#3451(swscale:new): sws_scale crashes in high resolutions when using mmx optimization
#3451: sws_scale crashes in high resolutions when using mmx optimization -------------------------------------+------------------------------------- Reporter: miro82 | Type: defect Status: new | Priority: normal Component: swscale | Version: Keywords: 4k vertical | unspecified flip mmx | Blocked By: Blocking: | Reproduced by developer: 0 Analyzed by developer: 0 | -------------------------------------+------------------------------------- Hi, sws_scale crashes in high resolutions (4k & 8k) when using mmx optimization and a negative stride for the input frame. I'm developing a render engine that uses openGL to render hi-res video for planetariums. OpenGL is storing the pixel information vertically flipped compared to FFmpeg so I'm flipping the frame in sws_scale. The code i'm using is: {{{ // Fill picture with image // pixels is a unsigned char buffer avpicture_fill((AVPicture*)tmpFrame, pixels, PIX_FMT_BGR24, mWidth, mHeight); // Flipping frame tmpFrame->data[0] += tmpFrame->linesize[0]*(mHeight-1); // Flipping frame tmpFrame->linesize[0] = -tmpFrame->linesize[0]; //convert BGR24 to YUV420 int ret = sws_scale(sContext, tmpFrame->data, tmpFrame->linesize, 0, mHeight, outFrame->data, outFrame->linesize); }}} The code above works fine for resolutions up to 2k but crashes if I go above that. I'm using XCode and the trace looks like this: Error line: 1629 rgb2rgb_template.c Thread 1, Queue : com.apple.main-thread #0 0x0000000101a75ec8 in rgb24toyv12_mmxext at /Users/miran46/Downloads/ffmpeg/libswscale/x86/rgb2rgb_template.c:1629 #1 0x0000000101a47561 in bgr24ToYv12Wrapper at /Users/miran46/Downloads/ffmpeg/libswscale/swscale_unscaled.c:1314 #2 0x0000000101a416c8 in sws_scale at /Users/miran46/Downloads/ffmpeg/libswscale/swscale.c:1075 #3 0x0000000100006b64 in Encoder::addFrame(unsigned char*) at /Users/miran46/code/projects/FFMpegCapture/hmap/Encoder.cpp:160 This problem is resolved by building FFmpeg with: --disable-mmx When disabling mmx I have no problems to encode 4096 x 4096 HEVC video. The SwsContext is inited using the following code: {{{ sContext = sws_getContext(mWidth, mHeight, AV_PIX_FMT_BGR24, mWidth, mHeight, AV_PIX_FMT_YUV420P, SWS_FAST_BILINEAR, NULL, NULL, NULL); }}} Best regards, Miroslav -- Ticket URL: <https://trac.ffmpeg.org/ticket/3451> FFmpeg <https://ffmpeg.org> FFmpeg issue tracker
#3451: sws_scale crashes in high resolutions when using mmx optimization -------------------------------------+----------------------------------- Reporter: miro82 | Owner: Type: defect | Status: new Priority: normal | Component: swscale Version: unspecified | Resolution: Keywords: crash | Blocked By: Blocking: | Reproduced by developer: 0 Analyzed by developer: 0 | -------------------------------------+----------------------------------- Changes (by cehoyos): * keywords: 4k vertical flip mmx => crash Comment: Is the crash reproducible with {{{ffmpeg}}} (the application)? If not, please provide source code that allows to reproduce the problem. Apart from the backtrace, please also provide disassembly and register content as explained on http://ffmpeg.org/bugreports.html And please add the output of {{{ffmpeg}}} just to post version information and configure line. -- Ticket URL: <https://trac.ffmpeg.org/ticket/3451#comment:1> FFmpeg <https://ffmpeg.org> FFmpeg issue tracker
