I traced a stability issue on my Braswell nuc to a lack of visibility of PTE writes into the GTT by the GPU. (The smoking gun was GPU hangs with random fault addresses but correct command streams). Adding clflushes or kicking the chipset flush harder had no effect, only disabling the WC cache for the GSM like for Broxton removed the hangs.
Signed-off-by: Chris Wilson <[email protected]> Cc: [email protected] --- drivers/gpu/drm/i915/i915_gem_gtt.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/i915/i915_gem_gtt.c b/drivers/gpu/drm/i915/i915_gem_gtt.c index 10efdc11c2e6..1f685e630d73 100644 --- a/drivers/gpu/drm/i915/i915_gem_gtt.c +++ b/drivers/gpu/drm/i915/i915_gem_gtt.c @@ -2791,8 +2791,10 @@ static int ggtt_probe_common(struct drm_device *dev, * when the WC buffer is flushed, so we can't use it, but have to * resort to an uncached mapping. The WC issue is easily caught by the * readback check when writing GTT PTE entries. + * + * Hmm, it seems that BSW also has a similar deficiency... */ - if (IS_BROXTON(dev)) + if (IS_CHERRYVIEW(dev_priv) || IS_BROXTON(dev_priv)) dev_priv->gtt.gsm = ioremap_nocache(gtt_phys_addr, gtt_size); else dev_priv->gtt.gsm = ioremap_wc(gtt_phys_addr, gtt_size); -- 2.5.0 -- To unsubscribe from this list: send the line "unsubscribe stable" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html
