Re: [PATCH 12/15] writeback: remove writeback_control.more_io

[Jan Kara]

On Wed 13-07-11 03:57:40, Hugh Dickins wrote:
> On Tue, 12 Jul 2011, Hugh Dickins wrote:
> > On Tue, 12 Jul 2011, Hugh Dickins wrote:
> > > On Mon, 11 Jul 2011, Wu Fengguang wrote:
> > > > 
> > > > It's relatively easy to confirm, by reusing the below trace event to
> > > > show the inode (together with its state) being requeued.
> > > > 
> > > > If this is the root cause, it may equally be fixed by
> > > > 
> > > > -			requeue_io(inode, wb);
> > > > +			redirty_tail(inode, wb);
> > > > 
> > > > which would be useful in case the bug is so deadly that it's no longer
> > > > possible to do tracing.
> > > 
> > > I checked again this morning that I could reproduce it on two machines,
> > > one went in a few minutes, the other within the hour.  Then I made that
> > > patch changing the requeue_io to redirty_tail, and left home with them
> > > running the test with the new kernel: we'll see at the end of the day
> > > how they fared.
> > 
> > I think that fixes it.  The x86_64 is still running with that, but the
> > ppc64 gave up fairly early, hitting freeze in __slab_free() instead.
> > 
> > I've now, I believe, reconstituted what ChristophL intended from the
> > mm_types.h struct page patch he posted (which applied neither to mmotm,
> > nor to Pekka's for-next, so far as I could tell: maybe cl did some
> > intermediate tidying of some of the random indentation).  So now
> > testing that with redirty_tail on ppc64: will report in 9 hours.
> 
> Same result as before.  The x86_64 is still going fine, but the ppc64
> again seized up in __slab_free() after two and a half hours of load.
> 
> I think we should assume that your -requeue_io +redirty_tail is a good
> fix for the writeback freeze (if you can reassure us, that it does not
> risk postponing some writes indefinitely), and I move over to the other
> thread to pursue the struct page __slab_free() freeze.
  Well, I_FREEING or I_WILL_FREE inodes are written back by iput_final()
and it is reclaim code that is responsible for eventually removing them. So
writeback code can safely ignore them. I_NEW inodes should move out of this
state when they are fully set up and in the writeback round following that,
we will consider them for writeback. So I think the change really makes
sense.

								Honza
-- 
Jan Kara <jack@xxxxxxx>
SUSE Labs, CR
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/