[PATCH v5 00/16] drm/panfrost: Misc improvements

[PATCH v5 00/16] drm/panfrost: Misc improvements

[Boris Brezillon]

Hello,

This is a merge of [1] and [2] since the second series depends on
patches in the preparatory series.

Main changes in this v5:
* Document what's excepted in the ->timedout_job() hook
* Add a patch increasing the AS_ACTIVE polling timeout
* Fix a few minor things here and there (see each commit for a detailed
  changelog) and collect R-b/A-b tags

Regards,

Boris

Boris Brezillon (15):
  drm/sched: Document what the timedout_job method should do
  drm/sched: Allow using a dedicated workqueue for the timeout/fault tdr
  drm/panfrost: Make ->run_job() return an ERR_PTR() when appropriate
  drm/panfrost: Get rid of the unused JS_STATUS_EVENT_ACTIVE definition
  drm/panfrost: Drop the pfdev argument passed to
    panfrost_exception_name()
  drm/panfrost: Do the exception -> string translation using a table
  drm/panfrost: Expose a helper to trigger a GPU reset
  drm/panfrost: Use a threaded IRQ for job interrupts
  drm/panfrost: Simplify the reset serialization logic
  drm/panfrost: Make sure job interrupts are masked before resetting
  drm/panfrost: Disable the AS on unhandled page faults
  drm/panfrost: Reset the GPU when the AS_ACTIVE bit is stuck
  drm/panfrost: Don't reset the GPU on job faults unless we really have
    to
  drm/panfrost: Kill in-flight jobs on FD close
  drm/panfrost: Increase the AS_ACTIVE polling timeout

Steven Price (1):
  drm/panfrost: Queue jobs on the hardware

 drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c  |   2 +-
 drivers/gpu/drm/etnaviv/etnaviv_sched.c    |   3 +-
 drivers/gpu/drm/lima/lima_sched.c          |   3 +-
 drivers/gpu/drm/panfrost/panfrost_device.c | 139 +++--
 drivers/gpu/drm/panfrost/panfrost_device.h |  91 ++-
 drivers/gpu/drm/panfrost/panfrost_gpu.c    |   2 +-
 drivers/gpu/drm/panfrost/panfrost_job.c    | 630 +++++++++++++++------
 drivers/gpu/drm/panfrost/panfrost_mmu.c    |  43 +-
 drivers/gpu/drm/panfrost/panfrost_regs.h   |   3 -
 drivers/gpu/drm/scheduler/sched_main.c     |  14 +-
 drivers/gpu/drm/v3d/v3d_sched.c            |  10 +-
 include/drm/gpu_scheduler.h                |  37 +-
 12 files changed, 721 insertions(+), 256 deletions(-)

-- 
2.31.1

[PATCH v5 01/16] drm/sched: Document what the timedout_job method should do

[Boris Brezillon]

The documentation is a bit vague and doesn't really describe what the
->timedout_job() is expected to do. Let's add a few more details.

v5:
* New patch

Suggested-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
---
 include/drm/gpu_scheduler.h | 14 ++++++++++++++
 1 file changed, 14 insertions(+)

diff --git a/include/drm/gpu_scheduler.h b/include/drm/gpu_scheduler.h
index 10225a0a35d0..65700511e074 100644
--- a/include/drm/gpu_scheduler.h
+++ b/include/drm/gpu_scheduler.h
@@ -239,6 +239,20 @@ struct drm_sched_backend_ops {
 	 * @timedout_job: Called when a job has taken too long to execute,
 	 * to trigger GPU recovery.
 	 *
+	 * This method is called in a workqueue context.
+	 *
+	 * Drivers typically issue a reset to recover from GPU hangs, and this
+	 * procedure usually follows the following workflow:
+	 *
+	 * 1. Stop the scheduler using drm_sched_stop(). This will park the
+	 *    scheduler thread and cancel the timeout work, guaranteeing that
+	 *    nothing is queued while we reset the hardware queue
+	 * 2. Try to gracefully stop non-faulty jobs (optional)
+	 * 3. Issue a GPU reset (driver-specific)
+	 * 4. Re-submit jobs using drm_sched_resubmit_jobs()
+	 * 5. Restart the scheduler using drm_sched_start(). At that point, new
+	 *    jobs can be queued, and the scheduler thread is unblocked
+	 *
 	 * Return DRM_GPU_SCHED_STAT_NOMINAL, when all is normal,
 	 * and the underlying driver has started or completed recovery.
 	 *
-- 
2.31.1

[PATCH v5 02/16] drm/sched: Allow using a dedicated workqueue for the timeout/fault tdr

[Boris Brezillon]

Mali Midgard/Bifrost GPUs have 3 hardware queues but only a global GPU
reset. This leads to extra complexity when we need to synchronize timeout
works with the reset work. One solution to address that is to have an
ordered workqueue at the driver level that will be used by the different
schedulers to queue their timeout work. Thanks to the serialization
provided by the ordered workqueue we are guaranteed that timeout
handlers are executed sequentially, and can thus easily reset the GPU
from the timeout handler without extra synchronization.

v5:
* Add a new paragraph to the timedout_job() method

v3:
* New patch

v4:
* Actually use the timeout_wq to queue the timeout work

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Steven Price <steven.price@arm.com>
Reviewed-by: Lucas Stach <l.stach@pengutronix.de>
Cc: Qiang Yu <yuq825@gmail.com>
Cc: Emma Anholt <emma@anholt.net>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: "Christian König" <christian.koenig@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c |  2 +-
 drivers/gpu/drm/etnaviv/etnaviv_sched.c   |  3 ++-
 drivers/gpu/drm/lima/lima_sched.c         |  3 ++-
 drivers/gpu/drm/panfrost/panfrost_job.c   |  3 ++-
 drivers/gpu/drm/scheduler/sched_main.c    | 14 +++++++++-----
 drivers/gpu/drm/v3d/v3d_sched.c           | 10 +++++-----
 include/drm/gpu_scheduler.h               | 23 ++++++++++++++++++++++-
 7 files changed, 43 insertions(+), 15 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c
index 47ea46859618..532636ea20bc 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c
@@ -488,7 +488,7 @@ int amdgpu_fence_driver_init_ring(struct amdgpu_ring *ring,
 
 	r = drm_sched_init(&ring->sched, &amdgpu_sched_ops,
 			   num_hw_submission, amdgpu_job_hang_limit,
-			   timeout, sched_score, ring->name);
+			   timeout, NULL, sched_score, ring->name);
 	if (r) {
 		DRM_ERROR("Failed to create scheduler on ring %s.\n",
 			  ring->name);
diff --git a/drivers/gpu/drm/etnaviv/etnaviv_sched.c b/drivers/gpu/drm/etnaviv/etnaviv_sched.c
index 19826e504efc..feb6da1b6ceb 100644
--- a/drivers/gpu/drm/etnaviv/etnaviv_sched.c
+++ b/drivers/gpu/drm/etnaviv/etnaviv_sched.c
@@ -190,7 +190,8 @@ int etnaviv_sched_init(struct etnaviv_gpu *gpu)
 
 	ret = drm_sched_init(&gpu->sched, &etnaviv_sched_ops,
 			     etnaviv_hw_jobs_limit, etnaviv_job_hang_limit,
-			     msecs_to_jiffies(500), NULL, dev_name(gpu->dev));
+			     msecs_to_jiffies(500), NULL, NULL,
+			     dev_name(gpu->dev));
 	if (ret)
 		return ret;
 
diff --git a/drivers/gpu/drm/lima/lima_sched.c b/drivers/gpu/drm/lima/lima_sched.c
index ecf3267334ff..dba8329937a3 100644
--- a/drivers/gpu/drm/lima/lima_sched.c
+++ b/drivers/gpu/drm/lima/lima_sched.c
@@ -508,7 +508,8 @@ int lima_sched_pipe_init(struct lima_sched_pipe *pipe, const char *name)
 	INIT_WORK(&pipe->recover_work, lima_sched_recover_work);
 
 	return drm_sched_init(&pipe->base, &lima_sched_ops, 1,
-			      lima_job_hang_limit, msecs_to_jiffies(timeout),
+			      lima_job_hang_limit,
+			      msecs_to_jiffies(timeout), NULL,
 			      NULL, name);
 }
 
diff --git a/drivers/gpu/drm/panfrost/panfrost_job.c b/drivers/gpu/drm/panfrost/panfrost_job.c
index 682f2161b999..8ff79fd49577 100644
--- a/drivers/gpu/drm/panfrost/panfrost_job.c
+++ b/drivers/gpu/drm/panfrost/panfrost_job.c
@@ -626,7 +626,8 @@ int panfrost_job_init(struct panfrost_device *pfdev)
 
 		ret = drm_sched_init(&js->queue[j].sched,
 				     &panfrost_sched_ops,
-				     1, 0, msecs_to_jiffies(JOB_TIMEOUT_MS),
+				     1, 0,
+				     msecs_to_jiffies(JOB_TIMEOUT_MS), NULL,
 				     NULL, "pan_js");
 		if (ret) {
 			dev_err(pfdev->dev, "Failed to create scheduler: %d.", ret);
diff --git a/drivers/gpu/drm/scheduler/sched_main.c b/drivers/gpu/drm/scheduler/sched_main.c
index c0a2f8f8d472..3e180f0d4305 100644
--- a/drivers/gpu/drm/scheduler/sched_main.c
+++ b/drivers/gpu/drm/scheduler/sched_main.c
@@ -232,7 +232,7 @@ static void drm_sched_start_timeout(struct drm_gpu_scheduler *sched)
 {
 	if (sched->timeout != MAX_SCHEDULE_TIMEOUT &&
 	    !list_empty(&sched->pending_list))
-		schedule_delayed_work(&sched->work_tdr, sched->timeout);
+		queue_delayed_work(sched->timeout_wq, &sched->work_tdr, sched->timeout);
 }
 
 /**
@@ -244,7 +244,7 @@ static void drm_sched_start_timeout(struct drm_gpu_scheduler *sched)
  */
 void drm_sched_fault(struct drm_gpu_scheduler *sched)
 {
-	mod_delayed_work(system_wq, &sched->work_tdr, 0);
+	mod_delayed_work(sched->timeout_wq, &sched->work_tdr, 0);
 }
 EXPORT_SYMBOL(drm_sched_fault);
 
@@ -270,7 +270,7 @@ unsigned long drm_sched_suspend_timeout(struct drm_gpu_scheduler *sched)
 	 * Modify the timeout to an arbitrarily large value. This also prevents
 	 * the timeout to be restarted when new submissions arrive
 	 */
-	if (mod_delayed_work(system_wq, &sched->work_tdr, MAX_SCHEDULE_TIMEOUT)
+	if (mod_delayed_work(sched->timeout_wq, &sched->work_tdr, MAX_SCHEDULE_TIMEOUT)
 			&& time_after(sched_timeout, now))
 		return sched_timeout - now;
 	else
@@ -294,7 +294,7 @@ void drm_sched_resume_timeout(struct drm_gpu_scheduler *sched,
 	if (list_empty(&sched->pending_list))
 		cancel_delayed_work(&sched->work_tdr);
 	else
-		mod_delayed_work(system_wq, &sched->work_tdr, remaining);
+		mod_delayed_work(sched->timeout_wq, &sched->work_tdr, remaining);
 
