scsi: killing requests for dead queue

scsi: killing requests for dead queue

[Luck, Tony]

First saw these messages on next-20111102 (but my previous build from linux-next
was next-20111017 ... so a big gap).  Still there today in next-20111103. Not seeing
these messages from Linus 3.2 merge window builds.

lspci says:
----------
20:01.0 SCSI storage controller: LSI Logic / Symbios Logic 53c1030 PCI-X Fusion-MPT Dual Ultra320 SCSI (rev 08)
20:01.1 SCSI storage controller: LSI Logic / Symbios Logic 53c1030 PCI-X Fusion-MPT Dual Ultra320 SCSI (rev 08)

extract from dmesg:
------------------
Fusion MPT base driver 3.04.20
Copyright (c) 1999-2008 LSI Corporation
Fusion MPT SPI Host driver 3.04.20
GSI 27 (level, low) -> CPU 0 (0x0000) vector 54
mptspi 0000:20:01.0: PCI INT A -> GSI 27 (level, low) -> IRQ 54
mptbase: ioc0: Initiating bringup
ioc0: LSI53C1030 C0: Capabilities={Initiator,Target}
scsi0 : ioc0: LSI53C1030 C0, FwRev=01032341h, Ports=1, MaxQ=255, IRQ=54
scsi 0:0:0:0: Direct-Access     HP 36.4G MAU3036NC        HPC2 PQ: 0 ANSI: 3
scsi target0:0:0: Beginning Domain Validation
scsi target0:0:0: Ending Domain Validation
scsi target0:0:0: FAST-160 WIDE SCSI 320.0 MB/s DT IU QAS RTI WRFLOW PCOMP (6.25 ns, offset 127)
sd 0:0:0:0: Attached scsi generic sg0 type 0
sd 0:0:0:0: [sda] 71132960 512-byte logical blocks: (36.4 GB/33.9 GiB)
scsi 0:0:1:0: Direct-Access     HP 36.4G MAU3036NC        HPC2 PQ: 0 ANSI: 3
scsi target0:0:1: Beginning Domain Validation
sd 0:0:0:0: [sda] Write Protect is off
sd 0:0:0:0: [sda] Mode Sense: cf 00 10 08
sd 0:0:0:0: [sda] Write cache: disabled, read cache: enabled, supports DPO and FUA
scsi target0:0:1: Ending Domain Validation
scsi target0:0:1: FAST-160 WIDE SCSI 320.0 MB/s DT IU QAS RTI WRFLOW PCOMP (6.25 ns, offset 127)
sd 0:0:1:0: Attached scsi generic sg1 type 0
sd 0:0:1:0: [sdb] 71132960 512-byte logical blocks: (36.4 GB/33.9 GiB)
 sda: sda1
sd 0:0:1:0: [sdb] Write Protect is off
sd 0:0:1:0: [sdb] Mode Sense: cf 00 10 08
scsi: killing requests for dead queue
scsi: killing requests for dead queue
sd 0:0:1:0: [sdb] Write cache: disabled, read cache: enabled, supports DPO and FUA
scsi: killing requests for dead queue
scsi: killing requests for dead queue
scsi: killing requests for dead queue
scsi: killing requests for dead queue
sd 0:0:0:0: [sda] Attached SCSI disk
scsi: killing requests for dead queue
scsi: killing requests for dead queue
scsi: killing requests for dead queue
scsi: killing requests for dead queue

bunch more "scsi: killing requests for dead queue" messages. But system
does eventually come up (and the messages stop).

Re: scsi: killing requests for dead queue

[Bart Van Assche]

On Thursday 03 November 2011 22:23:24 you wrote:
> First saw these messages on next-20111102 (but my previous build from linux-next
> was next-20111017 ... so a big gap).  Still there today in next-20111103. Not seeing
> these messages from Linus 3.2 merge window builds.
> 
> scsi: killing requests for dead queue

Does the patch below help ?

[PATCH] [SCSI] Silence queue cleanup code

Commit 3308511 causes the SCSI queue cleanup code to be invoked
for every killed queue. Since the SCSI scanning code can create
and destroy SCSI queues repeatedly while probing LUNs, the
following message can be printed several times during boot:

    scsi: killing requests for dead queue

This confuses users, so change the severity of that message to
KERN_DEBUG.

See also http://marc.info/?l=linux-scsi&m=132035543616061.

Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Reported-by: Tony Luck <tony.luck@intel.com>
Cc: <stable@kernel.org>
Cc: James Bottomley <JBottomley@Parallels.com>
---
 drivers/scsi/scsi_lib.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/drivers/scsi/scsi_lib.c b/drivers/scsi/scsi_lib.c
index b4d43ae..79ce802 100644
--- a/drivers/scsi/scsi_lib.c
+++ b/drivers/scsi/scsi_lib.c
@@ -1489,7 +1489,7 @@ static void scsi_request_fn(struct request_queue *q)
 	struct request *req;
 
 	if (!sdev) {
-		printk("scsi: killing requests for dead queue\n");
+		pr_debug("scsi: killing requests for dead queue\n");
 		while ((req = blk_peek_request(q)) != NULL)
 			scsi_kill_request(req, q);
 		return;
-- 
1.7.3.4

RE: scsi: killing requests for dead queue

[Luck, Tony]

> Commit 3308511 causes the SCSI queue cleanup code to be invoked
> for every killed queue. Since the SCSI scanning code can create
> and destroy SCSI queues repeatedly while probing LUNs, the
> following message can be printed several times during boot:
>
>    scsi: killing requests for dead queue

Does this message show up in some other situation (like when
a device has failed?). Would you actually want to see these
messages then?  If so, demoting them to debug sounds like
the wrong thing to do.

Would it be better to clean things up when destroying queues
during LUN probes?

-Tony

[apologies if these questions are naïve - scsi is all a black
box to me]

Re: scsi: killing requests for dead queue

[Bart Van Assche]

On Fri, Nov 4, 2011 at 7:07 PM, Luck, Tony <tony.luck@intel.com> wrote:
> > Commit 3308511 causes the SCSI queue cleanup code to be invoked
> > for every killed queue. Since the SCSI scanning code can create
> > and destroy SCSI queues repeatedly while probing LUNs, the
> > following message can be printed several times during boot:
> >
> >    scsi: killing requests for dead queue
>
> Does this message show up in some other situation (like when
> a device has failed?). Would you actually want to see these
> messages then?  If so, demoting them to debug sounds like
> the wrong thing to do.

I might be wrong but I doubt that anyone is interested in these
messages. It's not even mentioned in these messages which queue is
being cleaned up.

Bart.
--
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

RE: scsi: killing requests for dead queue

[Luck, Tony]

> I might be wrong but I doubt that anyone is interested in these
> messages. It's not even mentioned in these messages which queue is
> being cleaned up.

Your patch does indeed get rid of the 56 messages from my boot log.

-Tony