[Bug 222118] multimedia/handbrake handbrake dumps core when navigating to select source

[bugzilla-noreply at freebsd.org]

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=222118

--- Comment #25 from naito.yuichiro at gmail.com ---
While I'm testing clang built Handbrake on i386,
Handbrake crashes when I open MPEG-4 format H.264 encoded file.

Stack trace was shown as follows.

```
Core was generated by `ghb'.
Program terminated with signal SIGBUS, Bus error.
#0  0x085f9232 in ff_deblock_v_luma_8_avx ()
[Current thread is 1 (LWP 102012)]
(gdb) bt
#0  0x085f9232 in ff_deblock_v_luma_8_avx ()
#1  0xffffffff in ?? ()
#2  0xffffffff in ?? ()
#3  0xffffffff in ?? ()
#4  0x2e418000 in ?? ()
#5  0x084fb0c3 in loop_filter ()
#6  0x084f94d8 in decode_slice ()
#7  0x084f8eb4 in ff_h264_execute_decode_slices ()
#8  0x0820efbc in h264_decode_frame ()
#9  0x083a4831 in avcodec_decode_video2 ()
#10 0x083a4f75 in do_decode ()
#11 0x083a4f0d in avcodec_send_packet ()
#12 0x081861c8 in try_decode_frame ()
#13 0x08184c30 in avformat_find_stream_info ()
#14 0x080c60ed in hb_stream_open ()
#15 0x080e44a5 in ScanFunc ()
#16 0x080b0a9c in hb_thread_func ()
#17 0x2a2acdda in ?? () from /lib/libthr.so.3
#18 0x00000000 in ?? ()
(gdb) info registers
eax            0xfb3f095c       -79754916
ecx            0x10             16
edx            0x5              5
ebx            0x2              2
esp            0xfb3f08dc       0xfb3f08dc
ebp            0xfb3f0a58       0xfb3f0a58
esi            0xfb3f092c       -79754964
edi            0x3              3
eip            0x85f9232        0x85f9232 <ff_deblock_v_luma_8_avx+34>
eflags         0x10283          [ CF SF IF RF ]
cs             0x33             51
ss             0x3b             59
ds             0x3b             59
es             0x3b             59
fs             0x13             19
gs             0x1b             27
(gdb) disassemble ff_deblock_v_luma_8_avx
Dump of assembler code for function ff_deblock_v_luma_8_avx:
   0x085f9210 <+0>:    push   %ebx
   0x085f9211 <+1>:    push   %esi
   0x085f9212 <+2>:    sub    $0x24,%esp
   0x085f9215 <+5>:    mov    0x30(%esp),%eax
   0x085f9219 <+9>:    mov    0x34(%esp),%ecx
   0x085f921d <+13>:   mov    0x38(%esp),%edx
   0x085f9221 <+17>:   mov    0x3c(%esp),%ebx
   0x085f9225 <+21>:   mov    0x40(%esp),%esi
   0x085f9229 <+25>:   lea    (%ecx,%ecx,2),%esi
   0x085f922c <+28>:   dec    %edx
   0x085f922d <+29>:   neg    %esi
   0x085f922f <+31>:   dec    %ebx
   0x085f9230 <+32>:   add    %eax,%esi
=> 0x085f9232 <+34>:   vmovdqa (%esi,%ecx,1),%xmm0
   0x085f9237 <+39>:   vmovdqa (%esi,%ecx,2),%xmm1
   0x085f923c <+44>:   vmovdqa (%eax),%xmm2
   0x085f9240 <+48>:   vmovdqa (%eax,%ecx,1),%xmm3
   0x085f9245 <+53>:   vmovd  %edx,%xmm4
   0x085f9249 <+57>:   vmovd  %ebx,%xmm5
   0x085f924d <+61>:   vpshuflw $0x0,%xmm4,%xmm4
   0x085f9252 <+66>:   vpunpcklqdq %xmm4,%xmm4,%xmm4
   0x085f9256 <+70>:   vpshuflw $0x0,%xmm5,%xmm5
   0x085f925b <+75>:   vpunpcklqdq %xmm5,%xmm5,%xmm5
   0x085f925f <+79>:   vpackuswb %xmm4,%xmm4,%xmm4
   0x085f9263 <+83>:   vpackuswb %xmm5,%xmm5,%xmm5
   0x085f9267 <+87>:   vpsubusb %xmm1,%xmm2,%xmm6
   0x085f926b <+91>:   vpsubusb %xmm2,%xmm1,%xmm7
   0x085f926f <+95>:   vpor   %xmm6,%xmm7,%xmm7
   0x085f9273 <+99>:   vpsubusb %xmm4,%xmm7,%xmm7
   0x085f9277 <+103>:  vpsubusb %xmm0,%xmm1,%xmm6
   0x085f927b <+107>:  vpsubusb %xmm1,%xmm0,%xmm4
   0x085f927f <+111>:  vpor   %xmm6,%xmm4,%xmm4
   0x085f9283 <+115>:  vpsubusb %xmm5,%xmm4,%xmm4
   0x085f9287 <+119>:  vpor   %xmm4,%xmm7,%xmm7
   0x085f928b <+123>:  vpsubusb %xmm3,%xmm2,%xmm6
   0x085f928f <+127>:  vpsubusb %xmm2,%xmm3,%xmm4
---Type <return> to continue, or q <return> to quit---q
Quit
```

It seems some data structure was misaligned.
What I have done is the file attached `patch-clang-handbrake-i386`.
I tried to pass -mstackrealign option to clang,
but nothing changed.
It's hard for me to fix this problem.

-- 
You are receiving this mail because:
You are the assignee for the bug.