2011/7/4 Michael Walle [off-list ref]:

Am Donnerstag 30 Juni 2011, 22:58:15 schrieb Bj?rn Forsman:

quoted

Hi Michael,

[..]

quoted

So, the power switch also seem to suffer from some interrupt overload;
10154532 IRQs for Power-auto Switch is not right...right?

Have you seen this behaviour?

Nope, the interrupt behaves normally for me. But i'm running this patch with
kernel 2.6.32 and tested it briefly with 2.6.39.

Could you verify, if you are triggering the bug Simon posted above?

Yes, it must be the same bug. I reverted
e59347a1d15c0b1d9fdc510520f8fa78d7d19a5b (arm: orion: Use generic irq
chip) on top of "3.0-rc5 + this LS-XHL patch" and the IRQ issues are
gone.

$ cat /proc/interrupts   # after toggling the buttons and setting fan
speed to 0 a few times
...
104:          4  orion_gpio_irq  GPIO fan alarm
105:          4  orion_gpio_irq  Function Button
106:          6  orion_gpio_irq  Power-auto Switch
107:          4  orion_gpio_irq  Power-on Switch

No system hang/reboot/lockups :-)

Compare that with /proc/interrupts from the bugged version; it lists
"-" instead of "orion_gpio_irq" as IRQ handler:

104:     100000         -  GPIO fan alarm
105:          0         -  Function Button
106:          0         -  Power-auto Switch
107:          0         -  Power-on Switch

Best regards,
Bj?rn Forsman