7xxx_freebsd.c:
Handle changes in the transfer negotiation setting API
to filter incoming requests. No filtering is necessary
for "goal" requests from the XPT.
Set the SCB_CDB32_PTR flag when queing a transaction with
a large cdb.
In ahc_timeout, only take action if the active SCB is
the timedout SCB. This deals with the case of two
transactions to the same device with different timeout
values.
Use ahc_qinfifo_requeu_tail() instead of home grown
version.
aic7xxx_inline.h:
Honor SCB_CDB32_PTR when queuing a new request.
aic7xxx_pci.c:
Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
|
|
|
ahc_qinfifo_requeue_tail(ahc, scb);
|
|
|
|
|
ahc_outb(ahc, SCBPTR, saved_scbptr);
|
2000-09-16 20:02:28 +00:00
|
|
|
scb->io_ctx->ccb_h.timeout_ch =
|
|
|
|
|
timeout(ahc_timeout, (caddr_t)scb, 2 * hz);
|
This is an MFC candidate.
ahc_eisa.c:
Change aic7770_map_int to take an additional irq parameter.
Although we can get the irq from the eisa dev under FreeBSD,
we can't do this under linux, so the OSM interface must supply
this.
ahc_pci.c:
Move ahc_power_state_change() to the OSM. This allows us to
use a platform supplied function that does the same thing.
-current will move to the FreeBSD native API in the near
future.
aic7770.c:
Sync up with core changes to support Linux EISA.
We now store a 2 bit primary channel number rather
than a bit flag that only allows b to be the primary
channel. Adjust for this change.
aic7xxx.c:
Namespace and staticization cleanup. All exported symbols
use an "ahc_" prefix to avoid collisions with other modules.
Correct a logic bug that prevented us from dropping
ATN during some exceptional conditions during message
processing.
Take advantage of a new flag managed by the sequencer
that indicates if an SCB fetch is in progress. If so,
the currently selected SCB needs to be returned to the
free list to prevent an SCB leak. This leak is a rarity
and would only occur if a bus reset or timeout resulting
in a bus reset occurred in the middle of an SCB fetch.
Don't attempt to perform ULTRA transfers on ultra capable
adapters missing the external precision resistor required
for ultra speeds. I've never encountered an adapter
configured this way, but better safe than sorry.
Handle the case of 5MHz user sync rate set as "0" instead of 0x1c
in scratch ram.
If we lookup a period of 0 in our table (async), clear the scsi offset.
aic7xxx.h:
Adjust for the primary channel being represented as
a 2 bit integer in the flags member of the ahc softc.
Cleanup the flags definitions so that comment blocks are
not cramped.
Update seeprom definitions to correctly reflect the fact
that the primary channel is represented as a 2 bit integer.
Add AHC_ULTRA_DIASABLED softc flag to denote controllers
missing the external precision resistor.
aic7xxx.reg:
Add DFCACHETH to the definition of DFSTATUS for completness sake.
Add SEQ_FLAGS2 which currently only contains the SCB_DMA
(SCB DMA in progress) flag.
aic7xxx.seq:
Correct a problem when one lun has a disconnected untagged
transaction and another lun has disconnected tagged transactions.
Just because an entry is found in the untagged table doesn't
mean that it will match. If the match on the lun fails, cleanup
the SCB (return it to the disconnected list or free it), and snoop
for a tag message. Before this change, we reported an unsolicited
reselection. This bug was introduced about a month ago during an
overly aggressive optimization pass on the reselection code.
When cleaning up an SCB, we can't just blindly free the SCB. In
the paging case, if the SCB came off of the disconnected list, its
state may never have been updated in host memory. So, check the
disconnected bit in SCB_CONTROL and return the SCB to the disconnected
list if appropriate.
Manage the SCB_DMA flag of SEQ_FLAGS2.
More carefully shutdown the S/G dma engine in all cases by using
a subroutine. Supposedly not doing this can cause an arbiter hang
on some ULTRA2 chips.
Formatting cleanup.
On some chips, at least the aic7856, the transition from
MREQPEND to HDONE can take a full 4 clock cycles. Test
HDONE one more time to avoid this race. We only want our
FIFO hung recovery code to execute when the engine is
really hung.
aic7xxx_93cx6.c:
Sync perforce ids.
aic7xxx_freebsd.c:
Adjust for the primary channel being a 2 bit integer
rather than a flag for 'B' channel being the primary.
Namespace cleanup.
Unpause the sequencer in one error recovery path that
neglected to do so. This could have caused us to perform
a bus reset when a recovery message might have otherwise been
successful.
aic7xxx_freebsd.h:
Use AHC_PCI_CONFIG for controlling compilation of PCI
support consistently throughout the driver.
Move ahc_power_state_change() to OSM.
aic7xxx_inline.h
Namespace cleanup.
Adjust our interrupt handler so it will work in the edge
interrupt case. We must process all interrupt sources
when the interrupt fires or risk not ever getting an
interrupt again. This involves marking the fact
that we are relying on an edge interrupt in ahc->flags
and checking for this condition in addition to the
AHC_ALL_INTERRUPTS flag. This fixes hangs on the
284X and any other aic7770 installation where level
interrupts are not available.
aic7xxx_pci.c:
Move the powerstate manipulation code into the OSM. Several
OSes now provide this functionality natively.
Take another shot at using the data stored in scratch ram
if the SCB2 signature is correct and no SEEPROM data is
available. In the past this failed if external SCB ram
was configured because the memory port was locked. We
now release the memory port prior to testing the values
in SCB2 and re-acquire it prior to doing termination control.
Adjust for new 2 bit primary channel setting.
Trust the STPWLEVEL setting on v 3.X BIOSes too.
Configure any 785X ID in the same fashion and assume
that any device with a rev id of 1 or higher has the
PCI 2.1 retry bug.
2001-03-11 06:34:17 +00:00
|
|
|
ahc_unpause(ahc);
|
2000-09-16 20:02:28 +00:00
|
|
|
} else {
|
|
|
|
|
/* Go "immediatly" to the bus reset */
|
|
|
|
|
/* This shouldn't happen */
|
|
|
|
|
ahc_set_recoveryscb(ahc, scb);
|
|
|
|
|
ahc_print_path(ahc, scb);
|
|
|
|
|
printf("SCB %d: Immediate reset. "
|
|
|
|
|
"Flags = 0x%x\n", scb->hscb->tag,
|
|
|
|
|
scb->flags);
|
|
|
|
|
goto bus_reset;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
ahc_unlock(ahc, &s);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static void
|
|
|
|
|
ahc_abort_ccb(struct ahc_softc *ahc, struct cam_sim *sim, union ccb *ccb)
|
|
|
|
|
{
|
|
|
|
|
union ccb *abort_ccb;
|
|
|
|
|
|
|
|
|
|
abort_ccb = ccb->cab.abort_ccb;
|
|
|
|
|
switch (abort_ccb->ccb_h.func_code) {
|
|
|
|
|
case XPT_ACCEPT_TARGET_IO:
|
|
|
|
|
case XPT_IMMED_NOTIFY:
|
|
|
|
|
case XPT_CONT_TARGET_IO:
|
|
|
|
|
{
|
2001-03-29 00:36:35 +00:00
|
|
|
struct ahc_tmode_tstate *tstate;
|
|
|
|
|
struct ahc_tmode_lstate *lstate;
|
2000-09-16 20:02:28 +00:00
|
|
|
struct ccb_hdr_slist *list;
|
|
|
|
|
cam_status status;
|
|
|
|
|
|
|
|
|
|
status = ahc_find_tmode_devs(ahc, sim, abort_ccb, &tstate,
|
|
|
|
|
&lstate, TRUE);
|
|
|
|
|
|
|
|
|
|
if (status != CAM_REQ_CMP) {
|
|
|
|
|
ccb->ccb_h.status = status;
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (abort_ccb->ccb_h.func_code == XPT_ACCEPT_TARGET_IO)
|
|
|
|
|
list = &lstate->accept_tios;
|
|
|
|
|
else if (abort_ccb->ccb_h.func_code == XPT_IMMED_NOTIFY)
|
|
|
|
|
list = &lstate->immed_notifies;
|
|
|
|
|
else
|
|
|
|
|
list = NULL;
|
|
|
|
|
|
|
|
|
|
if (list != NULL) {
|
|
|
|
|
struct ccb_hdr *curelm;
|
|
|
|
|
int found;
|
|
|
|
|
|
|
|
|
|
curelm = SLIST_FIRST(list);
|
|
|
|
|
found = 0;
|
|
|
|
|
if (curelm == &abort_ccb->ccb_h) {
|
|
|
|
|
found = 1;
|
|
|
|
|
SLIST_REMOVE_HEAD(list, sim_links.sle);
|
|
|
|
|
} else {
|
|
|
|
|
while(curelm != NULL) {
|
|
|
|
|
struct ccb_hdr *nextelm;
|
|
|
|
|
|
|
|
|
|
nextelm =
|
|
|
|
|
SLIST_NEXT(curelm, sim_links.sle);
|
|
|
|
|
|
|
|
|
|
if (nextelm == &abort_ccb->ccb_h) {
|
|
|
|
|
found = 1;
|
|
|
|
|
SLIST_NEXT(curelm,
|
|
|
|
|
sim_links.sle) =
|
|
|
|
|
SLIST_NEXT(nextelm,
|
|
|
|
|
sim_links.sle);
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
curelm = nextelm;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (found) {
|
|
|
|
|
abort_ccb->ccb_h.status = CAM_REQ_ABORTED;
|
|
|
|
|
xpt_done(abort_ccb);
|
|
|
|
|
ccb->ccb_h.status = CAM_REQ_CMP;
|
|
|
|
|
} else {
|
aic7xxx.c:
Filter incoming transfer negotiation requests to ensure they
never exceed the settings specified by the user.
