We wait upon jiffies, but report the time elapsed using a
high-resolution timer. This discrepancy can lead to us timing out the
wait prior to us reporting the elapsed time as complete.
This restores the squelching lost in commit e95433c73a11 ("drm/i915:
Rearrange i915_wait_request() accounting with callers").
Fixes: e95433c73a11 ("drm/i915: Rearrange i915_wait_request() accounting with
callers")
Signed-off-by: Chris Wilson <ch...@chris-wilson.co.uk>
Cc: Matthew Auld <matthew.william.a...@gmail.com>
Cc: Joonas Lahtinen <joonas.lahti...@linux.intel.com>
Cc: <drm-intel-fi...@lists.freedesktop.org> # v4.10-rc1+
Cc: sta...@vger.kernel.org
---
drivers/gpu/drm/i915/i915_gem.c | 10 ++++++++++
1 file changed, 10 insertions(+)
diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index 98c483cf805c..6a159ed7ed08 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -3116,6 +3116,16 @@ i915_gem_wait_ioctl(struct drm_device *dev, void *data,
struct drm_file *file)
args->timeout_ns -= ktime_to_ns(ktime_sub(ktime_get(), start));
if (args->timeout_ns < 0)
args->timeout_ns = 0;
+
+ /*
+ * Apparently ktime isn't accurate enough and occasionally has a
+ * bit of mismatch in the jiffies<->nsecs<->ktime loop. So patch
+ * things up to make the test happy. We allow up to 1 jiffy.
+ *
+ * This is a regrssion from the timespec->ktime conversion.
+ */
+ if (ret == -ETIME && !nsecs_to_jiffies(args->timeout_ns))
+ args->timeout_ns = 0;
}
i915_gem_object_put(obj);
--
2.11.0
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
