Jeff Downs [EMAIL PROTECTED] added the comment:
I thought LGPL portions of swscale are meant to be the LGPL replacement for
img_convert? At least that was intent when img_convert was deprecated and we
told even LGPL users to switch to swscale
Jeff Downs [EMAIL PROTECTED] added the comment:
On Fri, 15 Aug 2008, Piers O'Hanlon wrote:
Piers O'Hanlon [EMAIL PROTECTED] added the comment:
Thanks for continued interest in this issue.
The streams I have are tcpdump/wireshark files containing RTP encapsulated
H.264
with in-band
Jeff Downs [EMAIL PROTECTED] added the comment:
Actually, after reading the spec 10 more times, I think this can legitimately
occur...
IDR pictures are always ref, and they do not get sliding window (8.2.5.1, #2).
So if you had a stream with only IDR pictures (no other reference pictures
Jeff Downs [EMAIL PROTECTED] added the comment:
Reported states fixed.
--
status: open - closed
__
FFmpeg issue tracker [EMAIL PROTECTED]
https://roundup.mplayerhq.hu/roundup/ffmpeg/issue460
__
Jeff Downs [EMAIL PROTECTED] added the comment:
Reported states fixed.
--
status: open - closed
__
FFmpeg issue tracker [EMAIL PROTECTED]
https://roundup.mplayerhq.hu/roundup/ffmpeg/issue460
__
On Fri, 1 Aug 2008, Paul Kendall wrote:
Paul Kendall [EMAIL PROTECTED] added the comment:
GDB output as requested.
Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 0xa6ffdb90 (LWP 5484)]
0xb6f1f537 in hl_decode_mb_complex (h=0x83d6850) at h264.c:2314
warning:
Jeff Downs [EMAIL PROTECTED] added the comment:
On Sat, 2 Aug 2008, Paul Kendall wrote:
I have a sample that shows this, do you want a sample attached to the ticket?
or should I upload it elsewhere?
Follow instructions at http://ffmpeg.mplayerhq.hu/bugreports.html
What about the analysis
Jeff Downs [EMAIL PROTECTED] added the comment:
Retested and shown to be fixed in recent svn, likely by r14296 (fix for issue
335).
--
status: open - closed
substatus: open - fixed
__
FFmpeg issue tracker [EMAIL PROTECTED]
https
Jeff Downs [EMAIL PROTECTED] added the comment:
Tested and shown to be fixed in recent svn, likely by r14296 (fix for issue
335).
--
status: open - closed
substatus: open - fixed
__
FFmpeg issue tracker [EMAIL PROTECTED]
https
Jeff Downs [EMAIL PROTECTED] added the comment:
No more hang here, either, but may be worth keeping this sample in mind as it
trips several issues. When I looked at this before, the hang was an infinite
loop created through heap corruption clobbering some loop control variables.
Valgrind
Jeff Downs [EMAIL PROTECTED] added the comment:
If num_ref_frames is zero, there are no reference frames (0 does not mean do
whatever you want, it means zero reference frames) and thus the code you're
citing should never be executed (only executed for reference frames). If it is,
seems to me
Jeff Downs [EMAIL PROTECTED] added the comment:
ffmpeg requires make 3.81 or higher. If you have 3.81 or higher, please reopen
and provide make version.
--
status: new - closed
substatus: new - invalid
__
FFmpeg issue tracker [EMAIL
Jeff Downs [EMAIL PROTECTED] added the comment:
This is caused by handling of the SPS cropping rectangle.
In short, cropping is applied by modifying MpegEncContext.width/height.
Due to later mpegvideo.c init code, this results in an incorrect
MpegEncContext.mb_width begin set (MPV_common_init
Jeff Downs [EMAIL PROTECTED] added the comment:
Preliminary analysis... the decoder is leaking delayed output pictures and
eventually running out of Pictures to decode into.
Pictures are being removed from delayed_pic[] without having the reference
member cleared from DELAYED_PIC_REF to zero
Jeff Downs [EMAIL PROTECTED] added the comment:
r12790 working fine on Linux here with stated configure line. Please try with a
clean checkout if you have not already. If that does not work, please provide
more details including make --version.
Only one platform necessary (Linux preferred
Jeff Downs [EMAIL PROTECTED] added the comment:
Patch to limit number of reference frames to that specified in SPS.
Also adds an error return when the SPS has a reference frame count too big for
us to handle.
__
FFmpeg issue tracker [EMAIL
Jeff Downs [EMAIL PROTECTED] added the comment:
This now aborts:
[h264 @ 0x84451b0]Internal error, picture buffer overflow
This is caused by h264 decoder passing B-frames to picture allocation routines
in mpegvideo.c, despite that it has no reference frames available.
--
nosy
Jeff Downs [EMAIL PROTECTED] added the comment:
Reproduced here using simply ./configure make ./ffmpeg
gcc version 4.1.1
Linux cascade 2.6.23-gentoo #1 SMP Wed Oct 31 17:21:57 EDT 2007 x86_64 Dual-Core
AMD Opteron(tm) Processor 2214 AuthenticAMD GNU/Linux
--
status: new - open
Jeff Downs [EMAIL PROTECTED] added the comment:
Looks like recent changes to add linker flags -Bsymbolic and nocopyreloc has
caused static x86-64 builds to fail in this manner.
Moving the addition of these linker flags into an if enabled shared block in
configure fixes static builds.
However, I
Jeff Downs [EMAIL PROTECTED] added the comment:
This is the same as issue 284.
--
status: new - closed
substatus: new - duplicate
__
FFmpeg issue tracker [EMAIL PROTECTED]
https://roundup.mplayerhq.hu/roundup/ffmpeg/issue282
__
20 matches
Mail list logo