In restart sequencer attempt to deal with a bug in the aic7895.
If a third party reset occurs at just the right time, the
stack register can lock up. When restarting the sequencer
after handling the SCSI reset, poke SEQADDR1 before resting
the sequencers program counter.
When something strange happens, dump the card's transaction
state via ahc_dump_card_state(). This should aid in debugging.
Handle request sense transactions via the QINFIFO instead of
attaching them to the waiting queue directly. The waiting
queue consumes card SCB resources and, in the pathological case
of every target on the bus beating our selection attemps and
issuing a check condition, could have caused us to run out
of SCBs. I have never seen this happen, and only early
cards with 3 or 4 SCBs had any real chance of ever getting
into this state.
Add additional sequencer interrupt codes to support firmware
diagnostics. The diagnostic code is enabled with the
AHC_DEBUG_SEQUENCER kernel option.
Make it possible to switch into and out of target mode on
the fly. The card comes up by default as an initiator but
will switch into target mode as soon as an enable lun operation
is performed. As always, target mode behavior is gated
by the AHC_TMODE_ENABLE kernel option so most users will
not be affected by this change.
In ahc_update_target_msg_request(), also issue a new
request if the ppr_options have changed.
Never issue a PPR as a target. It is forbidden by the spec.
Correct a bug in ahc_parse_msg() that prevented us from
responding to PPR messages as a target.
Mark SCBs that are on the untagged queue with a flag instead
of checking several fields in the SCB to see if the SCB should
be on the queue. This makes it easier for things like automatic
request sense requests to be queued without touching the
untagged queues even though they are untagged requests.
When dealing with ignore wide residue messages that occur
in the middle of a transfer, reset HADDR, not SHADDR for
non-ultra2 chips. Although SHADDR is where the firmware
fetches the ending transfer address for a save data pointers
request, it is readonly. Setting HADDR has the side effect
of also updating SHADDR.
Cleanup the output of ahc_dump_card_state() by nulling out the
free scb list in the non-paging case. The free list is only
used if we must page SCBs.
Correct the transmission of cdbs > 12 bytes in length. When
swapping HSCBs prior to notifing the sequencer of the new
transaction, the bus address pointer for the cdb must also
be recalculated to reflect its new location. We now defer
the calculation of the cdb address until just before queing
it to the card.
When pulling transfer negotiation settings out of scratch
ram, convert 5MHz/clock doubled settings to 10MHz.
Add a new function ahc_qinfifo_requeue_tail() for use by
error recovery actions and auto-request sense operations.
These operations always occur when the sequencer is paused,
so we can avoid the extra expense incurred in the normal
SCB queue method.
Use the BMOV instruction for all single byte moves on
controllers that support it. The bmov instruction is
twice as fast as an AND with an immediate of 0xFF as
is used on older controllers.
Correct a few bugs in ahc_dump_card_state(). If we have
hardware assisted queue registers, use them to get the
sequencer's idea of the head of the queue. When enumerating
the untagged queue, it helps to use the correct index for
the queue.
aic7xxx.h:
Indicate via a feature flag, which controllers can take
on both the target and the initiator role at the same time.
Add the AHC_SEQUENCER_DEBUG flag.
Add the SCB_CDB32_PTR flag used for dealing with cdbs
with lengths between 13 and 32 bytes.
Add new prototypes.
aic7xxx.reg:
Allow the SCSIBUSL register to be written to. This is
required to fix a selection timeout problem on the 7892/99.
Cleanup the sequencer interrupt codes so that all debugging
codes are grouped at the end of the list.
Correct the definition of the ULTRA_ENB and DISC_DSB locations
in scratch ram. This prevented the driver from properly honoring
these settings when no serial eeprom was available.
Remove an unused sequencer flag.
aic7xxx.seq:
Just before a potential select-out, clear the SCSIBUSL
register. Occasionally, during a selection timeout, the
contents of the register may be presented on the bus,
causing much confusion.
Add sequencer diagnostic code to detect software and or
hardware bugs. The code attempts to verify most list
operations so any corruption is caught before it occurs.
We also track information about why a particular reconnection
request was rejected.
Don't clobber the digital REQ/ACK filter setting in SXFRCTL0
when clearing the channel.
Fix a target mode bug that would cause us to return busy
status instead of queue full in respnse to a tagged transaction.
Cleanup the overrun case. It turns out that by simply
butting the chip in bitbucket mode, it will ack any
bytes until the phase changes. This drasticaly simplifies
things.
Prior to leaving the data phase, make sure that the S/G
preload queue is empty.
Remove code to place a request sense request on the waiting
queue. This is all handled by the kernel now.
Change the semantics of "findSCB". In the past, findSCB
ensured that a freshly paged in SCB appeared on the disconnected
list. The problem with this is that there is no guarantee that
the paged in SCB is for a disconnected transation. We now
defer any list manipulation to the caller who usually discards
the SCB via the free list.
Inline some busy target table operations.
Add a critical section to protect adding an SCB to
the disconnected list.
aic7xxx_freebsd.c:
Handle changes in the transfer negotiation setting API
to filter incoming requests. No filtering is necessary
for "goal" requests from the XPT.
Set the SCB_CDB32_PTR flag when queing a transaction with
a large cdb.
In ahc_timeout, only take action if the active SCB is
the timedout SCB. This deals with the case of two
transactions to the same device with different timeout
values.
Use ahc_qinfifo_requeu_tail() instead of home grown
version.
aic7xxx_inline.h:
Honor SCB_CDB32_PTR when queuing a new request.
aic7xxx_pci.c:
Use the maximum data fifo threshold for all chips.
2000-10-31 18:43:29 +00:00
|
|
|
xpt_print_path(abort_ccb->ccb_h.path);
|
2000-09-16 20:02:28 +00:00
|
|
|
printf("Not found\n");
|
|
|
|
|
ccb->ccb_h.status = CAM_PATH_INVALID;
|
|
|
|
|
}
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
/* FALLTHROUGH */
|
|
|
|
|
}
|
|
|
|
|
case XPT_SCSI_IO:
|
|
|
|
|
/* XXX Fully implement the hard ones */
|
|
|
|
|
ccb->ccb_h.status = CAM_UA_ABORT;
|
|
|
|
|
break;
|
|
|
|
|
default:
|
|
|
|
|
ccb->ccb_h.status = CAM_REQ_INVALID;
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
xpt_done(ccb);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
void
|
2000-09-22 22:18:05 +00:00
|
|
|
ahc_send_async(struct ahc_softc *ahc, char channel, u_int target,
|
2001-05-15 19:41:12 +00:00
|
|
|
u_int lun, ac_code code, void *opt_arg)
|
2000-09-16 20:02:28 +00:00
|
|
|
{
|
|
|
|
|
struct ccb_trans_settings cts;
|
|
|
|
|
struct cam_path *path;
|
|
|
|
|
void *arg;
|
|
|
|
|
int error;
|
|
|
|
|
|
|
|
|
|
arg = NULL;
|
2000-09-22 22:18:05 +00:00
|
|
|
error = ahc_create_path(ahc, channel, target, lun, &path);
|
2000-09-16 20:02:28 +00:00
|
|
|
|
|
|
|
|
if (error != CAM_REQ_CMP)
|
|
|
|
|
return;
|
|
|
|
|
|
|
|
|
|
switch (code) {
|
|
|
|
|
case AC_TRANSFER_NEG:
|
2001-05-15 19:41:12 +00:00
|
|
|
{
|
2000-09-16 20:02:28 +00:00
|
|
|
#ifdef AHC_NEW_TRAN_SETTINGS
|
2001-05-15 19:41:12 +00:00
|
|
|
struct ccb_trans_settings_scsi *scsi;
|
|
|
|
|
|
2000-09-16 20:02:28 +00:00
|
|
|
cts.type = CTS_TYPE_CURRENT_SETTINGS;
|
2001-05-15 19:41:12 +00:00
|
|
|
scsi = &cts.proto_specific.scsi;
|
2000-09-16 20:02:28 +00:00
|
|
|
#else
|
|
|
|
|
cts.flags = CCB_TRANS_CURRENT_SETTINGS;
|
|
|
|
|
#endif
|
|
|
|
|
cts.ccb_h.path = path;
|
2000-09-22 22:18:05 +00:00
|
|
|
cts.ccb_h.target_id = target;
|
|
|
|
|
cts.ccb_h.target_lun = lun;
|
|
|
|
|
ahc_get_tran_settings(ahc, channel == 'A' ? ahc->our_id
|
|
|
|
|
: ahc->our_id_b,
|
|
|
|
|
channel, &cts);
|
2000-09-16 20:02:28 +00:00
|
|
|
arg = &cts;
|
2001-05-15 19:41:12 +00:00
|
|
|
#ifdef AHC_NEW_TRAN_SETTINGS
|
|
|
|
|
scsi->valid &= ~CTS_SCSI_VALID_TQ;
|
|
|
|
|
scsi->flags &= ~CTS_SCSI_FLAGS_TAG_ENB;
|
|
|
|
|
#else
|
|
|
|
|
cts.valid &= ~CCB_TRANS_TQ_VALID;
|
|
|
|
|
cts.flags &= ~CCB_TRANS_TAG_ENB;
