Bugzilla Bug 172235

Here is why

lsmod

snd_aoa_codec_onyx     15680  2 
snd_aoa_fabric_layout    14120  0 
snd_aoa                20896  2 snd_aoa_codec_onyx,snd_aoa_fabric_layout
snd_aoa_i2sbus         24420  1 
snd_pcm                95044  1 snd_aoa_i2sbus
snd_timer              27396  1 snd_pcm
snd_page_alloc         11368  1 snd_pcm
snd                    61636  6
snd_aoa_codec_onyx,snd_aoa_fabric_layout,snd_aoa,snd_aoa_i2sbus,snd_pcm,snd_timer
snd_aoa_soundbus        8068  2 snd_aoa_fabric_layout,snd_aoa_i2sbus


the script tries to unload every module w/out expressed deps and
snd_aoa_codec_onyx is locked w/out an expressed module

and that's what happen when you try to unload the right module with alsasound
script still locked on

rmmod --wait snd_aoa_codec_onyx

rmmod snd_aoa_fabric_layout

Unable to handle kernel paging request for data at address 0x00000000
Faulting instruction address: 0xf249c178
Oops: Kernel access of bad area, sig: 11 [#1]
PREEMPT 
Modules linked in: snd_aoa_fabric_layout snd_aoa snd_aoa_i2sbus snd_pcm
snd_timer snd_page_alloc snd snd_aoa_soundbus genrtc pcmcia bcm43xx
firmware_class rng_core ieee80211softmac ieee80211 ieee80211_crypt yenta_socket
rsrc_nonstatic pcmcia_core eth1394
NIP: F249C178 LR: F249C170 CTR: 00000000
REGS: ebd8be00 TRAP: 0300   Not tainted  (2.6.20-gentoo-r4)
MSR: 00009032 <EE,ME,IR,DR>  CR: 24000488  XER: 00000000
DAR: 00000000, DSISR: 40000000
TASK = ec342650[10452] 'rmmod' THREAD: ebd8a000
GPR00: F249C170 EBD8BEB0 EC342650 EFD22D90 C0008B88 00000007 00000000 00000000 
GPR08: 0006FC00 00000000 24000448 00001032 00000000 1001B334 00000000 10010000 
GPR16: 00000000 100D3070 100C0000 100A0000 100C0000 100D53A0 10014008 00000000 
GPR24: 7FEBA6B0 00000000 00000880 7FEBA700 00000000 C0444148 00000880 FFFFFFC0 
NIP [F249C178] aoa_fabric_unregister+0x44/0x90 [snd_aoa]
LR [F249C170] aoa_fabric_unregister+0x3c/0x90 [snd_aoa]
Call Trace:
[EBD8BEB0] [F249C170] aoa_fabric_unregister+0x3c/0x90 [snd_aoa] (unreliable)
[EBD8BEC0] [F24B8E6C] aoa_fabric_layout_exit+0x24/0x698 [snd_aoa_fabric_layout]
[EBD8BED0] [C005172C] sys_delete_module+0x1b8/0x1f4
[EBD8BF40] [C0012E28] ret_from_syscall+0x0/0x38
--- Exception: c01 at 0xff3e234
    LR = 0x100022f4
Instruction dump:
80091148 7f801800 40be0060 3d20f24a 81290f1c 48000018 801f0048 2f800000 
419e0008 4bfffe95 813f0040 3be9ffc0 <801f0040> 2f800000 419e0008 7c00022c 

On IRC you figured out that the oops only happens when you unload the modules
in a certain order. Is the same still true on the latest development kernel,
currently v2.6.21-rc5?

I cannot reproduce it with later kernels, the issue about modules presenting
their lockers is still valid (and still hangs alsasound script)

the suspend/resume issue remains but seems unrelated to aoa

gentoo-sources-2.6.21 is now in portage