| Summary: | [Update Request] Update ffmpeg to fix several security issues | ||
|---|---|---|---|
| Product: | Mageia | Reporter: | Funda Wang <fundawang> |
| Component: | Security | Assignee: | Funda Wang <fundawang> |
| Status: | RESOLVED FIXED | QA Contact: | |
| Severity: | normal | ||
| Priority: | Normal | CC: | anssi.hannula, doktor5000, luigiwalser, marja11, qa-bugs, shlomif |
| Version: | 1 | Keywords: | TRACKER |
| Target Milestone: | --- | ||
| Hardware: | All | ||
| OS: | Linux | ||
| Whiteboard: | |||
| Source RPM: | ffmpeg-0.6.6-0.1.mga1 | CVE: | |
| Status comment: | |||
| Bug Depends on: | 6483, 6484, 6485, 6486, 6955, 6956 | ||
| Bug Blocks: | |||
|
Description
Funda Wang
2012-06-12 11:15:38 CEST
gstreamer0.10-ffmpeg, mplayer, and avidemux should also be updated if we are updating ffmpeg. See the tracker bug for the previous ffmpeg update (Bug 4146). We should also make separate bugs for QA for these different packages. CC:
(none) =>
luigiwalser Also, I don't believe the list of CVEs you listed for blender/ffmpeg 0.5.10 is correct. That looks like the CVE list from ffmpeg 0.5.8. The upstream changelogs show the following. For ffmpeg 0.5.10, included in the blender update [1]: - dpcm: ignore extra unpaired bytes in stereo streams (CVE-2011-3951) - h264: Add check for invalid chroma_format_idc (CVE-2012-0851) - adpcm: ADPCM Electronic Arts has always two channels (CVE-2012-0852) - kmvc: Check palsize (CVE-2011-3952) - several other bugfixes For ffmpeg 0.6.6, which will affect mplayer, gstreamer0.10-ffmpeg, and avidemux also once they are built, we have [2]: - dv: Fix small stack overread related to CVE-2011-3929 and CVE-2011-3936. - vqavideo: return error if image size is not a multiple of block size (CVE-2012-0947) - dpcm: ignore extra unpaired bytes in stereo streams (CVE-2011-3951) - aacsbr: prevent out of bounds memcpy() (CVE-2012-0850) - h264: Add check for invalid chroma_format_idc (CVE-2012-0851) - adpcm: ADPCM Electronic Arts has always two channels (CVE-2012-0852) - shorten: check for realloc failure (CVE-2012-0858) - shorten: Use separate pointers for the allocated memory for decoded samples (CVE-2012-0858) - kmvc: Check palsize (CVE-2011-3952) - several other bugfixes Unfortunately the ffmpeg 0.10.4 changelog [3] (which would affect Mageia 2) has no helpful information, so I don't know if we need an update there. [1] - http://git.videolan.org/?p=ffmpeg.git;a=blob;f=Changelog;h=3a66ab0b888799d3f0b48fea868c85f3e6454c05;hb=9eaec5b8f010c805fd8e77216a1ec67eb20b1466 [2] - http://git.videolan.org/?p=ffmpeg.git;a=blob;f=Changelog;h=6f753216f5383eb296802efe1dbd3eea0ed589af;hb=62133b38ed043b57eeecbe7fc8b6f187fd92e5e0 [3] - http://git.videolan.org/?p=ffmpeg.git;a=blob;f=Changelog;h=17ca5c9e3881ffa4fc040bef5f7d2868f9b84836;hb=8c0c0e9eb3341fe42a2a9315cef5af21e94c4855
David Walser
2012-06-12 14:41:09 CEST
CC:
(none) =>
doktor5000
David Walser
2012-06-12 14:41:52 CEST
CC:
(none) =>
shlomif
Manuel Hiebel
2012-06-14 13:06:22 CEST
Component:
RPM Packages =>
Security This doesn't seem ready for QA yet? Indeed, and I didn't know Funda wasn't even CC'd on this :o( I'll assign back to him. CC:
(none) =>
qa-bugs When it's ready could you create separate bugs for blender, avidemux etc please Digging through the git logs, it looks like in the 0.10 branch, previous fixes for CVE-2012-0851 and CVE-2011-3937 were accidentally reverted, and they were put back in 0.10.4. So that means we need to update ffmpeg in Mageia 2 as well. - h264: Add check for invalid chroma_format_idc (CVE-2012-0851) - h263dec: Disallow width/height changing with frame threads (CVE-2011-3937) Do we have any packages in Mageia 2 still building an internal copy of ffmpeg? Ugh, it looks like avidemux comes with an internal copy of ffmpeg-0.9. The following CVEs have been fixed since, the current 0.9 is 0.9.2. - vqavideodev: Check image dimensions, fixes out of heap array read (CVE-2012-0947) - vorbis: make sure ch is non zero before calling vorbis_residue_decode (CVE-2011-3895) - ogg: Avoid the possibility to read out-of-bounds of a static global array in Vorbis decoding (CVE-2011-3893) - mkv: Fix a bug where a pointer was cached to an array that might later move due to a realloc() (CVE-2011-3893) We should update its internal copy to that if it can't build against system 0.10.
