Re: [PATCH 02/15] ide-tape: remove back-to-back REQUEST_SENSE detection

[Tejun Heo <tj@kernel.org>]

Bartlomiej Zolnierkiewicz wrote:
> Can't we just apply them both? :)

I don't think Borislav's patch is correct.

> Could it be that we just need to take care if this case:
> 
>         if (rq->cmd[13] & REQ_IDETAPE_PC2) {
>                 idetape_media_access_finished(drive);
>                 return ide_stopped;
>         }
> 
> [all other code-paths set pc before calling ide_tape_issue_pc()]

drive->pc is not supposed to be set in do_request().  It should point
to the previous pc structure for REQUEST_SENSE retry handling.  Note
that in this case the previous command pc structure is guaranteed to
be valid.  It's a fragile and confusing logic but, well, that's how
the current code works.

Thanks.

-- 
tejun