Re: [Qemu-devel] [Bug 1332297] Re: qemu-img: crash on check of an image with large value in the 'size' header field

[Kevin Wolf <kwolf@redhat.com>]

Am 25.06.2014 um 11:54 hat M.Kustova geschrieben:
> On Wed, Jun 25, 2014 at 1:42 PM, Kevin Wolf <kwolf@redhat.com> wrote:
> > Am 25.06.2014 um 11:32 hat M.Kustova geschrieben:
> >> On Tue, Jun 24, 2014 at 7:36 PM, Kevin Wolf <kwolf@redhat.com> wrote:
> >> > Am 24.06.2014 um 15:19 hat M.Kustova geschrieben:
> >> >> On Mon, Jun 23, 2014 at 12:02 PM, Stefan Hajnoczi <stefanha@gmail.com> wrote:
> >> >> > On Thu, Jun 19, 2014 at 07:19:55PM -0000, Maria Kustova wrote:
> >> >> >> The bug description missed qemu-img error:
> >> >> >>
> >> >> >> (process:12283): GLib-ERROR **: gmem.c:110: failed to allocate
> >> >> >> 18446744059294601304 bytes
> >> >> >
> >> >> > Thanks, there has been recent work by Kevin Wolf to handle memory
> >> >> > allocation failures gracefully without terminating QEMU.  This sounds
> >> >> > like a candidate for g_try_malloc() and friends.
> >> >> >
> >> >> > Does the following patch series solve the problem?
> >> >> > https://lists.gnu.org/archive/html/qemu-devel/2014-06/msg01275.html
> >> >>
> >> >> These patches are conflicting with current master. So I can't test
> >> >> them as they are.
> >> >>
> >> >> Do you have a developer repository or branch containing these patches,
> >> >> so I could test it on the pre-release base?
> >> >
> >> > I'm just about to send a new version, I'll keep you CCed there.
> >>
> >> "[PATCH v4 21/21] qcow2: Return useful error code in refcount_init()"
> >> is still broken for the current master.
> >
> > In which way? I can cleanly apply the whole patch series on master (even
> > tried applying the emails from my inbox to be sure).
> 
> Beginning from line #49 in master:
> 
>     if (s->refcount_table_size > 0) {
>         BLKDBG_EVENT(bs->file, BLKDBG_REFTABLE_LOAD);
>         ret = bdrv_pread(bs->file, s->refcount_table_offset,
> 
> The patch:
> 
>    if (s->refcount_table_size > 0) {^M
>          if (s->refcount_table == NULL) {^M
> +            ret = -ENOMEM;^M
>              goto fail;^M
>          }^M
>          BLKDBG_EVENT(bs->file, BLKDBG_REFTABLE_LOAD);^M
>          ret = bdrv_pread(bs->file, s->refcount_table_offset,^M
> 
> At least master version doesn't have this condition.

It is code added in patch 11 of the same series.

Kevin