#3451: sws_scale crashes in high resolutions when using mmx optimization -------------------------------------+----------------------------------- Reporter: miro82 | Owner: Type: defect | Status: new Priority: normal | Component: swscale Version: unspecified | Resolution: Keywords: crash | Blocked By: Blocking: | Reproduced by developer: 0 Analyzed by developer: 0 | -------------------------------------+----------------------------------- Comment (by miro82): I'm not using the ffmpeg application. The crash seems only to occur in OS X. Tried the same code (x64) in Visual Studio 2013 without any problems. I have simplified the code for reproducing the problem and attached the files. In this case I'm only trying to flip, convert and encode a single 3840x2160 frame. I'm don't have gdb but here is the output from lldb: {{{ Miroslavs-MacBook-Pro:bin miran46$ lldb mmx_test2 Current executable set to 'mmx_test2' (x86_64). (lldb) r Process 78369 launched: '/Users/miran46/code/projects/FFMpegCapture/mmx_test2_build/bin/mmx_test2' (x86_64) x265 [info]: using cpu capabilities: MMX2 SSE2Fast SSSE3 SSE4.2 AVX AVX2 FMA3 LZCNT BMI2 x265 [info]: Main profile, Level-5.1 (High tier) x265 [info]: WPP streams / pool / frames : 34 / 4 / 1 x265 [info]: CU size : 64 x265 [info]: Max RQT depth inter / intra : 1 / 1 x265 [info]: ME / range / subpel / merge : hex / 57 / 2 / 2 x265 [info]: Keyframe min / max / scenecut : 25 / 250 / 40 x265 [info]: Lookahead / bframes / badapt : 20 / 4 / 2 x265 [info]: b-pyramid / weightp / refs : 1 / 1 / 3 x265 [info]: Rate Control / AQ-Strength / CUTree : ABR-104857 kbps / 1.0 / 1 x265 [info]: tools: rect amp rd=3 lft sao-lcu sign-hide Process 78369 stopped * thread #1: tid = 0x13ab7f, 0x0000000100fdcab8 libswscale.2.dylib`___lldb_unnamed_function410$$libswscale.2.dylib + 440, queue = 'com.apple.main-thread', stop reason = EXC_BAD_ACCESS (code=1, address=0x10c203000) frame #0: 0x0000000100fdcab8 libswscale.2.dylib`___lldb_unnamed_function410$$libswscale.2.dylib + 440 libswscale.2.dylib`___lldb_unnamed_function410$$libswscale.2.dylib + 440: -> 0x100fdcab8: movd 0x15(%rbx,%rdx), %mm3 0x100fdcabd: punpcklbw %mm7, %mm2 0x100fdcac0: punpcklbw %mm7, %mm3 0x100fdcac3: pmaddwd %mm6, %mm4 (lldb) bt * thread #1: tid = 0x13ab7f, 0x0000000100fdcab8 libswscale.2.dylib`___lldb_unnamed_function410$$libswscale.2.dylib + 440, queue = 'com.apple.main-thread', stop reason = EXC_BAD_ACCESS (code=1, address=0x10c203000) * frame #0: 0x0000000100fdcab8 libswscale.2.dylib`___lldb_unnamed_function410$$libswscale.2.dylib + 440 frame #1: 0x0000000100fc4d49 libswscale.2.dylib`___lldb_unnamed_function303$$libswscale.2.dylib + 137 frame #2: 0x0000000100fc18b7 libswscale.2.dylib`sws_scale + 2919 frame #3: 0x0000000100004694 mmx_test2`Encoder::addFrame(this=0x0000000100005170, pixels=0x000000010aa48000) + 212 at Encoder.cpp:150 frame #4: 0x00000001000049d9 mmx_test2`main + 345 at main.cpp:21 (lldb) disassemble --pc libswscale.2.dylib`___lldb_unnamed_function410$$libswscale.2.dylib + 440: -> 0x100fdcab8: movd 0x15(%rbx,%rdx), %mm3 0x100fdcabd: punpcklbw %mm7, %mm2 0x100fdcac0: punpcklbw %mm7, %mm3 0x100fdcac3: pmaddwd %mm6, %mm4 (lldb) register read --all General Purpose Registers: rax = 0xfffffffffffffff8 rbx = 0x000000010c203000 rcx = 0x0000000000000000 rdx = 0xffffffffffffffe8 rdi = 0x000000000000086c rsi = 0x0000000000000f00 rbp = 0x0000000000000780 rsp = 0x00007fff5fbff8b0 r8 = 0x0000000000000f00 r9 = 0x0000000000002d00 r10 = 0x0000000000000000 r11 = 