Re: [PATCH] block: replace __getblk_slow misfix by grow_dev_page fix

[Richard W.M. Jones]

On Tue, Aug 21, 2012 at 06:33:45PM -0700, Hugh Dickins wrote:
> Jeff,
> 
> Your commit 91f68c89d8f3 ("block: fix infinite loop in __getblk_slow"),
> already gone into 3.* stable, is not good.  Could you and your testers
> please give this alternative a try - I think it should work, and have
> started it on a few days' memory load on 3.5, but not tried your case.

I'll try your patch later today, but the test is very simple:

- Create a 1024 byte empty (all zeroes) partition and then try to mount it.

Failure was that the mount command would go into an infinite loop
using 100% CPU.  Success is that it gives an error.

More about it here including a very simple libguestfs script to
reproduce it:

  https://bugzilla.redhat.com/show_bug.cgi?id=835019#c0

and a virtual disk image that also demonstrates the problem without
needing anything except a VM running the test kernel:

  https://bugzilla.redhat.com/show_bug.cgi?id=835019#c1

Rich.

-- 
Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones
New in Fedora 11: Fedora Windows cross-compiler. Compile Windows
programs, test, and build Windows installers. Over 70 libraries supprt'd
http://fedoraproject.org/wiki/MinGW http://www.annexia.org/fedora_mingw
--
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/