RE: Frequent RESETs with 2.6.16 megaraid_sas

["Patro, Sumant" <Sumant.Patro@lsi.com>]

Hello Andrew,

	The patch that you have mentioned is a critical bug fix and must
be applied.
	
Thanks,

Sumant
 

-----Original Message-----
From: linux-kernel-owner@vger.kernel.org
[mailto:linux-kernel-owner@vger.kernel.org] On Behalf Of Andrew Moise
Sent: Sunday, October 15, 2006 2:12 PM
To: linux-kernel@vger.kernel.org
Subject: Frequent RESETs with 2.6.16 megaraid_sas

  I'm running a 2.6.16 kernel (Debian 2.6.16-18) on a Dell PE 2950, and
I used to get frequent warnings like this under heavy write load:

Oct  2 14:36:01 localhost kernel: sd 0:2:1:0: megasas: RESET -55455
cmd=2a Oct  2 14:36:01 localhost kernel: megasas: reset successful Oct
2 14:36:31 localhost kernel: sd 0:2:1:0: megasas: RESET -70369 cmd=2a
Oct  2 14:36:31 localhost kernel: megasas: reset successful Oct  2
14:37:02 localhost kernel: sd 0:2:1:0: megasas: RESET -83487 cmd=2a Oct
2 14:37:02 localhost kernel: megasas: reset successful Oct  2 14:37:32
localhost kernel: sd 0:2:1:0: megasas: RESET -95079 cmd=2a Oct  2
14:37:32 localhost kernel: megasas: reset successful Oct  2 14:38:02
localhost kernel: sd 0:2:1:0: megasas: RESET -105361 cmd=2a Oct  2
14:38:02 localhost kernel: megasas: reset successful Oct  2 14:38:33
localhost kernel: sd 0:2:1:0: megasas: RESET -115613 cmd=2a Oct  2
14:38:33 localhost kernel: megasas: reset successful Oct  2 14:38:33
localhost kernel: sd 0:2:1:0: SCSI error: return code= 0x6000000 Oct  2
14:38:33 localhost kernel: end_request: I/O error, 
dev sdb, sector 2927091007 Oct  2 14:38:33 localhost kernel: Buffer I/O
error on device sdb1, logical block 731772736 Oct  2 14:38:33 localhost
kernel: lost page write due to I/O error on sdb1 Oct  2 14:39:03
localhost kernel: sd 0:2:1:0: megasas: RESET -125667 cmd=2a Oct  2
14:39:03 localhost kernel: megasas: reset successful Oct  2 14:39:33
localhost kernel: sd 0:2:1:0: megasas: RESET -135588 cmd=2a Oct  2
14:39:33 localhost kernel: megasas: [ 0]waiting for 1 commands to
complete Oct  2 14:39:34 localhost kernel: megasas: reset successful

  I saw in some posting that someone with this problem had worked around
it by reducing BLKDEV_MAX_RQ to 8.  I did that, and it's been working
well for me for some weeks, but I happened to notice the following in a
recent megaraid patch:

--- linux-2.6.16-2.6.16/drivers/scsi/megaraid/megaraid_sas.c
2006-03-20 00:53:29.000000000 -0500
+++ linux-2.6.19-rc2drivers/scsi/megaraid/megaraid_sas.c
2006-10-13 12:25:04.000000000 -0400

@@ -1716,6 +1823,12 @@
         * Get various operational parameters from status register
         */
        instance->max_fw_cmds =
instance->instancet->read_fw_status_reg(reg_set) & 0x00FFFF;
+       /*
+        * Reduce the max supported cmds by 1. This is to ensure that
the
+        * reply_q_sz (1 more than the max cmd that driver may send)
+        * does not exceed max cmds that the FW can support
+        */
+       instance->max_fw_cmds = instance->max_fw_cmds-1;
        instance->max_num_sge =
(instance->instancet->read_fw_status_reg(reg_set) & 0xFF0000) >>
                                        0x10;
        /*

  I'd prefer to stick with an older, distro-supported kernel, and I'd
also like to avoid importing big recent changes to the megaraid driver
into my production server :-).  However, it looks to me like this one
line might be the right fix for the problem that the BLKDEV_MAX_RQ hack
is a workaround for.  I'm considering applying it during downtime in the
near future and seeing how it works out.
  Would anyone knowledgeable care to comment on the wisdom of this
approach?  I won't hold you responsible if my filesystem explodes :-).
 I'm just not familiar enough with storage drivers to know if what I'm
considering doing makes sense.
  Cheers.  Please CC any replies to me, as I'm not on the list.  Thanks.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel"
in the body of a message to majordomo@vger.kernel.org More majordomo
info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/