 	spin_unlock(&sched->job_list_lock);
 }
@@ -837,6 +837,8 @@ static int drm_sched_main(void *param)
  * @hw_submission: number of hw submissions that can be in flight
  * @hang_limit: number of times to allow a job to hang before dropping it
  * @timeout: timeout value in jiffies for the scheduler
+ * @timeout_wq: workqueue to use for timeout work. If NULL, the system_wq is
+ *		used
  * @score: optional score atomic shared with other schedulers
  * @name: name used for debugging
  *
@@ -844,7 +846,8 @@ static int drm_sched_main(void *param)
  */
 int drm_sched_init(struct drm_gpu_scheduler *sched,
 		   const struct drm_sched_backend_ops *ops,
-		   unsigned hw_submission, unsigned hang_limit, long timeout,
+		   unsigned hw_submission, unsigned hang_limit,
+		   long timeout, struct workqueue_struct *timeout_wq,
 		   atomic_t *score, const char *name)
 {
 	int i, ret;
@@ -852,6 +855,7 @@ int drm_sched_init(struct drm_gpu_scheduler *sched,
 	sched->hw_submission_limit = hw_submission;
 	sched->name = name;
 	sched->timeout = timeout;
+	sched->timeout_wq = timeout_wq ? : system_wq;
 	sched->hang_limit = hang_limit;
 	sched->score = score ? score : &sched->_score;
 	for (i = DRM_SCHED_PRIORITY_MIN; i < DRM_SCHED_PRIORITY_COUNT; i++)
diff --git a/drivers/gpu/drm/v3d/v3d_sched.c b/drivers/gpu/drm/v3d/v3d_sched.c
index 8992480c88fa..a39bdd5cfc4f 100644
--- a/drivers/gpu/drm/v3d/v3d_sched.c
+++ b/drivers/gpu/drm/v3d/v3d_sched.c
@@ -402,7 +402,7 @@ v3d_sched_init(struct v3d_dev *v3d)
 	ret = drm_sched_init(&v3d->queue[V3D_BIN].sched,
 			     &v3d_bin_sched_ops,
 			     hw_jobs_limit, job_hang_limit,
-			     msecs_to_jiffies(hang_limit_ms),
+			     msecs_to_jiffies(hang_limit_ms), NULL,
 			     NULL, "v3d_bin");
 	if (ret) {
 		dev_err(v3d->drm.dev, "Failed to create bin scheduler: %d.", ret);
@@ -412,7 +412,7 @@ v3d_sched_init(struct v3d_dev *v3d)
 	ret = drm_sched_init(&v3d->queue[V3D_RENDER].sched,
 			     &v3d_render_sched_ops,
 			     hw_jobs_limit, job_hang_limit,
-			     msecs_to_jiffies(hang_limit_ms),
+			     msecs_to_jiffies(hang_limit_ms), NULL,
 			     NULL, "v3d_render");
 	if (ret) {
 		dev_err(v3d->drm.dev, "Failed to create render scheduler: %d.",
@@ -424,7 +424,7 @@ v3d_sched_init(struct v3d_dev *v3d)
 	ret = drm_sched_init(&v3d->queue[V3D_TFU].sched,
 			     &v3d_tfu_sched_ops,
 			     hw_jobs_limit, job_hang_limit,
-			     msecs_to_jiffies(hang_limit_ms),
+			     msecs_to_jiffies(hang_limit_ms), NULL,
 			     NULL, "v3d_tfu");
 	if (ret) {
 		dev_err(v3d->drm.dev, "Failed to create TFU scheduler: %d.",
@@ -437,7 +437,7 @@ v3d_sched_init(struct v3d_dev *v3d)
 		ret = drm_sched_init(&v3d->queue[V3D_CSD].sched,
 				     &v3d_csd_sched_ops,
 				     hw_jobs_limit, job_hang_limit,
-				     msecs_to_jiffies(hang_limit_ms),
+				     msecs_to_jiffies(hang_limit_ms), NULL,
 				     NULL, "v3d_csd");
 		if (ret) {
 			dev_err(v3d->drm.dev, "Failed to create CSD scheduler: %d.",
@@ -449,7 +449,7 @@ v3d_sched_init(struct v3d_dev *v3d)
 		ret = drm_sched_init(&v3d->queue[V3D_CACHE_CLEAN].sched,
 				     &v3d_cache_clean_sched_ops,
 				     hw_jobs_limit, job_hang_limit,
-				     msecs_to_jiffies(hang_limit_ms),
+				     msecs_to_jiffies(hang_limit_ms), NULL,
 				     NULL, "v3d_cache_clean");
 		if (ret) {
 			dev_err(v3d->drm.dev, "Failed to create CACHE_CLEAN scheduler: %d.",
diff --git a/include/drm/gpu_scheduler.h b/include/drm/gpu_scheduler.h
index 65700511e074..9c41904ffd83 100644
--- a/include/drm/gpu_scheduler.h
+++ b/include/drm/gpu_scheduler.h
@@ -253,6 +253,24 @@ struct drm_sched_backend_ops {
 	 * 5. Restart the scheduler using drm_sched_start(). At that point, new
 	 *    jobs can be queued, and the scheduler thread is unblocked
 	 *
+	 * Note that some GPUs have distinct hardware queues but need to reset
+	 * the GPU globally, which requires extra synchronization between the
+	 * timeout handler of the different &drm_gpu_scheduler. One way to
+	 * achieve this synchronization is to create an ordered workqueue
+	 * (using alloc_ordered_workqueue()) at the driver level, and pass this
+	 * queue to drm_sched_init(), to guarantee that timeout handlers are
+	 * executed sequentially. The above workflow needs to be slightly
+	 * adjusted in that case:
+	 *
+	 * 1. Stop all schedulers impacted by the reset using drm_sched_stop()
+	 * 2. Try to gracefully stop non-faulty jobs on all queues impacted by
+	 *    the reset (optional)
+	 * 3. Issue a GPU reset on all faulty queues (driver-specific)
+	 * 4. Re-submit jobs on all schedulers impacted by the reset using
+	 *    drm_sched_resubmit_jobs()
+	 * 5. Restart all schedulers that were stopped in step #1 using
+	 *    drm_sched_start()
+	 *
 	 * Return DRM_GPU_SCHED_STAT_NOMINAL, when all is normal,
 	 * and the underlying driver has started or completed recovery.
 	 *
@@ -283,6 +301,7 @@ struct drm_sched_backend_ops {
  *                 finished.
  * @hw_rq_count: the number of jobs currently in the hardware queue.
  * @job_id_count: used to assign unique id to the each job.
+ * @timeout_wq: workqueue used to queue @work_tdr
  * @work_tdr: schedules a delayed call to @drm_sched_job_timedout after the
  *            timeout interval is over.
  * @thread: the kthread on which the scheduler which run.
@@ -307,6 +326,7 @@ struct drm_gpu_scheduler {
 	wait_queue_head_t		job_scheduled;
 	atomic_t			hw_rq_count;
 	atomic64_t			job_id_count;
+	struct workqueue_struct		*timeout_wq;
 	struct delayed_work		work_tdr;
 	struct task_struct		*thread;
 	struct list_head		pending_list;
@@ -320,7 +340,8 @@ struct drm_gpu_scheduler {
 
 int drm_sched_init(struct drm_gpu_scheduler *sched,
 		   const struct drm_sched_backend_ops *ops,
-		   uint32_t hw_submission, unsigned hang_limit, long timeout,
+		   uint32_t hw_submission, unsigned hang_limit,
+		   long timeout, struct workqueue_struct *timeout_wq,
 		   atomic_t *score, const char *name);
 
 void drm_sched_fini(struct drm_gpu_scheduler *sched);
-- 
2.31.1

[PATCH v5 03/16] drm/panfrost: Make ->run_job() return an ERR_PTR() when appropriate

[Boris Brezillon]

If the fence creation fail, we can return the error pointer directly.
The core will update the fence error accordingly.

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Steven Price <steven.price@arm.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
---
 drivers/gpu/drm/panfrost/panfrost_job.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/panfrost/panfrost_job.c b/drivers/gpu/drm/panfrost/panfrost_job.c
index 8ff79fd49577..d6c9698bca3b 100644
--- a/drivers/gpu/drm/panfrost/panfrost_job.c
+++ b/drivers/gpu/drm/panfrost/panfrost_job.c
@@ -355,7 +355,7 @@ static struct dma_fence *panfrost_job_run(struct drm_sched_job *sched_job)
 
 	fence = panfrost_fence_create(pfdev, slot);
 	if (IS_ERR(fence))
-		return NULL;
+		return fence;
 
 	if (job->done_fence)
 		dma_fence_put(job->done_fence);
-- 
2.31.1

[PATCH v5 04/16] drm/panfrost: Get rid of the unused JS_STATUS_EVENT_ACTIVE definition

[Boris Brezillon]

Exception types will be defined as an enum.

v4:
* Fix typo in the commit message

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Steven Price <steven.price@arm.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
---
 drivers/gpu/drm/panfrost/panfrost_regs.h | 3 ---
 1 file changed, 3 deletions(-)

diff --git a/drivers/gpu/drm/panfrost/panfrost_regs.h b/drivers/gpu/drm/panfrost/panfrost_regs.h
index eddaa62ad8b0..151cfebd80a0 100644
--- a/drivers/gpu/drm/panfrost/panfrost_regs.h
+++ b/drivers/gpu/drm/panfrost/panfrost_regs.h
@@ -261,9 +261,6 @@
 #define JS_COMMAND_SOFT_STOP_1		0x06	/* Execute SOFT_STOP if JOB_CHAIN_FLAG is 1 */
 #define JS_COMMAND_HARD_STOP_1		0x07	/* Execute HARD_STOP if JOB_CHAIN_FLAG is 1 */
 
-#define JS_STATUS_EVENT_ACTIVE		0x08
-
-
 /* MMU regs */
 #define MMU_INT_RAWSTAT			0x2000
 #define MMU_INT_CLEAR			0x2004
-- 
2.31.1

[PATCH v5 05/16] drm/panfrost: Drop the pfdev argument passed to panfrost_exception_name()

[Boris Brezillon]

Currently unused. We'll add it back if we need per-GPU definitions.

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Steven Price <steven.price@arm.com>
---
 drivers/gpu/drm/panfrost/panfrost_device.c | 2 +-
 drivers/gpu/drm/panfrost/panfrost_device.h | 2 +-
 drivers/gpu/drm/panfrost/panfrost_gpu.c    | 2 +-
 drivers/gpu/drm/panfrost/panfrost_job.c    | 2 +-
 drivers/gpu/drm/panfrost/panfrost_mmu.c    | 2 +-
 5 files changed, 5 insertions(+), 5 deletions(-)

diff --git a/drivers/gpu/drm/panfrost/panfrost_device.c b/drivers/gpu/drm/panfrost/panfrost_device.c
index fbcf5edbe367..bce6b0aff05e 100644
--- a/drivers/gpu/drm/panfrost/panfrost_device.c
+++ b/drivers/gpu/drm/panfrost/panfrost_device.c
@@ -292,7 +292,7 @@ void panfrost_device_fini(struct panfrost_device *pfdev)
 	panfrost_clk_fini(pfdev);
 }
 
-const char *panfrost_exception_name(struct panfrost_device *pfdev, u32 exception_code)
+const char *panfrost_exception_name(u32 exception_code)
 {
 	switch (exception_code) {
 		/* Non-Fault Status code */
diff --git a/drivers/gpu/drm/panfrost/panfrost_device.h b/drivers/gpu/drm/panfrost/panfrost_device.h
index 4c6bdea5537b..ade8a1974ee9 100644
--- a/drivers/gpu/drm/panfrost/panfrost_device.h
+++ b/drivers/gpu/drm/panfrost/panfrost_device.h
@@ -172,6 +172,6 @@ void panfrost_device_reset(struct panfrost_device *pfdev);
 int panfrost_device_resume(struct device *dev);
 int panfrost_device_suspend(struct device *dev);
 
-const char *panfrost_exception_name(struct panfrost_device *pfdev, u32 exception_code);
+const char *panfrost_exception_name(u32 exception_code);
 
 #endif
diff --git a/drivers/gpu/drm/panfrost/panfrost_gpu.c b/drivers/gpu/drm/panfrost/panfrost_gpu.c
index 2aae636f1cf5..ec59f15940fb 100644
--- a/drivers/gpu/drm/panfrost/panfrost_gpu.c
+++ b/drivers/gpu/drm/panfrost/panfrost_gpu.c
@@ -33,7 +33,7 @@ static irqreturn_t panfrost_gpu_irq_handler(int irq, void *data)
 		address |= gpu_read(pfdev, GPU_FAULT_ADDRESS_LO);
 
 		dev_warn(pfdev->dev, "GPU Fault 0x%08x (%s) at 0x%016llx\n",
-			 fault_status & 0xFF, panfrost_exception_name(pfdev, fault_status),
+			 fault_status & 0xFF, panfrost_exception_name(fault_status),
 			 address);
 
 		if (state & GPU_IRQ_MULTIPLE_FAULT)
diff --git a/drivers/gpu/drm/panfrost/panfrost_job.c b/drivers/gpu/drm/panfrost/panfrost_job.c
index d6c9698bca3b..3cd1aec6c261 100644
--- a/drivers/gpu/drm/panfrost/panfrost_job.c
+++ b/drivers/gpu/drm/panfrost/panfrost_job.c
@@ -500,7 +500,7 @@ static irqreturn_t panfrost_job_irq_handler(int irq, void *data)
 
 			dev_err(pfdev->dev, "js fault, js=%d, status=%s, head=0x%x, tail=0x%x",
 				j,
-				panfrost_exception_name(pfdev, job_read(pfdev, JS_STATUS(j))),
+				panfrost_exception_name(job_read(pfdev, JS_STATUS(j))),
 				job_read(pfdev, JS_HEAD_LO(j)),
 				job_read(pfdev, JS_TAIL_LO(j)));
 
diff --git a/drivers/gpu/drm/panfrost/panfrost_mmu.c b/drivers/gpu/drm/panfrost/panfrost_mmu.c
index d76dff201ea6..b4f0c673cd7f 100644
--- a/drivers/gpu/drm/panfrost/panfrost_mmu.c
+++ b/drivers/gpu/drm/panfrost/panfrost_mmu.c
@@ -676,7 +676,7 @@ static irqreturn_t panfrost_mmu_irq_handler_thread(int irq, void *data)
 				"TODO",
 				fault_status,
 				(fault_status & (1 << 10) ? "DECODER FAULT" : "SLAVE FAULT"),
-				exception_type, panfrost_exception_name(pfdev, exception_type),
+				exception_type, panfrost_exception_name(exception_type),
 				access_type, access_type_name(pfdev, fault_status),
 				source_id);
 
-- 
2.31.1

[PATCH v5 06/16] drm/panfrost: Do the exception -> string translation using a table

[Boris Brezillon]

Do the exception -> string translation using a table. This way we get
rid of those magic numbers and can easily add new fields if we need
to attach extra information to exception types.

v4:
* Don't expose exception type to userspace
* Merge the enum definition and the enum -> string table declaration
  in the same patch

v3:
* Drop the error field

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Steven Price <steven.price@arm.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
---
 drivers/gpu/drm/panfrost/panfrost_device.c | 130 +++++++++++++--------
 drivers/gpu/drm/panfrost/panfrost_device.h |  69 +++++++++++
 2 files changed, 152 insertions(+), 47 deletions(-)

diff --git a/drivers/gpu/drm/panfrost/panfrost_device.c b/drivers/gpu/drm/panfrost/panfrost_device.c
index bce6b0aff05e..736854542b05 100644
--- a/drivers/gpu/drm/panfrost/panfrost_device.c
+++ b/drivers/gpu/drm/panfrost/panfrost_device.c
@@ -292,55 +292,91 @@ void panfrost_device_fini(struct panfrost_device *pfdev)
 	panfrost_clk_fini(pfdev);
 }
 