0x0000000109e6af00 r12 = 0x00000001083beb40 r13 = 0x000000010a653000 r14 = 0xfffffffffffff100 r15 = 0x000000010c200300 rip = 0x0000000100fdcab8 libswscale.2.dylib`___lldb_unnamed_function410$$libswscale.2.dylib + 440 rflags = 0x0000000000010282 cs = 0x000000000000002b fs = 0x00000000ffff0000 gs = 0x00000000ffff0000 eax = 0xfffffff8 ebx = 0x0c203000 ecx = 0x00000000 edx = 0xffffffe8 edi = 0x0000086c esi = 0x00000f00 ebp = 0x00000780 esp = 0x5fbff8b0 r8d = 0x00000f00 r9d = 0x00002d00 r10d = 0x00000000 r11d = 0x09e6af00 r12d = 0x083beb40 r13d = 0x0a653000 r14d = 0xfffff100 r15d = 0x0c200300 ax = 0xfff8 bx = 0x3000 cx = 0x0000 dx = 0xffe8 di = 0x086c si = 0x0f00 bp = 0x0780 sp = 0xf8b0 r8w = 0x0f00 r9w = 0x2d00 r10w = 0x0000 r11w = 0xaf00 r12w = 0xeb40 r13w = 0x3000 r14w = 0xf100 r15w = 0x0300 ah = 0xff bh = 0x30 ch = 0x00 dh = 0xff al = 0xf8 bl = 0x00 cl = 0x00 dl = 0xe8 dil = 0x6c sil = 0x00 bpl = 0x80 spl = 0xb0 r8l = 0x00 r9l = 0x00 r10l = 0x00 r11l = 0x00 r12l = 0x40 r13l = 0x00 r14l = 0x00 r15l = 0x00 Floating Point Registers: fctrl = 0x037f fstat = 0x0000 ftag = 0xff fop = 0x0000 fioff = 0x00000000 fiseg = 0x0000 fooff = 0x00000000 foseg = 0x0000 mxcsr = 0x00001fa0 mxcsrmask = 0x0000ffff stmm0 = {0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0xff 0xff} stmm1 = {0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0xff 0xff} stmm2 = {0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0xff 0xff} stmm3 = {0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0xff 0xff} stmm4 = {0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0xff 0xff} stmm5 = {0x01 0x00 0x01 0x00 0x01 0x00 0x01 0x00 0xff 0xff} stmm6 = {0x88 0x0c 0x87 0x40 0xde 0x20 0x00 0x00 0xff 0xff} stmm7 = {0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0xff 0xff} ymm0 = {0x00 0x0f 0x00 0x00 0x80 0x07 0x00 0x00 0x80 0x07 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} ymm1 = {0x00 0x2d 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} ymm2 = {0x00 0x00 0x00 0x00 0x00 0x00 0xe0 0x43 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} ymm3 = {0x00 0x00 0x00 0x00 0x00 0x00 0xf0 0x3f 0x03 0x00 0x80 0x00 0x00 0x03 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} ymm4 = {0x00 0x00 0x00 0x00 0x00 0x00 0xf0 0x3f 0x03 0x00 0x80 0x00 0x00 0x03 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} ymm5 = {0x00 0x00 0x00 0x00 0x00 0x00 0xf0 0x3f 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} ymm6 = {0xae 0x72 0x46 0xe8 0x8f 0x1d 0xe4 0x3f 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} ymm7 = {0x6b 0xc8 0xb8 0xbe 0xd3 0xb9 0x0b 0x40 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} ymm8 = {0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} ymm9 = {0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} ymm10 = {0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} ymm11 = {0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} ymm12 = {0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} ymm13 = {0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} ymm14 = {0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} ymm15 = {0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} xmm0 = {0x00 0x0f 0x00 0x00 0x80 0x07 0x00 0x00 0x80 0x07 0x00 0x00 0x00 0x00 0x00 0x00} xmm1 = {0x00 0x2d 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} xmm2 = {0x00 0x00 0x00 0x00 0x00 0x00 0xe0 0x43 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} xmm3 = {0x00 0x00 0x00 0x00 0x00 