[Intel-gfx] [RFC] drm/i915: _wait_for might be called when irq is off
Chris Wilson
chris at chris-wilson.co.uk
Thu Apr 2 08:27:24 PDT 2015
On Thu, Apr 02, 2015 at 07:21:40PM +0530, deepak.s at linux.intel.com wrote:
> From: Deepak S <deepak.s at linux.intel.com>
>
> Sometimes, i915 might call _wait_for when irq is disabled.
> If the cpu is the main cpu to process jiffies, jiffies
> wouldn't be increased as this cpu disables irq. Then,
> time_after(jiffies, timeout__) becomes meaningless. If
> gunit doesn't work now, kernel wouldn't exit as the timeout
> doesn't work.
>
> The patch fixes it by using sched_clock instead of jiffies.
sched_clock() requires irq disabled, or at least so the header claims,
at the very least it would require preemption disabled - definitely not
for our general waits of many ms. Also local_clock() would seem to be the
right choice in these tight loops.
I think you want a specialised macro (if any) that is very aware of the
constraints it is running under. I would hope we never have to busy spin
with interrupts disabled. And so I want such code that does to blatantly
obvious and scrutinised carefully.
-Chris
--
Chris Wilson, Intel Open Source Technology Centre
More information about the Intel-gfx
mailing list