|
|
|
|
|
#endif
|
|
|
|
|
if (opt_arg == NULL)
|
|
|
|
|
break;
|
|
|
|
|
if (*((ahc_queue_alg *)opt_arg) == AHC_QUEUE_TAGGED)
|
|
|
|
|
#ifdef AHC_NEW_TRAN_SETTINGS
|
|
|
|
|
scsi->flags |= ~CTS_SCSI_FLAGS_TAG_ENB;
|
|
|
|
|
scsi->valid |= CTS_SCSI_VALID_TQ;
|
|
|
|
|
#else
|
|
|
|
|
cts.flags |= CCB_TRANS_TAG_ENB;
|
|
|
|
|
cts.valid |= CCB_TRANS_TQ_VALID;
|
|
|
|
|
#endif
|
2000-09-16 20:02:28 +00:00
|
|
|
break;
|
2001-05-15 19:41:12 +00:00
|
|
|
}
|
2000-09-16 20:02:28 +00:00
|
|
|
case AC_SENT_BDR:
|
|
|
|
|
case AC_BUS_RESET:
|
|
|
|
|
break;
|
|
|
|
|
default:
|
|
|
|
|
panic("ahc_send_async: Unexpected async event");
|
|
|
|
|
}
|
|
|
|
|
xpt_async(code, path, arg);
|
2000-11-06 20:05:38 +00:00
|
|
|
xpt_free_path(path);
|
2000-09-16 20:02:28 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
void
|
|
|
|
|
ahc_platform_set_tags(struct ahc_softc *ahc,
|
|
|
|
|
struct ahc_devinfo *devinfo, int enable)
|
|
|
|
|
{
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
int
|
|
|
|
|
ahc_platform_alloc(struct ahc_softc *ahc, void *platform_arg)
|
|
|
|
|
{
|
2000-10-29 15:47:16 +00:00
|
|
|
ahc->platform_data = malloc(sizeof(struct ahc_platform_data), M_DEVBUF,
|
|
|
|
|
M_NOWAIT | M_ZERO);
|
2000-09-16 20:02:28 +00:00
|
|
|
if (ahc->platform_data == NULL)
|
|
|
|
|
return (ENOMEM);
|
|
|
|
|
return (0);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
void
|
|
|
|
|
ahc_platform_free(struct ahc_softc *ahc)
|
|
|
|
|
{
|
ahc_eisa.c:
ahc_pci.c:
Add detach support.
Make use of soft allocated on our behalf by newbus.
For PCI devices, disable the mapping type we aren't
using for extra protection from rogue code.
aic7xxx_93cx6.c:
aic7xxx_93cx6.h:
Sync perforce IDs.
aic7xxx_freebsd.c:
Capture the eventhandle returned by EVENTHANDER_REGISTER
so we can kill the handler off during detach.
Use AHC_* constants instead of hard coded numbers in a
few more places.
Test PPR option state when deciding to "really" negotiate
when the CAM_NEGOTIATE flag is passed in a CCB.
Make use of core "ahc_pause_and_flushwork" routine in our
timeout handler rather than re-inventing this code.
Cleanup all of our resources (really!) in ahc_platform_free().
We should be all set to become a module now.
Implement the core ahc_detach() routine shared by all of
the FreeBSD front-ends.
aic7xxx_freebsd.h:
Softc storage for our event handler.
Null implementation for the ahc_platform_flushwork() OSM
callback. FreeBSD doesn't need this as XPT callbacks are
safe from all contexts and are done directly in ahc_done().
aic7xxx_inline.h:
Implement new lazy interrupt scheme. To avoid an extra
PCI bus read, we first check our completion queues to
see if any work has completed. If work is available, we
assume that this is the source of the interrupt and skip
reading INTSTAT. Any remaining interrupt status will be
cleared by a second call to the interrupt handler should
the interrupt line still be asserted. This drops the
interrupt handler down to a single PCI bus read in the
common case of I/O completion. This is the same overhead
as in the not so distant past, but the extra sanity of
perforning a PCI read after clearing the command complete
interrupt and before running the completion queue to avoid
missing command complete interrupts added a cycle.
aic7xxx.c:
During initialization, be sure to initialize all scratch
ram locations before they are read to avoid parity errors.
In this case, we use a new function, ahc_unbusy_tcl() to
initialize the scratch ram busy target table.
Replace instances of ahc_index_busy_tcl() used to unbusy
a tcl without looking at the old value with ahc_unbusy_tcl().
Modify ahc_sent_msg so that it can find single byte messages.
ahc_sent_msg is now used to determine if a transfer negotiation
attempt resulted in a bus free.
Be more careful in filtering out only the SCSI interrupts
of interest in ahc_handle_scsiint.
Rearrange interrupt clearing code to ensure that at least
one PCI transaction occurrs after hitting CLRSINT1 and
writting to CLRINT. CLRSINT1 writes take a bit to
take effect, and the re-arrangement provides sufficient
delay to ensure the write to CLRINT is effective. The
old code might report a spurious interrupt on some "fast"
chipsets.
export ahc-update_target_msg_request for use by OSM code.
If a target does not respond to our ATN request, clear
it once we move to a non-message phase. This avoids
sending a MSG_NOOP in some later message out phase.
Use max lun and max target constants instead of
hard-coded values.
Use softc storage built into our device_t under FreeBSD.
Fix a bug in ahc_free() that caused us to delete
resources that were not allocated.
Clean up any tstate/lstate info in ahc_free().
Clear the powerdown state in ahc_reset() so that
registers can be accessed.
Add a preliminary function for pausing the chip and
processing any posted work.
Add a preliminary suspend and resume functions.
aic7xxx.h:
Limit the number of supported luns to 64. We don't
support information unit transfers, so this is the
maximum that makes sense for these chips.
Add a new flag AHC_ALL_INTERRUPTS that forces the
processing of all interrupt state in a single invokation
of ahc_intr(). When the flag is not set, we use the
lazy interrupt handling scheme.
Add data structures to store controller state while
we are suspended.
Use constants instead of hard coded values where appropriate.
Correct some harmless "unsigned/signed" conflicts.
aic7xxx.seq:
Only perform the SCSIBUSL fix on ULTRA2 or newer controllers.
Older controllers seem to be confused by this.
In target mode, ignore PHASEMIS during data phases.
This bit seems to be flakey on U160 controllers acting
in target mode.
aic7xxx_pci.c:
Add support for the 29160C CPCI adapter.
Add definitions for subvendor ID information
available for devices with the "9005" vendor id.
We currently use this information to determine
if a multi-function device doesn't have the second
channel hooked up on a board.
Add rudimentary power mode code so we can put the
controller into the D0 state. In the future this
will be an OSM callback so that in FreeBSD we don't
duplicate functionality provided by the PCI code.
The powerstate code was added after I'd completed
my regression tests on this code.
Only capture "left over BIOS state" if the POWRDN
setting is not set in HCNTRL.
In target mode, don't bother sending incremental
CRC data.
2000-12-20 01:11:37 +00:00
|
|
|
struct ahc_platform_data *pdata;
|
2000-09-16 20:02:28 +00:00
|
|
|
|
ahc_eisa.c:
ahc_pci.c:
Add detach support.
Make use of soft allocated on our behalf by newbus.
For PCI devices, disable the mapping type we aren't
using for extra protection from rogue code.
aic7xxx_93cx6.c:
aic7xxx_93cx6.h:
Sync perforce IDs.
aic7xxx_freebsd.c:
Capture the eventhandle returned by EVENTHANDER_REGISTER
so we can kill the handler off during detach.
Use AHC_* constants instead of hard coded numbers in a
few more places.
Test PPR option state when deciding to "really" negotiate
when the CAM_NEGOTIATE flag is passed in a CCB.
Make use of core "ahc_pause_and_flushwork" routine in our
timeout handler rather than re-inventing this code.
Cleanup all of our resources (really!) in ahc_platform_free().
We should be all set to become a module now.
Implement the core ahc_detach() routine shared by all of
the FreeBSD front-ends.
aic7xxx_freebsd.h:
Softc storage for our event handler.
Null implementation for the ahc_platform_flushwork() OSM
callback. FreeBSD doesn't need this as XPT callbacks are
safe from all contexts and are done directly in ahc_done().
aic7xxx_inline.h:
Implement new lazy interrupt scheme. To avoid an extra
PCI bus read, we first check our completion queues to
see if any work has completed. If work is available, we
assume that this is the source of the interrupt and skip
reading INTSTAT. Any remaining interrupt status will be
cleared by a second call to the interrupt handler should
the interrupt line still be asserted. This drops the
interrupt handler down to a single PCI bus read in the
common case of I/O completion. This is the same overhead
as in the not so distant past, but the extra sanity of
perforning a PCI read after clearing the command complete
interrupt and before running the completion queue to avoid
missing command complete interrupts added a cycle.
aic7xxx.c:
During initialization, be sure to initialize all scratch
ram locations before they are read to avoid parity errors.
