Opened 14 years ago

Closed 14 years ago

#1824 closed defect (invalid)

mythtranscode segfault (MPEG2)

Reported by: kent Owned by: Isaac Richards
Priority: minor Milestone: unknown
Component: mythtv Version:
Severity: medium Keywords:
Cc: Ticket locked: no


All (at least the ones I've tested) of my recordings seem to cause mythtranscode to segfault. This hasn't always been the case, however it has been happening for a while now (I'll try to track down when it started happening).

It appears to be a different issue to [1651] as this happens with and without a cutlist. In addition to this, the segfault occurs at the beginning and the temporary file is 0 bytes.

I'm using DVB-T in Australia, and the problem is not limited to a particular channel. I can provide a portion of a recording if required.

Backtrace is attached, and the command used to generate it was: mythtranscode -i /mythstore/recordings/1001_20060507200000.mpg --mpeg2 -l -v all

Cheers, -Kent

Attachments (2)

gdb.txt (11.9 KB) - added by kent 14 years ago.
Backtrace of mythtranscode segfault
valgrind.txt (33.0 KB) - added by kent 14 years ago.
Valgrind output

Download all attachments as: .zip

Change History (9)

Changed 14 years ago by kent

Attachment: gdb.txt added

Backtrace of mythtranscode segfault

comment:1 Changed 14 years ago by kent

I forgot to mention, this is with SVN rev 9932.

comment:2 Changed 14 years ago by Isaac Richards

Resolution: invalid
Status: newclosed

Need the recording to debug this. Please reopen when it's available for download somewhere.

comment:3 Changed 14 years ago by kent

Resolution: invalid
Status: closedreopened

The first 20mb (part of the stuff I was trying to chop out) can be found here. My net connection has been a little flakey lately, so if it doesn't work pls try again a little later.

comment:4 Changed 14 years ago by Isaac Richards

Resolution: worksforme
Status: reopenedclosed

Can't repro with the sample there, sorry.

While crashed in gdb, assuming it's the same backtrace as posted above, please do:

print vf->pkt print *info print info->gop

(one after the other, of course) and send back the output.. Please reopen after attaching that stuff.

I'd also like to see a log of it run through valgrind (just valgrind -v mythtranscode blahblahblah), if possible.

comment:5 Changed 14 years ago by anonymous

Resolution: worksforme
Status: closedreopened

Installing valgrind now. Here is the output from gdb prints while crashed:

(gdb) print vf->pkt $1 = {pts = 2807880302, dts = 2807880302, data = 0x818d730 "", size = 21836, stream_index = 0, flags = 0, duration = 2, destruct = 0xb749e32b <av_destruct_packet>, priv = 0x0, pos = 564}

(gdb) print *info $2 = {sequence = 0x80d8738, gop = 0x0, current_picture = 0x80d87b0, current_picture_2nd = 0x0, current_fbuf = 0x80d8880, display_picture = 0x80d87b0, display_picture_2nd = 0x0, display_fbuf = 0x80d8880, discard_fbuf = 0x80d8880, user_data = 0x0, user_data_len = 0}

(gdb) print info->gop $3 = (const mpeg2_gop_t *) 0x0 (gdb)

Changed 14 years ago by kent

Attachment: valgrind.txt added

Valgrind output

comment:6 Changed 14 years ago by kent

Well, I think this may have something to do with my compiler setup. If compile the most recent SVN on a seperate myth setup, then copy libmyth* to the systems that are showing the problem, mythtranscode no longer segfaults.

comment:7 Changed 14 years ago by kent

Resolution: invalid
Status: reopenedclosed

Previously I had tried a make clean as well as a make distclean and the problem still occurred. In an effort to track down exactly what was causing the issue I deleted my source tree and manually went through and deleted all the myth related libraries and binaries that existed on the system. One, or a combination of those steps seems to have fixed the problem, as after a clean checkout of the code I can now successfully transcode the clip linked above which previously cause the segfault.

Note: See TracTickets for help on using tickets.