Re: Linux 2.6.10-rc2 OOPS on boot with 3ware + reiserfs

[Jens Axboe <axboe@suse.de>]

On Thu, Nov 18 2004, James Bottomley wrote:
> On Thu, 2004-11-18 at 13:10, Jens Axboe wrote:
> > It is a double requeue. The SCSI path looks really messy (and buggy
> > there). What happens is that the host queuecommand sets DID_ERROR and
> > calls scsi_done() on the command, which may decide the commands need
> > retrying and thus requeue it. Upon return from queuecommand, the SCSI
> > layer initiates a requeue of the request because queuecommand returned
> > 1. Double requeue, request list is now screwed.
> > 
> > James, it looks like the queuecommand returns need an overhaul so it's
> > clear who does what and when.
> 
> Hmm, I thought the Documentation/scsi/scsi_mid_low_api.txt was quite
> clear on this:
> 
>  *      Command ownership.  If the driver returns zero, it owns the
>  *      command and must take responsibility for ensuring the 'done'
>  *      callback is executed.  Note: the driver may call done before
>  *      returning zero, but after it has called done, it may not
>  *      return any value other than zero.  If the driver makes a
>  *      non-zero return, it must not execute the command's done
>  *      callback at any time.
> 
> The 3ware driver is clearly in violation by calling done and then
> returning one.

It's clearly not the only one, the first 3 I looked at all did it.
That's the big mess. I'll do an audit.

-- 
Jens Axboe