After being bitten by unsafe usage of disable_irq from within that irq's
handler [1], I went through and used Coccinelle to audit the rest of
drivers/ too. This turned up 9 more unsafe usages.
As Haavard pointed out in that thread, disable_irq waits for running
handlers of an irq to complete before returning. When a handler
disable_irq's its own irq then we see a deadlock.
Note that this wasn't actually observed in the wild until the recent
merge of
commit 3aa551c9b4c40018f0e261a178e3d25478dc04a9
Author: Thomas Gleixner <[email protected]>
Date: Mon Mar 23 18:28:15 2009 +0100
genirq: add threaded interrupt handler support
Add support for threaded interrupt handlers
Regards,
--Ben.
[1] http://lkml.org/lkml/2009/4/15/498
Hi Ben,
On Wednesday 15 April 2009 22:54:05 Ben Nizette wrote:
> After being bitten by unsafe usage of disable_irq from within that irq's
> handler [1], I went through and used Coccinelle to audit the rest of
> drivers/ too. This turned up 9 more unsafe usages.
>
> As Haavard pointed out in that thread, disable_irq waits for running
> handlers of an irq to complete before returning. When a handler
> disable_irq's its own irq then we see a deadlock.
>
> Note that this wasn't actually observed in the wild until the recent
> merge of
>
> commit 3aa551c9b4c40018f0e261a178e3d25478dc04a9
> Author: Thomas Gleixner <[email protected]>
> Date: Mon Mar 23 18:28:15 2009 +0100
>
> genirq: add threaded interrupt handler support
>
> Add support for threaded interrupt handlers
>
>
Applied all 4 patches affecting input devices, they should show up next
time Linus pulls from me. Thank you.
--
Dmitry