Hi,<br><br>I was also able to reproduce the fault.<br><br>Seems like
i_src_stride is corrupt when mc_luma is called. The fault seems to lie
in the mc_weight asm functions for NEON, since the attached patch seems
to remove the problem (I'm not suggesting to apply that patch, just
pointing out the problem). <br>I'll try to look into where exactly things go wrong, if no one suspects something particular already.<br><br>//Stefan<br><br><div class="gmail_quote">On Sun, Mar 3, 2013 at 8:10 AM, Marat Dukhan <span dir="ltr"><<a href="mailto:maratek@gmail.com" target="_blank">maratek@gmail.com</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div class="im"><div>Program received signal SIGSEGV, Segmentation fault.</div><div>0x00077694 in x264_pixel_avg2_w8_neon ()</div>
</div><div>(gdb) bt</div><div>#0 0x00077694 in x264_pixel_avg2_w8_neon ()</div><div>#1 0x000736d2 in mc_luma_neon (</div>
<div> dst=0x13b618 "\320\311\326\335\322\316\330", <incomplete sequence \326>,</div><div> i_dst_stride=32, src=<optimized out>, i_src_stride=<optimized out>,</div><div> mvx=-327, mvy=151, i_width=8, i_height=8, weight=0x137890)</div>
<div> at common/arm/mc-c.c:140</div><div>#2 0x0005f452 in x264_me_refine_qpel_rd (h=0x136600, m=0xbeffaf80,</div><div> i_lambda2=2925, i4=12, i_list=0) at encoder/me.c:1273</div><div>#3 0x00057288 in x264_macroblock_analyse (h=0x136600)</div>
<div> at encoder/analyse.c:3378</div><div>#4 0x00020828 in x264_slice_write (h=0x136600) at encoder/encoder.c:2351</div><div>#5 0x000213fa in x264_slices_write (h=0x136600) at encoder/encoder.c:2667</div><div>#6 0x00024532 in x264_encoder_encode (h=<optimized out>, pp_nal=0xbeffed10,</div>
<div> pi_nal=0xbeffed14, pic_in=<optimized out>, pic_out=0xbeffec50)</div><div> at encoder/encoder.c:3273</div><div>#7 0x0000ac88 in encode_frame (h=<optimized out>, hout=0x111008,</div><div> pic=0xbeffed80, last_dts=0xbeffeef0) at x264.c:1674</div>
<div>#8 0x0000c222 in encode (opt=0xbeffeeb8, param=0xbeffef08) at x264.c:1839</div><div>#9 main (argc=<optimized out>, argv=<optimized out>) at x264.c:295</div><span class="HOEnZb"><font color="#888888"><div>
<br></div><div>Marat</div></font></span><div class="HOEnZb"><div class="h5"><br><div class="gmail_quote">
On Sun, Mar 3, 2013 at 12:57 AM, Jason Garrett-Glaser <span dir="ltr"><<a href="mailto:jason@x264.com" target="_blank">jason@x264.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<div>On Sat, Mar 2, 2013 at 9:42 PM, Marat Dukhan <<a href="mailto:maratek@gmail.com" target="_blank">maratek@gmail.com</a>> wrote:<br>
> It doesn't crash if I configure with --disable-asm.<br>
> Still crashes if I configure with --extra-cflags="-O0 -mfpu=vfpv3".<br>
<br>
</div>Can you post a backtrace?<br>
<div><div><br>
Jason<br>
_______________________________________________<br>
x264-devel mailing list<br>
<a href="mailto:x264-devel@videolan.org" target="_blank">x264-devel@videolan.org</a><br>
<a href="http://mailman.videolan.org/listinfo/x264-devel" target="_blank">http://mailman.videolan.org/listinfo/x264-devel</a><br>
</div></div></blockquote></div><br>
</div></div><br>_______________________________________________<br>
x264-devel mailing list<br>
<a href="mailto:x264-devel@videolan.org">x264-devel@videolan.org</a><br>
<a href="http://mailman.videolan.org/listinfo/x264-devel" target="_blank">http://mailman.videolan.org/listinfo/x264-devel</a><br>
<br></blockquote></div><br>