Re: [PATCH] instrument ide-scsi in 2.5.68

[Douglas Gilbert <dougg@torque.net>]

Randy.Dunlap wrote:
> | Hi Doug-
> 
> | On Wed, 07 May 2003 23:21:43 +1000 Douglas Gilbert <dougg@torque.net> wrote:
> | 
> | | Randy.Dunlap wrote:
> | | > On Tue, 06 May 2003 18:38:39 +1000 Douglas Gilbert <dougg@torque.net> wrote:
> | | > 
> | | > | > Are there any docs on SCSI logging?
> | | 
> | | Do your failures only occur in an SMP environment?
> | 
> | I'll try 2.5.69 plus your patch ver. 2.
> 
> 2.5.69 + your ide-scsi ver. 2 patch, booted with "nosmp":
> 
>   hangs/locks up during boot sequence, immediately after this:
> 
> end_request: I/O error, dev hdc, sector 0
> scsi0 : SCSI host adapter emulation for IDE ATAPI devices
> ide-scsi: hdd: que 1, cmd = [ 12 0 0 0 24 0 ]
> ide-scsi: abort called for cmd serial #1
> ide-scsi: abort ignored
> ide-scsi: reset called for 1
> ide-scsi: device reset partially ignored
> ide-scsi: hdd: que 1, cmd = [ 0 0 0 0 0 0 ]
> ide-scsi: abort called for cmd serial #1
> ide-scsi: abort ignored
> scsi: Device offlined - not ready after error recovery: host 0 channel 0 id 0 lun 0
> 
> 
> same kernel without "nosmp", so it's dual-proc P4:
> 
>   I ran cdrecord, got to an error, cdrecord had a segfault.
> 
> Not much help I'm sorry to say.
> 
> Complete logs are here:
>   http://www.xenotime.net/linux/capt-2569-nosmp-locked.txt

Randy,
The "nosmp" failed on the very first SCSI command sent
to the device (36 byte INQUIRY during scan). Looks
like the 'rq->flags' were cleared (corruption) and
the end_request was routed through ide_end_request().

Good to see the scsi error handling took the device
offline. Did the device lock up or the whole machine?

> and
>   http://www.xenotime.net/linux/capt-2569-isc-dual-segv.txt

Got a bit further than last time but failed on the same
command, again with corrupted fields.

Perhaps you could send me your .config file so I can
set up my IDE subsystem the same way as yours.

Doug Gilbert