drm/i915: compute wait_ioctl timeout correctly
authorDaniel Vetter <daniel.vetter@ffwll.ch>
Thu, 4 Dec 2014 10:12:54 +0000 (11:12 +0100)
committerJani Nikula <jani.nikula@intel.com>
Fri, 5 Dec 2014 13:20:24 +0000 (15:20 +0200)
commit7bd0e226e313547add65c7e4b0ecf65cb981a3c6
treefe3e62205d3e573637a7098c51f3b47610ba1e8b
parentaf15d2ce5d497ba6535dead782cfadbcc83d90fe
drm/i915: compute wait_ioctl timeout correctly

We've lost the +1 required for correct timeouts in

commit 5ed0bdf21a85d78e04f89f15ccf227562177cbd9
Author: Thomas Gleixner <tglx@linutronix.de>
Date:   Wed Jul 16 21:05:06 2014 +0000

    drm: i915: Use nsec based interfaces

    Use ktime_get_raw_ns() and get rid of the back and forth timespec
    conversions.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: John Stultz <john.stultz@linaro.org>
So fix this up by reinstating our handrolled _timeout function. While
at it bother with handling MAX_JIFFIES.

v2: Convert to usecs (we don't care about the accuracy anyway) first
to avoid overflow issues Dave Gordon spotted.

v3: Drop the explicit MAX_JIFFY_OFFSET check, usecs_to_jiffies should
take care of that already. It might be a bit too enthusiastic about it
though.

v4: Chris has a much nicer color, so use his implementation.

This requires to export nsec_to_jiffies from time.c.

Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Dave Gordon <david.s.gordon@intel.com>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=82749
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: John Stultz <john.stultz@linaro.org>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Acked-by: John Stultz <john.stultz@linaro.org>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
drivers/gpu/drm/i915/i915_drv.h
drivers/gpu/drm/i915/i915_gem.c
kernel/time/time.c