0x00 0xf0 0x3f 0x03 0x00 0x80 0x00 0x00 0x03 0x00 0x00} xmm4 = {0x00 0x00 0x00 0x00 0x00 0x00 0xf0 0x3f 0x03 0x00 0x80 0x00 0x00 0x03 0x00 0x00} xmm5 = {0x00 0x00 0x00 0x00 0x00 0x00 0xf0 0x3f 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} xmm6 = {0xae 0x72 0x46 0xe8 0x8f 0x1d 0xe4 0x3f 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} xmm7 = {0x6b 0xc8 0xb8 0xbe 0xd3 0xb9 0x0b 0x40 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} xmm8 = {0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} xmm9 = {0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} xmm10 = {0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} xmm11 = {0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} xmm12 = {0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} xmm13 = {0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} xmm14 = {0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} xmm15 = {0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} Exception State Registers: trapno = 0x0000000e err = 0x00000004 faultvaddr = 0x000000010c203000 (lldb) }}} -- Ticket URL: <https://trac.ffmpeg.org/ticket/3451#comment:2> FFmpeg <https://ffmpeg.org> FFmpeg issue tracker
#3451: sws_scale crashes in high resolutions when using mmx optimization -------------------------------------+----------------------------------- Reporter: miro82 | Owner: Type: defect | Status: new Priority: normal | Component: swscale Version: unspecified | Resolution: Keywords: crash | Blocked By: Blocking: | Reproduced by developer: 0 Analyzed by developer: 0 | -------------------------------------+----------------------------------- Comment (by miro82): And my ffmpeg version & configuration is: {{{ ffmpeg version N-61143-g64e4bd7 Copyright (c) 2000-2014 the FFmpeg developers built on Mar 12 2014 21:25:49 with Apple LLVM version 5.1 (clang-503.0.38) (based on LLVM 3.4svn) configuration: --disable-static --enable-shared --disable-yasm --disable-iconv --enable-libx265 --enable-libx264 --enable-gpl libavutil 52. 66.101 / 52. 66.101 libavcodec 55. 52.102 / 55. 52.102 libavformat 55. 33.101 / 55. 33.101 libavdevice 55. 11.100 / 55. 11.100 libavfilter 4. 3.100 / 4. 3.100 libswscale 2. 5.101 / 2. 5.101 libswresample 0. 18.100 / 0. 18.100 libpostproc 52. 3.100 / 52. 3.100 Hyper fast Audio and Video encoder }}} If I configure ffmpeg with --disable-mmx it doesn't crash anymore. -- Ticket URL: <https://trac.ffmpeg.org/ticket/3451#comment:3> FFmpeg <https://ffmpeg.org> FFmpeg issue tracker
#3451: sws_scale crashes in high resolutions when using mmx optimization -------------------------------------+----------------------------------- Reporter: miro82 | Owner: Type: defect | Status: new Priority: normal | Component: swscale Version: unspecified | Resolution: Keywords: crash | Blocked By: Blocking: | Reproduced by developer: 0 Analyzed by developer: 0 | -------------------------------------+----------------------------------- Comment (by cehoyos): Please recompile with {{{--disable-stripping}}} (or {{{--disable-shared}}} to provide a useful backtrace. -- Ticket URL: <https://trac.ffmpeg.org/ticket/3451#comment:4> FFmpeg <https://ffmpeg.org> FFmpeg issue tracker
