drm/i915: Never fully mask the the EI up rps interrupt on SNB/IVB
authorVille Syrjälä <ville.syrjala@linux.intel.com>
Mon, 23 May 2016 14:42:48 +0000 (17:42 +0300)
committerChris Wilson <chris@chris-wilson.co.uk>
Thu, 28 Jul 2016 06:23:24 +0000 (07:23 +0100)
SNB (and IVB too I suppose) starts to misbehave if the GPU gets stuck
in an infinite batch buffer loop. The GPU apparently hogs something
critical and CPUs start to lose interrupts and whatnot. We can keep
the system limping along by unmasking some interrupts in
GEN6_PMINTRMSK. The EI up interrupt has been previously chosen for
that task, so let's never mask it.

v2: s/gen6_rps_pm_mask/gen6_sanitize_rps_pm_mask/ (Chris)

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=93122
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: http://patchwork.freedesktop.org/patch/msgid/1464014568-4529-1-git-send-email-ville.syrjala@linux.intel.com
Cc: stable@vger.kernel.org
drivers/gpu/drm/i915/intel_pm.c

index 64d628c915a39adadf94c3ecbbb813372fdc3cc7..e50505c67ef9185e497bd23fd4a52b58f692c3f9 100644 (file)
@@ -4892,7 +4892,8 @@ void gen6_rps_idle(struct drm_i915_private *dev_priv)
                else
                        gen6_set_rps(dev_priv, dev_priv->rps.idle_freq);
                dev_priv->rps.last_adj = 0;
-               I915_WRITE(GEN6_PMINTRMSK, 0xffffffff);
+               I915_WRITE(GEN6_PMINTRMSK,
+                          gen6_sanitize_rps_pm_mask(dev_priv, ~0));
        }
        mutex_unlock(&dev_priv->rps.hw_lock);