On 2026-11-16 10:24, Christian König wrote:
On 11/13/26 21:40, Kuehling, Felix wrote:
On 2026-11-13 15:29, Shahyan Soltani wrote:
Previously, the root PT was never cleared after setting vm->is_npa =
true within amdgpu_vm_make_npa(). This resulted in the flag
adev->gmc.no_retry_flags to never be applied for the NPA VMs.

Call amdgpu_vm_pt_clear() after vm->is_npa = true is set and the bo has
been pinned/validated. This allows the root PT of the NPA VM to be
cleared with the correct value.

If the pinning or reservation of the bo failed, the vm->is_npa flag will
no longer remain as true. This is to avoid conditionals outside of
amdgpu_vm_make_npa() that depend on the vm->is_npa flag to determine if
the VM is a correctly initialized NPA VM.

Warn if the VM already has mappings, since clearing the root PT will
break existing mappings. Do a heavy weight TLB flush after the root PT
clear to avoid any stale translations being left behind from a previous
table.

Remove the redundant second setting of vm->is_npa = true.

v2:
- Add the WARN_ON sanity check regarding an empty VM. (Felix)
- Change the return handling to only set vm->is_npa = true upon a
    successful pin (Felix)
- Add the heavy weight TLB flush after the root PT clear (Philip)

Suggested-by: Christian König<[email protected]>
Signed-off-by: Shahyan Soltani<[email protected]>
---
   drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 13 ++++++++++---
   1 file changed, 10 insertions(+), 3 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c 
b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
index 94b1cfe05e3e..c7187e0f99b3 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
@@ -2820,20 +2820,27 @@ int amdgpu_vm_make_compute(struct amdgpu_device *adev, 
struct amdgpu_vm *vm)
    */
   int amdgpu_vm_make_npa(struct amdgpu_device *adev, struct amdgpu_vm *vm)
   {
-    int r = amdgpu_vm_make_compute(adev, vm);
+    int r;
   +    r = amdgpu_vm_make_compute(adev, vm);
       if (r)
           return r;
-    vm->is_npa = true;
+
       r = amdgpu_bo_reserve(vm->root.bo, false);
       if (r)
           return r;
You need to reserve fence slots both for pin and clear after locking the root 
PD.

       r = amdgpu_bo_pin(vm->root.bo, AMDGPU_GEM_DOMAIN_VRAM);
+    if (!r) {
+        /* Check if VM mappings are empty, clearing would break existing 
mapping */
+        WARN_ON(!RB_EMPTY_ROOT(&vm->va.rb_root));
You should probably move that to the beginning as if 
(WARN_ON(!RB_EMPTY_ROOT(&vm->va.rb_root))) return -EINVAL;

+        vm->is_npa = true;
+        r = amdgpu_vm_pt_clear(adev, vm, to_amdgpu_bo_vm(vm->root.bo), false);
+        amdgpu_ualink_flush_tlb(adev, TLB_FLUSH_HEAVYWEIGHT);
The TLB flush isn't really needed IMO. We don't expect there to be any previous 
mappings that would need to be flushed. If there were previous mappings, we're 
in a broken state anyway, and the WARN_ON would tell you about it.
Completely agree. Sashiko-bot also complained that this wouldn't even compile.
I agree that there is no mapping on VMID 15 beforehand (which raises the question of why we clear the page tables in the first place?) My concern is stale entries in the cache: after we set up the VMID 15 page table address, we could still get an invalid translation.

Sashiko-bot complained "premature TLB flush in amdgpu_vm_make_npa()", probably means we should flush TLB after vmid 15 pt setup, like this? amdgpu_ualink_manager_start() { .... amdgpu_vm_init amdgpu_vm_make_npa amdgpu_vm_pt_clearsetup_vm_pt_regs amdgpu_ualink_flush_tlb ... } Regards,
Philip

Regards,
Christian.

With the flush_tlb removed, this patch is

Reviewed-by: Felix Kuehling<[email protected]>


+    }
       amdgpu_bo_unreserve(vm->root.bo);
       if (r)
           return r;
   -    vm->is_npa = true;
       vm->need_tlb_fence = false;
         return 0;

Reply via email to