[PATCH] drm/msm: drop cache sync hack

[PATCH] drm/msm: drop cache sync hack

[Rob Clark]

From: Rob Clark <robdclark@chromium.org>

Now that it isn't causing problems to use dma_map/unmap, we can drop the
hack of using dma_sync in certain cases.

Signed-off-by: Rob Clark <robdclark@chromium.org>
---
 drivers/gpu/drm/msm/msm_gem.c | 18 ++++--------------
 1 file changed, 4 insertions(+), 14 deletions(-)

diff --git a/drivers/gpu/drm/msm/msm_gem.c b/drivers/gpu/drm/msm/msm_gem.c
index b2f49152b4d4..3cb7aeb93fd3 100644
--- a/drivers/gpu/drm/msm/msm_gem.c
+++ b/drivers/gpu/drm/msm/msm_gem.c
@@ -52,26 +52,16 @@ static void sync_for_device(struct msm_gem_object *msm_obj)
 {
 	struct device *dev = msm_obj->base.dev->dev;
 
-	if (get_dma_ops(dev) && IS_ENABLED(CONFIG_ARM64)) {
-		dma_sync_sg_for_device(dev, msm_obj->sgt->sgl,
-			msm_obj->sgt->nents, DMA_BIDIRECTIONAL);
-	} else {
-		dma_map_sg(dev, msm_obj->sgt->sgl,
-			msm_obj->sgt->nents, DMA_BIDIRECTIONAL);
-	}
+	dma_map_sg(dev, msm_obj->sgt->sgl,
+		msm_obj->sgt->nents, DMA_BIDIRECTIONAL);
 }
 
 static void sync_for_cpu(struct msm_gem_object *msm_obj)
 {
 	struct device *dev = msm_obj->base.dev->dev;
 
-	if (get_dma_ops(dev) && IS_ENABLED(CONFIG_ARM64)) {
-		dma_sync_sg_for_cpu(dev, msm_obj->sgt->sgl,
-			msm_obj->sgt->nents, DMA_BIDIRECTIONAL);
-	} else {
-		dma_unmap_sg(dev, msm_obj->sgt->sgl,
-			msm_obj->sgt->nents, DMA_BIDIRECTIONAL);
-	}
+	dma_unmap_sg(dev, msm_obj->sgt->sgl,
+		msm_obj->sgt->nents, DMA_BIDIRECTIONAL);
 }
 
 /* allocate pages from VRAM carveout, used when no IOMMU: */
-- 
2.26.2

Re: [PATCH] drm/msm: drop cache sync hack

[Christoph Hellwig]

On Sat, Aug 22, 2020 at 10:52:54AM -0700, Rob Clark wrote:
> From: Rob Clark <robdclark@chromium.org>
> 
> Now that it isn't causing problems to use dma_map/unmap, we can drop the
> hack of using dma_sync in certain cases.

Great to see!  What did solve the problems?

Re: [PATCH] drm/msm: drop cache sync hack

[Rob Clark]

On Mon, Aug 24, 2020 at 11:52 PM Christoph Hellwig <hch@infradead.org> wrote:
>
> On Sat, Aug 22, 2020 at 10:52:54AM -0700, Rob Clark wrote:
> > From: Rob Clark <robdclark@chromium.org>
> >
> > Now that it isn't causing problems to use dma_map/unmap, we can drop the
> > hack of using dma_sync in certain cases.
>
> Great to see!  What did solve the problems?

should be 0e764a01015dfebff8a8ffd297d74663772e248a ("iommu/arm-smmu:
Allow client devices to select direct mapping")

I still need to confirm whether qcom_iommu needs a similar thing, but
I think it is ok as the iommu phandle link is down one level on the
'mdp' device, rather than attached to the toplevel drm device.

BR,
-R