Open picarica opened 4 months ago
i thought maybe the issue this is a laptop so i tried using optimus but same thing
00:00:19 - FFmpeg: [h264 @ 0x7fd322801f50] Decoding frame, 18093 bytes, 1 slices
/usr/bin/prime-run: line 2: 4360 Segmentation fault (core dumped) __NV_PRIME_RENDER_OFFLOAD=1 __VK_LAYER_NV_optimus=NVIDIA_only __GLX_VENDOR_LIBRARY_NAME=nvidia "$@"
Can you run it under gdb
and provide the output of the bt
command after it crashes?
Please also provide the version of ffmpeg/libavcodec and libplacebo that you used when compiling Moonlight. You will probably need the very latest libplacebo and ffmpeg releases for things to work properly as Vulkan Video support is still maturing in libraries and drivers.
gdb-moonlight.log o got the gdb output, but i am confused wdym about that bt command? not sure i am familiar with it
here is my ffmpeg build flags and also version 6.1.1 ffmpegbuild.log
and i believe libvacodec is bundled with ffmpeg in gentoo so i have it probably
/usr/lib64/libavcodec.so
/usr/lib64/libavcodec.so.60.31.102
/usr/lib64/libavcodec.so.60
/usr/lib64/pkgconfig/libavcodec.pc
/usr/lib64/vlc/plugins/codec/libavcodec_plugin.so
/usr/share/man/man3/libavcodec.3.bz2
/usr/lib/libavcodec.so
/usr/lib/libavcodec.so.60.31.102
/usr/lib/libavcodec.so.60
/usr/lib/pkgconfig/libavcodec.pc
/usr/include/libavcodec
i mean some vulkan gmes are working fine like gmod or tf2
I'am also having crash on amd laptop with gtx 3050:
00:01:22 - FFmpeg: [hevc @ 0x72e52018a580] Allocating 4096 bytes in bind index 0 for video session 00:01:22 - FFmpeg: [hevc @ 0x72e52018a580] Allocating 16384 bytes in bind index 1 for video session 00:01:22 - FFmpeg: [hevc @ 0x72e52018a580] Allocating 71303168 bytes in bind index 2 for video session 00:01:22 - FFmpeg: [hevc @ 0x72e52018a580] Allocating 146931712 bytes in bind index 3 for video session 00:01:22 - FFmpeg: [hevc @ 0x72e52018a580] Vulkan decoder initialization sucessful 00:01:22 - FFmpeg: [hevc @ 0x72e52018a580] Output frame with POC 0. 00:01:22 - FFmpeg: [hevc @ 0x72e52018a580] Created frame parameters: 1 SPS 1 PPS 1 VPS 00:01:22 - FFmpeg: [hevc @ 0x72e52018a580] Decoding frame, 30129 bytes, 1 slices fish: Job 1, 'moonlight' terminated by signal SIGSEGV (Address boundary error)
OS: CachyOS, moonlight 6.0.1, ffmpeg 7.0.1, wayland
P.s. in x11 harware decoding works
CachyOS
intresting for me in X11 it doesnt work, ill try wayland
Also, hw decoding works in flathub version
Snaps build on edge
have just this week added vulkan/libplacebo support, and started crashing with seg fault in hardware video decoding (see https://github.com/maxiberta/moonlight-snap/issues/21). Software video decoding works fine. Right before crashing, the new stream window pops up for a split second and a notification sound from Sunshine can sometimes be heard.
$ snap run --gdbserver moonlight stream PowerStation2 Desktop
...
Thread 7 "FFDecoder" received signal SIGSEGV, Segmentation fault.
[Switching to Thread 1716777.1717357]
0x00007d87d9c2463b in ?? () from target:/snap/moonlight/2528/usr/lib/libavcodec.so.61
(gdb) bt
#0 0x00007d87d9c2463b in ?? () from target:/snap/moonlight/2528/usr/lib/libavcodec.so.61
#1 0x00007d87d9c2ae2c in ?? () from target:/snap/moonlight/2528/usr/lib/libavcodec.so.61
#2 0x00007d87d9c31d5a in ?? () from target:/snap/moonlight/2528/usr/lib/libavcodec.so.61
#3 0x00007d87d9ba89d8 in ?? () from target:/snap/moonlight/2528/usr/lib/libavcodec.so.61
#4 0x00007d87d9aa330b in ?? () from target:/snap/moonlight/2528/usr/lib/libavcodec.so.61
#5 0x00007d87d9aa39da in avcodec_send_packet () from target:/snap/moonlight/2528/usr/lib/libavcodec.so.61
#6 0x00005732e115017b in FFmpegVideoDecoder::submitDecodeUnit (this=0x7d878405ed40, du=0x7d86f00008e0) at streaming/video/ffmpeg.cpp:1803
#7 0x00005732e11507a3 in FFmpegVideoDecoder::decoderThreadProc (this=this@entry=0x7d878405ed40) at streaming/video/ffmpeg.cpp:1578
#8 0x00005732e11508bd in FFmpegVideoDecoder::decoderThreadProcThunk (context=context@entry=0x7d878405ed40) at streaming/video/ffmpeg.cpp:1560
#9 0x00007d87d9e4788c in SDL_RunThread (thread=0x7d87842bed90) at /root/parts/sdl/build/src/thread/SDL_thread.c:333
#10 0x00007d87d9eef62d in RunThread (data=<optimized out>) at /root/parts/sdl/build/src/thread/pthread/SDL_systhread.c:76
#11 0x00007d87d6e27ac3 in ?? () from target:/lib/x86_64-linux-gnu/libc.so.6
#12 0x00007d87d6eb9850 in ?? () from target:/lib/x86_64-linux-gnu/libc.so.6
(gdb)
I've also tried https://github.com/cgutman/FFmpeg.git branch moonlight_7_0_1
(instead of upstream) with same result.
I think you're hitting https://trac.ffmpeg.org/ticket/10847 which is fixed upstream by https://git.videolan.org/?p=ffmpeg.git;a=commitdiff;h=ca591e6b50ee616fc36711cc66d594e51209dddb
@maxiberta I pushed moonlight_7_0_2
to https://github.com/cgutman/FFmpeg.git with this fix cherry-picked. See if that resolves the crash in the snap.
@cgutman yes, it works now!! Thanks a lot.
how would i try the fix in linux ? download this ffmpeg library somehow and tell moonlight to use it ?
Describe the bug moonlight crashed
Steps to reproduce launch install
Moonlight settings (please complete the following information)
Client PC details (please complete the following information)
Server PC details (please complete the following information)
Moonlight Logs (please attach)
Additional context not sure why does this , itred same on nvidia quadro k2200 gpu, is this issue with ffmpeg? maybe i have it compiled with abd options but i can encode and decode using ffmpeg using gpu cuda acceleration