Re: [PATCH] drm/amdkfd: relax checks for over allocation of save area

[Philip Yang]

The size up-limit check is not needed because user buffer validation 
will fail if the size is larger than the cwsr_bo or cwsr svm range size 
allocated.

Reviewed-by: Philip.Yang<philip.y...@amd.com>
On 2025-11-10 09:50, Kim, Jonathan wrote:
[Public]

Ping

-----Original Message-----
From: Kim, Jonathan <jonathan....@amd.com>
Sent: Friday, November 7, 2025 7:57 PM
To: amd-gfx@lists.freedesktop.org
Cc: Deucher, Alexander <alexander.deuc...@amd.com>; Kuehling, Felix
<felix.kuehl...@amd.com>; Six, Lancelot <lancelot....@amd.com>; Yang, Philip
<philip.y...@amd.com>; Kim, Jonathan <jonathan....@amd.com>
Subject: [PATCH] drm/amdkfd: relax checks for over allocation of save area

Over allocation of save area is not fatal, only under allocation is.
ROCm has various components that independently claim authority over save
area size.

Unless KFD decides to claim single authority, relax size checks.

v2: remove warning

Signed-off-by: Jonathan Kim <jonathan....@amd.com>
---
  drivers/gpu/drm/amd/amdkfd/kfd_queue.c | 12 ++++++------
  1 file changed, 6 insertions(+), 6 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_queue.c
b/drivers/gpu/drm/amd/amdkfd/kfd_queue.c
index a65c67cf56ff..f1e7583650c4 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_queue.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_queue.c
@@ -297,16 +297,16 @@ int kfd_queue_acquire_buffers(struct
kfd_process_device *pdd, struct queue_prope
               goto out_err_unreserve;
       }

-     if (properties->ctx_save_restore_area_size != topo_dev-
node_props.cwsr_size) {
-             pr_debug("queue cwsr size 0x%x not equal to node cwsr size
0x%x\n",
+     if (properties->ctx_save_restore_area_size < topo_dev-
node_props.cwsr_size) {
+             pr_debug("queue cwsr size 0x%x not sufficient for node cwsr size
0x%x\n",
                       properties->ctx_save_restore_area_size,
                       topo_dev->node_props.cwsr_size);
               err = -EINVAL;
               goto out_err_unreserve;
       }

-     total_cwsr_size = (topo_dev->node_props.cwsr_size + topo_dev-
node_props.debug_memory_size)
-                       * NUM_XCC(pdd->dev->xcc_mask);
+     total_cwsr_size = (properties->ctx_save_restore_area_size +
+                        topo_dev->node_props.debug_memory_size) *
NUM_XCC(pdd->dev->xcc_mask);
       total_cwsr_size = ALIGN(total_cwsr_size, PAGE_SIZE);

       err = kfd_queue_buffer_get(vm, (void *)properties-
ctx_save_restore_area_address,
@@ -352,8 +352,8 @@ int kfd_queue_release_buffers(struct kfd_process_device
*pdd, struct queue_prope
       topo_dev = kfd_topology_device_by_id(pdd->dev->id);
       if (!topo_dev)
               return -EINVAL;
-     total_cwsr_size = (topo_dev->node_props.cwsr_size + topo_dev-
node_props.debug_memory_size)
-                       * NUM_XCC(pdd->dev->xcc_mask);
+     total_cwsr_size = (properties->ctx_save_restore_area_size +
+                        topo_dev->node_props.debug_memory_size) *
NUM_XCC(pdd->dev->xcc_mask);
       total_cwsr_size = ALIGN(total_cwsr_size, PAGE_SIZE);

       kfd_queue_buffer_svm_put(pdd, properties-
ctx_save_restore_area_address, total_cwsr_size);
--
2.34.1