[FFmpeg-trac] #9009(ffmpeg:new): SIGINT ignored while awaiting RTP data
FFmpeg
trac at avcodec.org
Sat Jun 19 20:30:40 EEST 2021
#9009: SIGINT ignored while awaiting RTP data
--------------------------------------+----------------------------------
Reporter: Juan Navarro | Owner: (none)
Type: defect | Status: new
Priority: normal | Component: ffmpeg
Version: git-master | Resolution:
Keywords: sigint, exit | Blocked By:
Blocking: | Reproduced by developer: 0
Analyzed by developer: 0 |
--------------------------------------+----------------------------------
Comment (by Andriy Gelman):
> Thank you. I read the answer by Michael Niedermayer, but didn't
understand the implications of what they say. Is it a rejection of the
patch?
>
> Also according to the comment, do you think the oficial method to stop a
FFmpeg process is to always send '''two''' signals, and that's guaranteed
to work? As far as user experience goes, it is a bit weird, but an easy
workaround nonetheless.
The way it's done now is that we exit with one SIGINT during transcode()
function (ffmpeg.c:
4803) - assuming we are not stuck somewhere.
If the patch is added, the exit can actually be triggered in the muxer
somewhere during the ff_interupt_cb() call, even if we are not in an
infinite loop. This behavior is a bit different, and I guess you can
potentially lose some data on the input side (at least that's how I
understood Michael's answer).
I think what Michael was proposing was to add a smart way to detect if the
muxer is actually stuck and decide whether to exit accordingly.
--
Ticket URL: <https://trac.ffmpeg.org/ticket/9009#comment:10>
FFmpeg <https://ffmpeg.org>
FFmpeg issue tracker
More information about the FFmpeg-trac
mailing list