aic94xx breaks with SATA drives that have medium errors

Begin forwarded message:

Date: Mon, 27 Nov 2006 23:28:18 +0200
From: Dan Aloni <[email protected]>
To: [email protected]
Subject: aic94xx breaks with SATA drives that have medium errors

Hello,

I'm currently testing the aic94xx driver from the latest git version of Linux 2.6.19-rc generic x86_64 port merged with the aic94xx-sas-2.6 git, on a Supermicro X7DB3 board.

It seems that the driver breaks badly when my SATA drives have medium errors.

I deliberatly cause medium errors in order to test the error handling of the driver coupled with the contoller.

Everything works okay until I perform a read I/O to the media-error-causing location. Immediately I get:

aic94xx: escb_tasklet_complete: phy2: REQ_TASK_ABORT

But the I/O only returns to the SCSI layer after its full designated timeout, instead of returning quickly with MEDIUM_ERROR.

After that particular I/O fails, every I/O to the driver will immediately return as aborted. Unloading and loading the driver reverses the problem but may crash the kernel not long after printing this:

Nov 28 02:13:58 pro210 kernel: aic94xx: Uh-oh! Pending is not empty!
Nov 28 02:13:58 pro210 kernel: aic94xx: freeing from pending
Nov 28 02:13:58 pro210 kernel: aic94xx: Adaptec aic94xx SAS/SATA driver version 1.0.2 unloaded
Nov 28 02:14:01 pro210 kernel:
Nov 28 02:14:01 pro210 kernel: Call Trace:
Nov 28 02:14:01 pro210 kernel:  [<ffffffff80271313>] dump_trace+0xb3/0x450
Nov 28 02:14:01 pro210 kernel:  [<ffffffff802716f3>] show_trace+0x43/0x60
Nov 28 02:14:01 pro210 kernel:  [<ffffffff80271725>] dump_stack+0x15/0x20
Nov 28 02:14:01 pro210 kernel:  [<ffffffff802c5327>] kmem_cache_destroy+0xa7/0x110
Nov 28 02:14:01 pro210 kernel:  [<ffffffff8804f500>] :libsas:sas_class_exit+0x10/0x12
Nov 28 02:14:01 pro210 kernel:  [<ffffffff802a96a0>] sys_delete_module+0x220/0x280
Nov 28 02:14:01 pro210 kernel:  [<ffffffff8026411e>] system_call+0x7e/0x83
Nov 28 02:14:01 pro210 kernel:  [<00002b82def05959>]
Nov 28 02:14:01 pro210 kernel:
Nov 28 02:14:05 pro210 kernel: kmem_cache_create: duplicate cache sas_task
Nov 28 02:14:05 pro210 kernel:
Nov 28 02:14:05 pro210 kernel: Call Trace:
Nov 28 02:14:05 pro210 kernel:  [<ffffffff80271313>] dump_trace+0xb3/0x450
Nov 28 02:14:05 pro210 kernel:  [<ffffffff802716f3>] show_trace+0x43/0x60
Nov 28 02:14:05 pro210 kernel:  [<ffffffff80271725>] dump_stack+0x15/0x20
Nov 28 02:14:05 pro210 kernel:  [<ffffffff8023acf8>] kmem_cache_create+0x578/0x5c0
Nov 28 02:14:05 pro210 kernel:  [<ffffffff8803d022>] :libsas:sas_class_init+0x22/0x34
Nov 28 02:14:05 pro210 kernel:  [<ffffffff802ab7d6>] sys_init_module+0x1956/0x1ba0
Nov 28 02:14:05 pro210 kernel:  [<ffffffff8026411e>] system_call+0x7e/0x83
Nov 28 02:14:05 pro210 kernel:  [<00002aaabe27ea4c>]
Nov 28 02:14:05 pro210 kernel:

- Dan
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

Interesting that you get REQ_TASK_ABORT for a media error...

Yep. The abort function doesn't know how to tell libata to abort the command. I suppose the "proper" thing to do would be to modify sas_ata_task_done to check if the SAS_TASK_ABORTED or SAS_TASK_INITIATOR_ABORTED flags are set and send some sort of ATA error code back that would cause a retry. Though, I don't see why the sequencer sends back REQ_TASK_ABORT--presumably the drive generates some media error data that could be fed to libata.

Yep. Side effect of above. I'll send you a patch later today when I get this sorted out. In any case, thank you for testing out the driver! :)

--D