In this case, we use a new function, ahc_unbusy_tcl() to
initialize the scratch ram busy target table.
Replace instances of ahc_index_busy_tcl() used to unbusy
a tcl without looking at the old value with ahc_unbusy_tcl().
Modify ahc_sent_msg so that it can find single byte messages.
ahc_sent_msg is now used to determine if a transfer negotiation
attempt resulted in a bus free.
Be more careful in filtering out only the SCSI interrupts
of interest in ahc_handle_scsiint.
Rearrange interrupt clearing code to ensure that at least
one PCI transaction occurrs after hitting CLRSINT1 and
writting to CLRINT. CLRSINT1 writes take a bit to
take effect, and the re-arrangement provides sufficient
delay to ensure the write to CLRINT is effective. The
old code might report a spurious interrupt on some "fast"
chipsets.
export ahc-update_target_msg_request for use by OSM code.
If a target does not respond to our ATN request, clear
it once we move to a non-message phase. This avoids
sending a MSG_NOOP in some later message out phase.
Use max lun and max target constants instead of
hard-coded values.
Use softc storage built into our device_t under FreeBSD.
Fix a bug in ahc_free() that caused us to delete
resources that were not allocated.
Clean up any tstate/lstate info in ahc_free().
Clear the powerdown state in ahc_reset() so that
registers can be accessed.
Add a preliminary function for pausing the chip and
processing any posted work.
Add a preliminary suspend and resume functions.
aic7xxx.h:
Limit the number of supported luns to 64. We don't
support information unit transfers, so this is the
maximum that makes sense for these chips.
Add a new flag AHC_ALL_INTERRUPTS that forces the
processing of all interrupt state in a single invokation
of ahc_intr(). When the flag is not set, we use the
lazy interrupt handling scheme.
Add data structures to store controller state while
we are suspended.
Use constants instead of hard coded values where appropriate.
Correct some harmless "unsigned/signed" conflicts.
aic7xxx.seq:
Only perform the SCSIBUSL fix on ULTRA2 or newer controllers.
Older controllers seem to be confused by this.
In target mode, ignore PHASEMIS during data phases.
This bit seems to be flakey on U160 controllers acting
in target mode.
aic7xxx_pci.c:
Add support for the 29160C CPCI adapter.
Add definitions for subvendor ID information
available for devices with the "9005" vendor id.
We currently use this information to determine
if a multi-function device doesn't have the second
channel hooked up on a board.
Add rudimentary power mode code so we can put the
controller into the D0 state. In the future this
will be an OSM callback so that in FreeBSD we don't
duplicate functionality provided by the PCI code.
The powerstate code was added after I'd completed
my regression tests on this code.
Only capture "left over BIOS state" if the POWRDN
setting is not set in HCNTRL.
In target mode, don't bother sending incremental
CRC data.
2000-12-20 01:11:37 +00:00
|
|
|
pdata = ahc->platform_data;
|
|
|
|
|
if (pdata != NULL) {
|
|
|
|
|
if (pdata->regs != NULL)
|
2000-09-16 20:02:28 +00:00
|
|
|
bus_release_resource(ahc->dev_softc,
|
ahc_eisa.c:
ahc_pci.c:
Add detach support.
Make use of soft allocated on our behalf by newbus.
For PCI devices, disable the mapping type we aren't
using for extra protection from rogue code.
aic7xxx_93cx6.c:
aic7xxx_93cx6.h:
Sync perforce IDs.
aic7xxx_freebsd.c:
Capture the eventhandle returned by EVENTHANDER_REGISTER
so we can kill the handler off during detach.
Use AHC_* constants instead of hard coded numbers in a
few more places.
Test PPR option state when deciding to "really" negotiate
when the CAM_NEGOTIATE flag is passed in a CCB.
Make use of core "ahc_pause_and_flushwork" routine in our
timeout handler rather than re-inventing this code.
Cleanup all of our resources (really!) in ahc_platform_free().
We should be all set to become a module now.
Implement the core ahc_detach() routine shared by all of
the FreeBSD front-ends.
aic7xxx_freebsd.h:
Softc storage for our event handler.
Null implementation for the ahc_platform_flushwork() OSM
callback. FreeBSD doesn't need this as XPT callbacks are
safe from all contexts and are done directly in ahc_done().
aic7xxx_inline.h:
Implement new lazy interrupt scheme. To avoid an extra
PCI bus read, we first check our completion queues to
see if any work has completed. If work is available, we
assume that this is the source of the interrupt and skip
reading INTSTAT. Any remaining interrupt status will be
cleared by a second call to the interrupt handler should
the interrupt line still be asserted. This drops the
interrupt handler down to a single PCI bus read in the
common case of I/O completion. This is the same overhead
as in the not so distant past, but the extra sanity of
perforning a PCI read after clearing the command complete
interrupt and before running the completion queue to avoid
missing command complete interrupts added a cycle.
aic7xxx.c:
During initialization, be sure to initialize all scratch
ram locations before they are read to avoid parity errors.
In this case, we use a new function, ahc_unbusy_tcl() to
initialize the scratch ram busy target table.
Replace instances of ahc_index_busy_tcl() used to unbusy
a tcl without looking at the old value with ahc_unbusy_tcl().
Modify ahc_sent_msg so that it can find single byte messages.
ahc_sent_msg is now used to determine if a transfer negotiation
attempt resulted in a bus free.
Be more careful in filtering out only the SCSI interrupts
of interest in ahc_handle_scsiint.
Rearrange interrupt clearing code to ensure that at least
one PCI transaction occurrs after hitting CLRSINT1 and
writting to CLRINT. CLRSINT1 writes take a bit to
take effect, and the re-arrangement provides sufficient
delay to ensure the write to CLRINT is effective. The
old code might report a spurious interrupt on some "fast"
chipsets.
export ahc-update_target_msg_request for use by OSM code.
If a target does not respond to our ATN request, clear
it once we move to a non-message phase. This avoids
sending a MSG_NOOP in some later message out phase.
Use max lun and max target constants instead of
hard-coded values.
Use softc storage built into our device_t under FreeBSD.
Fix a bug in ahc_free() that caused us to delete
resources that were not allocated.
Clean up any tstate/lstate info in ahc_free().
Clear the powerdown state in ahc_reset() so that
registers can be accessed.
Add a preliminary function for pausing the chip and
processing any posted work.
Add a preliminary suspend and resume functions.
aic7xxx.h:
Limit the number of supported luns to 64. We don't
support information unit transfers, so this is the
maximum that makes sense for these chips.
Add a new flag AHC_ALL_INTERRUPTS that forces the
processing of all interrupt state in a single invokation
of ahc_intr(). When the flag is not set, we use the
lazy interrupt handling scheme.
Add data structures to store controller state while
we are suspended.
Use constants instead of hard coded values where appropriate.
Correct some harmless "unsigned/signed" conflicts.
aic7xxx.seq:
Only perform the SCSIBUSL fix on ULTRA2 or newer controllers.
Older controllers seem to be confused by this.
In target mode, ignore PHASEMIS during data phases.
This bit seems to be flakey on U160 controllers acting
in target mode.
aic7xxx_pci.c:
Add support for the 29160C CPCI adapter.
Add definitions for subvendor ID information
available for devices with the "9005" vendor id.
We currently use this information to determine
if a multi-function device doesn't have the second
channel hooked up on a board.
Add rudimentary power mode code so we can put the
controller into the D0 state. In the future this
will be an OSM callback so that in FreeBSD we don't
duplicate functionality provided by the PCI code.
The powerstate code was added after I'd completed
my regression tests on this code.
Only capture "left over BIOS state" if the POWRDN
setting is not set in HCNTRL.
In target mode, don't bother sending incremental
CRC data.
2000-12-20 01:11:37 +00:00
|
|
|
pdata->regs_res_type,
|
|
|
|
|
pdata->regs_res_id,
|
|
|
|
|
pdata->regs);
|
2000-09-16 20:02:28 +00:00
|
|
|
|
ahc_eisa.c:
ahc_pci.c:
Add detach support.
Make use of soft allocated on our behalf by newbus.
For PCI devices, disable the mapping type we aren't
using for extra protection from rogue code.
aic7xxx_93cx6.c:
aic7xxx_93cx6.h:
Sync perforce IDs.
aic7xxx_freebsd.c:
Capture the eventhandle returned by EVENTHANDER_REGISTER
so we can kill the handler off during detach.
Use AHC_* constants instead of hard coded numbers in a
few more places.
Test PPR option state when deciding to "really" negotiate
when the CAM_NEGOTIATE flag is passed in a CCB.
Make use of core "ahc_pause_and_flushwork" routine in our
timeout handler rather than re-inventing this code.
Cleanup all of our resources (really!) in ahc_platform_free().
We should be all set to become a module now.
