Re: [PATCH V1] accel/amdxdna: fix open_ref leak and stale client pointer on dma map failure

[Lizhi Hou]

Applied to drm-misc-fixes

On 7/7/26 08:09, Max Zhen wrote:

On 7/6/2026 Mon 22:56, Lizhi Hou wrote:

amdxdna_gem_obj_open() increments open_ref before attempting to set up
the DMA address mapping.  When amdxdna_dma_map_bo() fails, the function

returned immediately without rolling back either change made on the first

open (open_ref == 1 path).

Fix it by decrementing open_ref and clearing abo->client on the error path.

Fixes: ece3e8980907 ("accel/amdxdna: Allow forcing IOVA-based DMA via module parameter")

Signed-off-by: Lizhi Hou <lizhi....@amd.com>

Reviewed-by: Max Zhen <max.z...@amd.com>

---
  drivers/accel/amdxdna/amdxdna_gem.c | 5 ++++-
  1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/drivers/accel/amdxdna/amdxdna_gem.c b/drivers/accel/amdxdna/amdxdna_gem.c

index a4bd9e86bc8d..3f03d2db1603 100644
--- a/drivers/accel/amdxdna/amdxdna_gem.c
+++ b/drivers/accel/amdxdna/amdxdna_gem.c

@@ -663,8 +663,11 @@ static int amdxdna_gem_obj_open(struct drm_gem_object *gobj, struct drm_file *fi

      /* No need to set up dma addr mapping in PASID mode. */
      if (!amdxdna_pasid_on(abo->client)) {
          ret = amdxdna_dma_map_bo(xdna, abo);
-        if (ret)
+        if (ret) {
+            abo->open_ref--;
+            abo->client = NULL;
              return ret;
+        }
      }
        amdxdna_gem_add_bo_usage(abo);