On 17/04/15 22:21, yu.dai@xxxxxxxxx wrote: > From: Sagar Kamble <sagar.a.kamble@xxxxxxxxx> > > The firmware loader will use GuC DMA engine to move data from > ggtt to WOPCM. Need to take forcewake before GuC loading. > > Issue: VIZ-4884 > Change-Id: Ie422fc1e122933b161ff63cab23622197e6bba54 > Signed-off-by: Sagar Kamble <sagar.a.kamble@xxxxxxxxx> > --- > drivers/gpu/drm/i915/intel_guc_loader.c | 3 +++ > 1 file changed, 3 insertions(+) > > diff --git a/drivers/gpu/drm/i915/intel_guc_loader.c b/drivers/gpu/drm/i915/intel_guc_loader.c > index d1fd11e..f19eb90 100644 > --- a/drivers/gpu/drm/i915/intel_guc_loader.c > +++ b/drivers/gpu/drm/i915/intel_guc_loader.c > @@ -340,6 +340,7 @@ static int guc_load_ucode(struct drm_device *dev) > int ret; > > WARN_ON(!mutex_is_locked(&dev->struct_mutex)); > + intel_uncore_forcewake_get(dev_priv, FORCEWAKE_ALL); > > ret = i915_gem_obj_ggtt_pin(guc_fw->uc_fw_obj, 0, 0); > if (ret) > @@ -388,6 +389,8 @@ out: > if (pinned) > i915_gem_object_ggtt_unpin(guc_fw->uc_fw_obj); > > + intel_uncore_forcewake_put(dev_priv, FORCEWAKE_ALL); > + > return ret; > } This doesn't need to be a separate patch; it can be squashed into the initial creation of the GuC loader, so we don't (even temporarily) have a broken version in the codebase. We keep the authorship, of course :) And maybe turn the commit message into a code comment in the affected function? .Dave. _______________________________________________ Intel-gfx mailing list Intel-gfx@xxxxxxxxxxxxxxxxxxxxx http://lists.freedesktop.org/mailman/listinfo/intel-gfx
