Re: 2.6.19-rc1 regression: unable to read dvd's

[Jens Axboe <jens.axboe@oracle.com>]

On Thu, Oct 12 2006, Mike Galbraith wrote:
> On Thu, 2006-10-12 at 08:47 -0700, Alex Romosan wrote:
> > Jens Axboe <jens.axboe@oracle.com> writes:
> > 
> > > Argh damn, it needs this on top of it as well. Your second problem
> > > likely stems from that missing bit, please retest with this one applied
> > > as well.
> > >
> > > diff --git a/drivers/ide/ide-cd.c b/drivers/ide/ide-cd.c
> > > index e7513e5..bddfebd 100644
> > > --- a/drivers/ide/ide-cd.c
> > > +++ b/drivers/ide/ide-cd.c
> > > @@ -716,7 +716,7 @@ static int cdrom_decode_status(ide_drive
> > >  		ide_error(drive, "request sense failure", stat);
> > >  		return 1;
> > >  
> > > -	} else if (blk_pc_request(rq)) {
> > > +	} else if (blk_pc_request(rq) || rq->cmd_type == REQ_TYPE_ATA_PC) {
> > >  		/* All other functions, except for READ. */
> > >  		unsigned long flags;
> > >  
> > 
> > no more strange messages but, once again, i am not able to read movie
> > dvd's with the above patch applied.
> 
> Hmm.  Xine still works fine here.
> 
> I tried starting xine with no dvd in the drive for grins, and _without_
> this patch, I had to resort to SysRq-E to regain control of my box, and
> that still took quite a while.  I got no oops, but a zillion IO retries
> and sector blah messages.  DoSed me bigtime.  With this patch, I just
> got the expected can't open failure.

Yeah, the problem if you don't have this extra one-liner is that error
handling gets totally screwed. Everything should be fine in Linus' tree
now, he has everything.

-- 
Jens Axboe