[elrepo-devel] kmod-alsa update request ...

[Phil Perry]

Phil Perry wrote:
> Fabian Arrotin wrote:
>> Hi guys (and girls ;-) )
>>
>> I had issues to get the sound working on my Sony Vaio (but i had too 
>> much work / not enough free time to have a deeper look) but alsa-project 
>> released a new alsa-driver package yesterday (1.0.23)
>> I've just bumped the version in your existing kmod-alsa spec (from 
>> 1.0.20 to 1.0.23) and i have installed it on my vaio.
>> It now works as expected and i have now sound :
>>
>> [arrfab at init0 ~]$ rpm -q kmod-alsa
>> kmod-alsa-1.0.23-1.el5.elrepo.x86_64
>>
>> ;-)
>>
> 
> Hi Fabian,
> 
> That's excellent news. I had tried to build both 1.0.21 and 1.0.22, but 
> both failed to compile (which was fixable with a little effort), but 
> then caused kernel panics when I tested locally - hence why I haven't 
> updated kmod-alsa since 1.0.20.
> 
> I've just pulled the 1.0.23 source and it compiles cleanly against 
> RHEL5.5 (which is still using alsa-driver 1.0.14rc3 kernel modules), so 
> we're looking good for an update.
> 

I've built updated kmod-alsa-1.0.23-2.el5.elrepo packages and they are 
in the elrepo-testing repository:

http://elrepo.org/linux/testing/el5/

It's built against kernel-2.6.18-164.el5 and is kABI compatible with all 
RHEL5 kernels up to and including RHEL5.5.

I've not had too much joy testing them. On my i386 laptop (Intel HDA 
with SigmaTel STAC92XX codec), I get a kernel panic and the system fails 
to boot. Note that because the the package is kABI-tracking, this means 
all RHEL5 kernels will now panic and the system is left in an unbootable 
state. Either have a non-kABI compatibile kernel installed (e.g, one of 
the rt kernels) that you can boot or have a rescue disk to hand, boot 
and manually delete the offending module(s). This isn't a packaging 
issue - I see exactly the same behaviour on that machine when compiling 
and installing from source. You just need to be aware that if a kmod 
module does cause a kernel panic, it has the potential to leave all 
installed kernel unbootable.

Testing on another Intel HDA box (x86_64, with a different codec) was a 
little less dramatic and passed without incident. In both cases sound 
was already working with the default kernel drivers.

Please test and feedback either here or to the bug tracker.

Thanks.

Phil