g_object_set and thread safetry

[Nicolas Dufresne]

Le ven. 2 oct. 2020 08 h 15, Guy Morand <g.morand at scewo.ch> a écrit :

> Hallo gstreamers developers,
>
> I'm writing a small daemon to stream the camera and send the stream
> over RTP. The user can configure the stream and mirror the image. To
> do that, while the stream is running, I make a call from the main
> thread similar to:
>
> g_object_set(_video_flip, "video-direction", GST_VIDEO_FLIP_HORIZONTAL,
> NULL);
>
> Where _video_flip is a GstElement* built with
> gst_element_factory_make("videoflip", "videoflip")
>
> Sometime, the stream stops and the application "crashes" with the message:
> ERROR:gstvideoflip.c:954:gst_video_flip_y422: code should not be reached
>
> Back trace with gdb gives me this relevant gstreamer stack trace:
> __GI_raise (sig=sig at entry=0x6) at ../sysdeps/unix/sysv/linux/raise.c:50
> 50 ../sysdeps/unix/sysv/linux/raise.c: No such file or directory.
> gdb$ bt
> #0  0x00007ffff6d597bb in __GI_raise (sig=sig at entry=0x6) at
> ../sysdeps/unix/sysv/linux/raise.c:50
> #1  0x00007ffff6d44535 in __GI_abort () at abort.c:79
> #2  0x00007ffff7480dc3 in  () at /usr/lib/x86_64-linux-gnu/libglib-2.0.so.0
> #3  0x00007ffff74da67a in g_assertion_message_expr () at
> /usr/lib/x86_64-linux-gnu/libglib-2.0.so.0
> #4  0x00007ffff4c4f5ef in gst_video_flip_y422 (videoflip=<optimized
> out>, dest=<optimized out>, src=<optimized out>) at gstvideoflip.c:954
> #5  0x00007ffff4c4d893 in gst_video_flip_transform_frame
> (vfilter=0x555555858790, in_frame=0x7ffff1470530,
> out_frame=0x7ffff14707d0) at gstvideoflip.c:1139
> #6  0x00007ffff4ba9e03 in  () at
> /usr/lib/x86_64-linux-gnu/libgstvideo-1.0.so.0
> #7  0x00007ffff4b3c609 in default_generate_output
> (trans=0x555555858790, outbuf=0x7ffff1470ae0) at
> gstbasetransform.c:2132
> #8  0x00007ffff4b3bcd6 in gst_base_transform_chain (pad=<optimized
> out>, parent=0x555555858790, buffer=<optimized out>) at
> gstbasetransform.c:2285
> #9  0x00007ffff7651c3a in gst_pad_chain_data_unchecked
> (data=0x7fffd401b190, type=4112, pad=0x555555852780) at gstpad.c:4322
> #10 0x00007ffff7651c3a in gst_pad_push_data
> (pad=pad at entry=0x555555853560, type=type at entry=4112,
> data=data at entry=0x7fffd401b190) at gstpad.c:4578
> #11 0x00007ffff7659ed2 in gst_pad_push (pad=0x555555853560,
> buffer=0x7fffd401b190) at gstpad.c:4697
> #12 0x00007ffff4b3bdfd in gst_base_transform_chain (pad=<optimized
> out>, parent=0x5555558602c0, buffer=<optimized out>) at
> gstbasetransform.c:2321
> #13 0x00007ffff7651c3a in gst_pad_chain_data_unchecked
> (data=0x7fffd401b190, type=4112, pad=0x555555853310) at gstpad.c:4322
> #14 0x00007ffff7651c3a in gst_pad_push_data
> (pad=pad at entry=0x555555852090, type=type at entry=4112,
> data=data at entry=0x7fffd401b190) at gstpad.c:4578
> #15 0x00007ffff7659ed2 in gst_pad_push (pad=pad at entry=0x555555852090,
> buffer=0x7fffd401b190) at gstpad.c:4697
> #16 0x00007ffff4b37ce5 in gst_base_src_loop (pad=0x555555852090) at
> gstbasesrc.c:2957
> #17 0x00007ffff7686f41 in gst_task_func (task=0x55555588c290) at
> gsttask.c:332
> #18 0x00007ffff74dbdb3 in  () at /usr/lib/x86_64-linux-gnu/libglib-2.0.so.0
> #19 0x00007ffff74db415 in  () at /usr/lib/x86_64-linux-gnu/libglib-2.0.so.0
> #20 0x00007ffff720dfa3 in start_thread (arg=<optimized out>) at
> pthread_create.c:486
> #21 0x00007ffff6e1b4cf in clone () at
> ../sysdeps/unix/sysv/linux/x86_64/clone.S:95
>
> I'm wondering if that is a thread safety issue and if I'm allowed to
> do it this way. Any pointer?
>

You are quite likely correct, this looks like this property have a bug
regarding thread safety / runtime-safety. The code need to be inspected and
corrected. You should open an issue on gitlab to start with.


> Best regards,
>
> Guy Morand
> _______________________________________________
> gstreamer-devel mailing list
> gstreamer-devel at lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/gstreamer-devel
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/gstreamer-devel/attachments/20201002/1f1840d6/attachment.htm>