[x264-devel] x264 rev 146 segmentation fault

Tuukka Toivonen tuukkat at ee.oulu.fi
Wed Mar 2 18:51:48 CET 2005


I get segmentation fault from x264 with some videos.
Optimizations levels or CPU-specific things (e.g. -DARCH_X86) don't seem to 
make a difference, but I am linking x264 with Electric Fence:
-       $(CC) $(CFLAGS) -o x264 x264.o libx264.a -lm
+       $(CC) $(CFLAGS) -o x264 x264.o libx264.a -lefence -lm

One example test video is here:
http://www.ee.oulu.fi/~tuukkat/tmp/sg2ac-720x576.yuv.bz2
(please make a copy, I don't want to leave it there for a long time)

Sorry, no patch this time, but here's a gdb trace:
> gdb ./x264
(gdb) set args -v --frames 100 --fps 30 -i 9999 -b 0 -c -r 1 -q 28 -A all 
--subme 5 -o test.264 /home/tuukkat/std-video/sg2ac-720x576.yuv 720x576
(gdb) run
Starting program: /tmp/x264/x264 -v --frames 100 --fps 30 -i 9999 -b 0 -c 
-r 1 -q 28 -A all --subme 5 -o test.264 
/home/tuukkat/std-video/sg2ac-720x576.yuv 720x576
[Thread debugging using libthread_db enabled]
[New Thread 16384 (LWP 23878)]

   Electric Fence 2.1 Copyright (C) 1987-1998 Bruce Perens.
x264 [info]: using cpu capabilities
x264 [debug]: frame=   0 QP=25 NAL=3 Slice:I Poc:0   I4x4:839  I16x16:781  P:0    SKIP:0    size=26960 bytes PSNR Y:41.22 U:52.70 V:50.90
x264 [debug]: frame=   1 QP=28 NAL=2 Slice:P Poc:2   I4x4:340  I16x16:763  P:435  SKIP:82   size=15397 bytes PSNR Y:38.63 U:49.87 V:48.30
....
x264 [debug]: frame=  66 QP=28 NAL=2 Slice:P Poc:132 I4x4:424  I16x16:700  P:288  SKIP:208  size=10721 bytes PSNR Y:39.88 U:50.27 V:47.49
x264 [debug]: frame=  67 QP=28 NAL=2 Slice:P Poc:134 I4x4:614  I16x16:704  P:196  SKIP:106  size=13033 bytes PSNR Y:39.94 U:49.06 V:46.98
x264 [debug]: frame=  68 QP=28 NAL=2 Slice:P Poc:136 I4x4:781  I16x16:572  P:223  SKIP:44   size=15841 bytes PSNR Y:39.76 U:48.42 V:46.12

Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 16384 (LWP 23878)]
0x08064ec2 in x264_me_search_ref (h=0x402d1db0, m=0xbfffdf08, 
mvc=0xbfffdd00, i_mvc=4, p_fullpel_thresh=0x0) at encoder/me.c:113
113                 COST_MV( omx+2, omy   );
(gdb) display m->mvp[0]
2: m->mvp[0] = -2045
(gdb) display m->mvp[1]
3: m->mvp[1] = -1041
(gdb) display omx
4: omx = 1
(gdb) display omy
5: omy = -2

Looks like mvp is something ridiculous (uninitialized?).

-- 
This is the x264-devel mailing-list
To unsubscribe, go to: http://developers.videolan.org/lists.html



More information about the x264-devel mailing list