-const char *panfrost_exception_name(u32 exception_code)
-{
-	switch (exception_code) {
-		/* Non-Fault Status code */
-	case 0x00: return "NOT_STARTED/IDLE/OK";
-	case 0x01: return "DONE";
-	case 0x02: return "INTERRUPTED";
-	case 0x03: return "STOPPED";
-	case 0x04: return "TERMINATED";
-	case 0x08: return "ACTIVE";
-		/* Job exceptions */
-	case 0x40: return "JOB_CONFIG_FAULT";
-	case 0x41: return "JOB_POWER_FAULT";
-	case 0x42: return "JOB_READ_FAULT";
-	case 0x43: return "JOB_WRITE_FAULT";
-	case 0x44: return "JOB_AFFINITY_FAULT";
-	case 0x48: return "JOB_BUS_FAULT";
-	case 0x50: return "INSTR_INVALID_PC";
-	case 0x51: return "INSTR_INVALID_ENC";
-	case 0x52: return "INSTR_TYPE_MISMATCH";
-	case 0x53: return "INSTR_OPERAND_FAULT";
-	case 0x54: return "INSTR_TLS_FAULT";
-	case 0x55: return "INSTR_BARRIER_FAULT";
-	case 0x56: return "INSTR_ALIGN_FAULT";
-	case 0x58: return "DATA_INVALID_FAULT";
-	case 0x59: return "TILE_RANGE_FAULT";
-	case 0x5A: return "ADDR_RANGE_FAULT";
-	case 0x60: return "OUT_OF_MEMORY";
-		/* GPU exceptions */
-	case 0x80: return "DELAYED_BUS_FAULT";
-	case 0x88: return "SHAREABILITY_FAULT";
-		/* MMU exceptions */
-	case 0xC1: return "TRANSLATION_FAULT_LEVEL1";
-	case 0xC2: return "TRANSLATION_FAULT_LEVEL2";
-	case 0xC3: return "TRANSLATION_FAULT_LEVEL3";
-	case 0xC4: return "TRANSLATION_FAULT_LEVEL4";
-	case 0xC8: return "PERMISSION_FAULT";
-	case 0xC9 ... 0xCF: return "PERMISSION_FAULT";
-	case 0xD1: return "TRANSTAB_BUS_FAULT_LEVEL1";
-	case 0xD2: return "TRANSTAB_BUS_FAULT_LEVEL2";
-	case 0xD3: return "TRANSTAB_BUS_FAULT_LEVEL3";
-	case 0xD4: return "TRANSTAB_BUS_FAULT_LEVEL4";
-	case 0xD8: return "ACCESS_FLAG";
-	case 0xD9 ... 0xDF: return "ACCESS_FLAG";
-	case 0xE0 ... 0xE7: return "ADDRESS_SIZE_FAULT";
-	case 0xE8 ... 0xEF: return "MEMORY_ATTRIBUTES_FAULT";
+#define PANFROST_EXCEPTION(id) \
+	[DRM_PANFROST_EXCEPTION_ ## id] = { \
+		.name = #id, \
 	}
 
-	return "UNKNOWN";
+struct panfrost_exception_info {
+	const char *name;
+};
+
+static const struct panfrost_exception_info panfrost_exception_infos[] = {
+	PANFROST_EXCEPTION(OK),
+	PANFROST_EXCEPTION(DONE),
+	PANFROST_EXCEPTION(INTERRUPTED),
+	PANFROST_EXCEPTION(STOPPED),
+	PANFROST_EXCEPTION(TERMINATED),
+	PANFROST_EXCEPTION(KABOOM),
+	PANFROST_EXCEPTION(EUREKA),
+	PANFROST_EXCEPTION(ACTIVE),
+	PANFROST_EXCEPTION(JOB_CONFIG_FAULT),
+	PANFROST_EXCEPTION(JOB_POWER_FAULT),
+	PANFROST_EXCEPTION(JOB_READ_FAULT),
+	PANFROST_EXCEPTION(JOB_WRITE_FAULT),
+	PANFROST_EXCEPTION(JOB_AFFINITY_FAULT),
+	PANFROST_EXCEPTION(JOB_BUS_FAULT),
+	PANFROST_EXCEPTION(INSTR_INVALID_PC),
+	PANFROST_EXCEPTION(INSTR_INVALID_ENC),
+	PANFROST_EXCEPTION(INSTR_TYPE_MISMATCH),
+	PANFROST_EXCEPTION(INSTR_OPERAND_FAULT),
+	PANFROST_EXCEPTION(INSTR_TLS_FAULT),
+	PANFROST_EXCEPTION(INSTR_BARRIER_FAULT),
+	PANFROST_EXCEPTION(INSTR_ALIGN_FAULT),
+	PANFROST_EXCEPTION(DATA_INVALID_FAULT),
+	PANFROST_EXCEPTION(TILE_RANGE_FAULT),
+	PANFROST_EXCEPTION(ADDR_RANGE_FAULT),
+	PANFROST_EXCEPTION(IMPRECISE_FAULT),
+	PANFROST_EXCEPTION(OOM),
+	PANFROST_EXCEPTION(OOM_AFBC),
+	PANFROST_EXCEPTION(UNKNOWN),
+	PANFROST_EXCEPTION(DELAYED_BUS_FAULT),
+	PANFROST_EXCEPTION(GPU_SHAREABILITY_FAULT),
+	PANFROST_EXCEPTION(SYS_SHAREABILITY_FAULT),
+	PANFROST_EXCEPTION(GPU_CACHEABILITY_FAULT),
+	PANFROST_EXCEPTION(TRANSLATION_FAULT_0),
+	PANFROST_EXCEPTION(TRANSLATION_FAULT_1),
+	PANFROST_EXCEPTION(TRANSLATION_FAULT_2),
+	PANFROST_EXCEPTION(TRANSLATION_FAULT_3),
+	PANFROST_EXCEPTION(TRANSLATION_FAULT_4),
+	PANFROST_EXCEPTION(TRANSLATION_FAULT_IDENTITY),
+	PANFROST_EXCEPTION(PERM_FAULT_0),
+	PANFROST_EXCEPTION(PERM_FAULT_1),
+	PANFROST_EXCEPTION(PERM_FAULT_2),
+	PANFROST_EXCEPTION(PERM_FAULT_3),
+	PANFROST_EXCEPTION(TRANSTAB_BUS_FAULT_0),
+	PANFROST_EXCEPTION(TRANSTAB_BUS_FAULT_1),
+	PANFROST_EXCEPTION(TRANSTAB_BUS_FAULT_2),
+	PANFROST_EXCEPTION(TRANSTAB_BUS_FAULT_3),
+	PANFROST_EXCEPTION(ACCESS_FLAG_0),
+	PANFROST_EXCEPTION(ACCESS_FLAG_1),
+	PANFROST_EXCEPTION(ACCESS_FLAG_2),
+	PANFROST_EXCEPTION(ACCESS_FLAG_3),
+	PANFROST_EXCEPTION(ADDR_SIZE_FAULT_IN0),
+	PANFROST_EXCEPTION(ADDR_SIZE_FAULT_IN1),
+	PANFROST_EXCEPTION(ADDR_SIZE_FAULT_IN2),
+	PANFROST_EXCEPTION(ADDR_SIZE_FAULT_IN3),
+	PANFROST_EXCEPTION(ADDR_SIZE_FAULT_OUT0),
+	PANFROST_EXCEPTION(ADDR_SIZE_FAULT_OUT1),
+	PANFROST_EXCEPTION(ADDR_SIZE_FAULT_OUT2),
+	PANFROST_EXCEPTION(ADDR_SIZE_FAULT_OUT3),
+	PANFROST_EXCEPTION(MEM_ATTR_FAULT_0),
+	PANFROST_EXCEPTION(MEM_ATTR_FAULT_1),
+	PANFROST_EXCEPTION(MEM_ATTR_FAULT_2),
+	PANFROST_EXCEPTION(MEM_ATTR_FAULT_3),
+	PANFROST_EXCEPTION(MEM_ATTR_NONCACHE_0),
+	PANFROST_EXCEPTION(MEM_ATTR_NONCACHE_1),
+	PANFROST_EXCEPTION(MEM_ATTR_NONCACHE_2),
+	PANFROST_EXCEPTION(MEM_ATTR_NONCACHE_3),
+};
+
+const char *panfrost_exception_name(u32 exception_code)
+{
+	if (WARN_ON(exception_code >= ARRAY_SIZE(panfrost_exception_infos) ||
+		    !panfrost_exception_infos[exception_code].name))
+		return "Unknown exception type";
+
+	return panfrost_exception_infos[exception_code].name;
 }
 
 void panfrost_device_reset(struct panfrost_device *pfdev)
diff --git a/drivers/gpu/drm/panfrost/panfrost_device.h b/drivers/gpu/drm/panfrost/panfrost_device.h
index ade8a1974ee9..4c876476268f 100644
--- a/drivers/gpu/drm/panfrost/panfrost_device.h
+++ b/drivers/gpu/drm/panfrost/panfrost_device.h
@@ -172,6 +172,75 @@ void panfrost_device_reset(struct panfrost_device *pfdev);
 int panfrost_device_resume(struct device *dev);
 int panfrost_device_suspend(struct device *dev);
 
+enum drm_panfrost_exception_type {
+	DRM_PANFROST_EXCEPTION_OK = 0x00,
+	DRM_PANFROST_EXCEPTION_DONE = 0x01,
+	DRM_PANFROST_EXCEPTION_INTERRUPTED = 0x02,
+	DRM_PANFROST_EXCEPTION_STOPPED = 0x03,
+	DRM_PANFROST_EXCEPTION_TERMINATED = 0x04,
+	DRM_PANFROST_EXCEPTION_KABOOM = 0x05,
+	DRM_PANFROST_EXCEPTION_EUREKA = 0x06,
+	DRM_PANFROST_EXCEPTION_ACTIVE = 0x08,
+	DRM_PANFROST_EXCEPTION_JOB_CONFIG_FAULT = 0x40,
+	DRM_PANFROST_EXCEPTION_JOB_POWER_FAULT = 0x41,
+	DRM_PANFROST_EXCEPTION_JOB_READ_FAULT = 0x42,
+	DRM_PANFROST_EXCEPTION_JOB_WRITE_FAULT = 0x43,
+	DRM_PANFROST_EXCEPTION_JOB_AFFINITY_FAULT = 0x44,
+	DRM_PANFROST_EXCEPTION_JOB_BUS_FAULT = 0x48,
+	DRM_PANFROST_EXCEPTION_INSTR_INVALID_PC = 0x50,
+	DRM_PANFROST_EXCEPTION_INSTR_INVALID_ENC = 0x51,
+	DRM_PANFROST_EXCEPTION_INSTR_TYPE_MISMATCH = 0x52,
+	DRM_PANFROST_EXCEPTION_INSTR_OPERAND_FAULT = 0x53,
+	DRM_PANFROST_EXCEPTION_INSTR_TLS_FAULT = 0x54,
+	DRM_PANFROST_EXCEPTION_INSTR_BARRIER_FAULT = 0x55,
+	DRM_PANFROST_EXCEPTION_INSTR_ALIGN_FAULT = 0x56,
+	DRM_PANFROST_EXCEPTION_DATA_INVALID_FAULT = 0x58,
+	DRM_PANFROST_EXCEPTION_TILE_RANGE_FAULT = 0x59,
+	DRM_PANFROST_EXCEPTION_ADDR_RANGE_FAULT = 0x5a,
+	DRM_PANFROST_EXCEPTION_IMPRECISE_FAULT = 0x5b,
+	DRM_PANFROST_EXCEPTION_OOM = 0x60,
+	DRM_PANFROST_EXCEPTION_OOM_AFBC = 0x61,
+	DRM_PANFROST_EXCEPTION_UNKNOWN = 0x7f,
+	DRM_PANFROST_EXCEPTION_DELAYED_BUS_FAULT = 0x80,
+	DRM_PANFROST_EXCEPTION_GPU_SHAREABILITY_FAULT = 0x88,
+	DRM_PANFROST_EXCEPTION_SYS_SHAREABILITY_FAULT = 0x89,
+	DRM_PANFROST_EXCEPTION_GPU_CACHEABILITY_FAULT = 0x8a,
+	DRM_PANFROST_EXCEPTION_TRANSLATION_FAULT_0 = 0xc0,
+	DRM_PANFROST_EXCEPTION_TRANSLATION_FAULT_1 = 0xc1,
+	DRM_PANFROST_EXCEPTION_TRANSLATION_FAULT_2 = 0xc2,
+	DRM_PANFROST_EXCEPTION_TRANSLATION_FAULT_3 = 0xc3,
+	DRM_PANFROST_EXCEPTION_TRANSLATION_FAULT_4 = 0xc4,
+	DRM_PANFROST_EXCEPTION_TRANSLATION_FAULT_IDENTITY = 0xc7,
+	DRM_PANFROST_EXCEPTION_PERM_FAULT_0 = 0xc8,
+	DRM_PANFROST_EXCEPTION_PERM_FAULT_1 = 0xc9,
+	DRM_PANFROST_EXCEPTION_PERM_FAULT_2 = 0xca,
+	DRM_PANFROST_EXCEPTION_PERM_FAULT_3 = 0xcb,
+	DRM_PANFROST_EXCEPTION_TRANSTAB_BUS_FAULT_0 = 0xd0,
+	DRM_PANFROST_EXCEPTION_TRANSTAB_BUS_FAULT_1 = 0xd1,
+	DRM_PANFROST_EXCEPTION_TRANSTAB_BUS_FAULT_2 = 0xd2,
+	DRM_PANFROST_EXCEPTION_TRANSTAB_BUS_FAULT_3 = 0xd3,
+	DRM_PANFROST_EXCEPTION_ACCESS_FLAG_0 = 0xd8,
+	DRM_PANFROST_EXCEPTION_ACCESS_FLAG_1 = 0xd9,
+	DRM_PANFROST_EXCEPTION_ACCESS_FLAG_2 = 0xda,
+	DRM_PANFROST_EXCEPTION_ACCESS_FLAG_3 = 0xdb,
+	DRM_PANFROST_EXCEPTION_ADDR_SIZE_FAULT_IN0 = 0xe0,
+	DRM_PANFROST_EXCEPTION_ADDR_SIZE_FAULT_IN1 = 0xe1,
+	DRM_PANFROST_EXCEPTION_ADDR_SIZE_FAULT_IN2 = 0xe2,
+	DRM_PANFROST_EXCEPTION_ADDR_SIZE_FAULT_IN3 = 0xe3,
+	DRM_PANFROST_EXCEPTION_ADDR_SIZE_FAULT_OUT0 = 0xe4,
+	DRM_PANFROST_EXCEPTION_ADDR_SIZE_FAULT_OUT1 = 0xe5,
+	DRM_PANFROST_EXCEPTION_ADDR_SIZE_FAULT_OUT2 = 0xe6,
+	DRM_PANFROST_EXCEPTION_ADDR_SIZE_FAULT_OUT3 = 0xe7,
+	DRM_PANFROST_EXCEPTION_MEM_ATTR_FAULT_0 = 0xe8,
+	DRM_PANFROST_EXCEPTION_MEM_ATTR_FAULT_1 = 0xe9,
+	DRM_PANFROST_EXCEPTION_MEM_ATTR_FAULT_2 = 0xea,
+	DRM_PANFROST_EXCEPTION_MEM_ATTR_FAULT_3 = 0xeb,
+	DRM_PANFROST_EXCEPTION_MEM_ATTR_NONCACHE_0 = 0xec,
+	DRM_PANFROST_EXCEPTION_MEM_ATTR_NONCACHE_1 = 0xed,
+	DRM_PANFROST_EXCEPTION_MEM_ATTR_NONCACHE_2 = 0xee,
+	DRM_PANFROST_EXCEPTION_MEM_ATTR_NONCACHE_3 = 0xef,
+};
+
 const char *panfrost_exception_name(u32 exception_code);
 
 #endif
-- 
2.31.1

[PATCH v5 07/16] drm/panfrost: Expose a helper to trigger a GPU reset

[Boris Brezillon]

Expose a helper to trigger a GPU reset so we can easily trigger reset
operations outside the job timeout handler.

