OSDN Git Service

drm/amdgpu: fix scatter-gather mapping with user pages
authorShane Francis <bigbeeshane@gmail.com>
Wed, 25 Mar 2020 09:07:40 +0000 (09:07 +0000)
committerAlex Deucher <alexander.deucher@amd.com>
Wed, 25 Mar 2020 16:10:40 +0000 (12:10 -0400)
Calls to dma_map_sg may return less segments / entries than requested
if they fall on page bounderies. The old implementation did not
support this use case.

Fixes: be62dbf554c5 ("iommu/amd: Convert AMD iommu driver to the dma-iommu api")
Bug: https://bugzilla.kernel.org/show_bug.cgi?id=206461
Bug: https://bugzilla.kernel.org/show_bug.cgi?id=206895
Bug: https://gitlab.freedesktop.org/drm/amd/issues/1056
Signed-off-by: Shane Francis <bigbeeshane@gmail.com>
Reviewed-by: Michael J. Ruhl <michael.j.ruhl@intel.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200325090741.21957-3-bigbeeshane@gmail.com
Cc: stable@vger.kernel.org
drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c

index dee4462..c6e9885 100644 (file)
@@ -974,7 +974,7 @@ static int amdgpu_ttm_tt_pin_userptr(struct ttm_tt *ttm)
        /* Map SG to device */
        r = -ENOMEM;
        nents = dma_map_sg(adev->dev, ttm->sg->sgl, ttm->sg->nents, direction);
-       if (nents != ttm->sg->nents)
+       if (nents == 0)
                goto release_sg;
 
        /* convert SG to linear array of pages and dma addresses */