HEnquist / camilladsp

A flexible cross-platform IIR and FIR engine for crossovers, room correction etc.
https://henquist.github.io/
GNU General Public License v3.0
505 stars 48 forks source link

Camiladsp 2.0 stop working on loopback ( 32bit os) #310

Closed melomane63 closed 5 months ago

melomane63 commented 6 months ago

Using CamillaDSP 2.0.0, pi zero W 2, Raspberry Pi OS lite 32bit ( latest or legacy ).

Camiladsp 2.0 stop working after somes seconds if input is on loopback ( hw:Loopback,0 ) this is the log out:

2023-12-15 11:36:15.912336 INFO [src/bin.rs:683] CamillaDSP version 2.0.0 2023-12-15 11:36:15.912596 INFO [src/bin.rs:684] Running on linux, arm 2023-12-15 11:36:16.079361 INFO [src/alsadevice.rs:648] Capture device supports rate adjust 2023-12-15 11:36:16.118391 INFO [src/alsadevice.rs:142] PB: Starting playback from Prepared state 2023-12-15 11:36:19.191169 WARN [src/alsadevice.rs:281] Capture device failed while waiting for available frames, error: ALSA function 'snd_pcm_wait' failed with error 'EPIPE: Broken pipe' 2023-12-15 11:36:19.191409 ERROR [src/bin.rs:307] Capture error: ALSA function 'snd_pcm_wait' failed with error 'EPIPE: Broken pipe'

Work fine on 64bit os with the same .yml config file .

HEnquist commented 6 months ago

Can you please run this with trace log level and post the output, together with the full config file?

melomane63 commented 6 months ago

camilladsp.log

config files.tar.gz

i hope this is what you want

HEnquist commented 6 months ago

Thanks! I see some of these lines: 2023-12-15 21:57:30.597097 TRACE [src/alsadevice.rs:274] Capture waited for Some(39.145833ms), ready 39 ms is about twice what it should be. It looks as if sometimes the event that signals that the loopback is ready goes missing. There have been similar issues before, and if I remember correctly it seemed linked to then the CPU governor adjusted the CPU frequency. Could you try locking the CPU frequency to a single value?

melomane63 commented 6 months ago

Set governor on Performance, no luck. camilladsp.log

HEnquist commented 6 months ago

Ok, well it was worth a try. Thanks for testing!

HEnquist commented 6 months ago

Don't have any new clever ideas unfortunately. Does it work better with a 2x or 4x larger chunksize?

melomane63 commented 6 months ago

Very quick tested, it seem work with 4096. Not the time now, i test in deep after tomorrow Where is the différence under 32 and 64 bit ?

Le 19 décembre 2023 21:25:49 GMT+01:00, Henrik Enquist @.***> a écrit :

Don't have any new clever ideas unfortunately. Does it work better with a 2x or 4x larger chunksize?

-- Reply to this email directly or view it on GitHub: https://github.com/HEnquist/camilladsp/issues/310#issuecomment-1863426197 You are receiving this because you authored the thread.

Message ID: @.***>

HEnquist commented 6 months ago

Ok! There shouldn't be any difference between how this behaves on 32 and 64-bit systems. But the kernel may be somewhat different, not sure how similar these versions are.

HEnquist commented 5 months ago

Does this still happen on the latest release?

melomane63 commented 5 months ago

i don't know, i don't use anymore 32 bit version. sorry.

HEnquist commented 5 months ago

Ok no worries. Let's close this issue then.