[PATCH 02/72] drm/i915: Sleep around performing iommu unmaps on Tigerlake

[Chris Wilson]

Tigerlake is plagued by spontaneous DMAR faults [reason 7, next page
table ptr is invalid] which lead to GPU hangs. These faults occur when
an iommu map is immediately reused. Adding further clflushes and
barriers around either the GTT PTE or iommu PTE updates do not prevent
the faults. So far the only effect has been from inducing a delay
between reuse of the iommu on the GPU, and applying the delay at the
iommu map allows for the smallest stable delay.

Note that such a delay clearly does not fix the root cause, and so
should only be a bandaid until a complete solution is found. The delay
was determined by running igt/gem_exec_fence/parallel in a loop for a
few hours (unpatched MTBF is about 10s).

Testcase: igt/gem_exec_fence/parallel
Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala at linux.intel.com>
Acked-by: Mika Kuoppala <mika.kuoppala at linux.intel.com>
---
 drivers/gpu/drm/i915/gt/intel_ggtt.c | 11 ++++++++++-
 1 file changed, 10 insertions(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/i915/gt/intel_ggtt.c b/drivers/gpu/drm/i915/gt/intel_ggtt.c
index cf94525be2c1..f5b981443117 100644
--- a/drivers/gpu/drm/i915/gt/intel_ggtt.c
+++ b/drivers/gpu/drm/i915/gt/intel_ggtt.c
@@ -101,7 +101,16 @@ static bool needs_idle_maps(struct drm_i915_private *i915)
 	 * Query intel_iommu to see if we need the workaround. Presumably that
 	 * was loaded first.
 	 */
-	return IS_GEN(i915, 5) && IS_MOBILE(i915) && intel_vtd_active();
+	if (!intel_vtd_active())
+		return false;
+
+	if (IS_GEN(i915, 5) && IS_MOBILE(i915))
+		return true;
+
+	if (IS_GEN(i915, 12))
+		return true; /* XXX DMAR fault reason 7 */
+
+	return false;
 }
 
 void i915_ggtt_suspend(struct i915_ggtt *ggtt)
-- 
2.20.1