David Walser
2012-06-16 21:22:32 CEST
CC:
(none) =>
anssi.hannula For ffmpeg 0.6.6, here's a more complete CVE list:
- nsvdec: Fix use of uninitialized streams, Be more careful with
av_malloc(), nsvdec: Propagate errors (CVE-2011-3940)
- dv: Fix small stack overread, check stype, Fix null pointer
dereference due to ach=0 (CVE-2011-3929 and CVE-2011-3936)
- atrac3: Fix crash in tonal component decoding (CVE-2012-0853)
- mjpegbdec: Fix overflow in SOS (CVE-2011-3947)
- kgv1dec: Increase offsets array size so it is large enough
(CVE-2011-3945)
- vqavideo: return error if image size is not a multiple of block size
(CVE-2012-0947)
- dpcm: ignore extra unpaired bytes in stereo streams (CVE-2011-3951)
- aacsbr: prevent out of bounds memcpy() (CVE-2012-0850)
- h264: Add check for invalid chroma_format_idc (CVE-2012-0851)
- adpcm: ADPCM Electronic Arts has always two channels (CVE-2012-0852)
- shorten: check for realloc failure (CVE-2012-0858)
- shorten: Use separate pointers for the allocated memory for decoded
samples (CVE-2012-0858)
- kmvc: Check palsize (CVE-2011-3952)
- several other bugfixes
David Walser
2012-06-17 00:15:16 CEST
Depends on:
(none) =>
6483
David Walser
2012-06-17 00:26:19 CEST
Depends on:
(none) =>
6484
David Walser
2012-06-17 00:33:45 CEST
Depends on:
(none) =>
6485
David Walser
2012-06-17 01:02:28 CEST
Depends on:
(none) =>
6486 OK I've converted this into a tracking bug. Here's the current status. Mageia 1: ffmpeg - Bug 6484 mplayer - Bug 6483 blender - Bug 6485 gstreamer0.10-ffmpeg - not fixed avidemux - not fixed (needs fixes from ffmpeg 0.6.6) Mageia 2: ffmpeg - Bug 6486 avidemux - not fixed (needs fixes from ffmpeg 0.9.2) (In reply to comment #10) > OK I've converted this into a tracking bug. Here's the current status. > setting TRACKER keyword Keywords:
(none) =>
TRACKER Apparently gstreamer0.10-ffmpeg was changed to use external ffmpeg in the previous update (Bug 4152). If so, all that needs fixed is avidemux. I'm looking through the avidemux 2.5.6 (from Mageia 2/Cauldron) tarball, and I don't see any ffmpeg code. However, I saw this in the avidemux 2.5.6 release announcement [1]: *Updated the FFmpeg libraries (version 0.9) Am I missing something? [1] - http://fixounet.free.fr/avidemux/news.html Looking at the avidemux from Mageia 1 ChangeLog, it's missing the fixes from ffmpeg 0.6.5 (included in the previous mplayer update in January) as well. (In reply to comment #13) > I'm looking through the avidemux 2.5.6 (from Mageia 2/Cauldron) tarball, and I > don't see any ffmpeg code. However, I saw this in the avidemux 2.5.6 release > announcement [1]: > *Updated the FFmpeg libraries (version 0.9) > > Am I missing something? > > [1] - http://fixounet.free.fr/avidemux/news.html It's a bundled tarball in avidemux/ADM_libraries (thanks Florian).
David Walser
2012-08-05 02:38:45 CEST
Depends on:
(none) =>
6955
David Walser
2012-08-05 02:38:51 CEST
Depends on:
(none) =>
6956 Final list. Mageia 1: ffmpeg - Bug 6484 mplayer - Bug 6483 blender - Bug 6485 avidemux - Bug 6955 Mageia 2: ffmpeg - Bug 6486 avidemux - Bug 6956 All better now :o) Status:
NEW =>
RESOLVED |