drm/ttm: Fix ttm_bo_swapout() infinite LRU walk on swapout failure

When ttm_tt_swapout() fails, the current code calls
ttm_resource_add_bulk_move() followed by ttm_resource_move_to_lru_tail()
to restore the resource's bulk_move membership.

However, ttm_resource_move_to_lru_tail() places the resource at the tail
of the LRU list which, relative to the walk cursor's hitch node (placed
immediately after the resource when it was yielded), puts the resource
*in front of the* the hitch. The next list_for_each_entry_continue() from
the hitch finds the same resource again, causing an infinite loop.

Fix by deferring del_bulk_move to the success path only.

On the success path, TTM_TT_FLAG_SWAPPED has just been set by
ttm_tt_swapout() but the resource is still tracked in the bulk_move range,
so ttm_resource_del_bulk_move()'s !ttm_resource_unevictable() guard would
incorrectly skip the removal. Introduce
ttm_resource_del_bulk_move_unevictable() which bypasses that guard.

Reported-by: Jatin Kataria <jkataria@netflix.com>
Fixes: fc5d96670eb2 ("drm/ttm: Move swapped objects off the manager's LRU list")
Cc: Christian König <christian.koenig@amd.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Cc: <dri-devel@lists.freedesktop.org>
Cc: <stable@vger.kernel.org> # v6.13+
Assisted-by: GitHub_Copilot:claude-sonnet-4.6
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Tested-by: Boqun Feng <boqun@kernel.org>
Link: https://patch.msgid.link/20260428094442.16985-1-thomas.hellstrom@linux.intel.com

diff --git a/drivers/gpu/drm/ttm/ttm_bo.c b/drivers/gpu/drm/ttm/ttm_bo.c
index d85f0a37ac35f36ce526519ad5d4395752211f06..2934017055429ca15b22207d2806dd29cc95a44d 100644 (file)
--- a/drivers/gpu/drm/ttm/ttm_bo.c
+++ b/drivers/gpu/drm/ttm/ttm_bo.c
@@ -1177,17 +1177,13 @@ ttm_bo_swapout_cb(struct ttm_lru_walk *walk, struct ttm_buffer_object *bo)
                bdev->funcs->swap_notify(bo);
 
        if (ttm_tt_is_populated(tt)) {
-               spin_lock(&bdev->lru_lock);
-               ttm_resource_del_bulk_move(bo->resource, bo);
-               spin_unlock(&bdev->lru_lock);
-
                ret = ttm_tt_swapout(bdev, tt, swapout_walk->gfp_flags);
-
-               spin_lock(&bdev->lru_lock);
-               if (ret)
-                       ttm_resource_add_bulk_move(bo->resource, bo);
-               ttm_resource_move_to_lru_tail(bo->resource);
-               spin_unlock(&bdev->lru_lock);
+               if (!ret) {
+                       spin_lock(&bdev->lru_lock);
+                       ttm_resource_del_bulk_move_unevictable(bo->resource, bo);
+                       ttm_resource_move_to_lru_tail(bo->resource);
+                       spin_unlock(&bdev->lru_lock);
+               }
        }
 
 out:

diff --git a/drivers/gpu/drm/ttm/ttm_resource.c b/drivers/gpu/drm/ttm/ttm_resource.c
index 9f36631d48b617a8a0260ee114ebf5a8edb07fdb..0e5f1582f13d8466e17b9cbfb3827e9667f045c0 100644 (file)
--- a/drivers/gpu/drm/ttm/ttm_resource.c
+++ b/drivers/gpu/drm/ttm/ttm_resource.c
@@ -292,6 +292,19 @@ void ttm_resource_del_bulk_move(struct ttm_resource *res,
                ttm_lru_bulk_move_del(bo->bulk_move, res);
 }
 
+/*
+ * Remove a resource from its bulk_move, bypassing the unevictable check.
+ * Use only when the resource is known to still be tracked in the range despite
+ * the BO having just become unevictable; asserts that this is the case.
+ */
+void ttm_resource_del_bulk_move_unevictable(struct ttm_resource *res,
+                                           struct ttm_buffer_object *bo)
+{
+       WARN_ON_ONCE(!ttm_resource_unevictable(res, bo));
+       if (bo->bulk_move)
+               ttm_lru_bulk_move_del(bo->bulk_move, res);
+}
+
 /* Move a resource to the LRU or bulk tail */
 void ttm_resource_move_to_lru_tail(struct ttm_resource *res)
 {

diff --git a/include/drm/ttm/ttm_resource.h b/include/drm/ttm/ttm_resource.h
index 33e80f30b8b822682be80ef4dcacf42dbd9a5dd0..a5d386583fb6e0e6a2d7ec5e25b12f054cd27e4e 100644 (file)
--- a/include/drm/ttm/ttm_resource.h
+++ b/include/drm/ttm/ttm_resource.h
@@ -448,6 +448,8 @@ void ttm_resource_add_bulk_move(struct ttm_resource *res,
                                struct ttm_buffer_object *bo);
 void ttm_resource_del_bulk_move(struct ttm_resource *res,
                                struct ttm_buffer_object *bo);
+void ttm_resource_del_bulk_move_unevictable(struct ttm_resource *res,
+                                           struct ttm_buffer_object *bo);
 void ttm_resource_move_to_lru_tail(struct ttm_resource *res);
 
 void ttm_resource_init(struct ttm_buffer_object *bo,