This is a note to let you know that I've just added the patch titled
drm/ttm: fix ttm_bo_cleanup_refs_or_queue once more
to the 4.14-stable tree which can be found at: http://www.kernel.org/git/?p=linux/kernel/git/stable/stable-queue.git%3Ba=su...
The filename of the patch is: drm-ttm-fix-ttm_bo_cleanup_refs_or_queue-once-more.patch and it can be found in the queue-4.14 subdirectory.
If you, or anyone else, feels it should not be added to the stable tree, please let stable@vger.kernel.org know about it.
From 378e2d5b504fe0231c557751e58b80fcf717cc20 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Christian=20K=C3=B6nig?= christian.koenig@amd.com Date: Mon, 4 Sep 2017 20:58:45 +0200 Subject: drm/ttm: fix ttm_bo_cleanup_refs_or_queue once more MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit
From: Christian König christian.koenig@amd.com
commit 378e2d5b504fe0231c557751e58b80fcf717cc20 upstream.
With shared reservation objects __ttm_bo_reserve() can easily fail even on destroyed BOs. This prevents correct handling when we need to individualize the reservation object.
Fix this by individualizing the object before even trying to reserve it.
Signed-off-by: Christian König christian.koenig@amd.com Acked-by: Chunming Zhou david1.zhou@amd.com Signed-off-by: Alex Deucher alexander.deucher@amd.com Signed-off-by: Lyude Paul lyude@redhat.com Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org
--- drivers/gpu/drm/ttm/ttm_bo.c | 32 +++++++++++++++++--------------- 1 file changed, 17 insertions(+), 15 deletions(-)
--- a/drivers/gpu/drm/ttm/ttm_bo.c +++ b/drivers/gpu/drm/ttm/ttm_bo.c @@ -440,28 +440,29 @@ static void ttm_bo_cleanup_refs_or_queue struct ttm_bo_global *glob = bo->glob; int ret;
+ ret = ttm_bo_individualize_resv(bo); + if (ret) { + /* Last resort, if we fail to allocate memory for the + * fences block for the BO to become idle + */ + reservation_object_wait_timeout_rcu(bo->resv, true, false, + 30 * HZ); + spin_lock(&glob->lru_lock); + goto error; + } + spin_lock(&glob->lru_lock); ret = __ttm_bo_reserve(bo, false, true, NULL); - if (!ret) { - if (!ttm_bo_wait(bo, false, true)) { + if (reservation_object_test_signaled_rcu(&bo->ttm_resv, true)) { ttm_bo_del_from_lru(bo); spin_unlock(&glob->lru_lock); + if (bo->resv != &bo->ttm_resv) + reservation_object_unlock(&bo->ttm_resv); ttm_bo_cleanup_memtype_use(bo); - return; }
- ret = ttm_bo_individualize_resv(bo); - if (ret) { - /* Last resort, if we fail to allocate memory for the - * fences block for the BO to become idle and free it. - */ - spin_unlock(&glob->lru_lock); - ttm_bo_wait(bo, true, true); - ttm_bo_cleanup_memtype_use(bo); - return; - } ttm_bo_flush_all_fences(bo);
/* @@ -474,11 +475,12 @@ static void ttm_bo_cleanup_refs_or_queue ttm_bo_add_to_lru(bo); }
- if (bo->resv != &bo->ttm_resv) - reservation_object_unlock(&bo->ttm_resv); __ttm_bo_unreserve(bo); } + if (bo->resv != &bo->ttm_resv) + reservation_object_unlock(&bo->ttm_resv);
+error: kref_get(&bo->list_kref); list_add_tail(&bo->ddestroy, &bdev->ddestroy); spin_unlock(&glob->lru_lock);
Patches currently in stable-queue which might be from christian.koenig@amd.com are
queue-4.14/drm-ttm-fix-ttm_bo_cleanup_refs_or_queue-once-more.patch queue-4.14/drm-amdgpu-potential-uninitialized-variable-in-amdgpu_vce_ring_parse_cs.patch queue-4.14/drm-amdgpu-properly-allocate-vm-invalidate-eng-v2.patch queue-4.14/drm-amdgpu-potential-uninitialized-variable-in-amdgpu_vm_update_directories.patch queue-4.14/drm-amdgpu-fix-error-handling-in-amdgpu_bo_do_create.patch queue-4.14/dma-buf-make-reservation_object_copy_fences-rcu-save.patch