On Wed, 15 Mar 2006, Michael Stuebiger wrote:
> We are using kernel version 2.6.15.6 and we noticed some odd behavior of the qla2xxx driver.
> Sometimes the machines boot without incident and at other times the following error occurs:
>
> klogd 1.4.1#17, log source = /proc/kmsg started.
> Inspecting /boot/System.map-2.6.15.6
> Loaded 38338 symbols from /boot/System.map-2.6.15.6.
> Symbols match kernel version 2.6.15.
> No module symbols loaded - kernel modules not enabled.
> tainted 2.6.15.6 #1
> [ 62.255008] RIP: 0010:[__delay+8/16] <ffffffff80270489>{__delay+8}
> [ 62.255150] RSP: 0000:ffff81007ffdbab0 EFLAGS: 00000283
> [ 62.255344] RAX: 00000000e06fa013 RBX: ffff810001ec0488 RCX: 00000000e06f9f0f
> [ 62.255460] RDX: 000000000000001c RSI: 0000000000000246 RDI: 0000000000004a38
> [ 62.255576] RBP: 00000000fffc3e81 R08: 0000000000000000 R09: 0000000000000000
> [ 62.255692] R10: 0000000000000000 R11: 0000000000000000 R12: ffff81007ffdbb58
> [ 62.255809] R13: 0000000000000000 R14: 0000000000800000 R15: ffffc20000004000
> [ 62.255926] FS: 0000000000000000(0000) GS:ffffffff807d0800(0000) knlGS:0000000000000000
> [ 62.256077] CS: 0010 DS: 0018 ES: 0018 CR0: 000000008005003b
> [ 62.256187] CR2: 0000000000000000 CR3: 0000000000101000 CR4: 00000000000006e0
> [ 62.256303]
> [ 62.256304] Call Trace:<ffffffff80319a89>{qla2x00_mailbox_command+554} <ffffffff80319b60>{qla2x00_mailbox_command+769}
> [ 62.256693] <ffffffff8031a908>{qla2x00_login_fabric+110} <ffffffff8031a542>{qla2x00_get_port_database+381}
> [ 62.257072] <ffffffff8031866a>{qla2x00_fabric_login+62} <ffffffff80318300>{qla2x00_find_all_fabric_devs+1126}
This looks like a soft-lockup backtrace occuring during mbx command
completion polling. This issue was addressed post-2.6.15:
[SCSI] qla2xxx: Use msleep() as delay during ISP polling.
http://gitlnx/cgi-bin/gitweb.cgi?p=avasquez/qla2xxx-upstream.git;a=commitdiff;h=03ab2eabed74708c0c74163723007e8e72244219;hp=e038a1be22bef90795c5239d1563609b89c3aa18
> The machines are Dual AMD Opteron servers with a "QLogic Corp. QLA2312 Fibre Channel Adapter (rev 02)".
> Since the machines in Question are fileservers we naturally are eager to know what the cause of this is.
>
> You?ll find the config of the Kernel in question attached to this email.
Regards,
Andrew Vasquez
-
: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
