On Thu, Jan 05, 2023 at 01:56:56PM +0000, Daniel Lee wrote: > I fetched HEAD from cvs just now and built GENERIC.MP. > Can confirm the issue is still present. I am only building > the kernel and not building the userland (the userland is > still from the latest snapshot.) >
The code under #ifdef UAUDIO_DEBUG only prints values already in memory and shouldn't affect device behavior. Furthermore during playback it doesn't print anything at all to not change timing of requests. Could you try to see if initializing the device without playing anything triggers the problem: stop sndiod and just open the device in full-duplex: cat <>/dev/audio0 then hit ^C and check dmesg for errors. Depending on the result, you could also try to figure out which code chunk under #ifdef UAUDIO_DEBUG "fixes" the device by disabling one by one each chunk FYI, the functions below are called during playback and recording, so if just opening the device doesn't trigger the problem, they are good candidates to start with. uaudio_pdata_copy uaudio_pdata_calcsizes uaudio_pdata_xfer uaudio_pdata_intr uaudio_rdata_xfer uaudio_rdata_intr my 2 cents
