From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail.linuxfoundation.org ([140.211.169.12]:56476 "EHLO mail.linuxfoundation.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932333AbcKQHO5 (ORCPT ); Thu, 17 Nov 2016 02:14:57 -0500 Subject: Patch "drm/amd: fix scheduler fence teardown order v2" has been added to the 4.8-stable tree To: christian.koenig@amd.com, alexander.deucher@amd.com, gregkh@linuxfoundation.org, notasas@gmail.com Cc: , From: Date: Thu, 17 Nov 2016 08:15:08 +0100 Message-ID: <1479366908222224@kroah.com> MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Sender: stable-owner@vger.kernel.org List-ID: This is a note to let you know that I've just added the patch titled drm/amd: fix scheduler fence teardown order v2 to the 4.8-stable tree which can be found at: http://www.kernel.org/git/?p=linux/kernel/git/stable/stable-queue.git;a=summary The filename of the patch is: drm-amd-fix-scheduler-fence-teardown-order-v2.patch and it can be found in the queue-4.8 subdirectory. If you, or anyone else, feels it should not be added to the stable tree, please let know about it. >>From c24784f01549ecdf23fc00d0588423bcf8956714 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Christian=20K=C3=B6nig?= Date: Fri, 28 Oct 2016 17:04:07 +0200 Subject: drm/amd: fix scheduler fence teardown order v2 MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit From: Christian König commit c24784f01549ecdf23fc00d0588423bcf8956714 upstream. Some fences might be alive even after we have stopped the scheduler leading to warnings about leaked objects from the SLUB allocator. Fix this by allocating/freeing the SLUB allocator from the module init/fini functions just like we do it for hw fences. v2: make variable static, add link to bug Fixes: https://bugs.freedesktop.org/show_bug.cgi?id=97500 Reported-by: Grazvydas Ignotas Signed-off-by: Christian König Reviewed-by: Alex Deucher (v1) Signed-off-by: Alex Deucher Signed-off-by: Greg Kroah-Hartman --- drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c | 2 ++ drivers/gpu/drm/amd/scheduler/gpu_scheduler.c | 13 ------------- drivers/gpu/drm/amd/scheduler/gpu_scheduler.h | 6 +++--- drivers/gpu/drm/amd/scheduler/sched_fence.c | 19 +++++++++++++++++++ 4 files changed, 24 insertions(+), 16 deletions(-) --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c @@ -605,6 +605,7 @@ static int __init amdgpu_init(void) { amdgpu_sync_init(); amdgpu_fence_slab_init(); + amd_sched_fence_slab_init(); if (vgacon_text_force()) { DRM_ERROR("VGACON disables amdgpu kernel modesetting.\n"); return -EINVAL; @@ -624,6 +625,7 @@ static void __exit amdgpu_exit(void) drm_pci_exit(driver, pdriver); amdgpu_unregister_atpx_handler(); amdgpu_sync_fini(); + amd_sched_fence_slab_fini(); amdgpu_fence_slab_fini(); } --- a/drivers/gpu/drm/amd/scheduler/gpu_scheduler.c +++ b/drivers/gpu/drm/amd/scheduler/gpu_scheduler.c @@ -34,9 +34,6 @@ static bool amd_sched_entity_is_ready(st static void amd_sched_wakeup(struct amd_gpu_scheduler *sched); static void amd_sched_process_job(struct fence *f, struct fence_cb *cb); -struct kmem_cache *sched_fence_slab; -atomic_t sched_fence_slab_ref = ATOMIC_INIT(0); - /* Initialize a given run queue struct */ static void amd_sched_rq_init(struct amd_sched_rq *rq) { @@ -618,13 +615,6 @@ int amd_sched_init(struct amd_gpu_schedu INIT_LIST_HEAD(&sched->ring_mirror_list); spin_lock_init(&sched->job_list_lock); atomic_set(&sched->hw_rq_count, 0); - if (atomic_inc_return(&sched_fence_slab_ref) == 1) { - sched_fence_slab = kmem_cache_create( - "amd_sched_fence", sizeof(struct amd_sched_fence), 0, - SLAB_HWCACHE_ALIGN, NULL); - if (!sched_fence_slab) - return -ENOMEM; - } /* Each scheduler will run on a seperate kernel thread */ sched->thread = kthread_run(amd_sched_main, sched, sched->name); @@ -645,7 +635,4 @@ void amd_sched_fini(struct amd_gpu_sched { if (sched->thread) kthread_stop(sched->thread); - rcu_barrier(); - if (atomic_dec_and_test(&sched_fence_slab_ref)) - kmem_cache_destroy(sched_fence_slab); } --- a/drivers/gpu/drm/amd/scheduler/gpu_scheduler.h +++ b/drivers/gpu/drm/amd/scheduler/gpu_scheduler.h @@ -30,9 +30,6 @@ struct amd_gpu_scheduler; struct amd_sched_rq; -extern struct kmem_cache *sched_fence_slab; -extern atomic_t sched_fence_slab_ref; - /** * A scheduler entity is a wrapper around a job queue or a group * of other entities. Entities take turns emitting jobs from their @@ -145,6 +142,9 @@ void amd_sched_entity_fini(struct amd_gp struct amd_sched_entity *entity); void amd_sched_entity_push_job(struct amd_sched_job *sched_job); +int amd_sched_fence_slab_init(void); +void amd_sched_fence_slab_fini(void); + struct amd_sched_fence *amd_sched_fence_create( struct amd_sched_entity *s_entity, void *owner); void amd_sched_fence_scheduled(struct amd_sched_fence *fence); --- a/drivers/gpu/drm/amd/scheduler/sched_fence.c +++ b/drivers/gpu/drm/amd/scheduler/sched_fence.c @@ -27,6 +27,25 @@ #include #include "gpu_scheduler.h" +static struct kmem_cache *sched_fence_slab; + +int amd_sched_fence_slab_init(void) +{ + sched_fence_slab = kmem_cache_create( + "amd_sched_fence", sizeof(struct amd_sched_fence), 0, + SLAB_HWCACHE_ALIGN, NULL); + if (!sched_fence_slab) + return -ENOMEM; + + return 0; +} + +void amd_sched_fence_slab_fini(void) +{ + rcu_barrier(); + kmem_cache_destroy(sched_fence_slab); +} + struct amd_sched_fence *amd_sched_fence_create(struct amd_sched_entity *entity, void *owner) { Patches currently in stable-queue which might be from christian.koenig@amd.com are queue-4.8/drm-amdgpu-fix-sched-fence-slab-teardown.patch queue-4.8/drm-radeon-disable-runtime-pm-in-certain-cases.patch queue-4.8/drm-amd-fix-scheduler-fence-teardown-order-v2.patch queue-4.8/drm-amdgpu-disable-runtime-pm-in-certain-cases.patch