#3451: sws_scale crashes in high resolutions when using mmx optimization -------------------------------------+----------------------------------- Reporter: miro82 | Owner: Type: defect | Status: new Priority: normal | Component: swscale Version: unspecified | Resolution: Keywords: crash | Blocked By: Blocking: | Reproduced by developer: 0 Analyzed by developer: 0 | -------------------------------------+----------------------------------- Comment (by miro82): I discovered that --enable-shared must be set in order to trigger the crash. This time FFmpeg was configured and build with: {{{ ./configure --disable-yasm --disable-iconv --enable-libx265 --enable- libx264 --enable-gpl --enable-shared --disable-stripping }}} Debug output from the application: {{{ Miroslavs-MacBook-Pro:bin miran46$ lldb mmx_test2 Current executable set to 'mmx_test2' (x86_64). (lldb) r Process 29388 launched: '/Users/miran46/code/projects/FFMpegCapture/mmx_test2_build/bin/mmx_test2' (x86_64) x265 [info]: using cpu capabilities: MMX2 SSE2Fast SSSE3 SSE4.2 AVX AVX2 FMA3 LZCNT BMI2 x265 [info]: Main profile, Level-5.1 (High tier) x265 [info]: WPP streams / pool / frames : 34 / 4 / 1 x265 [info]: CU size : 64 x265 [info]: Max RQT depth inter / intra : 1 / 1 x265 [info]: ME / range / subpel / merge : hex / 57 / 2 / 2 x265 [info]: Keyframe min / max / scenecut : 25 / 250 / 40 x265 [info]: Lookahead / bframes / badapt : 20 / 4 / 2 x265 [info]: b-pyramid / weightp / refs : 1 / 1 / 3 x265 [info]: Rate Control / AQ-Strength / CUTree : ABR-104857 kbps / 1.0 / 1 x265 [info]: tools: rect amp rd=3 lft sao-lcu sign-hide Process 29388 stopped * thread #1: tid = 0x14f790, 0x000000010112e8b8 libswscale.2.dylib`rgb24toyv12_mmxext(src=0x000000010c396300, ydst=0x000000010a000000, udst=0x000000010a7e9000, vdst=0x000000010abdd080, width=3840, height=2160, lumStride=<unavailable>, chromStride=<unavailable>, srcStride=<unavailable>, rgb2yuv=0x0000000109805340) + 440 at rgb2rgb_template.c:1629, queue = 'com.apple.main-thread', stop reason = EXC_BAD_ACCESS (code=1, address=0x10c399000) frame #0: 0x000000010112e8b8 libswscale.2.dylib`rgb24toyv12_mmxext(src=0x000000010c396300, ydst=0x000000010a000000, udst=0x000000010a7e9000, vdst=0x000000010abdd080, width=3840, height=2160, lumStride=<unavailable>, chromStride=<unavailable>, srcStride=<unavailable>, rgb2yuv=0x0000000109805340) + 440 at rgb2rgb_template.c:1629 1626 for (y=0; y<height-2; y+=2) { 1627 int i; 1628 for (i=0; i<2; i++) { -> 1629 __asm__ volatile( 1630 "mov %2, %%"REG_a" \n\t" 1631 "movq "BGR2Y_IDX"(%3), %%mm6 \n\t" 1632 "movq "MANGLE(ff_w1111)", %%mm5 \n\t" (lldb) bt * thread #1: tid = 0x14f790, 0x000000010112e8b8 libswscale.2.dylib`rgb24toyv12_mmxext(src=0x000000010c396300, ydst=0x000000010a000000, udst=0x000000010a7e9000, vdst=0x000000010abdd080, width=3840, height=2160, lumStride=<unavailable>, chromStride=<unavailable>, srcStride=<unavailable>, rgb2yuv=0x0000000109805340) + 440 at rgb2rgb_template.c:1629, queue = 'com.apple.main-thread', stop reason = EXC_BAD_ACCESS (code=1, address=0x10c399000) * frame #0: 0x000000010112e8b8 libswscale.2.dylib`rgb24toyv12_mmxext(src=0x000000010c396300, ydst=0x000000010a000000, udst=0x000000010a7e9000, vdst=0x000000010abdd080, width=3840, height=2160, lumStride=<unavailable>, chromStride=<unavailable>, srcStride=<unavailable>, rgb2yuv=0x0000000109805340) + 440 at rgb2rgb_template.c:1629 frame #1: 0x0000000101116b49 libswscale.2.dylib`bgr24ToYv12Wrapper(c=0x0000000109801200, src=<unavailable>, srcStride=<unavailable>, srcSliceY=0, srcSliceH=2160, dst=0x00007fff5fbffa00, dstStride=0x00007fff5fbff9e0) + 137 at swscale_unscaled.c:1314 frame #2: 0x00000001011136b7 