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Steven Price <steven.price@arm.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
---
 drivers/gpu/drm/panfrost/panfrost_device.h | 8 ++++++++
 drivers/gpu/drm/panfrost/panfrost_job.c    | 4 +---
 2 files changed, 9 insertions(+), 3 deletions(-)

diff --git a/drivers/gpu/drm/panfrost/panfrost_device.h b/drivers/gpu/drm/panfrost/panfrost_device.h
index 4c876476268f..f2190f90be75 100644
--- a/drivers/gpu/drm/panfrost/panfrost_device.h
+++ b/drivers/gpu/drm/panfrost/panfrost_device.h
@@ -243,4 +243,12 @@ enum drm_panfrost_exception_type {
 
 const char *panfrost_exception_name(u32 exception_code);
 
+static inline void
+panfrost_device_schedule_reset(struct panfrost_device *pfdev)
+{
+	/* Schedule a reset if there's no reset in progress. */
+	if (!atomic_xchg(&pfdev->reset.pending, 1))
+		schedule_work(&pfdev->reset.work);
+}
+
 #endif
diff --git a/drivers/gpu/drm/panfrost/panfrost_job.c b/drivers/gpu/drm/panfrost/panfrost_job.c
index 3cd1aec6c261..be8f68f63974 100644
--- a/drivers/gpu/drm/panfrost/panfrost_job.c
+++ b/drivers/gpu/drm/panfrost/panfrost_job.c
@@ -458,9 +458,7 @@ static enum drm_gpu_sched_stat panfrost_job_timedout(struct drm_sched_job
 	if (!panfrost_scheduler_stop(&pfdev->js->queue[js], sched_job))
 		return DRM_GPU_SCHED_STAT_NOMINAL;
 
-	/* Schedule a reset if there's no reset in progress. */
-	if (!atomic_xchg(&pfdev->reset.pending, 1))
-		schedule_work(&pfdev->reset.work);
+	panfrost_device_schedule_reset(pfdev);
 
 	return DRM_GPU_SCHED_STAT_NOMINAL;
 }
-- 
2.31.1

[PATCH v5 08/16] drm/panfrost: Use a threaded IRQ for job interrupts

[Boris Brezillon]

This should avoid switching to interrupt context when the GPU is under
heavy use.

v3:
* Don't take the job_lock in panfrost_job_handle_irq()

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Acked-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Steven Price <steven.price@arm.com>
---
 drivers/gpu/drm/panfrost/panfrost_job.c | 53 ++++++++++++++++++-------
 1 file changed, 38 insertions(+), 15 deletions(-)

diff --git a/drivers/gpu/drm/panfrost/panfrost_job.c b/drivers/gpu/drm/panfrost/panfrost_job.c
index be8f68f63974..e0c479e67304 100644
--- a/drivers/gpu/drm/panfrost/panfrost_job.c
+++ b/drivers/gpu/drm/panfrost/panfrost_job.c
@@ -470,19 +470,12 @@ static const struct drm_sched_backend_ops panfrost_sched_ops = {
 	.free_job = panfrost_job_free
 };
 
-static irqreturn_t panfrost_job_irq_handler(int irq, void *data)
+static void panfrost_job_handle_irq(struct panfrost_device *pfdev, u32 status)
 {
-	struct panfrost_device *pfdev = data;
-	u32 status = job_read(pfdev, JOB_INT_STAT);
 	int j;
 
 	dev_dbg(pfdev->dev, "jobslot irq status=%x\n", status);
 
-	if (!status)
-		return IRQ_NONE;
-
-	pm_runtime_mark_last_busy(pfdev->dev);
-
 	for (j = 0; status; j++) {
 		u32 mask = MK_JS_MASK(j);
 
@@ -519,7 +512,6 @@ static irqreturn_t panfrost_job_irq_handler(int irq, void *data)
 		if (status & JOB_INT_MASK_DONE(j)) {
 			struct panfrost_job *job;
 
-			spin_lock(&pfdev->js->job_lock);
 			job = pfdev->jobs[j];
 			/* Only NULL if job timeout occurred */
 			if (job) {
@@ -531,21 +523,49 @@ static irqreturn_t panfrost_job_irq_handler(int irq, void *data)
 				dma_fence_signal_locked(job->done_fence);
 				pm_runtime_put_autosuspend(pfdev->dev);
 			}
-			spin_unlock(&pfdev->js->job_lock);
 		}
 
 		status &= ~mask;
 	}
+}
 
+static irqreturn_t panfrost_job_irq_handler_thread(int irq, void *data)
+{
+	struct panfrost_device *pfdev = data;
+	u32 status = job_read(pfdev, JOB_INT_RAWSTAT);
+
+	while (status) {
+		pm_runtime_mark_last_busy(pfdev->dev);
+
+		spin_lock(&pfdev->js->job_lock);
+		panfrost_job_handle_irq(pfdev, status);
+		spin_unlock(&pfdev->js->job_lock);
+		status = job_read(pfdev, JOB_INT_RAWSTAT);
+	}
+
+	job_write(pfdev, JOB_INT_MASK,
+		  GENMASK(16 + NUM_JOB_SLOTS - 1, 16) |
+		  GENMASK(NUM_JOB_SLOTS - 1, 0));
 	return IRQ_HANDLED;
 }
 
+static irqreturn_t panfrost_job_irq_handler(int irq, void *data)
+{
+	struct panfrost_device *pfdev = data;
+	u32 status = job_read(pfdev, JOB_INT_STAT);
+
+	if (!status)
+		return IRQ_NONE;
+
+	job_write(pfdev, JOB_INT_MASK, 0);
+	return IRQ_WAKE_THREAD;
+}
+
 static void panfrost_reset(struct work_struct *work)
 {
 	struct panfrost_device *pfdev = container_of(work,
 						     struct panfrost_device,
 						     reset.work);
-	unsigned long flags;
 	unsigned int i;
 	bool cookie;
 
@@ -575,7 +595,7 @@ static void panfrost_reset(struct work_struct *work)
 	/* All timers have been stopped, we can safely reset the pending state. */
 	atomic_set(&pfdev->reset.pending, 0);
 
-	spin_lock_irqsave(&pfdev->js->job_lock, flags);
+	spin_lock(&pfdev->js->job_lock);
 	for (i = 0; i < NUM_JOB_SLOTS; i++) {
 		if (pfdev->jobs[i]) {
 			pm_runtime_put_noidle(pfdev->dev);
@@ -583,7 +603,7 @@ static void panfrost_reset(struct work_struct *work)
 			pfdev->jobs[i] = NULL;
 		}
 	}
-	spin_unlock_irqrestore(&pfdev->js->job_lock, flags);
+	spin_unlock(&pfdev->js->job_lock);
 
 	panfrost_device_reset(pfdev);
 
@@ -610,8 +630,11 @@ int panfrost_job_init(struct panfrost_device *pfdev)
 	if (irq <= 0)
 		return -ENODEV;
 
-	ret = devm_request_irq(pfdev->dev, irq, panfrost_job_irq_handler,
-			       IRQF_SHARED, KBUILD_MODNAME "-job", pfdev);
+	ret = devm_request_threaded_irq(pfdev->dev, irq,
+					panfrost_job_irq_handler,
+					panfrost_job_irq_handler_thread,
+					IRQF_SHARED, KBUILD_MODNAME "-job",
+					pfdev);
 	if (ret) {
 		dev_err(pfdev->dev, "failed to request job irq");
 		return ret;
-- 
2.31.1

[PATCH v5 09/16] drm/panfrost: Simplify the reset serialization logic

[Boris Brezillon]

Now that we can pass our own workqueue to drm_sched_init(), we can use
an ordered workqueue on for both the scheduler timeout tdr and our own
reset work (which we use when the reset is not caused by a fault/timeout
on a specific job, like when we have AS_ACTIVE bit stuck). This
guarantees that the timeout handlers and reset handler can't run
concurrently which drastically simplifies the locking.

v4:
* Actually pass the reset workqueue to drm_sched_init()
* Don't call cancel_work_sync() in panfrost_reset(). It will deadlock
  since it might be called from the reset work, which is executing and
  cancel_work_sync() will wait for the handler to return. Checking the
  reset pending status should avoid spurious resets

v3:
* New patch

Suggested-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Steven Price <steven.price@arm.com>
---
 drivers/gpu/drm/panfrost/panfrost_device.h |   6 +-
 drivers/gpu/drm/panfrost/panfrost_job.c    | 187 ++++++++-------------
 2 files changed, 72 insertions(+), 121 deletions(-)

diff --git a/drivers/gpu/drm/panfrost/panfrost_device.h b/drivers/gpu/drm/panfrost/panfrost_device.h
index f2190f90be75..59a487e8aba3 100644
--- a/drivers/gpu/drm/panfrost/panfrost_device.h
+++ b/drivers/gpu/drm/panfrost/panfrost_device.h
@@ -108,6 +108,7 @@ struct panfrost_device {
 	struct mutex sched_lock;
 
 	struct {
+		struct workqueue_struct *wq;
 		struct work_struct work;
 		atomic_t pending;
 	} reset;
@@ -246,9 +247,8 @@ const char *panfrost_exception_name(u32 exception_code);
 static inline void
 panfrost_device_schedule_reset(struct panfrost_device *pfdev)
 {
-	/* Schedule a reset if there's no reset in progress. */
-	if (!atomic_xchg(&pfdev->reset.pending, 1))
-		schedule_work(&pfdev->reset.work);
+	atomic_set(&pfdev->reset.pending, 1);
+	queue_work(pfdev->reset.wq, &pfdev->reset.work);
 }
 
 #endif
diff --git a/drivers/gpu/drm/panfrost/panfrost_job.c b/drivers/gpu/drm/panfrost/panfrost_job.c
index e0c479e67304..98193a557a2d 100644
--- a/drivers/gpu/drm/panfrost/panfrost_job.c
+++ b/drivers/gpu/drm/panfrost/panfrost_job.c
@@ -25,17 +25,8 @@
 #define job_write(dev, reg, data) writel(data, dev->iomem + (reg))
 #define job_read(dev, reg) readl(dev->iomem + (reg))
 
-enum panfrost_queue_status {
-	PANFROST_QUEUE_STATUS_ACTIVE,
-	PANFROST_QUEUE_STATUS_STOPPED,
-	PANFROST_QUEUE_STATUS_STARTING,
-	PANFROST_QUEUE_STATUS_FAULT_PENDING,
-};
-
 struct panfrost_queue_state {
 	struct drm_gpu_scheduler sched;
-	atomic_t status;
-	struct mutex lock;
 	u64 fence_context;
 	u64 emit_seqno;
 };
@@ -379,57 +370,72 @@ void panfrost_job_enable_interrupts(struct panfrost_device *pfdev)
 	job_write(pfdev, JOB_INT_MASK, irq_mask);
 }
 
-static bool panfrost_scheduler_stop(struct panfrost_queue_state *queue,
-				    struct drm_sched_job *bad)
+static void panfrost_reset(struct panfrost_device *pfdev,
+			   struct drm_sched_job *bad)
 {
-	enum panfrost_queue_status old_status;
-	bool stopped = false;
+	unsigned int i;
+	bool cookie;
 
-	mutex_lock(&queue->lock);
-	old_status = atomic_xchg(&queue->status,
-				 PANFROST_QUEUE_STATUS_STOPPED);
-	if (old_status == PANFROST_QUEUE_STATUS_STOPPED)
-		goto out;
+	if (!atomic_read(&pfdev->reset.pending))
+		return;
+
+	/* Stop the schedulers.
+	 *
+	 * FIXME: We temporarily get out of the dma_fence_signalling section
+	 * because the cleanup path generate lockdep splats when taking locks
+	 * to release job resources. We should rework the code to follow this
+	 * pattern:
+	 *
+	 *	try_lock
+	 *	if (locked)
+	 *		release
+	 *	else
+	 *		schedule_work_to_release_later
+	 */
+	for (i = 0; i < NUM_JOB_SLOTS; i++)
+		drm_sched_stop(&pfdev->js->queue[i].sched, bad);
+
+	cookie = dma_fence_begin_signalling();
 
-	WARN_ON(old_status != PANFROST_QUEUE_STATUS_ACTIVE);
-	drm_sched_stop(&queue->sched, bad);
 	if (bad)
 		drm_sched_increase_karma(bad);
 
-	stopped = true;
+	spin_lock(&pfdev->js->job_lock);
+	for (i = 0; i < NUM_JOB_SLOTS; i++) {
+		if (pfdev->jobs[i]) {
+			pm_runtime_put_noidle(pfdev->dev);
+			panfrost_devfreq_record_idle(&pfdev->pfdevfreq);
+			pfdev->jobs[i] = NULL;
+		}
+	}
+	spin_unlock(&pfdev->js->job_lock);
 
-	/*
-	 * Set the timeout to max so the timer doesn't get started
-	 * when we return from the timeout handler (restored in
-	 * panfrost_scheduler_start()).
+	panfrost_device_reset(pfdev);
+
+	/* GPU has been reset, we can cancel timeout/fault work that may have
+	 * been queued in the meantime and clear the reset pending bit.
 	 */
-	queue->sched.timeout = MAX_SCHEDULE_TIMEOUT;
+	atomic_set(&pfdev->reset.pending, 0);
+	for (i = 0; i < NUM_JOB_SLOTS; i++)
+		cancel_delayed_work(&pfdev->js->queue[i].sched.work_tdr);
 
-out:
-	mutex_unlock(&queue->lock);
 