Implement the core ahc_detach() routine shared by all of
the FreeBSD front-ends.
aic7xxx_freebsd.h:
Softc storage for our event handler.
Null implementation for the ahc_platform_flushwork() OSM
callback. FreeBSD doesn't need this as XPT callbacks are
safe from all contexts and are done directly in ahc_done().
aic7xxx_inline.h:
Implement new lazy interrupt scheme. To avoid an extra
PCI bus read, we first check our completion queues to
see if any work has completed. If work is available, we
assume that this is the source of the interrupt and skip
reading INTSTAT. Any remaining interrupt status will be
cleared by a second call to the interrupt handler should
the interrupt line still be asserted. This drops the
interrupt handler down to a single PCI bus read in the
common case of I/O completion. This is the same overhead
as in the not so distant past, but the extra sanity of
perforning a PCI read after clearing the command complete
interrupt and before running the completion queue to avoid
missing command complete interrupts added a cycle.
aic7xxx.c:
During initialization, be sure to initialize all scratch
ram locations before they are read to avoid parity errors.
In this case, we use a new function, ahc_unbusy_tcl() to
initialize the scratch ram busy target table.
Replace instances of ahc_index_busy_tcl() used to unbusy
a tcl without looking at the old value with ahc_unbusy_tcl().
Modify ahc_sent_msg so that it can find single byte messages.
ahc_sent_msg is now used to determine if a transfer negotiation
attempt resulted in a bus free.
Be more careful in filtering out only the SCSI interrupts
of interest in ahc_handle_scsiint.
Rearrange interrupt clearing code to ensure that at least
one PCI transaction occurrs after hitting CLRSINT1 and
writting to CLRINT. CLRSINT1 writes take a bit to
take effect, and the re-arrangement provides sufficient
delay to ensure the write to CLRINT is effective. The
old code might report a spurious interrupt on some "fast"
chipsets.
export ahc-update_target_msg_request for use by OSM code.
If a target does not respond to our ATN request, clear
it once we move to a non-message phase. This avoids
sending a MSG_NOOP in some later message out phase.
Use max lun and max target constants instead of
hard-coded values.
Use softc storage built into our device_t under FreeBSD.
Fix a bug in ahc_free() that caused us to delete
resources that were not allocated.
Clean up any tstate/lstate info in ahc_free().
Clear the powerdown state in ahc_reset() so that
registers can be accessed.
Add a preliminary function for pausing the chip and
processing any posted work.
Add a preliminary suspend and resume functions.
aic7xxx.h:
Limit the number of supported luns to 64. We don't
support information unit transfers, so this is the
maximum that makes sense for these chips.
Add a new flag AHC_ALL_INTERRUPTS that forces the
processing of all interrupt state in a single invokation
of ahc_intr(). When the flag is not set, we use the
lazy interrupt handling scheme.
Add data structures to store controller state while
we are suspended.
Use constants instead of hard coded values where appropriate.
Correct some harmless "unsigned/signed" conflicts.
aic7xxx.seq:
Only perform the SCSIBUSL fix on ULTRA2 or newer controllers.
Older controllers seem to be confused by this.
In target mode, ignore PHASEMIS during data phases.
This bit seems to be flakey on U160 controllers acting
in target mode.
aic7xxx_pci.c:
Add support for the 29160C CPCI adapter.
Add definitions for subvendor ID information
available for devices with the "9005" vendor id.
We currently use this information to determine
if a multi-function device doesn't have the second
channel hooked up on a board.
Add rudimentary power mode code so we can put the
controller into the D0 state. In the future this
will be an OSM callback so that in FreeBSD we don't
duplicate functionality provided by the PCI code.
The powerstate code was added after I'd completed
my regression tests on this code.
Only capture "left over BIOS state" if the POWRDN
setting is not set in HCNTRL.
In target mode, don't bother sending incremental
CRC data.
2000-12-20 01:11:37 +00:00
|
|
|
if (pdata->irq != NULL)
|
|
|
|
|
bus_release_resource(ahc->dev_softc,
|
|
|
|
|
pdata->irq_res_type,
|
|
|
|
|
0, pdata->irq);
|
|
|
|
|
|
|
|
|
|
if (pdata->sim_b != NULL) {
|
|
|
|
|
xpt_async(AC_LOST_DEVICE, pdata->path_b, NULL);
|
|
|
|
|
xpt_free_path(pdata->path_b);
|
|
|
|
|
xpt_bus_deregister(cam_sim_path(pdata->sim_b));
|
|
|
|
|
cam_sim_free(pdata->sim_b, /*free_devq*/TRUE);
|
|
|
|
|
}
|
|
|
|
|
if (pdata->sim != NULL) {
|
|
|
|
|
xpt_async(AC_LOST_DEVICE, pdata->path, NULL);
|
|
|
|
|
xpt_free_path(pdata->path);
|
|
|
|
|
xpt_bus_deregister(cam_sim_path(pdata->sim));
|
|
|
|
|
cam_sim_free(pdata->sim, /*free_devq*/TRUE);
|
|
|
|
|
}
|
|
|
|
|
if (pdata->eh != NULL)
|
|
|
|
|
EVENTHANDLER_DEREGISTER(shutdown_final, pdata->eh);
|
2000-09-16 20:02:28 +00:00
|
|
|
free(ahc->platform_data, M_DEVBUF);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
int
|
|
|
|
|
ahc_softc_comp(struct ahc_softc *lahc, struct ahc_softc *rahc)
|
|
|
|
|
{
|
|
|
|
|
/* We don't sort softcs under FreeBSD so report equal always */
|
|
|
|
|
return (0);
|
|
|
|
|
}
|
|
|
|
|
|
ahc_eisa.c:
ahc_pci.c:
Add detach support.
Make use of soft allocated on our behalf by newbus.
For PCI devices, disable the mapping type we aren't
using for extra protection from rogue code.
aic7xxx_93cx6.c:
aic7xxx_93cx6.h:
Sync perforce IDs.
aic7xxx_freebsd.c:
Capture the eventhandle returned by EVENTHANDER_REGISTER
so we can kill the handler off during detach.
Use AHC_* constants instead of hard coded numbers in a
few more places.
Test PPR option state when deciding to "really" negotiate
when the CAM_NEGOTIATE flag is passed in a CCB.
Make use of core "ahc_pause_and_flushwork" routine in our
timeout handler rather than re-inventing this code.
Cleanup all of our resources (really!) in ahc_platform_free().
We should be all set to become a module now.
Implement the core ahc_detach() routine shared by all of
the FreeBSD front-ends.
aic7xxx_freebsd.h:
Softc storage for our event handler.
Null implementation for the ahc_platform_flushwork() OSM
callback. FreeBSD doesn't need this as XPT callbacks are
safe from all contexts and are done directly in ahc_done().
aic7xxx_inline.h:
Implement new lazy interrupt scheme. To avoid an extra
PCI bus read, we first check our completion queues to
see if any work has completed. If work is available, we
assume that this is the source of the interrupt and skip
reading INTSTAT. Any remaining interrupt status will be
cleared by a second call to the interrupt handler should
the interrupt line still be asserted. This drops the
interrupt handler down to a single PCI bus read in the
common case of I/O completion. This is the same overhead
as in the not so distant past, but the extra sanity of
perforning a PCI read after clearing the command complete
interrupt and before running the completion queue to avoid
missing command complete interrupts added a cycle.
aic7xxx.c:
During initialization, be sure to initialize all scratch
ram locations before they are read to avoid parity errors.
In this case, we use a new function, ahc_unbusy_tcl() to
initialize the scratch ram busy target table.
Replace instances of ahc_index_busy_tcl() used to unbusy
a tcl without looking at the old value with ahc_unbusy_tcl().
Modify ahc_sent_msg so that it can find single byte messages.
ahc_sent_msg is now used to determine if a transfer negotiation
attempt resulted in a bus free.
Be more careful in filtering out only the SCSI interrupts
of interest in ahc_handle_scsiint.
Rearrange interrupt clearing code to ensure that at least
one PCI transaction occurrs after hitting CLRSINT1 and
writting to CLRINT. CLRSINT1 writes take a bit to
take effect, and the re-arrangement provides sufficient
delay to ensure the write to CLRINT is effective. The
old code might report a spurious interrupt on some "fast"
chipsets.
export ahc-update_target_msg_request for use by OSM code.
If a target does not respond to our ATN request, clear
it once we move to a non-message phase. This avoids
sending a MSG_NOOP in some later message out phase.
Use max lun and max target constants instead of
hard-coded values.
Use softc storage built into our device_t under FreeBSD.
Fix a bug in ahc_free() that caused us to delete
resources that were not allocated.
Clean up any tstate/lstate info in ahc_free().
Clear the powerdown state in ahc_reset() so that
registers can be accessed.
Add a preliminary function for pausing the chip and
processing any posted work.
Add a preliminary suspend and resume functions.
aic7xxx.h:
Limit the number of supported luns to 64. We don't
support information unit transfers, so this is the
maximum that makes sense for these chips.
Add a new flag AHC_ALL_INTERRUPTS that forces the
processing of all interrupt state in a single invokation
of ahc_intr(). When the flag is not set, we use the
lazy interrupt handling scheme.
Add data structures to store controller state while
we are suspended.
