At Sat, 11 Aug 2007 15:03:09 +0200, Claude Michel wrote:
Hi,
I have done come test of the new suspend/resume capabilities. However suspend failed to correctly work on my laptop (dell d630). You must note that there is a few trouble issues with sound on this laptop: the snd_hda_intel driver correctly detect a STAC9205 and get the default pin configuration from the bios. So far, so good ... however, pin B is by default configured as line out and is working as such. But, it does not correctly work with an external microphone as it need a vref. So, instead of using the bios settings, I use the windows default settings. This is not perfectly working : I usually get some glinches that disappear after a logout/login procedure. Nothing significant appears from a diff between /proc/asound/card0/codec0 before and after this procedure ...
Now let talk about the suspend failure. The suspend procedure (which was previously working) produce a bug (I was not able to catch the message at this stage. This is a partial and hand written report ..): Bug: unable to handle kernel NULL Pointer deferred at virtual address 0000 .... Call trace ... <f9546d24> hda_set_power_state+0x34/0x150 [snd_hda_intel] <f9547768> hda_call_code_suspend+0x39/0x60 [snd_hda_intel] <f955477b7> snd_hda_suspend+0x27/0x50 [snd_hda_intel]<f954603c> azx_suspend+0x60/0xd0 [snd_hda_intel] <c023b29ac> pci_device_suspend+0x2c/0x70 <c0f29effd> suspend_devide+0xdd/0x110 ....
Could you copy the exact output of these call traces?
By the way, I am running mandriva 2007.1 with kernel 2.6.17-14mdv.
I guess it's a problem with 2.6.17 kernel. Maybe the wrapper for the new kernel API in alsa-driver tree is somehow broken. Check compiler warnings especially for the compilation of hda_codec.c, too.
I didn't get oops, at least. Tthere was another bug that got fixed today, so please update to the latest version from hg.alsa-project.org (not from hg-mirror).
Takashi