Closed borine closed 1 year ago
codecov[bot]
commented
1 year ago Attention: 20 lines in your changes are missing coverage. Please review.
| Files | Coverage Δ | |
|---|---|---|
| src/asound/bluealsa-pcm.c | 76.33% <56.52%> (-0.71%) |
:arrow_down: |
... and 2 files with indirect coverage changes
:loudspeaker: Thoughts on this report? Let us know!.
borine
commented
1 year ago Hmm - the test for io->state == SND_PCM_STATE_SUSPENDED in that last commit was silly; there is no way the system can have set that state. Do you know if there is any way to test if the system has been suspended during a call to poll()? If not then perhaps we should just drop the stream and report -EIO for any timeout error.
arkq
commented
1 year ago Do you know if there is any way to test if the system has been suspended during a call to poll()
I have no idea how to do it properly... the only thing that comes to my mind is some heuristic based on wall clock time vs monotonic time check. But I guess that simple EIO will suffice here.
Implement support for applications that use snd_pcm_abort() or snd_pcm_nonblock(pcm, 2) to request that blocking calls are aborted by a signal. Also apply a timeout to the local drain so that an application can not be blocked indefinitely should the bluealsa server stop reading from the PCM FIFO.
This PR is an attempt to address the plugin issues revealed indirectly by #666