-	return stopped;
-}
+	/* Now resubmit jobs that were previously queued but didn't have a
+	 * chance to finish.
+	 * FIXME: We temporarily get out of the DMA fence signalling section
+	 * while resubmitting jobs because the job submission logic will
+	 * allocate memory with the GFP_KERNEL flag which can trigger memory
+	 * reclaim and exposes a lock ordering issue.
+	 */
+	dma_fence_end_signalling(cookie);
+	for (i = 0; i < NUM_JOB_SLOTS; i++)
+		drm_sched_resubmit_jobs(&pfdev->js->queue[i].sched);
+	cookie = dma_fence_begin_signalling();
 
-static void panfrost_scheduler_start(struct panfrost_queue_state *queue)
-{
-	enum panfrost_queue_status old_status;
+	for (i = 0; i < NUM_JOB_SLOTS; i++)
+		drm_sched_start(&pfdev->js->queue[i].sched, true);
 
-	mutex_lock(&queue->lock);
-	old_status = atomic_xchg(&queue->status,
-				 PANFROST_QUEUE_STATUS_STARTING);
-	WARN_ON(old_status != PANFROST_QUEUE_STATUS_STOPPED);
-
-	/* Restore the original timeout before starting the scheduler. */
-	queue->sched.timeout = msecs_to_jiffies(JOB_TIMEOUT_MS);
-	drm_sched_resubmit_jobs(&queue->sched);
-	drm_sched_start(&queue->sched, true);
-	old_status = atomic_xchg(&queue->status,
-				 PANFROST_QUEUE_STATUS_ACTIVE);
-	if (old_status == PANFROST_QUEUE_STATUS_FAULT_PENDING)
-		drm_sched_fault(&queue->sched);
-
-	mutex_unlock(&queue->lock);
+	dma_fence_end_signalling(cookie);
 }
 
 static enum drm_gpu_sched_stat panfrost_job_timedout(struct drm_sched_job
@@ -454,11 +460,8 @@ static enum drm_gpu_sched_stat panfrost_job_timedout(struct drm_sched_job
 		job_read(pfdev, JS_TAIL_LO(js)),
 		sched_job);
 
-	/* Scheduler is already stopped, nothing to do. */
-	if (!panfrost_scheduler_stop(&pfdev->js->queue[js], sched_job))
-		return DRM_GPU_SCHED_STAT_NOMINAL;
-
-	panfrost_device_schedule_reset(pfdev);
+	atomic_set(&pfdev->reset.pending, 1);
+	panfrost_reset(pfdev, sched_job);
 
 	return DRM_GPU_SCHED_STAT_NOMINAL;
 }
@@ -485,8 +488,6 @@ static void panfrost_job_handle_irq(struct panfrost_device *pfdev, u32 status)
 		job_write(pfdev, JOB_INT_CLEAR, mask);
 
 		if (status & JOB_INT_MASK_ERR(j)) {
-			enum panfrost_queue_status old_status;
-
 			job_write(pfdev, JS_COMMAND_NEXT(j), JS_COMMAND_NOP);
 
 			dev_err(pfdev->dev, "js fault, js=%d, status=%s, head=0x%x, tail=0x%x",
@@ -494,19 +495,7 @@ static void panfrost_job_handle_irq(struct panfrost_device *pfdev, u32 status)
 				panfrost_exception_name(job_read(pfdev, JS_STATUS(j))),
 				job_read(pfdev, JS_HEAD_LO(j)),
 				job_read(pfdev, JS_TAIL_LO(j)));
-
-			/*
-			 * When the queue is being restarted we don't report
-			 * faults directly to avoid races between the timeout
-			 * and reset handlers. panfrost_scheduler_start() will
-			 * call drm_sched_fault() after the queue has been
-			 * started if status == FAULT_PENDING.
-			 */
-			old_status = atomic_cmpxchg(&pfdev->js->queue[j].status,
-						    PANFROST_QUEUE_STATUS_STARTING,
-						    PANFROST_QUEUE_STATUS_FAULT_PENDING);
-			if (old_status == PANFROST_QUEUE_STATUS_ACTIVE)
-				drm_sched_fault(&pfdev->js->queue[j].sched);
+			drm_sched_fault(&pfdev->js->queue[j].sched);
 		}
 
 		if (status & JOB_INT_MASK_DONE(j)) {
@@ -561,56 +550,13 @@ static irqreturn_t panfrost_job_irq_handler(int irq, void *data)
 	return IRQ_WAKE_THREAD;
 }
 
-static void panfrost_reset(struct work_struct *work)
+static void panfrost_reset_work(struct work_struct *work)
 {
 	struct panfrost_device *pfdev = container_of(work,
 						     struct panfrost_device,
 						     reset.work);
-	unsigned int i;
-	bool cookie;
 
-	cookie = dma_fence_begin_signalling();
-	for (i = 0; i < NUM_JOB_SLOTS; i++) {
-		/*
-		 * We want pending timeouts to be handled before we attempt
-		 * to stop the scheduler. If we don't do that and the timeout
-		 * handler is in flight, it might have removed the bad job
-		 * from the list, and we'll lose this job if the reset handler
-		 * enters the critical section in panfrost_scheduler_stop()
-		 * before the timeout handler.
-		 *
-		 * Timeout is set to MAX_SCHEDULE_TIMEOUT - 1 because we need
-		 * something big enough to make sure the timer will not expire
-		 * before we manage to stop the scheduler, but we can't use
-		 * MAX_SCHEDULE_TIMEOUT because drm_sched_get_cleanup_job()
-		 * considers that as 'timer is not running' and will dequeue
-		 * the job without making sure the timeout handler is not
-		 * running.
-		 */
-		pfdev->js->queue[i].sched.timeout = MAX_SCHEDULE_TIMEOUT - 1;
-		cancel_delayed_work_sync(&pfdev->js->queue[i].sched.work_tdr);
-		panfrost_scheduler_stop(&pfdev->js->queue[i], NULL);
-	}
-
-	/* All timers have been stopped, we can safely reset the pending state. */
-	atomic_set(&pfdev->reset.pending, 0);
-
-	spin_lock(&pfdev->js->job_lock);
-	for (i = 0; i < NUM_JOB_SLOTS; i++) {
-		if (pfdev->jobs[i]) {
-			pm_runtime_put_noidle(pfdev->dev);
-			panfrost_devfreq_record_idle(&pfdev->pfdevfreq);
-			pfdev->jobs[i] = NULL;
-		}
-	}
-	spin_unlock(&pfdev->js->job_lock);
-
-	panfrost_device_reset(pfdev);
-
-	for (i = 0; i < NUM_JOB_SLOTS; i++)
-		panfrost_scheduler_start(&pfdev->js->queue[i]);
-
-	dma_fence_end_signalling(cookie);
+	panfrost_reset(pfdev, NULL);
 }
 
 int panfrost_job_init(struct panfrost_device *pfdev)
@@ -618,7 +564,7 @@ int panfrost_job_init(struct panfrost_device *pfdev)
 	struct panfrost_job_slot *js;
 	int ret, j, irq;
 
-	INIT_WORK(&pfdev->reset.work, panfrost_reset);
+	INIT_WORK(&pfdev->reset.work, panfrost_reset_work);
 
 	pfdev->js = js = devm_kzalloc(pfdev->dev, sizeof(*js), GFP_KERNEL);
 	if (!js)
@@ -640,15 +586,18 @@ int panfrost_job_init(struct panfrost_device *pfdev)
 		return ret;
 	}
 
-	for (j = 0; j < NUM_JOB_SLOTS; j++) {
-		mutex_init(&js->queue[j].lock);
+	pfdev->reset.wq = alloc_ordered_workqueue("panfrost-reset", 0);
+	if (!pfdev->reset.wq)
+		return -ENOMEM;
 
+	for (j = 0; j < NUM_JOB_SLOTS; j++) {
 		js->queue[j].fence_context = dma_fence_context_alloc(1);
 
 		ret = drm_sched_init(&js->queue[j].sched,
 				     &panfrost_sched_ops,
 				     1, 0,
-				     msecs_to_jiffies(JOB_TIMEOUT_MS), NULL,
+				     msecs_to_jiffies(JOB_TIMEOUT_MS),
+				     pfdev->reset.wq,
 				     NULL, "pan_js");
 		if (ret) {
 			dev_err(pfdev->dev, "Failed to create scheduler: %d.", ret);
@@ -664,6 +613,7 @@ int panfrost_job_init(struct panfrost_device *pfdev)
 	for (j--; j >= 0; j--)
 		drm_sched_fini(&js->queue[j].sched);
 
+	destroy_workqueue(pfdev->reset.wq);
 	return ret;
 }
 
@@ -676,9 +626,10 @@ void panfrost_job_fini(struct panfrost_device *pfdev)
 
 	for (j = 0; j < NUM_JOB_SLOTS; j++) {
 		drm_sched_fini(&js->queue[j].sched);
-		mutex_destroy(&js->queue[j].lock);
 	}
 
+	cancel_work_sync(&pfdev->reset.work);
+	destroy_workqueue(pfdev->reset.wq);
 }
 
 int panfrost_job_open(struct panfrost_file_priv *panfrost_priv)
-- 
2.31.1

[PATCH v5 10/16] drm/panfrost: Make sure job interrupts are masked before resetting

[Boris Brezillon]

This is not yet needed because we let active jobs be killed during by
the reset and we don't really bother making sure they can be restarted.
But once we start adding soft-stop support, controlling when we deal
with the remaining interrrupts and making sure those are handled before
the reset is issued gets tricky if we keep job interrupts active.

Let's prepare for that and mask+flush job IRQs before issuing a reset.

v4:
* Add a comment explaining why we WARN_ON(!job) in the irq handler
* Keep taking the job_lock when evicting stalled jobs

v3:
* New patch

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Steven Price <steven.price@arm.com>
---
 drivers/gpu/drm/panfrost/panfrost_job.c | 27 ++++++++++++++++++++-----
 1 file changed, 22 insertions(+), 5 deletions(-)

diff --git a/drivers/gpu/drm/panfrost/panfrost_job.c b/drivers/gpu/drm/panfrost/panfrost_job.c
index 98193a557a2d..4bd4d11377b7 100644
--- a/drivers/gpu/drm/panfrost/panfrost_job.c
+++ b/drivers/gpu/drm/panfrost/panfrost_job.c
@@ -34,6 +34,7 @@ struct panfrost_queue_state {
 struct panfrost_job_slot {
 	struct panfrost_queue_state queue[NUM_JOB_SLOTS];
 	spinlock_t job_lock;
+	int irq;
 };
 
 static struct panfrost_job *
@@ -400,6 +401,15 @@ static void panfrost_reset(struct panfrost_device *pfdev,
 	if (bad)
 		drm_sched_increase_karma(bad);
 
+	/* Mask job interrupts and synchronize to make sure we won't be
+	 * interrupted during our reset.
+	 */
+	job_write(pfdev, JOB_INT_MASK, 0);
+	synchronize_irq(pfdev->js->irq);
+
+	/* Schedulers are stopped and interrupts are masked+flushed, we don't
+	 * need to protect the 'evict unfinished jobs' lock with the job_lock.
+	 */
 	spin_lock(&pfdev->js->job_lock);
 	for (i = 0; i < NUM_JOB_SLOTS; i++) {
 		if (pfdev->jobs[i]) {
@@ -502,7 +512,14 @@ static void panfrost_job_handle_irq(struct panfrost_device *pfdev, u32 status)
 			struct panfrost_job *job;
 
 			job = pfdev->jobs[j];
-			/* Only NULL if job timeout occurred */
+			/* The only reason this job could be NULL is if the
+			 * job IRQ handler is called just after the
+			 * in-flight job eviction in the reset path, and
+			 * this shouldn't happen because the job IRQ has
+			 * been masked and synchronized when this eviction
+			 * happens.
+			 */
+			WARN_ON(!job);
 			if (job) {
 				pfdev->jobs[j] = NULL;
 
@@ -562,7 +579,7 @@ static void panfrost_reset_work(struct work_struct *work)
 int panfrost_job_init(struct panfrost_device *pfdev)
 {
 	struct panfrost_job_slot *js;
-	int ret, j, irq;
+	int ret, j;
 
 	INIT_WORK(&pfdev->reset.work, panfrost_reset_work);
 
@@ -572,11 +589,11 @@ int panfrost_job_init(struct panfrost_device *pfdev)
 
 	spin_lock_init(&js->job_lock);
 
-	irq = platform_get_irq_byname(to_platform_device(pfdev->dev), "job");
-	if (irq <= 0)
+	js->irq = platform_get_irq_byname(to_platform_device(pfdev->dev), "job");
+	if (js->irq <= 0)
 		return -ENODEV;
 
-	ret = devm_request_threaded_irq(pfdev->dev, irq,
+	ret = devm_request_threaded_irq(pfdev->dev, js->irq,
 					panfrost_job_irq_handler,
 					panfrost_job_irq_handler_thread,
 					IRQF_SHARED, KBUILD_MODNAME "-job",
-- 
2.31.1

[PATCH v5 11/16] drm/panfrost: Disable the AS on unhandled page faults

[Boris Brezillon]

If we don't do that, we have to wait for the job timeout to expire
before the fault jobs gets killed.

v3:
* Make sure the AS is re-enabled when new jobs are submitted to the
  context

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Steven Price <steven.price@arm.com>
---
 drivers/gpu/drm/panfrost/panfrost_device.h |  1 +
 drivers/gpu/drm/panfrost/panfrost_mmu.c    | 34 ++++++++++++++++++++--
 2 files changed, 32 insertions(+), 3 deletions(-)

diff --git a/drivers/gpu/drm/panfrost/panfrost_device.h b/drivers/gpu/drm/panfrost/panfrost_device.h
index 59a487e8aba3..2dc8c0d1d987 100644
--- a/drivers/gpu/drm/panfrost/panfrost_device.h
+++ b/drivers/gpu/drm/panfrost/panfrost_device.h
@@ -96,6 +96,7 @@ struct panfrost_device {
 	spinlock_t as_lock;
 	unsigned long as_in_use_mask;
 	unsigned long as_alloc_mask;
+	unsigned long as_faulty_mask;
 	struct list_head as_lru_list;
 
 	struct panfrost_job_slot *js;
diff --git a/drivers/gpu/drm/panfrost/panfrost_mmu.c b/drivers/gpu/drm/panfrost/panfrost_mmu.c
index b4f0c673cd7f..65e98c51cb66 100644
--- a/drivers/gpu/drm/panfrost/panfrost_mmu.c
+++ b/drivers/gpu/drm/panfrost/panfrost_mmu.c
@@ -154,6 +154,7 @@ u32 panfrost_mmu_as_get(struct panfrost_device *pfdev, struct panfrost_mmu *mmu)
 	as = mmu->as;
 	if (as >= 0) {
 		int en = atomic_inc_return(&mmu->as_count);
+		u32 mask = BIT(as) | BIT(16 + as);
 
