Hi Jesse, At Mon, 01 Jul 2002 09:04:05 -0400, Jesse Lackey wrote: > > Hello all, > > Maybe someone can help me... > I *think* I have alsa (0.9rc2) set up correctly. > > I have a laptop with a snd-es1968 chip built-in. This seems to be > working fine for both oss (test: xmms) and native (test: mpg321 -o alsa > <file.mp3>) playback. > > I also have a USB audio device: Roland US-30. Here's what's happening: > modprobe snd-usb-audio > (works fine) > Playing an mp3 with either xmms (via oss) or mpg321 -o alsa (native > alsa) works fine UNTIL I stop playback. In both cases, linux locks > instantly. I have to power-cycle. > > /var/log/messages: > > Jul 1 08:25:43 localhost kernel: ALSA usbaudio.c:1501: 6:1:5 : > non-supported sample bit 1 in 3 bytes > Jul 1 08:25:43 localhost kernel: ALSA usbaudio.c:1501: 6:1:6 : > non-supported sample bit 1 in 3 bytes > Jul 1 08:25:43 localhost kernel: ALSA usbaudio.c:1501: 6:2:5 : > non-supported sample bit 1 in 3 bytes > Jul 1 08:25:43 localhost kernel: ALSA usbaudio.c:1501: 6:2:6 : > non-supported sample bit 1 in 3 bytes these formats are supported on cvs version. please update both all alsa-driver, alsa-kernel and alsa-lib at the same time.
> The next message is the syslogd restart message when I reboot. > At first I thought this had something to do with using oss emulation > with the USB device, or some problem with having both the USB audio and > the es1968 device simultaneously (i.e. alsa configuration problem). But > this happens when it is just the USB device and with alsa-native playback. > > I'd be happy to add some further test/snd_printk() code to help debug > this. I know C/C++ and am familiar with both USB and audio, but am a > linux newbie. Takashi? My only concern is how to avoid having the > crashes eventually destroy something important on this machine, it is my > main workstation. generally, in the case of deadlock, it's difficult to guess from printk. this might be related also with the kernel version and config. which kernel are you using? SMP enabled? which kernel version? a rough guess from the code is that the driver caused the double unlinking of urbs. please try the attached patch. Takashi
usb-mask-fix.dif
Description: Binary data
