Hi, At Fri, 9 Nov 2001 23:14:23 -0500, Paul Winkler wrote: > > On Fri, Nov 09, 2001 at 04:35:45PM -0500, Paul Winkler wrote: > > I've got a laptop with a NeoMagic 256 soundcard built in. It pretty > > much sucks, but it's what I have. > (snip) > > I then put alsasound in /etc/init.d and it runs on startup ... but the > > system seems to have about a 50% possibility of hanging as soon as it > > tries to load the sound driver. Anybody else seen this? > > OK, I read the INSTALL file notes more carefully - this seems to be a > known problem with this card, but the solution is rather cryptic: > > Note: on some notebooks the buffer address cannot be detected > automatically, or causes hang-up during initialization. > In such a case, specify the buffer top address explicity via > snd_buffer_top option. > For example, > Sony F250: snd_buffer_top=0x25a800 > Sony F270: snd_buffer_top=0x272800 > The driver supports only ac97 codec. It's possible to force > to initialize/use ac97 although it's not detected. In such a > case, use snd_force_ac97=1 option. > > > So I've added snd_force_ac97=1 to my modules.conf... but how am I > supposed to know the address to use for snd_buffer_top? > Where does one find that information?? > I looked in the source for card-nm256.c hoping for some useful > comments, but there's nothing... The ALSA nm256 driver is derived from OSS/Free driver in kernel tree and there you find better docs. This chip is really really weird. And almost no info from Neomagic.
Anyway, plese try to load nm256 driver before starting X server. Once when you start X, the signature on VRAM will be lost, so the sound driver cannot detect the right position for audio buffer. If you've already run X, reboot the machine on the proper runlevel. When the driver is compiled with debug option, it will show the detected signature position or an error message, so that you can pass the option at the next time. > I guessed it might be as reported by /proc/iomem but there's a number > of address ranges there and I have no idea which corresponds to > "snd_buffer_top": Usually the address mentioned in INSTALL should work, e.g. even on Dell notebook 0x25a800 was Ok (sorry, I could test only with ALSA 0.5.x). (Hmm, I found a bug in snd_vaio_hack option. Fixed now.) It's possible that ALSA 0.9.0 doesn't work although 0.5.x does. I've heard that some notebooks with NM256 crash when invalid AC97 registers are accessed. Except for ac97 routine, there are very few changes between ALSA 0.5.x and 0.9.0. Takashi _______________________________________________ Alsa-user mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/alsa-user