 		/*
 		 * AS can be retained by active jobs or a perfcnt context,
@@ -162,6 +163,18 @@ u32 panfrost_mmu_as_get(struct panfrost_device *pfdev, struct panfrost_mmu *mmu)
 		WARN_ON(en >= (NUM_JOB_SLOTS + 1));
 
 		list_move(&mmu->list, &pfdev->as_lru_list);
+
+		if (pfdev->as_faulty_mask & mask) {
+			/* Unhandled pagefault on this AS, the MMU was
+			 * disabled. We need to re-enable the MMU after
+			 * clearing+unmasking the AS interrupts.
+			 */
+			mmu_write(pfdev, MMU_INT_CLEAR, mask);
+			mmu_write(pfdev, MMU_INT_MASK, ~pfdev->as_faulty_mask);
+			pfdev->as_faulty_mask &= ~mask;
+			panfrost_mmu_enable(pfdev, mmu);
+		}
+
 		goto out;
 	}
 
@@ -211,6 +224,7 @@ void panfrost_mmu_reset(struct panfrost_device *pfdev)
 	spin_lock(&pfdev->as_lock);
 
 	pfdev->as_alloc_mask = 0;
+	pfdev->as_faulty_mask = 0;
 
 	list_for_each_entry_safe(mmu, mmu_tmp, &pfdev->as_lru_list, list) {
 		mmu->as = -1;
@@ -662,7 +676,7 @@ static irqreturn_t panfrost_mmu_irq_handler_thread(int irq, void *data)
 		if ((status & mask) == BIT(as) && (exception_type & 0xF8) == 0xC0)
 			ret = panfrost_mmu_map_fault_addr(pfdev, as, addr);
 
-		if (ret)
+		if (ret) {
 			/* terminal fault, print info about the fault */
 			dev_err(pfdev->dev,
 				"Unhandled Page fault in AS%d at VA 0x%016llX\n"
@@ -680,14 +694,28 @@ static irqreturn_t panfrost_mmu_irq_handler_thread(int irq, void *data)
 				access_type, access_type_name(pfdev, fault_status),
 				source_id);
 
+			spin_lock(&pfdev->as_lock);
+			/* Ignore MMU interrupts on this AS until it's been
+			 * re-enabled.
+			 */
+			pfdev->as_faulty_mask |= mask;
+
+			/* Disable the MMU to kill jobs on this AS. */
+			panfrost_mmu_disable(pfdev, as);
+			spin_unlock(&pfdev->as_lock);
+		}
+
 		status &= ~mask;
 
 		/* If we received new MMU interrupts, process them before returning. */
 		if (!status)
-			status = mmu_read(pfdev, MMU_INT_RAWSTAT);
+			status = mmu_read(pfdev, MMU_INT_RAWSTAT) & ~pfdev->as_faulty_mask;
 	}
 
-	mmu_write(pfdev, MMU_INT_MASK, ~0);
+	spin_lock(&pfdev->as_lock);
+	mmu_write(pfdev, MMU_INT_MASK, ~pfdev->as_faulty_mask);
+	spin_unlock(&pfdev->as_lock);
+
 	return IRQ_HANDLED;
 };
 
-- 
2.31.1

[PATCH v5 12/16] drm/panfrost: Reset the GPU when the AS_ACTIVE bit is stuck

[Boris Brezillon]

Things are unlikely to resolve until we reset the GPU. Let's not wait
for other faults/timeout to happen to trigger this reset.

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Steven Price <steven.price@arm.com>
---
 drivers/gpu/drm/panfrost/panfrost_mmu.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/panfrost/panfrost_mmu.c b/drivers/gpu/drm/panfrost/panfrost_mmu.c
index 65e98c51cb66..5267c3a1f02f 100644
--- a/drivers/gpu/drm/panfrost/panfrost_mmu.c
+++ b/drivers/gpu/drm/panfrost/panfrost_mmu.c
@@ -36,8 +36,11 @@ static int wait_ready(struct panfrost_device *pfdev, u32 as_nr)
 	ret = readl_relaxed_poll_timeout_atomic(pfdev->iomem + AS_STATUS(as_nr),
 		val, !(val & AS_STATUS_AS_ACTIVE), 10, 1000);
 
-	if (ret)
+	if (ret) {
+		/* The GPU hung, let's trigger a reset */
+		panfrost_device_schedule_reset(pfdev);
 		dev_err(pfdev->dev, "AS_ACTIVE bit stuck\n");
+	}
 
 	return ret;
 }
-- 
2.31.1

[PATCH v5 13/16] drm/panfrost: Don't reset the GPU on job faults unless we really have to

[Boris Brezillon]

If we can recover from a fault without a reset there's no reason to
issue one.

v3:
* Drop the mention of Valhall requiring a reset on JOB_BUS_FAULT
* Set the fence error to -EINVAL instead of having per-exception
  error codes

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Steven Price <steven.price@arm.com>
---
 drivers/gpu/drm/panfrost/panfrost_device.c |  9 +++++++++
 drivers/gpu/drm/panfrost/panfrost_device.h |  2 ++
 drivers/gpu/drm/panfrost/panfrost_job.c    | 16 ++++++++++++++--
 3 files changed, 25 insertions(+), 2 deletions(-)

diff --git a/drivers/gpu/drm/panfrost/panfrost_device.c b/drivers/gpu/drm/panfrost/panfrost_device.c
index 736854542b05..f4e42009526d 100644
--- a/drivers/gpu/drm/panfrost/panfrost_device.c
+++ b/drivers/gpu/drm/panfrost/panfrost_device.c
@@ -379,6 +379,15 @@ const char *panfrost_exception_name(u32 exception_code)
 	return panfrost_exception_infos[exception_code].name;
 }
 
+bool panfrost_exception_needs_reset(const struct panfrost_device *pfdev,
+				    u32 exception_code)
+{
+	/* Right now, none of the GPU we support need a reset, but this
+	 * might change.
+	 */
+	return false;
+}
+
 void panfrost_device_reset(struct panfrost_device *pfdev)
 {
 	panfrost_gpu_soft_reset(pfdev);
diff --git a/drivers/gpu/drm/panfrost/panfrost_device.h b/drivers/gpu/drm/panfrost/panfrost_device.h
index 2dc8c0d1d987..d91f71366214 100644
--- a/drivers/gpu/drm/panfrost/panfrost_device.h
+++ b/drivers/gpu/drm/panfrost/panfrost_device.h
@@ -244,6 +244,8 @@ enum drm_panfrost_exception_type {
 };
 
 const char *panfrost_exception_name(u32 exception_code);
+bool panfrost_exception_needs_reset(const struct panfrost_device *pfdev,
+				    u32 exception_code);
 
 static inline void
 panfrost_device_schedule_reset(struct panfrost_device *pfdev)
diff --git a/drivers/gpu/drm/panfrost/panfrost_job.c b/drivers/gpu/drm/panfrost/panfrost_job.c
index 4bd4d11377b7..b0f4857ca084 100644
--- a/drivers/gpu/drm/panfrost/panfrost_job.c
+++ b/drivers/gpu/drm/panfrost/panfrost_job.c
@@ -498,14 +498,26 @@ static void panfrost_job_handle_irq(struct panfrost_device *pfdev, u32 status)
 		job_write(pfdev, JOB_INT_CLEAR, mask);
 
 		if (status & JOB_INT_MASK_ERR(j)) {
+			u32 js_status = job_read(pfdev, JS_STATUS(j));
+
 			job_write(pfdev, JS_COMMAND_NEXT(j), JS_COMMAND_NOP);
 
 			dev_err(pfdev->dev, "js fault, js=%d, status=%s, head=0x%x, tail=0x%x",
 				j,
-				panfrost_exception_name(job_read(pfdev, JS_STATUS(j))),
+				panfrost_exception_name(js_status),
 				job_read(pfdev, JS_HEAD_LO(j)),
 				job_read(pfdev, JS_TAIL_LO(j)));
-			drm_sched_fault(&pfdev->js->queue[j].sched);
+
+			/* If we need a reset, signal it to the timeout
+			 * handler, otherwise, update the fence error field and
+			 * signal the job fence.
+			 */
+			if (panfrost_exception_needs_reset(pfdev, js_status)) {
+				drm_sched_fault(&pfdev->js->queue[j].sched);
+			} else {
+				dma_fence_set_error(pfdev->jobs[j]->done_fence, -EINVAL);
+				status |= JOB_INT_MASK_DONE(j);
+			}
 		}
 
 		if (status & JOB_INT_MASK_DONE(j)) {
-- 
2.31.1

[PATCH v5 14/16] drm/panfrost: Kill in-flight jobs on FD close

[Boris Brezillon]

If the process who submitted these jobs decided to close the FD before
the jobs are done it probably means it doesn't care about the result.

v5:
* Add a panfrost_exception_is_fault() helper and the
  DRM_PANFROST_EXCEPTION_MAX_NON_FAULT value

v4:
* Don't disable/restore irqs when taking the job_lock (not needed since
  this lock is never taken from an interrupt context)

v3:
* Set fence error to ECANCELED when a TERMINATED exception is received

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
---
 drivers/gpu/drm/panfrost/panfrost_device.h |  7 ++++
 drivers/gpu/drm/panfrost/panfrost_job.c    | 42 ++++++++++++++++++----
 2 files changed, 43 insertions(+), 6 deletions(-)

diff --git a/drivers/gpu/drm/panfrost/panfrost_device.h b/drivers/gpu/drm/panfrost/panfrost_device.h
index d91f71366214..d2ee6e5fe5d8 100644
--- a/drivers/gpu/drm/panfrost/panfrost_device.h
+++ b/drivers/gpu/drm/panfrost/panfrost_device.h
@@ -183,6 +183,7 @@ enum drm_panfrost_exception_type {
 	DRM_PANFROST_EXCEPTION_KABOOM = 0x05,
 	DRM_PANFROST_EXCEPTION_EUREKA = 0x06,
 	DRM_PANFROST_EXCEPTION_ACTIVE = 0x08,
+	DRM_PANFROST_EXCEPTION_MAX_NON_FAULT = 0x3f,
 	DRM_PANFROST_EXCEPTION_JOB_CONFIG_FAULT = 0x40,
 	DRM_PANFROST_EXCEPTION_JOB_POWER_FAULT = 0x41,
 	DRM_PANFROST_EXCEPTION_JOB_READ_FAULT = 0x42,
@@ -243,6 +244,12 @@ enum drm_panfrost_exception_type {
 	DRM_PANFROST_EXCEPTION_MEM_ATTR_NONCACHE_3 = 0xef,
 };
 
+static inline bool
+panfrost_exception_is_fault(u32 exception_code)
+{
+	return exception_code > DRM_PANFROST_EXCEPTION_MAX_NON_FAULT;
+}
+
 const char *panfrost_exception_name(u32 exception_code);
 bool panfrost_exception_needs_reset(const struct panfrost_device *pfdev,
 				    u32 exception_code);
diff --git a/drivers/gpu/drm/panfrost/panfrost_job.c b/drivers/gpu/drm/panfrost/panfrost_job.c
index b0f4857ca084..d8e1bc227455 100644
--- a/drivers/gpu/drm/panfrost/panfrost_job.c
+++ b/drivers/gpu/drm/panfrost/panfrost_job.c
@@ -499,14 +499,21 @@ static void panfrost_job_handle_irq(struct panfrost_device *pfdev, u32 status)
 
 		if (status & JOB_INT_MASK_ERR(j)) {
 			u32 js_status = job_read(pfdev, JS_STATUS(j));
+			const char *exception_name = panfrost_exception_name(js_status);
 
 			job_write(pfdev, JS_COMMAND_NEXT(j), JS_COMMAND_NOP);
 
-			dev_err(pfdev->dev, "js fault, js=%d, status=%s, head=0x%x, tail=0x%x",
-				j,
-				panfrost_exception_name(js_status),
-				job_read(pfdev, JS_HEAD_LO(j)),
-				job_read(pfdev, JS_TAIL_LO(j)));
+			if (!panfrost_exception_is_fault(js_status)) {
+				dev_dbg(pfdev->dev, "js interrupt, js=%d, status=%s, head=0x%x, tail=0x%x",
+					j, exception_name,
+					job_read(pfdev, JS_HEAD_LO(j)),
+					job_read(pfdev, JS_TAIL_LO(j)));
+			} else {
+				dev_err(pfdev->dev, "js fault, js=%d, status=%s, head=0x%x, tail=0x%x",
+					j, exception_name,
+					job_read(pfdev, JS_HEAD_LO(j)),
+					job_read(pfdev, JS_TAIL_LO(j)));
+			}
 
