PNP PS/2 probing racy?

PNP PS/2 probing racy?

[Andi Kleen]

Hi,

I noticed that on one test system of machine when booting the same
kernel I get:

PNP: No PS/2 controller found. Probing ports directly.
serio: i8042 KBD port at 0x60,0x64 irq 1
serio: i8042 AUX port at 0x60,0x64 irq 12

and sometimes on another boot

PNP: PS/2 Controller [PNP0303:PS2K,PNP0f03:PS2M] at 0x60,0x64 irq 1,12

Since I presume the BIOS does not change, is it possible the Linux
procedure for this PNP probe is racy? I haven't checked the code so far.

Full boot logs for both cases:

http://halobates.de/config/dmesg-pnprace-1
http://halobates.de/config/dmesg-pnprace-2

-Andi
-- 
ak@linux.intel.com -- Speaking for myself only.

Re: PNP PS/2 probing racy?

[Dmitry Torokhov]

Hi Andi,

On Jan 28, 2010, at 5:54 AM, Andi Kleen <andi@firstfloor.org> wrote:

>
> Hi,
>
> I noticed that on one test system of machine when booting the same
> kernel I get:
>
> PNP: No PS/2 controller found. Probing ports directly.
> serio: i8042 KBD port at 0x60,0x64 irq 1
> serio: i8042 AUX port at 0x60,0x64 irq 12
>
> and sometimes on another boot
>
> PNP: PS/2 Controller [PNP0303:PS2K,PNP0f03:PS2M] at 0x60,0x64 irq 1,12
>
> Since I presume the BIOS does not change, is it possible the Linux
> procedure for this PNP probe is racy? I haven't checked the code so  
> far.
>

Are you using KVM with that box by any chance?

-- 
Dmitry

Re: PNP PS/2 probing racy?

[Andi Kleen]

On Thu, Jan 28, 2010 at 07:57:47AM -0800, Dmitry Torokhov wrote:
>> Hi,
>>
>> I noticed that on one test system of machine when booting the same
>> kernel I get:
>>
>> PNP: No PS/2 controller found. Probing ports directly.
>> serio: i8042 KBD port at 0x60,0x64 irq 1
>> serio: i8042 AUX port at 0x60,0x64 irq 12
>>
>> and sometimes on another boot
>>
>> PNP: PS/2 Controller [PNP0303:PS2K,PNP0f03:PS2M] at 0x60,0x64 irq 1,12
>>
>> Since I presume the BIOS does not change, is it possible the Linux
>> procedure for this PNP probe is racy? I haven't checked the code so far.
>>
>
> Are you using KVM with that box by any chance?

Yes, it uses a KVM, but the console was always on it.

-Andi

-- 
ak@linux.intel.com -- Speaking for myself only.

Re: PNP PS/2 probing racy?

[Dmitry Torokhov]

On Thu, Jan 28, 2010 at 05:22:48PM +0100, Andi Kleen wrote:
> On Thu, Jan 28, 2010 at 07:57:47AM -0800, Dmitry Torokhov wrote:
> >> Hi,
> >>
> >> I noticed that on one test system of machine when booting the same
> >> kernel I get:
> >>
> >> PNP: No PS/2 controller found. Probing ports directly.
> >> serio: i8042 KBD port at 0x60,0x64 irq 1
> >> serio: i8042 AUX port at 0x60,0x64 irq 12
> >>
> >> and sometimes on another boot
> >>
> >> PNP: PS/2 Controller [PNP0303:PS2K,PNP0f03:PS2M] at 0x60,0x64 irq 1,12
> >>
> >> Since I presume the BIOS does not change, is it possible the Linux
> >> procedure for this PNP probe is racy? I haven't checked the code so far.
> >>
> >
> > Are you using KVM with that box by any chance?
> 
> Yes, it uses a KVM, but the console was always on it.
> 

That is wierd... Was it on the whole time (including the time when BIOS
was doing the initialization)? Anyway i8042 expects to find all PNP
devices already enumerated by the time its initialization runs. When
you see that message about PS2 controller not found do you see PNP0303
and PNP0f03 if you do:

	for i in /sys/devices/pnp0/00*; do echo -n "$i: "; cat $i/id; done

?

-- 
Dmitry

Re: PNP PS/2 probing racy?

[Andi Kleen]

On Thu, Jan 28, 2010 at 10:24:05PM -0800, Dmitry Torokhov wrote:
> On Thu, Jan 28, 2010 at 05:22:48PM +0100, Andi Kleen wrote:
> > On Thu, Jan 28, 2010 at 07:57:47AM -0800, Dmitry Torokhov wrote:
> > >> Hi,
> > >>
> > >> I noticed that on one test system of machine when booting the same
> > >> kernel I get:
> > >>
> > >> PNP: No PS/2 controller found. Probing ports directly.
> > >> serio: i8042 KBD port at 0x60,0x64 irq 1
> > >> serio: i8042 AUX port at 0x60,0x64 irq 12
> > >>
> > >> and sometimes on another boot
> > >>
> > >> PNP: PS/2 Controller [PNP0303:PS2K,PNP0f03:PS2M] at 0x60,0x64 irq 1,12
> > >>
> > >> Since I presume the BIOS does not change, is it possible the Linux
> > >> procedure for this PNP probe is racy? I haven't checked the code so far.
> > >>
> > >
> > > Are you using KVM with that box by any chance?
> > 
> > Yes, it uses a KVM, but the console was always on it.
> > 
> 
> That is wierd... Was it on the whole time (including the time when BIOS
> was doing the initialization)? Anyway i8042 expects to find all PNP

I think so. That is of course the KVM might have a special state
when the display is off, but it's hard for me to detect that.
But I didn't switch the display

> devices already enumerated by the time its initialization runs. When
> you see that message about PS2 controller not found do you see PNP0303
> and PNP0f03 if you do:
> 
> 	for i in /sys/devices/pnp0/00*; do echo -n "$i: "; cat $i/id; done

I'll check next time (machine out of reach right now)

-Andi

-- 
ak@linux.intel.com -- Speaking for myself only.