[PATCH 2/2] drm/amdgpu: Improve huge page mapping update

[Philip Yang]

Update huge page mapping, ex 2MB address and size aligned, we alloc PTB
bo, and then free the PTB bo after updating PDE0 as PTE.

If fragment size >= parent_shift, don't alloc PT bo, because we will
update PDE entry, this will improve the huge page mapping update
by removing the extra PTB bo alloc and free.

Signed-off-by: Philip Yang <Philip.Yang at amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_vm_pt.c | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_pt.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_pt.c
index a3d609655ce3..ef3ef03e50ab 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_pt.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_pt.c
@@ -916,7 +916,11 @@ int amdgpu_vm_ptes_update(struct amdgpu_vm_update_params *params,
 		uint64_t incr, entry_end, pe_start;
 		struct amdgpu_bo *pt;
 
-		if (!params->unlocked) {
+		shift = amdgpu_vm_pt_level_shift(adev, cursor.level);
+		parent_shift = amdgpu_vm_pt_level_shift(adev, cursor.level - 1);
+
+		if (!params->unlocked &&
+		    (adev->asic_type < CHIP_VEGA10 || frag < parent_shift)) {
 			/* make sure that the page tables covering the
 			 * address range are actually allocated
 			 */
@@ -926,8 +930,6 @@ int amdgpu_vm_ptes_update(struct amdgpu_vm_update_params *params,
 				return r;
 		}
 
-		shift = amdgpu_vm_pt_level_shift(adev, cursor.level);
-		parent_shift = amdgpu_vm_pt_level_shift(adev, cursor.level - 1);
 		if (params->unlocked) {
 			/* Unlocked updates are only allowed on the leaves */
 			if (amdgpu_vm_pt_descendant(adev, &cursor))
-- 
2.35.1