Re: [Bug 12713] Hang on resume from hibernation, apparently as result of 213d9417fec

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

On Mon, Feb 16 2009, Michael S. Tsirkin wrote:
> On Mon, Feb 16, 2009 at 09:47:32AM +0100, Jens Axboe wrote:
> > On Mon, Feb 16 2009, Michael S. Tsirkin wrote:
> > > On Mon, Feb 16, 2009 at 9:27 AM, Jens Axboe <jens.axboe@oracle.com> wrote:
> > > > On Mon, Feb 16 2009, Michael S. Tsirkin wrote:
> > > >> On Mon, Feb 16, 2009 at 2:05 AM, Michael S. Tsirkin
> > > >> <m.s.tsirkin@gmail.com> wrote:
> > > >> > Summary: seem to need to revert 213d9417fec62ef4c3675621b9364a667954d4dd
> > > >> > to fix resume from hibernation. Bugzilla entry created:
> > > >> > http://bugzilla.kernel.org/show_bug.cgi?id=12713
> > > >>
> > > >> Looking over this, I see something strange in the commit in question:
> > > >>
> > > >> diff --git a/include/linux/bio.h b/include/linux/bio.h
> > > >> index 5175aa3..f53568c 100644
> > > >> --- a/include/linux/bio.h
> > > >> +++ b/include/linux/bio.h
> > > >> @@ -163,12 +163,15 @@ struct bio {
> > > >>  #define BIO_RW         0       /* Must match RW in req flags (blkdev.h) */
> > > >>  #define BIO_RW_AHEAD   1       /* Must match FAILFAST in req flags */
> > > >>  #define BIO_RW_BARRIER 2
> > > >> -#define BIO_RW_SYNC    3
> > > >> -#define BIO_RW_META    4
> > > >> -#define BIO_RW_DISCARD 5
> > > >> -#define BIO_RW_FAILFAST_DEV            6
> > > >> -#define BIO_RW_FAILFAST_TRANSPORT      7
> > > >> -#define BIO_RW_FAILFAST_DRIVER         8
> > > >> +#define BIO_RW_SYNCIO  3
> > > >> +#define BIO_RW_UNPLUG  4
> > > >> +#define BIO_RW_META    5
> > > >> +#define BIO_RW_DISCARD 6
> > > >> +#define BIO_RW_FAILFAST_DEV            7
> > > >> +#define BIO_RW_FAILFAST_TRANSPORT      8
> > > >> +#define BIO_RW_FAILFAST_DRIVER         9
> > > >> +
> > > >> +#define BIO_RW_SYNC    (BIO_RW_SYNCIO | BIO_RW_UNPLUG)
> > > >>
> > > >>  /*
> > > >>   * upper 16 bits of bi_rw define the io priority of this bio
> > > >>
> > > >> I haven't read the code in depth, but taking running numbers and doing
> > > >> bitwise "or"
> > > >> on them looks a bit strange to me.
> > > >> So here BIO_RW_SYNC is (3 | 4) = 7, that is the same as BIO_RW_FAILFAST_DEV.
> > > >> So for example bio_failfast_dev and bio_sync are the same.
> > > >>
> > > >> Jens, could you comment on this please? Is this intentional?
> > > >
> > > > That's clearly a braino, you can't OR the shift values of course. I'll
> > > > get it fixed up asap!
> > > 
> > > Cool, send me a patch, I'll test.
> > 
> > Try this. Not a huge fan of this approach, but it should bring behaviour
> > back to what it used to be.
> 
> I verified that applying this patch on top of 2.6.29-rc5 fixes
> hibernation for me.

Thanks for testing and confirming, it's queued up for inclusion asap.

-- 
Jens Axboe