Use constants instead of hard coded values where appropriate.
Correct some harmless "unsigned/signed" conflicts.
aic7xxx.seq:
Only perform the SCSIBUSL fix on ULTRA2 or newer controllers.
Older controllers seem to be confused by this.
In target mode, ignore PHASEMIS during data phases.
This bit seems to be flakey on U160 controllers acting
in target mode.
aic7xxx_pci.c:
Add support for the 29160C CPCI adapter.
Add definitions for subvendor ID information
available for devices with the "9005" vendor id.
We currently use this information to determine
if a multi-function device doesn't have the second
channel hooked up on a board.
Add rudimentary power mode code so we can put the
controller into the D0 state. In the future this
will be an OSM callback so that in FreeBSD we don't
duplicate functionality provided by the PCI code.
The powerstate code was added after I'd completed
my regression tests on this code.
Only capture "left over BIOS state" if the POWRDN
setting is not set in HCNTRL.
In target mode, don't bother sending incremental
CRC data.
2000-12-20 01:11:37 +00:00
|
|
|
int
|
|
|
|
|
ahc_detach(device_t dev)
|
|
|
|
|
{
|
|
|
|
|
struct ahc_softc *ahc;
|
2002-08-31 06:43:55 +00:00
|
|
|
u_long l;
|
ahc_eisa.c:
ahc_pci.c:
Add detach support.
Make use of soft allocated on our behalf by newbus.
For PCI devices, disable the mapping type we aren't
using for extra protection from rogue code.
aic7xxx_93cx6.c:
aic7xxx_93cx6.h:
Sync perforce IDs.
aic7xxx_freebsd.c:
Capture the eventhandle returned by EVENTHANDER_REGISTER
so we can kill the handler off during detach.
Use AHC_* constants instead of hard coded numbers in a
few more places.
Test PPR option state when deciding to "really" negotiate
when the CAM_NEGOTIATE flag is passed in a CCB.
Make use of core "ahc_pause_and_flushwork" routine in our
timeout handler rather than re-inventing this code.
Cleanup all of our resources (really!) in ahc_platform_free().
We should be all set to become a module now.
Implement the core ahc_detach() routine shared by all of
the FreeBSD front-ends.
aic7xxx_freebsd.h:
Softc storage for our event handler.
Null implementation for the ahc_platform_flushwork() OSM
callback. FreeBSD doesn't need this as XPT callbacks are
safe from all contexts and are done directly in ahc_done().
aic7xxx_inline.h:
Implement new lazy interrupt scheme. To avoid an extra
PCI bus read, we first check our completion queues to
see if any work has completed. If work is available, we
assume that this is the source of the interrupt and skip
reading INTSTAT. Any remaining interrupt status will be
cleared by a second call to the interrupt handler should
the interrupt line still be asserted. This drops the
interrupt handler down to a single PCI bus read in the
common case of I/O completion. This is the same overhead
as in the not so distant past, but the extra sanity of
perforning a PCI read after clearing the command complete
interrupt and before running the completion queue to avoid
missing command complete interrupts added a cycle.
aic7xxx.c:
During initialization, be sure to initialize all scratch
ram locations before they are read to avoid parity errors.
In this case, we use a new function, ahc_unbusy_tcl() to
initialize the scratch ram busy target table.
Replace instances of ahc_index_busy_tcl() used to unbusy
a tcl without looking at the old value with ahc_unbusy_tcl().
Modify ahc_sent_msg so that it can find single byte messages.
ahc_sent_msg is now used to determine if a transfer negotiation
attempt resulted in a bus free.
Be more careful in filtering out only the SCSI interrupts
of interest in ahc_handle_scsiint.
Rearrange interrupt clearing code to ensure that at least
one PCI transaction occurrs after hitting CLRSINT1 and
writting to CLRINT. CLRSINT1 writes take a bit to
take effect, and the re-arrangement provides sufficient
delay to ensure the write to CLRINT is effective. The
old code might report a spurious interrupt on some "fast"
chipsets.
export ahc-update_target_msg_request for use by OSM code.
If a target does not respond to our ATN request, clear
it once we move to a non-message phase. This avoids
sending a MSG_NOOP in some later message out phase.
Use max lun and max target constants instead of
hard-coded values.
Use softc storage built into our device_t under FreeBSD.
Fix a bug in ahc_free() that caused us to delete
resources that were not allocated.
Clean up any tstate/lstate info in ahc_free().
Clear the powerdown state in ahc_reset() so that
registers can be accessed.
Add a preliminary function for pausing the chip and
processing any posted work.
Add a preliminary suspend and resume functions.
aic7xxx.h:
Limit the number of supported luns to 64. We don't
support information unit transfers, so this is the
maximum that makes sense for these chips.
Add a new flag AHC_ALL_INTERRUPTS that forces the
processing of all interrupt state in a single invokation
of ahc_intr(). When the flag is not set, we use the
lazy interrupt handling scheme.
Add data structures to store controller state while
we are suspended.
Use constants instead of hard coded values where appropriate.
Correct some harmless "unsigned/signed" conflicts.
aic7xxx.seq:
Only perform the SCSIBUSL fix on ULTRA2 or newer controllers.
Older controllers seem to be confused by this.
In target mode, ignore PHASEMIS during data phases.
This bit seems to be flakey on U160 controllers acting
in target mode.
aic7xxx_pci.c:
Add support for the 29160C CPCI adapter.
Add definitions for subvendor ID information
available for devices with the "9005" vendor id.
We currently use this information to determine
if a multi-function device doesn't have the second
channel hooked up on a board.
Add rudimentary power mode code so we can put the
controller into the D0 state. In the future this
will be an OSM callback so that in FreeBSD we don't
duplicate functionality provided by the PCI code.
The powerstate code was added after I'd completed
my regression tests on this code.
Only capture "left over BIOS state" if the POWRDN
setting is not set in HCNTRL.
In target mode, don't bother sending incremental
CRC data.
2000-12-20 01:11:37 +00:00
|
|
|
u_long s;
|
|
|
|
|
|
2002-08-31 06:43:55 +00:00
|
|
|
ahc_list_lock(&l);
|
ahc_eisa.c:
ahc_pci.c:
Add detach support.
Make use of soft allocated on our behalf by newbus.
For PCI devices, disable the mapping type we aren't
using for extra protection from rogue code.
aic7xxx_93cx6.c:
aic7xxx_93cx6.h:
Sync perforce IDs.
aic7xxx_freebsd.c:
Capture the eventhandle returned by EVENTHANDER_REGISTER
so we can kill the handler off during detach.
Use AHC_* constants instead of hard coded numbers in a
few more places.
Test PPR option state when deciding to "really" negotiate
when the CAM_NEGOTIATE flag is passed in a CCB.
Make use of core "ahc_pause_and_flushwork" routine in our
timeout handler rather than re-inventing this code.
Cleanup all of our resources (really!) in ahc_platform_free().
We should be all set to become a module now.
Implement the core ahc_detach() routine shared by all of
the FreeBSD front-ends.
aic7xxx_freebsd.h:
Softc storage for our event handler.
Null implementation for the ahc_platform_flushwork() OSM
callback. FreeBSD doesn't need this as XPT callbacks are
safe from all contexts and are done directly in ahc_done().
aic7xxx_inline.h:
Implement new lazy interrupt scheme. To avoid an extra
PCI bus read, we first check our completion queues to
see if any work has completed. If work is available, we
assume that this is the source of the interrupt and skip
reading INTSTAT. Any remaining interrupt status will be
cleared by a second call to the interrupt handler should
the interrupt line still be asserted. This drops the
interrupt handler down to a single PCI bus read in the
common case of I/O completion. This is the same overhead
as in the not so distant past, but the extra sanity of
perforning a PCI read after clearing the command complete
interrupt and before running the completion queue to avoid
missing command complete interrupts added a cycle.
aic7xxx.c:
During initialization, be sure to initialize all scratch
ram locations before they are read to avoid parity errors.
In this case, we use a new function, ahc_unbusy_tcl() to
initialize the scratch ram busy target table.
Replace instances of ahc_index_busy_tcl() used to unbusy
a tcl without looking at the old value with ahc_unbusy_tcl().
Modify ahc_sent_msg so that it can find single byte messages.
ahc_sent_msg is now used to determine if a transfer negotiation
attempt resulted in a bus free.
Be more careful in filtering out only the SCSI interrupts
of interest in ahc_handle_scsiint.
Rearrange interrupt clearing code to ensure that at least
one PCI transaction occurrs after hitting CLRSINT1 and
writting to CLRINT. CLRSINT1 writes take a bit to
take effect, and the re-arrangement provides sufficient
delay to ensure the write to CLRINT is effective. The
old code might report a spurious interrupt on some "fast"
chipsets.
export ahc-update_target_msg_request for use by OSM code.
If a target does not respond to our ATN request, clear
it once we move to a non-message phase. This avoids
sending a MSG_NOOP in some later message out phase.
Use max lun and max target constants instead of
hard-coded values.
Use softc storage built into our device_t under FreeBSD.
Fix a bug in ahc_free() that caused us to delete
resources that were not allocated.
Clean up any tstate/lstate info in ahc_free().
Clear the powerdown state in ahc_reset() so that
registers can be accessed.
