[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Patch] shm cleanups



Andrea Arcangeli <andrea@suse.de> writes:

> Christoph Rohland <hans-christoph.rohland@sap.com> writes:
> 
> >         if (!(page_map = prepare_highmem_swapout(page_map)))
> > -               goto check_table;
> > +               goto failed;
> 
> This fragment isn't correct. You may fail too early and so you may get
> a task killed due OOM even if you still have lots of regular pages
> queued in a shm segment.

Yes I know it is questionable, but if prepare_highmem_swapout fails we
are in the highmem area and probably most of the rest of shm is also
there. So we only consume a lot of CPU if going on and calling
prepare_highmem_swapout again and again..

prepare_highmem_swapout should not fail (and does not fail very often
in 2.3.26-pre2) In 2.3.25 the machine effectivly locked up when it
failed (and it failed very often).

On 2.3.26-pre2 I do not see a difference between the two versions. So
I do not know which way will be better.

        Christoph
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://humbolt.geo.uu.nl/Linux-MM/