Hello, while testing audio on a allwinner A20 I noticed two issues: - halt_output() may be called without calling trigger_output() first. This messes up the ref counting i awin_hdmi.c and ends up triggering a KASSERT() in awin_hdmi_poweron() (and also causes the HDMI screen to go to sleep unexpectedly). I have a patch to workaround this, but there is another problem: - cat some_audio.wav > /dev/audio1 (/dev/audio1 is: awinhdmiaudio0 at awinio0: HDMI 1.3 audio1 at awinhdmiaudio0: half duplex, playback, mmap awinhdmiaudio0: Virtual format configured - Format SLINEAR, precision 16, channe ) cause cat to hang on audiowr. This is because, I guess, trigger_output() is *not* called.
I guess that's not a general problem or no audio would work. any idea ? -- Manuel Bouyer <[email protected]> NetBSD: 26 ans d'experience feront toujours la difference --
