[5240] in linux-scsi channel archive
last_reset == 0 means nothing in scsi code
daemon@ATHENA.MIT.EDU (Andrea Arcangeli)
Mon Nov 30 06:24:47 1998
Date: Mon, 30 Nov 1998 11:48:59 +0100 (CET)
From: Andrea Arcangeli <andrea@e-mind.com>
To: Gerard Roudier <groudier@club-internet.fr>
cc: linux-kernel@vger.rutgers.edu, linux-scsi@vger.rutgers.edu
In-Reply-To: <Pine.LNX.3.96.981130001405.26425F-100000@dragon.bogus>
I developed a very very experimental patch. I have no idea if it can be
right. I can try it only with my ppa drive and works but I never had a
scsi reset with ppa... Soon I' ll hack ppa forcing it to cause a scsi
reset to see the code in action.
Comments?
BTW, the patch is only a snapshot, it will not apply cleanly to 2.1.130.
Index: drivers/scsi//hosts.c
===================================================================
RCS file: /var/cvs/linux/drivers/scsi/hosts.c,v
retrieving revision 1.1.1.1.2.3
diff -u -r1.1.1.1.2.3 hosts.c
--- hosts.c 1998/11/30 08:47:56 1.1.1.1.2.3
+++ hosts.c 1998/11/30 10:30:05
@@ -581,7 +581,8 @@
next_scsi_host++;
retval->host_queue = NULL;
retval->host_wait = NULL;
- retval->last_reset = jiffies - 3600*HZ; /* set a bit ago to be safe */
+ retval->resetting = 0;
+ /* retval->last_reset don' t need to be set because resetting is 0 -arca */
retval->irq = 0;
retval->dma_channel = 0xff;
Index: drivers/scsi//hosts.h
===================================================================
RCS file: /var/cvs/linux/drivers/scsi/hosts.h,v
retrieving revision 1.1.1.1
diff -u -r1.1.1.1 hosts.h
--- hosts.h 1998/11/19 23:02:39 1.1.1.1
+++ hosts.h 1998/11/30 10:27:41
@@ -320,6 +320,7 @@
/* public: */
unsigned short extra_bytes;
unsigned short host_no; /* Used for IOCTL_GET_IDLUN, /proc/scsi et al. */
+ int resetting; /* if set, it means that last_reset is a valid value */
unsigned long last_reset;
Index: drivers/scsi//scsi.c
===================================================================
RCS file: /var/cvs/linux/drivers/scsi/scsi.c,v
retrieving revision 1.1.1.1.2.1
diff -u -r1.1.1.1.2.1 scsi.c
--- scsi.c 1998/11/19 23:15:31 1.1.1.1.2.1
+++ scsi.c 1998/11/30 10:41:05
@@ -1297,7 +1297,7 @@
*/
timeout = host->last_reset + MIN_RESET_DELAY;
- if (time_before(jiffies, timeout)) {
+ if (host->resetting && time_before(jiffies, timeout)) {
int ticks_remaining = timeout - jiffies;
/*
* NOTE: This may be executed from within an interrupt
@@ -1310,7 +1310,7 @@
*/
spin_unlock_irq(&io_request_lock);
while (--ticks_remaining >= 0) mdelay(1+999/HZ);
- host->last_reset = jiffies - MIN_RESET_DELAY;
+ host->resetting = 0;
spin_lock_irq(&io_request_lock);
}
Index: drivers/scsi//scsi_obsolete.c
===================================================================
RCS file: /var/cvs/linux/drivers/scsi/scsi_obsolete.c,v
retrieving revision 1.1.1.1.2.2
diff -u -r1.1.1.1.2.2 scsi_obsolete.c
--- scsi_obsolete.c 1998/11/30 00:39:26 1.1.1.1.2.2
+++ scsi_obsolete.c 1998/11/30 10:41:43
@@ -607,8 +607,9 @@
if ((++SCpnt->retries) < SCpnt->allowed)
{
if ((SCpnt->retries >= (SCpnt->allowed >> 1))
- && time_after(jiffies, SCpnt->host->last_reset
- + MIN_RESET_PERIOD)
+ && (SCpnt->host->resetting &&
+ time_after(jiffies, SCpnt->host->last_reset
+ + MIN_RESET_PERIOD))
&& !(SCpnt->flags & WAS_RESET))
{
printk("scsi%d channel %d : resetting for second half of retries.\n",
@@ -934,6 +935,12 @@
}
host->last_reset = jiffies;
+ host->resetting = 1;
+ /*
+ * I suppose that the host reset callback will not play
+ * with the resetting field. We have just set the resetting
+ * flag here. -arca
+ */
temp = host->hostt->reset(SCpnt, reset_flags);
/*
This test allows the driver to introduce an additional bus
@@ -952,7 +959,13 @@
{
if (!host->block) host->host_busy++;
host->last_reset = jiffies;
+ host->resetting = 1;
SCpnt->flags |= (WAS_RESET | IS_RESETTING);
+ /*
+ * I suppose that the host reset callback will not play
+ * with the resetting field. We have just set the resetting
+ * flag here. -arca
+ */
temp = host->hostt->reset(SCpnt, reset_flags);
if (time_before(host->last_reset, jiffies) ||
(time_after(host->last_reset, jiffies + 20 * HZ)))
Can be right to consider the resetting expired when we send the first
normal scsi command to the lowlevel driver?
Andrea Arcangeli
-
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@vger.rutgers.edu