Add a preliminary function for pausing the chip and
processing any posted work.
Add a preliminary suspend and resume functions.
aic7xxx.h:
Limit the number of supported luns to 64. We don't
support information unit transfers, so this is the
maximum that makes sense for these chips.
Add a new flag AHC_ALL_INTERRUPTS that forces the
processing of all interrupt state in a single invokation
of ahc_intr(). When the flag is not set, we use the
lazy interrupt handling scheme.
Add data structures to store controller state while
we are suspended.
Use constants instead of hard coded values where appropriate.
Correct some harmless "unsigned/signed" conflicts.
aic7xxx.seq:
Only perform the SCSIBUSL fix on ULTRA2 or newer controllers.
Older controllers seem to be confused by this.
In target mode, ignore PHASEMIS during data phases.
This bit seems to be flakey on U160 controllers acting
in target mode.
aic7xxx_pci.c:
Add support for the 29160C CPCI adapter.
Add definitions for subvendor ID information
available for devices with the "9005" vendor id.
We currently use this information to determine
if a multi-function device doesn't have the second
channel hooked up on a board.
Add rudimentary power mode code so we can put the
controller into the D0 state. In the future this
will be an OSM callback so that in FreeBSD we don't
duplicate functionality provided by the PCI code.
The powerstate code was added after I'd completed
my regression tests on this code.
Only capture "left over BIOS state" if the POWRDN
setting is not set in HCNTRL.
In target mode, don't bother sending incremental
CRC data.
2000-12-20 01:11:37 +00:00
|
|
|
device_printf(dev, "detaching device\n");
|
|
|
|
|
ahc = device_get_softc(dev);
|
2002-08-31 06:43:55 +00:00
|
|
|
ahc = ahc_find_softc(ahc);
|
|
|
|
|
if (ahc == NULL) {
|
|
|
|
|
device_printf(dev, "aic7xxx already detached\n");
|
|
|
|
|
ahc_list_unlock(&l);
|
|
|
|
|
return (ENOENT);
|
|
|
|
|
}
|
ahc_eisa.c:
ahc_pci.c:
Add detach support.
Make use of soft allocated on our behalf by newbus.
For PCI devices, disable the mapping type we aren't
using for extra protection from rogue code.
aic7xxx_93cx6.c:
aic7xxx_93cx6.h:
Sync perforce IDs.
aic7xxx_freebsd.c:
Capture the eventhandle returned by EVENTHANDER_REGISTER
so we can kill the handler off during detach.
Use AHC_* constants instead of hard coded numbers in a
few more places.
Test PPR option state when deciding to "really" negotiate
when the CAM_NEGOTIATE flag is passed in a CCB.
Make use of core "ahc_pause_and_flushwork" routine in our
timeout handler rather than re-inventing this code.
Cleanup all of our resources (really!) in ahc_platform_free().
We should be all set to become a module now.
Implement the core ahc_detach() routine shared by all of
the FreeBSD front-ends.
aic7xxx_freebsd.h:
Softc storage for our event handler.
Null implementation for the ahc_platform_flushwork() OSM
callback. FreeBSD doesn't need this as XPT callbacks are
safe from all contexts and are done directly in ahc_done().
aic7xxx_inline.h:
Implement new lazy interrupt scheme. To avoid an extra
PCI bus read, we first check our completion queues to
see if any work has completed. If work is available, we
assume that this is the source of the interrupt and skip
reading INTSTAT. Any remaining interrupt status will be
cleared by a second call to the interrupt handler should
the interrupt line still be asserted. This drops the
interrupt handler down to a single PCI bus read in the
common case of I/O completion. This is the same overhead
as in the not so distant past, but the extra sanity of
perforning a PCI read after clearing the command complete
interrupt and before running the completion queue to avoid
missing command complete interrupts added a cycle.
aic7xxx.c:
During initialization, be sure to initialize all scratch
ram locations before they are read to avoid parity errors.
In this case, we use a new function, ahc_unbusy_tcl() to
initialize the scratch ram busy target table.
Replace instances of ahc_index_busy_tcl() used to unbusy
a tcl without looking at the old value with ahc_unbusy_tcl().
Modify ahc_sent_msg so that it can find single byte messages.
ahc_sent_msg is now used to determine if a transfer negotiation
attempt resulted in a bus free.
Be more careful in filtering out only the SCSI interrupts
of interest in ahc_handle_scsiint.
Rearrange interrupt clearing code to ensure that at least
one PCI transaction occurrs after hitting CLRSINT1 and
writting to CLRINT. CLRSINT1 writes take a bit to
take effect, and the re-arrangement provides sufficient
delay to ensure the write to CLRINT is effective. The
old code might report a spurious interrupt on some "fast"
chipsets.
export ahc-update_target_msg_request for use by OSM code.
If a target does not respond to our ATN request, clear
it once we move to a non-message phase. This avoids
sending a MSG_NOOP in some later message out phase.
Use max lun and max target constants instead of
hard-coded values.
Use softc storage built into our device_t under FreeBSD.
Fix a bug in ahc_free() that caused us to delete
resources that were not allocated.
Clean up any tstate/lstate info in ahc_free().
Clear the powerdown state in ahc_reset() so that
registers can be accessed.
Add a preliminary function for pausing the chip and
processing any posted work.
Add a preliminary suspend and resume functions.
aic7xxx.h:
Limit the number of supported luns to 64. We don't
support information unit transfers, so this is the
maximum that makes sense for these chips.
Add a new flag AHC_ALL_INTERRUPTS that forces the
processing of all interrupt state in a single invokation
of ahc_intr(). When the flag is not set, we use the
lazy interrupt handling scheme.
Add data structures to store controller state while
we are suspended.
Use constants instead of hard coded values where appropriate.
Correct some harmless "unsigned/signed" conflicts.
aic7xxx.seq:
Only perform the SCSIBUSL fix on ULTRA2 or newer controllers.
Older controllers seem to be confused by this.
In target mode, ignore PHASEMIS during data phases.
This bit seems to be flakey on U160 controllers acting
in target mode.
aic7xxx_pci.c:
Add support for the 29160C CPCI adapter.
Add definitions for subvendor ID information
available for devices with the "9005" vendor id.
We currently use this information to determine
if a multi-function device doesn't have the second
channel hooked up on a board.
Add rudimentary power mode code so we can put the
controller into the D0 state. In the future this
will be an OSM callback so that in FreeBSD we don't
duplicate functionality provided by the PCI code.
The powerstate code was added after I'd completed
my regression tests on this code.
Only capture "left over BIOS state" if the POWRDN
setting is not set in HCNTRL.
In target mode, don't bother sending incremental
CRC data.
2000-12-20 01:11:37 +00:00
|
|
|
ahc_lock(ahc, &s);
|
2002-08-31 06:43:55 +00:00
|
|
|
ahc_intr_enable(ahc, FALSE);
|
ahc_eisa.c:
ahc_pci.c:
Add detach support.
Make use of soft allocated on our behalf by newbus.
For PCI devices, disable the mapping type we aren't
using for extra protection from rogue code.
aic7xxx_93cx6.c:
aic7xxx_93cx6.h:
Sync perforce IDs.
aic7xxx_freebsd.c:
Capture the eventhandle returned by EVENTHANDER_REGISTER
so we can kill the handler off during detach.
Use AHC_* constants instead of hard coded numbers in a
few more places.
Test PPR option state when deciding to "really" negotiate
when the CAM_NEGOTIATE flag is passed in a CCB.
Make use of core "ahc_pause_and_flushwork" routine in our
timeout handler rather than re-inventing this code.
Cleanup all of our resources (really!) in ahc_platform_free().
We should be all set to become a module now.
Implement the core ahc_detach() routine shared by all of
the FreeBSD front-ends.
aic7xxx_freebsd.h:
Softc storage for our event handler.
Null implementation for the ahc_platform_flushwork() OSM
callback. FreeBSD doesn't need this as XPT callbacks are
safe from all contexts and are done directly in ahc_done().
aic7xxx_inline.h:
Implement new lazy interrupt scheme. To avoid an extra
PCI bus read, we first check our completion queues to
see if any work has completed. If work is available, we
assume that this is the source of the interrupt and skip
reading INTSTAT. Any remaining interrupt status will be
cleared by a second call to the interrupt handler should
the interrupt line still be asserted. This drops the
interrupt handler down to a single PCI bus read in the
common case of I/O completion. This is the same overhead
as in the not so distant past, but the extra sanity of
perforning a PCI read after clearing the command complete
interrupt and before running the completion queue to avoid
missing command complete interrupts added a cycle.
aic7xxx.c:
During initialization, be sure to initialize all scratch
ram locations before they are read to avoid parity errors.
In this case, we use a new function, ahc_unbusy_tcl() to
initialize the scratch ram busy target table.
Replace instances of ahc_index_busy_tcl() used to unbusy
a tcl without looking at the old value with ahc_unbusy_tcl().
Modify ahc_sent_msg so that it can find single byte messages.
ahc_sent_msg is now used to determine if a transfer negotiation
attempt resulted in a bus free.
Be more careful in filtering out only the SCSI interrupts
of interest in ahc_handle_scsiint.