libswscale.2.dylib`sws_scale(c=<unavailable>, srcSlice=<unavailable>, srcStride=<unavailable>, srcSliceY=<unavailable>, srcSliceH=<unavailable>, dst=<unavailable>, dstStride=<unavailable>) + 2919 at swscale.c:1101 frame #3: 0x0000000100004894 mmx_test2`Encoder::addFrame(unsigned char*) + 212 frame #4: 0x0000000100003769 mmx_test2`main + 345 (lldb) disassemble --pc libswscale.2.dylib`rgb24toyv12_mmxext + 440 at rgb2rgb_template.c:1629: -> 0x10112e8b8: movd 0x15(%rbx,%rdx), %mm3 0x10112e8bd: punpcklbw %mm7, %mm2 0x10112e8c0: punpcklbw %mm7, %mm3 0x10112e8c3: pmaddwd %mm6, %mm4 (lldb) info all-registers error: 'info' is not a valid command. (lldb) register read --all General Purpose Registers: rax = 0xfffffffffffffff8 rbx = 0x000000010c399000 rcx = 0x0000000000000000 rdx = 0xffffffffffffffe8 rdi = 0x000000000000086c rsi = 0x0000000000000f00 rbp = 0x0000000000000780 rsp = 0x00007fff5fbff8b0 r8 = 0x0000000000000f00 r9 = 0x0000000000002d00 r10 = 0x0000000000000000 r11 = 0x000000010a000f00 r12 = 0x0000000109805340 r13 = 0x000000010a7e9000 r14 = 0xfffffffffffff100 r15 = 0x000000010c396300 rip = 0x000000010112e8b8 libswscale.2.dylib`rgb24toyv12_mmxext + 440 at rgb2rgb_template.c:1629 rflags = 0x0000000000010282 cs = 0x000000000000002b fs = 0x00000000ffff0000 gs = 0x00000000ffff0000 eax = 0xfffffff8 ebx = 0x0c399000 ecx = 0x00000000 edx = 0xffffffe8 edi = 0x0000086c esi = 0x00000f00 ebp = 0x00000780 esp = 0x5fbff8b0 r8d = 0x00000f00 r9d = 0x00002d00 r10d = 0x00000000 r11d = 0x0a000f00 r12d = 0x09805340 r13d = 0x0a7e9000 r14d = 0xfffff100 r15d = 0x0c396300 ax = 0xfff8 bx = 0x9000 cx = 0x0000 dx = 0xffe8 di = 0x086c si = 0x0f00 bp = 0x0780 sp = 0xf8b0 r8w = 0x0f00 r9w = 0x2d00 r10w = 0x0000 r11w = 0x0f00 r12w = 0x5340 r13w = 0x9000 r14w = 0xf100 r15w = 0x6300 ah = 0xff bh = 0x90 ch = 0x00 dh = 0xff al = 0xf8 bl = 0x00 cl = 0x00 dl = 0xe8 dil = 0x6c sil = 0x00 bpl = 0x80 spl = 0xb0 r8l = 0x00 r9l = 0x00 r10l = 0x00 r11l = 0x00 r12l = 0x40 r13l = 0x00 r14l = 0x00 r15l = 0x00 Floating Point Registers: fctrl = 0x037f fstat = 0x0000 ftag = 0xff fop = 0x0000 fioff = 0x00000000 fiseg = 0x0000 fooff = 0x00000000 foseg = 0x0000 mxcsr = 0x00001fa0 mxcsrmask = 0x0000ffff stmm0 = {0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0xff 0xff} stmm1 = {0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0xff 0xff} stmm2 = {0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0xff 0xff} stmm3 = {0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0xff 0xff} stmm4 = {0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0xff 0xff} stmm5 = {0x01 0x00 0x01 0x00 0x01 0x00 0x01 0x00 0xff 0xff} stmm6 = {0x88 0x0c 0x87 0x40 0xde 0x20 0x00 0x00 0xff 0xff} stmm7 = {0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0xff 0xff} ymm0 = {0x00 0x0f 0x00 0x00 0x80 0x07 0x00 0x00 0x80 0x07 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} ymm1 = {0x00 0x2d 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} ymm2 = {0x00 0x00 0x00 0x00 0x00 0x00 0xe0 0x43 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} ymm3 = {0x66 0xa9 0x49 0x15 0x00 0x00 0x00 0x10 0xdf 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} ymm4 = {0x6b 0xae 0x54 0x16 0x00 0x00 0x00 0x10 0x30 0x01 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} ymm5 = {0x00 0x00 0x00 0x00 0x00 0x00 0xf0 0x3f 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} ymm6 = {0xae 0x72 0x46 0xe8 0x8f 0x1d 0xe4 0x3f 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} ymm7 = {0x6b 0xc8 0xb8 0xbe 0xd3 0xb9 0x0b 0x40 