Need for 'nomux' has been rising lately ...

Need for 'nomux' has been rising lately ...

[Jiri Kosina]

Hi Dmitry,

lately, I have been seeing a couple of bug reports in which touchpad of 
certain notebooks misbehaved, and i8042.nomux solved the problem. Most 
recently, this happened on [1]. Do you have any idea what might be the 
cause of the need for nomux rising so much lately? Apparently, some older 
kernels behaved correctly even without this workaround, at least on Acer 
Aspire 5710, as referenced in the bugreport.

Thanks.

[1] https://bugzilla.novell.com/show_bug.cgi?id=404881

-- 
Jiri Kosina
SUSE Labs

Re: Need for 'nomux' has been rising lately ...

[Dmitry Torokhov]

Hi Jiri,

On Tue, Oct 14, 2008 at 04:55:22PM +0200, Jiri Kosina wrote:
> Hi Dmitry,
> 
> lately, I have been seeing a couple of bug reports in which touchpad of 
> certain notebooks misbehaved, and i8042.nomux solved the problem. Most 
> recently, this happened on [1]. Do you have any idea what might be the 
> cause of the need for nomux rising so much lately? Apparently, some older 
> kernels behaved correctly even without this workaround, at least on Acer 
> Aspire 5710, as referenced in the bugreport.
> 

I am always happy to blame ACPI code (especially EC) for bad
iteraction with input devices ;), especially since there weren't any
significant changes to i8042 in any recent kernels. Do you think the
person reporting the failure on Acer could try bisecting it? Could
ALexey Starikovskiy have any ideas whether there is bad EC iteractions
stopping timely delivery of mouse interrupts?

-- 
Dmitry

Re: Need for 'nomux' has been rising lately ...

[Alexey Starikovskiy]

Dmitry Torokhov wrote:
> Hi Jiri,
> 
> On Tue, Oct 14, 2008 at 04:55:22PM +0200, Jiri Kosina wrote:
>> Hi Dmitry,
>>
>> lately, I have been seeing a couple of bug reports in which touchpad of 
>> certain notebooks misbehaved, and i8042.nomux solved the problem. Most 
>> recently, this happened on [1]. Do you have any idea what might be the 
>> cause of the need for nomux rising so much lately? Apparently, some older 
>> kernels behaved correctly even without this workaround, at least on Acer 
>> Aspire 5710, as referenced in the bugreport.
>>
> 
> I am always happy to blame ACPI code (especially EC) for bad
> iteraction with input devices ;), especially since there weren't any
> significant changes to i8042 in any recent kernels. Do you think the
> person reporting the failure on Acer could try bisecting it? Could
> ALexey Starikovskiy have any ideas whether there is bad EC iteractions
> stopping timely delivery of mouse interrupts?
> 
There was an issue with EC interrupt storm on several Acer notebooks lately,
hopefully solved (bug 11549).
Symptoms could be that keyboard/mouse are slow, missing keys, etc.
I suggest you should try the patch in 11549, or acpi-test tree, and see if your
problem goes away...

Regards,
Alex.

Re: Need for 'nomux' has been rising lately ...

[Jiri Kosina]

On Tue, 14 Oct 2008, Alexey Starikovskiy wrote:

> There was an issue with EC interrupt storm on several Acer notebooks 
> lately, hopefully solved (bug 11549). Symptoms could be that 
> keyboard/mouse are slow, missing keys, etc. I suggest you should try the 
> patch in 11549, or acpi-test tree, and see if your problem goes away...

I am not experiencing the problems myself, but there has been quite a 
number of reports in our bugzilla about this.

I have built the kernel package with your patch included to test, and will 
let you know if their report that the problem is fixed by that kernel.

Thanks,

-- 
Jiri Kosina
SUSE Labs