[FFmpeg-devel] [PATCH] Fix a couple of errors with bad Vorbis headers

Reimar Döffinger Reimar.Doeffinger
Wed Dec 29 17:06:43 CET 2010


On Wed, Dec 29, 2010 at 03:32:20AM -0500, Gregory Maxwell wrote:
> On Tue, Dec 28, 2010 at 11:19 PM, Alex Converse <alex.converse at gmail.com> wrote:
> > Patch by Chris Evans/Google Chrome
> >
> > Chrome review URL:
> > http://codereview.chromiumcodereview.org/5964011<http://codereview.chromium.org/5964011>
> > Chrome bug URL:
> > http://code.google.com/p/chromium/issues/detail?id=68815(Broken?)
> 
> I was going to test this patch to validate that it doesn't break any
> regular files, but I found that a current ffmpeg git checkout
> (3f9de618) doesn't play all the files in the public test vectors:
> 
> http://people.xiph.org/~xiphmont/test-vectors/vorbis/one-entry-codebook-test.ogg

After checking again, this test-case seems broken to me.
Specification says:
"it consists of a single codework[sic] of zero bits"
However the codebook is coded as unordered spares, and the codeword
length is get_bits(5) + 1, which makes it impossible to specify
a codeword of 0 bits.
As such I think our code correctly identifies this as an invalid stream,
since it contains a codebook with a single 1-bit codeword, resulting
in an underspecified tree which is forbidden.



More information about the ffmpeg-devel mailing list