Opened 10 years ago

Last modified 9 years ago

#1588 new defect

mencoder segmentation fault

Reported by: ttabyss@… Owned by: reimar
Priority: normal Component: core
Version: HEAD Severity: normal
Keywords: Cc: bpringlemeir@…
Blocked By: Blocking:
Reproduced by developer: Analyzed by developer:

Description

mencoder fails on AMD64 system. With the same input file, mencoder does not fail under x86.

The input file is part of an MPEG captured stream of a high definition ATSC over-the-air broadcast so the file may contain anomalies.

The system is an up-to-date Debian (Testing) system with a Core-I7 processor.

I have attempted to FTP a log and sample file but the support FTP server will not accept anonymous uploads to /MPlayer/incoming.

Attachments (2)

debug.log (43.1 KB) - added by ttabyss@… 10 years ago.
Contains gdb debug log of seg fault
cut-4.tar.bz (957.1 KB) - added by ttabyss@… 10 years ago.
sample gdb log and mpeg video file which seg-faults

Download all attachments as: .zip

Change History (7)

Changed 10 years ago by ttabyss@…

Contains gdb debug log of seg fault

comment:1 Changed 10 years ago by ttabyss@…

This file shows the seg-fault under AMD64.

Changed 10 years ago by ttabyss@…

sample gdb log and mpeg video file which seg-faults

comment:2 Changed 10 years ago by ttabyss@…

I don't think this is the same seg-fault as before. This may be a different problem caused by cutting the video file down to 1mb so it could be uploaded to Bugzilla.

The attached log and mpg file reproduce a seg-fault with mencoder anyway.

comment:3 Changed 10 years ago by reimar

The cut file has a different issue.
FOr the first one, try with -noslices

comment:4 Changed 9 years ago by bpringlemeir@…

Possibly related,

http://bugzilla.mplayerhq.hu/show_bug.cgi?id=1613

This is occuring on an x86 P-4. I think that the RENAME(hScale) is getting a bad pointer (from mpeg-ts). Removing the scale option may allow the ATSC stream to play.

I encountered this with a stream captured with atscap. I can replay the captured stream. The place where RENAME(hScale) has a SEGV is near an ac-tex damaged message.

In the debug.log attached to this ticket, I think that the line,

src=0x7ffff64b2840 <Address 0x7ffff64b2840 out of bounds>,

is important and GDB is not lying. The pointer is _really_ invalid. I have a similar case with,

$ gdb --version
GNU gdb 6.8-debian
This GDB was configured as "i486-linux-gnu".
$ gcc --version
gcc (Debian 4.3.2-1.1) 4.3.2

My mplayer/mencoder source has been fluid for the last few days (SVN head, Dec 10-16).

comment:5 Changed 9 years ago by bpringlemeir@…

  • Cc bpringlemeir@… added
Note: See TracTickets for help on using tickets.