drm/amdgpu: Add EXT_COHERENT support for APU and NUMA systems

[linux-2.6-microblaze.git] / drivers / gpu / drm / amd / amdgpu / gmc_v9_0.c

diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
index fee3141..b66c5f7 100644 (file)
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
@@ -1251,12 +1251,15 @@ static void gmc_v9_0_override_vm_pte_flags(struct amdgpu_device *adev,
                return;
        }
 
-       /* Only override mappings with MTYPE_NC, which is the safe default for
-        * cacheable memory.
+       /* MTYPE_NC is the same default and can be overridden.
+        * MTYPE_UC will be present if the memory is extended-coherent
+        * and can also be overridden.
         */
        if ((*flags & AMDGPU_PTE_MTYPE_VG10_MASK) !=
-           AMDGPU_PTE_MTYPE_VG10(MTYPE_NC)) {
-               dev_dbg_ratelimited(adev->dev, "MTYPE is not NC\n");
+           AMDGPU_PTE_MTYPE_VG10(MTYPE_NC) &&
+           (*flags & AMDGPU_PTE_MTYPE_VG10_MASK) !=
+           AMDGPU_PTE_MTYPE_VG10(MTYPE_UC)) {
+               dev_dbg_ratelimited(adev->dev, "MTYPE is not NC or UC\n");
                return;
        }
 
@@ -1283,15 +1286,23 @@ static void gmc_v9_0_override_vm_pte_flags(struct amdgpu_device *adev,
                            vm->mem_id, local_node, nid);
        if (nid == local_node) {
                uint64_t old_flags = *flags;
-               unsigned int mtype_local = MTYPE_RW;
+               if ((*flags & AMDGPU_PTE_MTYPE_VG10_MASK) ==
+                       AMDGPU_PTE_MTYPE_VG10(MTYPE_NC)) {
+                       unsigned int mtype_local = MTYPE_RW;
 
-               if (amdgpu_mtype_local == 1)
-                       mtype_local = MTYPE_NC;
-               else if (amdgpu_mtype_local == 2)
-                       mtype_local = MTYPE_CC;
+                       if (amdgpu_mtype_local == 1)
+                               mtype_local = MTYPE_NC;
+                       else if (amdgpu_mtype_local == 2)
+                               mtype_local = MTYPE_CC;
+
+                       *flags = (*flags & ~AMDGPU_PTE_MTYPE_VG10_MASK) |
+                                AMDGPU_PTE_MTYPE_VG10(mtype_local);
+               } else {
+                       /* MTYPE_UC case */
+                       *flags = (*flags & ~AMDGPU_PTE_MTYPE_VG10_MASK) |
+                                AMDGPU_PTE_MTYPE_VG10(MTYPE_CC);
+               }
 
-               *flags = (*flags & ~AMDGPU_PTE_MTYPE_VG10_MASK) |
-                        AMDGPU_PTE_MTYPE_VG10(mtype_local);
                dev_dbg_ratelimited(adev->dev, "flags updated from %llx to %llx\n",
                                    old_flags, *flags);
        }