From: Sergey Senozhatsky <senozhatsky@xxxxxxxxxxxx> commit ea86f3defd55f141a44146e66cbf8ffb683d60da upstream. We observed that some of virtio_gpu_object_shmem_init() allocations can be rather costly - order 6 - which can be difficult to fulfill under memory pressure conditions. Switch to kvmalloc_array() in virtio_gpu_object_shmem_init() and let the kernel vmalloc the entries array. Signed-off-by: Sergey Senozhatsky <senozhatsky@xxxxxxxxxxxx> Link: http://patchwork.freedesktop.org/patch/msgid/20201105014744.1662226-1-senozhatsky@xxxxxxxxxxxx Signed-off-by: Gerd Hoffmann <kraxel@xxxxxxxxxx> Signed-off-by: Doug Horn <doughorn@xxxxxxxxxx> Signed-off-by: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx> --- drivers/gpu/drm/virtio/virtgpu_vq.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) --- a/drivers/gpu/drm/virtio/virtgpu_vq.c +++ b/drivers/gpu/drm/virtio/virtgpu_vq.c @@ -865,9 +865,9 @@ int virtio_gpu_object_attach(struct virt } /* gets freed when the ring has consumed it */ - ents = kmalloc_array(obj->pages->nents, - sizeof(struct virtio_gpu_mem_entry), - GFP_KERNEL); + ents = kvmalloc_array(obj->pages->nents, + sizeof(struct virtio_gpu_mem_entry), + GFP_KERNEL); if (!ents) { DRM_ERROR("failed to allocate ent list\n"); return -ENOMEM;
