Re: 2.6.25-rc5.git4 regression PS/2 mouse not detected/working

[Balaji Rao <balajirrao@gmail.com>]

On Wednesday 02 April 2008 08:29:00 pm Dmitry Torokhov wrote:
> Hi Yanko,
> 
> On Wed, Apr 02, 2008 at 02:17:31AM +0300, Yanko Kaneti wrote:
> > On Tue, 2008-03-18 at 11:54 -0400, Dmitry Torokhov wrote:
> > > On Mon, Mar 17, 2008 at 08:58:43PM +0200, Yanko Kaneti wrote:
> > > > On Mon, 2008-03-17 at 14:55 -0400, Dmitry Torokhov wrote:
> > > > > On Mon, Mar 17, 2008 at 07:33:49PM +0200, Yanko Kaneti wrote:
> > > > > >
> > > > > > My PS/2 mouse stops being detected between kernels 2.6.24.3-12.fc8 
and
> > > > > > 2.6.25-0.121.rc5.git4.fc9 (which I believe do not carry any 
patches in
> > > > > > that area). Its an ASUS P5E-VM DO motherboard , bios revision 
0506. The
> > > > > > PS/2 keyboard still works.
> > > > > > 
> > > > > > Attached are both dmesg logs with i8042.debug=1, dmidecode, 
lspci -vvnn
> > > > > > Tried i8042.noloop and both the mouse and keyboard stop working.
> > > > > > 
> > > > > 
> > > > > Could you please send me dmesg from booting with i8042.noloop. I 
would
> > > > > expect this option to help in your case so I am curious what went 
wrong.
> > > > 
> > > > Attached
> > > > 
> > > 
> > > Hmm, it looks the kernel does the right thing in case of i8042.noloop,
> > > except that it does not work :(. Kernel triest to query the mouse but
> > > it does not respond.
> > > 
> > > Can you try taking contents of drivers/input/serio from 2.6.25 and
> > > copy it in .24? It will most likely compile... If that works I'm
> > > afraid I will ahve to ask you to do git biscect to find the offending
> > > commit.
> > 
> > Hey again,
> > 
> > I finally did the bisecting. It found the cuplrit to be
> > 
> > commit e3f37a54f690d3e64995ea7ecea08c5ab3070faf
> > Author: Balaji Rao <balajirrao@gmail.com>
> > Date:   Wed Jan 30 13:30:03 2008 +0100
> > 
> >     x86: assign IRQs to HPET timers
> >     
> > 
> > I tried reverting the change on the latest git with the attached patch
> > and my mouse and keyboard appear to be detected and working fine now.
> > The patch had to be stiched by hand and due to my lack of understanding
> > of the code might be breaking random other stuff.
> > 
> > FWIW the bios on this board has no knobs for HPET that I could find.
> > 
> > What gives?
> > 
> 
> Hmm, this is wierd and really outsize of my knowledge... Lets try copying
> Ingo, Thomas and Balaji, maybe they can help us out here.
> 
> Btw, what is in your /proc/interrupts?
> 
Hi,

Coincidentally even I hit upon this bug a couple of days ago after enabling 
HPET in my BIOS. A quick check told me that IRQ 12 which the patch assigns to 
the HPET device is used by the PS/2 mouse as well. This makes the mouse 
inactive.

I am working on this.. I am not sure if the PS/2 code is misbehaving though..

-- 
regards,
Balaji Rao
Dept. of Mechanical Engineering,
National Institute of Technology Karnataka, India