 			/* If we need a reset, signal it to the timeout
 			 * handler, otherwise, update the fence error field and
@@ -515,7 +522,16 @@ static void panfrost_job_handle_irq(struct panfrost_device *pfdev, u32 status)
 			if (panfrost_exception_needs_reset(pfdev, js_status)) {
 				drm_sched_fault(&pfdev->js->queue[j].sched);
 			} else {
-				dma_fence_set_error(pfdev->jobs[j]->done_fence, -EINVAL);
+				int error = 0;
+
+				if (js_status == DRM_PANFROST_EXCEPTION_TERMINATED)
+					error = -ECANCELED;
+				else if (panfrost_exception_is_fault(js_status))
+					error = -EINVAL;
+
+				if (error)
+					dma_fence_set_error(pfdev->jobs[j]->done_fence, error);
+
 				status |= JOB_INT_MASK_DONE(j);
 			}
 		}
@@ -681,10 +697,24 @@ int panfrost_job_open(struct panfrost_file_priv *panfrost_priv)
 
 void panfrost_job_close(struct panfrost_file_priv *panfrost_priv)
 {
+	struct panfrost_device *pfdev = panfrost_priv->pfdev;
 	int i;
 
 	for (i = 0; i < NUM_JOB_SLOTS; i++)
 		drm_sched_entity_destroy(&panfrost_priv->sched_entity[i]);
+
+	/* Kill in-flight jobs */
+	spin_lock(&pfdev->js->job_lock);
+	for (i = 0; i < NUM_JOB_SLOTS; i++) {
+		struct drm_sched_entity *entity = &panfrost_priv->sched_entity[i];
+		struct panfrost_job *job = pfdev->jobs[i];
+
+		if (!job || job->base.entity != entity)
+			continue;
+
+		job_write(pfdev, JS_COMMAND(i), JS_COMMAND_HARD_STOP);
+	}
+	spin_unlock(&pfdev->js->job_lock);
 }
 
 int panfrost_job_is_idle(struct panfrost_device *pfdev)
-- 
2.31.1

[PATCH v5 15/16] drm/panfrost: Queue jobs on the hardware

[Boris Brezillon]

From: Steven Price <steven.price@arm.com>

The hardware has a set of '_NEXT' registers that can hold a second job
while the first is executing. Make use of these registers to enqueue a
second job per slot.

v5:
* Fix a comment in panfrost_job_init()

v3:
* Fix the done/err job dequeuing logic to get a valid active state
* Only enable the second slot on GPUs supporting jobchain disambiguation
* Split interrupt handling in sub-functions

Signed-off-by: Steven Price <steven.price@arm.com>
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
---
 drivers/gpu/drm/panfrost/panfrost_device.h |   2 +-
 drivers/gpu/drm/panfrost/panfrost_job.c    | 468 +++++++++++++++------
 2 files changed, 351 insertions(+), 119 deletions(-)

diff --git a/drivers/gpu/drm/panfrost/panfrost_device.h b/drivers/gpu/drm/panfrost/panfrost_device.h
index d2ee6e5fe5d8..9f1f2a603208 100644
--- a/drivers/gpu/drm/panfrost/panfrost_device.h
+++ b/drivers/gpu/drm/panfrost/panfrost_device.h
@@ -101,7 +101,7 @@ struct panfrost_device {
 
 	struct panfrost_job_slot *js;
 
-	struct panfrost_job *jobs[NUM_JOB_SLOTS];
+	struct panfrost_job *jobs[NUM_JOB_SLOTS][2];
 	struct list_head scheduled_jobs;
 
 	struct panfrost_perfcnt *perfcnt;
diff --git a/drivers/gpu/drm/panfrost/panfrost_job.c b/drivers/gpu/drm/panfrost/panfrost_job.c
index d8e1bc227455..e89e6adef481 100644
--- a/drivers/gpu/drm/panfrost/panfrost_job.c
+++ b/drivers/gpu/drm/panfrost/panfrost_job.c
@@ -4,6 +4,7 @@
 #include <linux/delay.h>
 #include <linux/interrupt.h>
 #include <linux/io.h>
+#include <linux/iopoll.h>
 #include <linux/platform_device.h>
 #include <linux/pm_runtime.h>
 #include <linux/dma-resv.h>
@@ -140,9 +141,52 @@ static void panfrost_job_write_affinity(struct panfrost_device *pfdev,
 	job_write(pfdev, JS_AFFINITY_NEXT_HI(js), affinity >> 32);
 }
 
+static u32
+panfrost_get_job_chain_flag(const struct panfrost_job *job)
+{
+	struct panfrost_fence *f = to_panfrost_fence(job->done_fence);
+
+	if (!panfrost_has_hw_feature(job->pfdev, HW_FEATURE_JOBCHAIN_DISAMBIGUATION))
+		return 0;
+
+	return (f->seqno & 1) ? JS_CONFIG_JOB_CHAIN_FLAG : 0;
+}
+
+static struct panfrost_job *
+panfrost_dequeue_job(struct panfrost_device *pfdev, int slot)
+{
+	struct panfrost_job *job = pfdev->jobs[slot][0];
+
+	WARN_ON(!job);
+	pfdev->jobs[slot][0] = pfdev->jobs[slot][1];
+	pfdev->jobs[slot][1] = NULL;
+
+	return job;
+}
+
+static unsigned int
+panfrost_enqueue_job(struct panfrost_device *pfdev, int slot,
+		     struct panfrost_job *job)
+{
+	if (WARN_ON(!job))
+		return 0;
+
+	if (!pfdev->jobs[slot][0]) {
+		pfdev->jobs[slot][0] = job;
+		return 0;
+	}
+
+	WARN_ON(pfdev->jobs[slot][1]);
+	pfdev->jobs[slot][1] = job;
+	WARN_ON(panfrost_get_job_chain_flag(job) ==
+		panfrost_get_job_chain_flag(pfdev->jobs[slot][0]));
+	return 1;
+}
+
 static void panfrost_job_hw_submit(struct panfrost_job *job, int js)
 {
 	struct panfrost_device *pfdev = job->pfdev;
+	unsigned int subslot;
 	u32 cfg;
 	u64 jc_head = job->jc;
 	int ret;
@@ -168,7 +212,8 @@ static void panfrost_job_hw_submit(struct panfrost_job *job, int js)
 	 * start */
 	cfg |= JS_CONFIG_THREAD_PRI(8) |
 		JS_CONFIG_START_FLUSH_CLEAN_INVALIDATE |
-		JS_CONFIG_END_FLUSH_CLEAN_INVALIDATE;
+		JS_CONFIG_END_FLUSH_CLEAN_INVALIDATE |
+		panfrost_get_job_chain_flag(job);
 
 	if (panfrost_has_hw_feature(pfdev, HW_FEATURE_FLUSH_REDUCTION))
 		cfg |= JS_CONFIG_ENABLE_FLUSH_REDUCTION;
@@ -182,10 +227,17 @@ static void panfrost_job_hw_submit(struct panfrost_job *job, int js)
 		job_write(pfdev, JS_FLUSH_ID_NEXT(js), job->flush_id);
 
 	/* GO ! */
-	dev_dbg(pfdev->dev, "JS: Submitting atom %p to js[%d] with head=0x%llx",
-				job, js, jc_head);
 
-	job_write(pfdev, JS_COMMAND_NEXT(js), JS_COMMAND_START);
+	spin_lock(&pfdev->js->job_lock);
+	subslot = panfrost_enqueue_job(pfdev, js, job);
+	/* Don't queue the job if a reset is in progress */
+	if (!atomic_read(&pfdev->reset.pending)) {
+		job_write(pfdev, JS_COMMAND_NEXT(js), JS_COMMAND_START);
+		dev_dbg(pfdev->dev,
+			"JS: Submitting atom %p to js[%d][%d] with head=0x%llx AS %d",
+			job, js, subslot, jc_head, cfg & 0xf);
+	}
+	spin_unlock(&pfdev->js->job_lock);
 }
 
 static void panfrost_acquire_object_fences(struct drm_gem_object **bos,
@@ -343,7 +395,11 @@ static struct dma_fence *panfrost_job_run(struct drm_sched_job *sched_job)
 	if (unlikely(job->base.s_fence->finished.error))
 		return NULL;
 
-	pfdev->jobs[slot] = job;
+	/* Nothing to execute: can happen if the job has finished while
+	 * we were resetting the GPU.
+	 */
+	if (!job->jc)
+		return NULL;
 
 	fence = panfrost_fence_create(pfdev, slot);
 	if (IS_ERR(fence))
@@ -371,11 +427,218 @@ void panfrost_job_enable_interrupts(struct panfrost_device *pfdev)
 	job_write(pfdev, JOB_INT_MASK, irq_mask);
 }
 
-static void panfrost_reset(struct panfrost_device *pfdev,
-			   struct drm_sched_job *bad)
+static void panfrost_job_handle_err(struct panfrost_device *pfdev,
+				    struct panfrost_job *job,
+				    unsigned int js)
 {
-	unsigned int i;
+	u32 js_status = job_read(pfdev, JS_STATUS(js));
+	const char *exception_name = panfrost_exception_name(js_status);
+	bool signal_fence = true;
+
+	if (!panfrost_exception_is_fault(js_status)) {
+		dev_dbg(pfdev->dev, "js event, js=%d, status=%s, head=0x%x, tail=0x%x",
+			js, exception_name,
+			job_read(pfdev, JS_HEAD_LO(js)),
+			job_read(pfdev, JS_TAIL_LO(js)));
+	} else {
+		dev_err(pfdev->dev, "js fault, js=%d, status=%s, head=0x%x, tail=0x%x",
+			js, exception_name,
+			job_read(pfdev, JS_HEAD_LO(js)),
+			job_read(pfdev, JS_TAIL_LO(js)));
+	}
+
+	if (js_status == DRM_PANFROST_EXCEPTION_STOPPED) {
+		/* Update the job head so we can resume */
+		job->jc = job_read(pfdev, JS_TAIL_LO(js)) |
+			  ((u64)job_read(pfdev, JS_TAIL_HI(js)) << 32);
+
+		/* The job will be resumed, don't signal the fence */
+		signal_fence = false;
+	} else if (js_status == DRM_PANFROST_EXCEPTION_TERMINATED) {
+		/* Job has been hard-stopped, flag it as canceled */
+		dma_fence_set_error(job->done_fence, -ECANCELED);
+		job->jc = 0;
+	} else if (panfrost_exception_is_fault(js_status)) {
+		/* We might want to provide finer-grained error code based on
+		 * the exception type, but unconditionally setting to EINVAL
+		 * is good enough for now.
+		 */
+		dma_fence_set_error(job->done_fence, -EINVAL);
+		job->jc = 0;
+	}
+
+	panfrost_mmu_as_put(pfdev, job->file_priv->mmu);
+	panfrost_devfreq_record_idle(&pfdev->pfdevfreq);
+
+	if (signal_fence)
+		dma_fence_signal_locked(job->done_fence);
+
+	pm_runtime_put_autosuspend(pfdev->dev);
+
+	if (panfrost_exception_needs_reset(pfdev, js_status)) {
+		atomic_set(&pfdev->reset.pending, 1);
+		drm_sched_fault(&pfdev->js->queue[js].sched);
+	}
+}
+
+static void panfrost_job_handle_done(struct panfrost_device *pfdev,
+				     struct panfrost_job *job)
+{
+	/* Set ->jc to 0 to avoid re-submitting an already finished job (can
+	 * happen when we receive the DONE interrupt while doing a GPU reset).
+	 */
+	job->jc = 0;
+	panfrost_mmu_as_put(pfdev, job->file_priv->mmu);
+	panfrost_devfreq_record_idle(&pfdev->pfdevfreq);
+
+	dma_fence_signal_locked(job->done_fence);
+	pm_runtime_put_autosuspend(pfdev->dev);
+}
+
+static void panfrost_job_handle_irq(struct panfrost_device *pfdev, u32 status)
+{
+	struct panfrost_job *done[NUM_JOB_SLOTS][2] = {};
+	struct panfrost_job *failed[NUM_JOB_SLOTS] = {};
+	u32 js_state = 0, js_events = 0;
+	unsigned int i, j;
+
+	/* First we collect all failed/done jobs. */
+	while (status) {
+		u32 js_state_mask = 0;
+
+		for (j = 0; j < NUM_JOB_SLOTS; j++) {
+			if (status & MK_JS_MASK(j))
+				js_state_mask |= MK_JS_MASK(j);
+
+			if (status & JOB_INT_MASK_DONE(j)) {
+				if (done[j][0])
+					done[j][1] = panfrost_dequeue_job(pfdev, j);
+				else
+					done[j][0] = panfrost_dequeue_job(pfdev, j);
+			}
+
+			if (status & JOB_INT_MASK_ERR(j)) {
+				/* Cancel the next submission. Will be submitted
+				 * after we're done handling this failure if
+				 * there's no reset pending.
+				 */
+				job_write(pfdev, JS_COMMAND_NEXT(j), JS_COMMAND_NOP);
+				failed[j] = panfrost_dequeue_job(pfdev, j);
+			}
+		}
+
+		/* JS_STATE is sampled when JOB_INT_CLEAR is written.
+		 * For each BIT(slot) or BIT(slot + 16) bit written to
+		 * JOB_INT_CLEAR, the corresponding bits in JS_STATE
+		 * (BIT(slot) and BIT(slot + 16)) are updated, but this
+		 * is racy. If we only have one job done at the time we
+		 * read JOB_INT_RAWSTAT but the second job fails before we
+		 * clear the status, we end up with a status containing
+		 * only the DONE bit and consider both jobs as DONE since
+		 * JS_STATE reports both NEXT and CURRENT as inactive.
+		 * To prevent that, let's repeat this clear+read steps
+		 * until status is 0.
+		 */
+		job_write(pfdev, JOB_INT_CLEAR, status);
+		js_state &= ~js_state_mask;
+		js_state |= job_read(pfdev, JOB_INT_JS_STATE) & js_state_mask;
+		js_events |= status;
+		status = job_read(pfdev, JOB_INT_RAWSTAT);
+	}
+
+	/* Then we handle the dequeued jobs. */
+	for (j = 0; j < NUM_JOB_SLOTS; j++) {
+		if (!(js_events & MK_JS_MASK(j)))
+			continue;
+
+		if (failed[j]) {
+			panfrost_job_handle_err(pfdev, failed[j], j);
+		} else if (pfdev->jobs[j][0] && !(js_state & MK_JS_MASK(j))) {
+			/* When the current job doesn't fail, the JM dequeues
+			 * the next job without waiting for an ACK, this means
+			 * we can have 2 jobs dequeued and only catch the
+			 * interrupt when the second one is done. If both slots
+			 * are inactive, but one job remains in pfdev->jobs[j],
+			 * consider it done. Of course that doesn't apply if a
+			 * failure happened since we cancelled execution of the
+			 * job in _NEXT (see above).
+			 */
+			if (WARN_ON(!done[j][0]))
+				done[j][0] = panfrost_dequeue_job(pfdev, j);
+			else
+				done[j][1] = panfrost_dequeue_job(pfdev, j);
+		}
+
+		for (i = 0; i < ARRAY_SIZE(done[0]) && done[j][i]; i++)
+			panfrost_job_handle_done(pfdev, done[j][i]);
+	}
+
+	/* And finally we requeue jobs that were waiting in the second slot
+	 * and have been stopped if we detected a failure on the first slot.
+	 */
+	for (j = 0; j < NUM_JOB_SLOTS; j++) {
+		if (!(js_events & MK_JS_MASK(j)))
+			continue;
+
+		if (!failed[j] || !pfdev->jobs[j][0])
+			continue;
+
+		if (pfdev->jobs[j][0]->jc == 0) {
+			/* The job was cancelled, signal the fence now */
+			struct panfrost_job *canceled = panfrost_dequeue_job(pfdev, j);
+
+			dma_fence_set_error(canceled->done_fence, -ECANCELED);
+			panfrost_job_handle_done(pfdev, canceled);
+		} else if (!atomic_read(&pfdev->reset.pending)) {
+			/* Requeue the job we removed if no reset is pending */
+			job_write(pfdev, JS_COMMAND_NEXT(j), JS_COMMAND_START);
+		}
+	}
+}
+
+static void panfrost_job_handle_irqs(struct panfrost_device *pfdev)
+{
+	u32 status = job_read(pfdev, JOB_INT_RAWSTAT);
+
+	while (status) {
+		pm_runtime_mark_last_busy(pfdev->dev);
+
+		spin_lock(&pfdev->js->job_lock);
+		panfrost_job_handle_irq(pfdev, status);
+		spin_unlock(&pfdev->js->job_lock);
+		status = job_read(pfdev, JOB_INT_RAWSTAT);
+	}
+}
+
+static u32 panfrost_active_slots(struct panfrost_device *pfdev,
+				 u32 *js_state_mask, u32 js_state)
+{
+	u32 rawstat;
+
+	if (!(js_state & *js_state_mask))
+		return 0;
+
+	rawstat = job_read(pfdev, JOB_INT_RAWSTAT);
+	if (rawstat) {
+		unsigned int i;
+
+		for (i = 0; i < NUM_JOB_SLOTS; i++) {
+			if (rawstat & MK_JS_MASK(i))
+				*js_state_mask &= ~MK_JS_MASK(i);
+		}
+	}
+
+	return js_state & *js_state_mask;
+}
+
+static void
+panfrost_reset(struct panfrost_device *pfdev,
+	       struct drm_sched_job *bad)
+{
+	u32 js_state, js_state_mask = 0xffffffff;
+	unsigned int i, j;
 	bool cookie;
+	int ret;
 
