Ffmpeg segfault

My Channels DVR server version 2021.08.12.0054 is running under CentOS 8.4 on a 4 core / AMD A10-4600M APU.

I do not use Channels DVR outside my local network.

Whenever Channels DVR starts I see the following in /var/log/messages
Aug 12 01:28:24 myhost kernel: ffmpeg[31046]: segfault at 0 ip 0000000001976cf9 sp
00007ffdc5eec710 error 4 in ffmpeg[400000+48e7000]
Aug 12 01:28:24 myhost abrt-hook-ccpp[31049]: Process 31046 (ffmpeg) of user 1000 killed by
SIGSEGV - dumping core

The provided dynamically linked ffmpeg-dl finds all the relevant libraries:
2021.08.12.0054/ffmpeg-dl:
linux-vdso.so.1 (0x00007ffdd25fc000)
libm.so.6 => /lib64/libm.so.6 (0x00007f1470ad0000)
libpthread.so.0 => /lib64/libpthread.so.0 (0x00007f14708b0000)
libdl.so.2 => /lib64/libdl.so.2 (0x00007f14706a8000)
libc.so.6 => /lib64/libc.so.6 (0x00007f14702e0000)
/lib64/ld-linux-x86-64.so.2 (0x00007f1470e58000)

Currently there are certain streams that will not start from the guide when attempting to view on an iPad.

Is this happening only with hardware transcoding enabled?

Thank you for your prompt response.

Apologies, but I do not see a hardware transcoding switch on the 'Settings' page.

The segfault occurs when channels-dvr starts. From the abrt output, the command-line is

/usr/local/DVR/channels-dvr/2021.08.12.0054/ffmpeg -hide_banner -nostats -loglevel warning -loglevel quiet -init_hw_device vaapi=intel:/dev/dri/renderD128 -f lavfi -t 0.1 -i color=black:640x480 -f null -y /dev/null

It is using the statically linked ffmpeg.

Note: I'm using an AMD apu. Would it be more appropriate to use the arg: vaapi=amd:/dev/dri/renderD128?

I'm a novice in the video field.
Thanks!

Okay I see. If its just during boot then its harmless. Its probing to see what hardware options are available.

1 Like

Thanks!