Re: [PATCH v2] panic: keep blinking in spite of long spin timer mode

[TAMUKI Shoichi <tamuki@linet.gr.jp>]

Hello,

On Thu, 10 Jun 2010 12:30:33 +1000 Anton Blanchard wrote:

> > Please remind me why we're making this configurable at all.  Can't we
> > just hardwire the thing to 1Hz or something?  Add an
> > im_using_a_hypervisor boot option or something, if necessary?
> 
> I agree. The panic_blink() interface is quite painful and I have no idea
> why someone would want to configure the blink frequency of their keyboard
> LED when panicing. Maybe they want to match it to the beat of their techno
> music.

That's cool.  Joking aside, we might want to know which machine has
paniced among two or more overcrowded machines.  Maybe it is a good
idea to change the blinking speed of each machine so as to distinguish
whether the machine has paniced or not.

> Since the keyboard LED default is a transition every 0.5s, why don't we just
> remove i8042.panicblink and change all users (all 2 of them) to expect a 2 HZ
> call rate? The hypervisor case should be fine with 0.5s mdelays, so we end up
> removing that special case.

If the speed of blinking is slow enough (i.e. panicblink<=5), both the
native case and the hypervisor case should be fine, indeed.  However,
gta02_panic_blink(), one of panic_blink() users, expects a 10Hz call
rate at the native case.  IOW, it blinks at 5Hz (i.e. panicblink=36).
I think that the desirable blinking speed is different according to
devices.

> I would have said 1 HZ, but it seems like the default was chosen to be
> different to kdb:
> 
> 	/*
>          * We expect frequency to be about 1/2s. KDB uses about 1s.
>          * Make sure they are different.
> 	 */
> 
> No idea if that comment is still valid.

That comment maybe still valid according to the following:

On Sun, 30 May 2010 08:55:22 +0200 Andi Kleen wrote:

> Changing the frequency is ok, the only requirement is that it is
> visible and different from the frequency kdb uses.

Regards,
TAMUKI Shoichi