2004-04-15 22:37:04

by Mukker, Atul

[permalink] [raw]
Subject: RE: assertion failure with new megaraid beta driver leads to sche duling failure

Paul,

Random deletion has issues with this version of the driver. Sreenivas will
release the next version tomorrow morning with the fix.

Thanks
-Atul

> -----Original Message-----
> From: Paul Wagland [mailto:[email protected]]
> Sent: Thursday, April 15, 2004 5:57 PM
> To: Linux SCSI mailing list; Linux kernel mailing list
> Cc: Atul Mukker
> Subject: assertion failure with new megaraid beta driver leads to
> scheduling failure
>
>
> Hi all,
>
> Well, seems I have a penchance for causing trouble ;-)
>
> I was playing around with the new beta driver, and was trying
> to remove
> some logical drives using the dellmgr utility.
>
> Well, something between the driver and the BIOS got unhappy,
> and all of
> a sudden I got about 20 of the following lines:
>
> assertion failed:(spin_is_locked(adapter->host_lock)), file:
> drivers/scsi/megaraid.c, line: 3061:megaraid_abort_handler
>
> followed by (copied by hand... maybe I should set up a serial
> console :-):
>
> bad: scheduling while atomic!
> Call Trace:
> [<c011cd5f>] schedule+0x58f/0x5a0
> [<c01209a7>] __call_console_drivers+0x57/0x60
> [<c0108126>] __down+0x86/0x100
> [<c011cdc0>] default_wake_function+0x0/0x20
> [<f8832790>] megaraid_reset_handler+0x0/0xc0 [megaraid]
> [<c010833c>] __down_failed+0x8/0xc
> [<f8833fee>] .text.lock.megaraid_clib+0x5/0x17 [megaraid]
> [<f88327be>] megaraid_reset_handler+0x2e/0xc0 [megaraid]
>
> this was called from:
> scsi_try_bus_device_reset
> scsi_eh_bus_device_reset
> scsi_eh_ready_devs
> scsi_unjam_host
> default_wake_function
> scsi_error_handler
> scsi_error_handler
> kernel_thread_helper
>
>
>
> Thoughts and/or suggestions?
>
> Cheers,
> Paul
>


2004-04-16 05:45:27

by Paul Wagland

[permalink] [raw]
Subject: Re: assertion failure with new megaraid beta driver leads to sche duling failure

Atul,

On Apr 16, 2004, at 0:36, Mukker, Atul wrote:

> Paul,
>
> Random deletion has issues with this version of the driver. Sreenivas
> will
> release the next version tomorrow morning with the fix.

As I understand it you want this new driver to replace the one in the
kernel, correct?

If so, would it be possible to rename the output file to megaraid.o,
instead of megaraid_nm.o. The reason for this is that then the various
vendor tools that produce initrd images would then most likely not need
to be changed.

If not, then a new name is good ;-)

Cheers,
Paul


Attachments:
PGP.sig (186.00 B)
This is a digitally signed message part