Rearrange interrupt clearing code to ensure that at least
one PCI transaction occurrs after hitting CLRSINT1 and
writting to CLRINT. CLRSINT1 writes take a bit to
take effect, and the re-arrangement provides sufficient
delay to ensure the write to CLRINT is effective. The
old code might report a spurious interrupt on some "fast"
chipsets.
export ahc-update_target_msg_request for use by OSM code.
If a target does not respond to our ATN request, clear
it once we move to a non-message phase. This avoids
sending a MSG_NOOP in some later message out phase.
Use max lun and max target constants instead of
hard-coded values.
Use softc storage built into our device_t under FreeBSD.
Fix a bug in ahc_free() that caused us to delete
resources that were not allocated.
Clean up any tstate/lstate info in ahc_free().
Clear the powerdown state in ahc_reset() so that
registers can be accessed.
Add a preliminary function for pausing the chip and
processing any posted work.
Add a preliminary suspend and resume functions.
aic7xxx.h:
Limit the number of supported luns to 64. We don't
support information unit transfers, so this is the
maximum that makes sense for these chips.
Add a new flag AHC_ALL_INTERRUPTS that forces the
processing of all interrupt state in a single invokation
of ahc_intr(). When the flag is not set, we use the
lazy interrupt handling scheme.
Add data structures to store controller state while
we are suspended.
Use constants instead of hard coded values where appropriate.
Correct some harmless "unsigned/signed" conflicts.
aic7xxx.seq:
Only perform the SCSIBUSL fix on ULTRA2 or newer controllers.
Older controllers seem to be confused by this.
In target mode, ignore PHASEMIS during data phases.
This bit seems to be flakey on U160 controllers acting
in target mode.
aic7xxx_pci.c:
Add support for the 29160C CPCI adapter.
Add definitions for subvendor ID information
available for devices with the "9005" vendor id.
We currently use this information to determine
if a multi-function device doesn't have the second
channel hooked up on a board.
Add rudimentary power mode code so we can put the
controller into the D0 state. In the future this
will be an OSM callback so that in FreeBSD we don't
duplicate functionality provided by the PCI code.
The powerstate code was added after I'd completed
my regression tests on this code.
Only capture "left over BIOS state" if the POWRDN
setting is not set in HCNTRL.
In target mode, don't bother sending incremental
CRC data.
2000-12-20 01:11:37 +00:00
|
|
|
bus_teardown_intr(dev, ahc->platform_data->irq, ahc->platform_data->ih);
|
|
|
|
|
ahc_unlock(ahc, &s);
|
|
|
|
|
ahc_free(ahc);
|
2002-08-31 06:43:55 +00:00
|
|
|
ahc_list_unlock(&l);
|
ahc_eisa.c:
ahc_pci.c:
Add detach support.
Make use of soft allocated on our behalf by newbus.
For PCI devices, disable the mapping type we aren't
using for extra protection from rogue code.
aic7xxx_93cx6.c:
aic7xxx_93cx6.h:
Sync perforce IDs.
aic7xxx_freebsd.c:
Capture the eventhandle returned by EVENTHANDER_REGISTER
so we can kill the handler off during detach.
Use AHC_* constants instead of hard coded numbers in a
few more places.
Test PPR option state when deciding to "really" negotiate
when the CAM_NEGOTIATE flag is passed in a CCB.
Make use of core "ahc_pause_and_flushwork" routine in our
timeout handler rather than re-inventing this code.
Cleanup all of our resources (really!) in ahc_platform_free().
We should be all set to become a module now.
Implement the core ahc_detach() routine shared by all of
the FreeBSD front-ends.
aic7xxx_freebsd.h:
Softc storage for our event handler.
Null implementation for the ahc_platform_flushwork() OSM
callback. FreeBSD doesn't need this as XPT callbacks are
safe from all contexts and are done directly in ahc_done().
aic7xxx_inline.h:
Implement new lazy interrupt scheme. To avoid an extra
PCI bus read, we first check our completion queues to
see if any work has completed. If work is available, we
assume that this is the source of the interrupt and skip
reading INTSTAT. Any remaining interrupt status will be
cleared by a second call to the interrupt handler should
the interrupt line still be asserted. This drops the
interrupt handler down to a single PCI bus read in the
common case of I/O completion. This is the same overhead
as in the not so distant past, but the extra sanity of
perforning a PCI read after clearing the command complete
interrupt and before running the completion queue to avoid
missing command complete interrupts added a cycle.
aic7xxx.c:
During initialization, be sure to initialize all scratch
ram locations before they are read to avoid parity errors.
In this case, we use a new function, ahc_unbusy_tcl() to
initialize the scratch ram busy target table.
Replace instances of ahc_index_busy_tcl() used to unbusy
a tcl without looking at the old value with ahc_unbusy_tcl().
Modify ahc_sent_msg so that it can find single byte messages.
ahc_sent_msg is now used to determine if a transfer negotiation
attempt resulted in a bus free.
Be more careful in filtering out only the SCSI interrupts
of interest in ahc_handle_scsiint.
Rearrange interrupt clearing code to ensure that at least
one PCI transaction occurrs after hitting CLRSINT1 and
writting to CLRINT. CLRSINT1 writes take a bit to
take effect, and the re-arrangement provides sufficient
delay to ensure the write to CLRINT is effective. The
old code might report a spurious interrupt on some "fast"
chipsets.
export ahc-update_target_msg_request for use by OSM code.
If a target does not respond to our ATN request, clear
it once we move to a non-message phase. This avoids
sending a MSG_NOOP in some later message out phase.
Use max lun and max target constants instead of
hard-coded values.
Use softc storage built into our device_t under FreeBSD.
Fix a bug in ahc_free() that caused us to delete
resources that were not allocated.
Clean up any tstate/lstate info in ahc_free().
Clear the powerdown state in ahc_reset() so that
registers can be accessed.
Add a preliminary function for pausing the chip and
processing any posted work.
Add a preliminary suspend and resume functions.
aic7xxx.h:
Limit the number of supported luns to 64. We don't
support information unit transfers, so this is the
maximum that makes sense for these chips.
Add a new flag AHC_ALL_INTERRUPTS that forces the
processing of all interrupt state in a single invokation
of ahc_intr(). When the flag is not set, we use the
lazy interrupt handling scheme.
Add data structures to store controller state while
we are suspended.
Use constants instead of hard coded values where appropriate.
Correct some harmless "unsigned/signed" conflicts.
aic7xxx.seq:
Only perform the SCSIBUSL fix on ULTRA2 or newer controllers.
Older controllers seem to be confused by this.
In target mode, ignore PHASEMIS during data phases.
This bit seems to be flakey on U160 controllers acting
in target mode.
aic7xxx_pci.c:
Add support for the 29160C CPCI adapter.
Add definitions for subvendor ID information
available for devices with the "9005" vendor id.
We currently use this information to determine
if a multi-function device doesn't have the second
channel hooked up on a board.
Add rudimentary power mode code so we can put the
controller into the D0 state. In the future this
will be an OSM callback so that in FreeBSD we don't
duplicate functionality provided by the PCI code.
The powerstate code was added after I'd completed
my regression tests on this code.
Only capture "left over BIOS state" if the POWRDN
setting is not set in HCNTRL.
In target mode, don't bother sending incremental
CRC data.
2000-12-20 01:11:37 +00:00
|
|
|
return (0);
|
|
|
|
|
}
|
|
|
|
|
|
2000-09-16 20:02:28 +00:00
|
|
|
#if UNUSED
|
|
|
|
|
static void
|
|
|
|
|
ahc_dump_targcmd(struct target_cmd *cmd)
|
|
|
|
|
{
|
|
|
|
|
uint8_t *byte;
|
|
|
|
|
uint8_t *last_byte;
|
|
|
|
|
int i;
|
|
|
|
|
|
|
|
|
|
byte = &cmd->initiator_channel;
|
|
|
|
|
/* Debugging info for received commands */
|
|
|
|
|
last_byte = &cmd[1].initiator_channel;
|
|
|
|
|
|
|
|
|
|
i = 0;
|
|
|
|
|
while (byte < last_byte) {
|
|
|
|
|
if (i == 0)
|
|
|
|
|
printf("\t");
|
|
|
|
|
printf("%#x", *byte++);
|
|
|
|
|
i++;
|
|
|
|
|
if (i == 8) {
|
|
|
|
|
printf("\n");
|
|
|
|
|
i = 0;
|
|
|
|
|
} else {
|
|
|
|
|
printf(", ");
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
#endif
|
2001-05-15 19:41:12 +00:00
|
|
|
|
|
|
|
|
static int
|
|
|
|
|
ahc_modevent(module_t mod, int type, void *data)
|
|
|
|
|
{
|
|
|
|
|
/* XXX Deal with busy status on unload. */
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static moduledata_t ahc_mod = {
|
|
|
|
|
"ahc",
|
|
|
|
|
ahc_modevent,
|
|
|
|
|
NULL
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
DECLARE_MODULE(ahc, ahc_mod, SI_SUB_DRIVERS, SI_ORDER_MIDDLE);
|
|
|
|
|
MODULE_DEPEND(ahc, cam, 1, 1, 1);
|
|
|
|
|
MODULE_VERSION(ahc, 1);
|