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} ymm8 = {0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} ymm9 = {0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} ymm10 = {0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} ymm11 = {0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} ymm12 = {0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} ymm13 = {0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} ymm14 = {0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} ymm15 = {0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} xmm0 = {0x00 0x0f 0x00 0x00 0x80 0x07 0x00 0x00 0x80 0x07 0x00 0x00 0x00 0x00 0x00 0x00} xmm1 = {0x00 0x2d 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} xmm2 = {0x00 0x00 0x00 0x00 0x00 0x00 0xe0 0x43 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} xmm3 = {0x66 0xa9 0x49 0x15 0x00 0x00 0x00 0x10 0xdf 0x00 0x00 0x00 0x00 0x00 0x00 0x00} xmm4 = {0x6b 0xae 0x54 0x16 0x00 0x00 0x00 0x10 0x30 0x01 0x00 0x00 0x00 0x00 0x00 0x00} xmm5 = {0x00 0x00 0x00 0x00 0x00 0x00 0xf0 0x3f 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} xmm6 = {0xae 0x72 0x46 0xe8 0x8f 0x1d 0xe4 0x3f 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} xmm7 = {0x6b 0xc8 0xb8 0xbe 0xd3 0xb9 0x0b 0x40 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} xmm8 = {0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} xmm9 = {0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} xmm10 = {0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} xmm11 = {0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} xmm12 = {0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} xmm13 = {0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} xmm14 = {0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} xmm15 = {0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00} Exception State Registers: trapno = 0x0000000e err = 0x00000004 faultvaddr = 0x000000010c399000 (lldb) }}} If I simplify my code and remove all FFmpeg encoding and file-writing code then the application doesn't crash. The code below works while the attached code don't. {{{ extern "C" { #ifndef __STDC_CONSTANT_MACROS #define __STDC_CONSTANT_MACROS #endif #include <libavcodec/avcodec.h> #include <libavutil/imgutils.h> #include <libavformat/avformat.h> #include <libswscale/swscale.h> } #include <new> int main() { fprintf(stderr, "Running mmx test\n"); av_register_all(); int mWidth = 3840; int mHeight = 2160; int ret; uint8_t * pixels = new (std::nothrow) uint8_t[mWidth * mHeight * 3]; //fill buffer with zeros memset(pixels, 0, mWidth * mHeight * 3); SwsContext * sContext = NULL; //create context for frame convertion sContext = sws_getContext(mWidth, mHeight, AV_PIX_FMT_BGR24, mWidth, mHeight, AV_PIX_FMT_YUV420P, SWS_FAST_BILINEAR, NULL, NULL, NULL); if (!sContext) { fprintf(stderr, "Could not allocate frame convertion context!\n"); return EXIT_FAILURE; } //allocate frames AVFrame * tmpFrame = NULL; AVFrame * outFrame = NULL; outFrame = av_frame_alloc(); if (!outFrame) { fprintf(stderr, "Could not create output frame\n"); return false; } outFrame->format = PIX_FMT_YUV420P; outFrame->width = mWidth; outFrame->height = mHeight; ret = av_image_alloc(outFrame->data, outFrame->linesize, mWidth, mHeight, PIX_FMT_YUV420P, 32); if (ret < 0) { fprintf(stderr, "Could not allocate output frame\n"); return false; } tmpFrame = av_frame_alloc(); if (!tmpFrame) { fprintf(stderr, "Could not create swap