Re: [kirkstone 03/11] ffmpeg: fix for CVE-2022-3109


Martin Jansa
 

> Not sure why this didn't show up in my testing! 

It's shown only when do_patch task is really executed and it doesn't cause the do_patch to fail completely. So it's possible that you've built it once with the warning/error shown and then the next time you were doing the final test for kirkstone it was just re-using ffmpeg from sstate (without the need to re-executed do_patch again).

Maybe we should consider this issue to be fatal for do_fetch when patch-fuzz is in ERROR_QA (I even thought it was working like that at some time).

Regards,

On Mon, Jan 16, 2023 at 3:32 PM Steve Sakoman <steve@...> wrote:
On Mon, Jan 16, 2023 at 2:00 AM Martin Jansa <martin.jansa@...> wrote:
>
> This patch doesn't apply cleanly on ffmpeg-5.0.1:

Thanks for the review Martin.

Not sure why this didn't show up in my testing!  But since Richard
hasn't taken the pull request yet I will remove this patch from the
current pull request and move it to my next set of patches (along with
your fix).

Thanks!

Steve

>
> ERROR: ffmpeg-5.0.1-r0 do_patch: Fuzz detected:
>
> Applying patch 0001-avcodec-vp3-Add-missing-check-for-av_malloc.patch
> patching file libavcodec/vp3.c
> Hunk #1 succeeded at 2677 with fuzz 1 (offset -2 lines).
>
>
> The context lines in the patches can be updated with devtool:
>
>     devtool modify ffmpeg
>     devtool finish --force-patch-refresh ffmpeg <layer_path>
>
> Don't forget to review changes done by devtool!
>
> ERROR: ffmpeg-5.0.1-r0 do_patch: QA Issue: Patch log indicates that patches do not apply cleanly. [patch-fuzz]
>
> Narpat: Should I send a fix or will you handle that?
>
> On Thu, Jan 12, 2023 at 3:33 AM Steve Sakoman <steve@...> wrote:
>>
>> From: Narpat Mali <narpat.mali@...>
>>
>> An issue was discovered in the FFmpeg package, where vp3_decode_frame in libavcodec/vp3.c lacks check of
>> the return value of av_malloc() and will cause a null pointer dereference, impacting availability.
>>
>> CVE: CVE-2022-3109
>>
>> Upstream-Status: Backport [https://github.com/FFmpeg/FFmpeg/commit/656cb0450aeb73b25d7d26980af342b37ac4c568]
>>
>> Signed-off-by: Narpat Mali <narpat.mali@...>
>> Signed-off-by: Steve Sakoman <steve@...>
>> ---
>>  ...-vp3-Add-missing-check-for-av_malloc.patch | 44 +++++++++++++++++++
>>  .../recipes-multimedia/ffmpeg/ffmpeg_5.0.1.bb |  3 +-
>>  2 files changed, 46 insertions(+), 1 deletion(-)
>>  create mode 100644 meta/recipes-multimedia/ffmpeg/ffmpeg/0001-avcodec-vp3-Add-missing-check-for-av_malloc.patch
>>
>> diff --git a/meta/recipes-multimedia/ffmpeg/ffmpeg/0001-avcodec-vp3-Add-missing-check-for-av_malloc.patch b/meta/recipes-multimedia/ffmpeg/ffmpeg/0001-avcodec-vp3-Add-missing-check-for-av_malloc.patch
>> new file mode 100644
>> index 0000000000..94858a6cdd
>> --- /dev/null
>> +++ b/meta/recipes-multimedia/ffmpeg/ffmpeg/0001-avcodec-vp3-Add-missing-check-for-av_malloc.patch
>> @@ -0,0 +1,44 @@
>> +From 656cb0450aeb73b25d7d26980af342b37ac4c568 Mon Sep 17 00:00:00 2001
>> +From: Jiasheng Jiang <jiasheng@...>
>> +Date: Tue, 15 Feb 2022 17:58:08 +0800
>> +Subject: [PATCH] avcodec/vp3: Add missing check for av_malloc
>> +
>> +Since the av_malloc() may fail and return NULL pointer,
>> +it is needed that the 's->edge_emu_buffer' should be checked
>> +whether the new allocation is success.
>> +
>> +Fixes: d14723861b ("VP3: fix decoding of videos with stride > 2048")
>> +Reviewed-by: Peter Ross <pross@...>
>> +Signed-off-by: Jiasheng Jiang <jiasheng@...>
>> +
>> +CVE: CVE-2022-3109
>> +
>> +Upstream-Status: Backport [https://github.com/FFmpeg/FFmpeg/commit/656cb0450aeb73b25d7d26980af342b37ac4c568]
>> +
>> +Signed-off-by: Narpat Mali <narpat.mali@...>
>> +---
>> + libavcodec/vp3.c | 7 ++++++-
>> + 1 file changed, 6 insertions(+), 1 deletion(-)
>> +
>> +diff --git a/libavcodec/vp3.c b/libavcodec/vp3.c
>> +index e9ab54d736..e2418eb6fa 100644
>> +--- a/libavcodec/vp3.c
>> ++++ b/libavcodec/vp3.c
>> +@@ -2679,8 +2679,13 @@ static int vp3_decode_frame(AVCodecContext *avctx,
>> +                                         AV_GET_BUFFER_FLAG_REF)) < 0)
>> +         goto error;
>> +
>> +-    if (!s->edge_emu_buffer)
>> ++    if (!s->edge_emu_buffer) {
>> +         s->edge_emu_buffer = av_malloc(9 * FFABS(s->current_frame.f->linesize[0]));
>> ++        if (!s->edge_emu_buffer) {
>> ++            ret = AVERROR(ENOMEM);
>> ++            goto error;
>> ++        }
>> ++    }
>> +
>> +     if (s->keyframe) {
>> +         if (!s->theora) {
>> +--
>> +2.34.1
>> +
>> diff --git a/meta/recipes-multimedia/ffmpeg/ffmpeg_5.0.1.bb b/meta/recipes-multimedia/ffmpeg/ffmpeg_5.0.1.bb
>> index 95b4bf50ac..c5bebe9c2d 100644
>> --- a/meta/recipes-multimedia/ffmpeg/ffmpeg_5.0.1.bb
>> +++ b/meta/recipes-multimedia/ffmpeg/ffmpeg_5.0.1.bb
>> @@ -26,7 +26,8 @@ SRC_URI = "https://www.ffmpeg.org/releases/${BP}.tar.xz \
>>             file://0001-libavutil-include-assembly-with-full-path-from-sourc.patch \
>>             file://0001-avcodec-rpzaenc-stop-accessing-out-of-bounds-frame.patch \
>>             file://0001-avcodec-smcenc-stop-accessing-out-of-bounds-frame.patch \
>> -           "
>> +           file://0001-avcodec-vp3-Add-missing-check-for-av_malloc.patch \
>> +          "
>>
>>  SRC_URI[sha256sum] = "ef2efae259ce80a240de48ec85ecb062cecca26e4352ffb3fda562c21a93007b"
>>
>> --
>> 2.25.1
>>
>>
>>
>>

Join {openembedded-core@lists.openembedded.org to automatically receive all group messages.