On Mon, Oct 20, 2008 at 02:17:45PM +0100, Jeremy Morse wrote: > USB_DEBUG should help - you should also enable UHCI_DEBUG. In the > meantime I'll work out what that "overflow!" message indicates. > > You've mentioned this panic occurs when you're using mplayer - does it > fault immediately when you start mplayer, some time in the middle or > when mplayer stops?
Directly after 'starting playback'. > > Also attached is a workaround patch for ehci which stops NMIs from > being raised on my system, it'd be great if you could test this too. Heh, funny enough it talks about 768 frames (64 * 12) but the NMI still gets raised. Ok... what I did: I patched my sys/dev/usb/ehci.c; I rebuilt and moved over a new kernel (GENERIC.UVD) which looks like this now: (some device diff cut out, no context) -#ident "GENERIC-$Revision: 1.914 $" +#ident "GENERIC-$Revision: 1.910 $" -#options DIAGNOSTIC # expensive kernel consistency checks -#options DEBUG # expensive debugging checks/support +options DIAGNOSTIC # expensive kernel consistency checks +options DEBUG # expensive debugging checks/support -#makeoptions DEBUG="-g" # compile full symbol table +makeoptions DEBUG="-g" # compile full symbol table + +options UVIDEO_DEBUG +options USB_DEBUG +options UHCI_DEBUG +options EHCI_DEBUG I tried these four things: - boot generic.uvd -> crash (-8) - boot generic.uvd -c -> 1st attempt works (?), then crash (-9) - boot generic.uvd -1 -> crash (-A) - boot generic.uvd -1c -> 1st attempt works (?), then crash (-B) I was surprised to see situation improved on the uhci side, well, somewhat, by the ehci patch? I also cvs updated before building, who knows.. and ehci claims to hand over the video to uhci this way or another?? supplied stuff this time: http://ayaken.net/dmesg-9 http://ayaken.net/crash-8-bt.jpg http://ayaken.net/crash-8-dmesg1.jpg http://ayaken.net/crash-8-dmesg2.jpg http://ayaken.net/attempt9.jpg http://ayaken.net/crash-9-bt1.jpg http://ayaken.net/crash-9-bt2.jpg http://ayaken.net/crash-9-dmesg1.jpg http://ayaken.net/crash-9-dmesg2.jpg http://ayaken.net/crash-A-bt.jpg http://ayaken.net/crash-A-dmesg.jpg http://ayaken.net/attemptB.jpg http://ayaken.net/crash-B-bt.jpg http://ayaken.net/crash-B-dmesg1.jpg http://ayaken.net/crash-B-dmesg2.jpg The 'attempt' photos are me being baffled by the 1st attempt working so I had to shoot a pic as proof :) Although the output might indicate it didn't work at all. These two attempts (9 and B) also have different dmesg parts, of course. * phew * Enough crashin for today I believe. Regards, -Martin