frame\n"); return false; } tmpFrame->width = mWidth; tmpFrame->height = mHeight; tmpFrame->format = PIX_FMT_BGR24; //fill with random values //for(unsigned int i=0; i<mWidth * mHeight * 3;i++) // pixels[i]= static_cast<uint8_t>( rand()%256 ); //convert // Fill picture with image avpicture_fill((AVPicture*)tmpFrame, pixels, PIX_FMT_BGR24, mWidth, mHeight); // Flipping frame tmpFrame->data[0] += tmpFrame->linesize[0]*(mHeight-1); // Flipping frame tmpFrame->linesize[0] = -tmpFrame->linesize[0]; fprintf(stderr, "Converting to YUV420\n"); //convert BGR24 to YUV420 ret = sws_scale(sContext, tmpFrame->data, tmpFrame->linesize, 0, mHeight, outFrame->data, outFrame->linesize); if (ret < 0) { fprintf(stderr, "Failed to convert frame to YUV420!\n"); } //Cleanup if (tmpFrame) { av_frame_free(&tmpFrame); } if (outFrame) { av_freep(&outFrame->data[0]); av_frame_free(&outFrame); } if (sContext) sws_freeContext(sContext); delete [] pixels; // Exit program exit( EXIT_SUCCESS ); } }}} -- Ticket URL: <https://trac.ffmpeg.org/ticket/3451#comment:5> FFmpeg <https://ffmpeg.org> FFmpeg issue tracker
#3451: sws_scale crashes in high resolutions when using mmx optimization -------------------------------------+----------------------------------- Reporter: miro82 | Owner: Type: defect | Status: new Priority: important | Component: swscale Version: unspecified | Resolution: Keywords: crash osx | Blocked By: Blocking: | Reproduced by developer: 0 Analyzed by developer: 0 | -------------------------------------+----------------------------------- Changes (by cehoyos): * keywords: crash => crash osx * priority: normal => important Comment: Do you know if this is still reproducible? I don't have an OSX box and last time I had access, I forgot to test this. -- Ticket URL: <https://trac.ffmpeg.org/ticket/3451#comment:6> FFmpeg <https://ffmpeg.org> FFmpeg issue tracker
#3451: sws_scale crashes in high resolutions when using mmx optimization -------------------------------------+----------------------------------- Reporter: miro82 | Owner: Type: defect | Status: new Priority: important | Component: swscale Version: unspecified | Resolution: Keywords: crash osx | Blocked By: Blocking: | Reproduced by developer: 0 Analyzed by developer: 0 | -------------------------------------+----------------------------------- Changes (by gjdfgh): * cc: nfxjfg@… (added) Comment:
OpenGL is storing the pixel information vertically flipped compared to FFmpeg so I'm flipping the frame in sws_scale.
Uh, what? You can flip it on OpenGL with no additional effort. Just play with the texture coordinates. -- Ticket URL: <https://trac.ffmpeg.org/ticket/3451#comment:7> FFmpeg <https://ffmpeg.org> FFmpeg issue tracker
#3451: sws_scale crashes in high resolutions when using mmx optimization -------------------------------------+----------------------------------- Reporter: miro82 | Owner: Type: defect | Status: closed Priority: important | Component: swscale Version: unspecified | Resolution: fixed Keywords: crash osx | Blocked By: Blocking: | Reproduced by developer: 0 Analyzed by developer: 0 | -------------------------------------+----------------------------------- Changes (by michael): * status: new => closed * resolution: => fixed Comment: Should be fixed in 4388e78a0f022c8572996f9ab568a39b5f716f9d please reopen if not -- Ticket URL: <https://trac.ffmpeg.org/ticket/3451#comment:8> FFmpeg <https://ffmpeg.org> FFmpeg issue tracker
participants (1)
-
FFmpeg