drm/i915: Do one shot unclaimed mmio detection less frequently
We have done unclaimed register access check in normal
(mmio_debug=0) mode once per write. This adds probability
of finding the exact sequence where we did the bad access, but
also adds burden to each write.
As we have mmio_debug available for more fine grained analysis,
give up accuracy of detecting correct spot at the first occurrence
by doing the one shot detection and arming of mmio_debug in hangcheck
and in modeset. This removes the write path performance burden.
v2: Remove gratuitous DRM_DEBUG and return value, comments (Chris)
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Paulo Zanoni <przanoni@gmail.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1450250808-14864-1-git-send-email-mika.kuoppala@intel.com
diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c
index 738d0dc..3edf22ad 100644
--- a/drivers/gpu/drm/i915/intel_display.c
+++ b/drivers/gpu/drm/i915/intel_display.c
@@ -13684,6 +13684,19 @@
drm_atomic_state_free(state);
+ /* As one of the primary mmio accessors, KMS has a high likelihood
+ * of triggering bugs in unclaimed access. After we finish
+ * modesetting, see if an error has been flagged, and if so
+ * enable debugging for the next modeset - and hope we catch
+ * the culprit.
+ *
+ * XXX note that we assume display power is on at this point.
+ * This might hold true now but we need to add pm helper to check
+ * unclaimed only when the hardware is on, as atomic commits
+ * can happen also when the device is completely off.
+ */
+ intel_uncore_arm_unclaimed_mmio_detection(dev_priv);
+
return 0;
}