 	if (!atomic_read(&pfdev->reset.pending))
 		return;
@@ -407,21 +670,46 @@ static void panfrost_reset(struct panfrost_device *pfdev,
 	job_write(pfdev, JOB_INT_MASK, 0);
 	synchronize_irq(pfdev->js->irq);
 
-	/* Schedulers are stopped and interrupts are masked+flushed, we don't
-	 * need to protect the 'evict unfinished jobs' lock with the job_lock.
+	for (i = 0; i < NUM_JOB_SLOTS; i++) {
+		/* Cancel the next job and soft-stop the running job. */
+		job_write(pfdev, JS_COMMAND_NEXT(i), JS_COMMAND_NOP);
+		job_write(pfdev, JS_COMMAND(i), JS_COMMAND_SOFT_STOP);
+	}
+
+	/* Wait at most 10ms for soft-stops to complete */
+	ret = readl_poll_timeout(pfdev->iomem + JOB_INT_JS_STATE, js_state,
+				 !panfrost_active_slots(pfdev, &js_state_mask, js_state),
+				 10, 10000);
+
+	if (ret)
+		dev_err(pfdev->dev, "Soft-stop failed\n");
+
+	/* Handle the remaining interrupts before we reset. */
+	panfrost_job_handle_irqs(pfdev);
+
+	/* Remaining interrupts have been handled, but we might still have
+	 * stuck jobs. Let's make sure the PM counters stay balanced by
+	 * manually calling pm_runtime_put_noidle() and
+	 * panfrost_devfreq_record_idle() for each stuck job.
 	 */
 	spin_lock(&pfdev->js->job_lock);
 	for (i = 0; i < NUM_JOB_SLOTS; i++) {
-		if (pfdev->jobs[i]) {
+		for (j = 0; j < ARRAY_SIZE(pfdev->jobs[0]) && pfdev->jobs[i][j]; j++) {
 			pm_runtime_put_noidle(pfdev->dev);
 			panfrost_devfreq_record_idle(&pfdev->pfdevfreq);
-			pfdev->jobs[i] = NULL;
 		}
 	}
+	memset(pfdev->jobs, 0, sizeof(pfdev->jobs));
 	spin_unlock(&pfdev->js->job_lock);
 
+	/* Proceed with reset now. */
 	panfrost_device_reset(pfdev);
 
+	/* panfrost_device_reset() unmasks job interrupts, but we want to
+	 * keep them masked a bit longer.
+	 */
+	job_write(pfdev, JOB_INT_MASK, 0);
+
 	/* GPU has been reset, we can cancel timeout/fault work that may have
 	 * been queued in the meantime and clear the reset pending bit.
 	 */
@@ -429,7 +717,6 @@ static void panfrost_reset(struct panfrost_device *pfdev,
 	for (i = 0; i < NUM_JOB_SLOTS; i++)
 		cancel_delayed_work(&pfdev->js->queue[i].sched.work_tdr);
 
-
 	/* Now resubmit jobs that were previously queued but didn't have a
 	 * chance to finish.
 	 * FIXME: We temporarily get out of the DMA fence signalling section
@@ -442,9 +729,15 @@ static void panfrost_reset(struct panfrost_device *pfdev,
 		drm_sched_resubmit_jobs(&pfdev->js->queue[i].sched);
 	cookie = dma_fence_begin_signalling();
 
+	/* Restart the schedulers */
 	for (i = 0; i < NUM_JOB_SLOTS; i++)
 		drm_sched_start(&pfdev->js->queue[i].sched, true);
 
+	/* Re-enable job interrupts now that everything has been restarted. */
+	job_write(pfdev, JOB_INT_MASK,
+		  GENMASK(16 + NUM_JOB_SLOTS - 1, 16) |
+		  GENMASK(NUM_JOB_SLOTS - 1, 0));
+
 	dma_fence_end_signalling(cookie);
 }
 
@@ -476,6 +769,14 @@ static enum drm_gpu_sched_stat panfrost_job_timedout(struct drm_sched_job
 	return DRM_GPU_SCHED_STAT_NOMINAL;
 }
 
+static void panfrost_reset_work(struct work_struct *work)
+{
+	struct panfrost_device *pfdev;
+
+	pfdev = container_of(work, struct panfrost_device, reset.work);
+	panfrost_reset(pfdev, NULL);
+}
+
 static const struct drm_sched_backend_ops panfrost_sched_ops = {
 	.dependency = panfrost_job_dependency,
 	.run_job = panfrost_job_run,
@@ -483,100 +784,11 @@ static const struct drm_sched_backend_ops panfrost_sched_ops = {
 	.free_job = panfrost_job_free
 };
 
-static void panfrost_job_handle_irq(struct panfrost_device *pfdev, u32 status)
-{
-	int j;
-
-	dev_dbg(pfdev->dev, "jobslot irq status=%x\n", status);
-
-	for (j = 0; status; j++) {
-		u32 mask = MK_JS_MASK(j);
-
-		if (!(status & mask))
-			continue;
-
-		job_write(pfdev, JOB_INT_CLEAR, mask);
-
-		if (status & JOB_INT_MASK_ERR(j)) {
-			u32 js_status = job_read(pfdev, JS_STATUS(j));
-			const char *exception_name = panfrost_exception_name(js_status);
-
-			job_write(pfdev, JS_COMMAND_NEXT(j), JS_COMMAND_NOP);
-
-			if (!panfrost_exception_is_fault(js_status)) {
-				dev_dbg(pfdev->dev, "js interrupt, js=%d, status=%s, head=0x%x, tail=0x%x",
-					j, exception_name,
-					job_read(pfdev, JS_HEAD_LO(j)),
-					job_read(pfdev, JS_TAIL_LO(j)));
-			} else {
-				dev_err(pfdev->dev, "js fault, js=%d, status=%s, head=0x%x, tail=0x%x",
-					j, exception_name,
-					job_read(pfdev, JS_HEAD_LO(j)),
-					job_read(pfdev, JS_TAIL_LO(j)));
-			}
-
-			/* If we need a reset, signal it to the timeout
-			 * handler, otherwise, update the fence error field and
-			 * signal the job fence.
-			 */
-			if (panfrost_exception_needs_reset(pfdev, js_status)) {
-				drm_sched_fault(&pfdev->js->queue[j].sched);
-			} else {
-				int error = 0;
-
-				if (js_status == DRM_PANFROST_EXCEPTION_TERMINATED)
-					error = -ECANCELED;
-				else if (panfrost_exception_is_fault(js_status))
-					error = -EINVAL;
-
-				if (error)
-					dma_fence_set_error(pfdev->jobs[j]->done_fence, error);
-
-				status |= JOB_INT_MASK_DONE(j);
-			}
-		}
-
-		if (status & JOB_INT_MASK_DONE(j)) {
-			struct panfrost_job *job;
-
-			job = pfdev->jobs[j];
-			/* The only reason this job could be NULL is if the
-			 * job IRQ handler is called just after the
-			 * in-flight job eviction in the reset path, and
-			 * this shouldn't happen because the job IRQ has
-			 * been masked and synchronized when this eviction
-			 * happens.
-			 */
-			WARN_ON(!job);
-			if (job) {
-				pfdev->jobs[j] = NULL;
-
-				panfrost_mmu_as_put(pfdev, job->file_priv->mmu);
-				panfrost_devfreq_record_idle(&pfdev->pfdevfreq);
-
-				dma_fence_signal_locked(job->done_fence);
-				pm_runtime_put_autosuspend(pfdev->dev);
-			}
-		}
-
-		status &= ~mask;
-	}
-}
-
 static irqreturn_t panfrost_job_irq_handler_thread(int irq, void *data)
 {
 	struct panfrost_device *pfdev = data;
-	u32 status = job_read(pfdev, JOB_INT_RAWSTAT);
-
-	while (status) {
-		pm_runtime_mark_last_busy(pfdev->dev);
-
-		spin_lock(&pfdev->js->job_lock);
-		panfrost_job_handle_irq(pfdev, status);
-		spin_unlock(&pfdev->js->job_lock);
-		status = job_read(pfdev, JOB_INT_RAWSTAT);
-	}
 
+	panfrost_job_handle_irqs(pfdev);
 	job_write(pfdev, JOB_INT_MASK,
 		  GENMASK(16 + NUM_JOB_SLOTS - 1, 16) |
 		  GENMASK(NUM_JOB_SLOTS - 1, 0));
@@ -595,26 +807,24 @@ static irqreturn_t panfrost_job_irq_handler(int irq, void *data)
 	return IRQ_WAKE_THREAD;
 }
 
-static void panfrost_reset_work(struct work_struct *work)
-{
-	struct panfrost_device *pfdev = container_of(work,
-						     struct panfrost_device,
-						     reset.work);
-
-	panfrost_reset(pfdev, NULL);
-}
-
 int panfrost_job_init(struct panfrost_device *pfdev)
 {
 	struct panfrost_job_slot *js;
+	unsigned int nentries = 2;
 	int ret, j;
 
-	INIT_WORK(&pfdev->reset.work, panfrost_reset_work);
+	/* All GPUs have two entries per queue, but without jobchain
+	 * disambiguation stopping the right job in the close path is tricky,
+	 * so let's just advertise one entry in that case.
+	 */
+	if (!panfrost_has_hw_feature(pfdev, HW_FEATURE_JOBCHAIN_DISAMBIGUATION))
+		nentries = 1;
 
 	pfdev->js = js = devm_kzalloc(pfdev->dev, sizeof(*js), GFP_KERNEL);
 	if (!js)
 		return -ENOMEM;
 
+	INIT_WORK(&pfdev->reset.work, panfrost_reset_work);
 	spin_lock_init(&js->job_lock);
 
 	js->irq = platform_get_irq_byname(to_platform_device(pfdev->dev), "job");
@@ -640,7 +850,7 @@ int panfrost_job_init(struct panfrost_device *pfdev)
 
 		ret = drm_sched_init(&js->queue[j].sched,
 				     &panfrost_sched_ops,
-				     1, 0,
+				     nentries, 0,
 				     msecs_to_jiffies(JOB_TIMEOUT_MS),
 				     pfdev->reset.wq,
 				     NULL, "pan_js");
@@ -707,12 +917,34 @@ void panfrost_job_close(struct panfrost_file_priv *panfrost_priv)
 	spin_lock(&pfdev->js->job_lock);
 	for (i = 0; i < NUM_JOB_SLOTS; i++) {
 		struct drm_sched_entity *entity = &panfrost_priv->sched_entity[i];
-		struct panfrost_job *job = pfdev->jobs[i];
+		int j;
 
-		if (!job || job->base.entity != entity)
-			continue;
+		for (j = ARRAY_SIZE(pfdev->jobs[0]) - 1; j >= 0; j--) {
+			struct panfrost_job *job = pfdev->jobs[i][j];
+			u32 cmd;
 
-		job_write(pfdev, JS_COMMAND(i), JS_COMMAND_HARD_STOP);
+			if (!job || job->base.entity != entity)
+				continue;
+
+			if (j == 1) {
+				/* Try to cancel the job before it starts */
+				job_write(pfdev, JS_COMMAND_NEXT(i), JS_COMMAND_NOP);
+				/* Reset the job head so it doesn't get restarted if
+				 * the job in the first slot failed.
+				 */
+				job->jc = 0;
+			}
+
+			if (panfrost_has_hw_feature(pfdev, HW_FEATURE_JOBCHAIN_DISAMBIGUATION)) {
+				cmd = panfrost_get_job_chain_flag(job) ?
+				      JS_COMMAND_HARD_STOP_1 :
+				      JS_COMMAND_HARD_STOP_0;
+			} else {
+				cmd = JS_COMMAND_HARD_STOP;
+			}
+
+			job_write(pfdev, JS_COMMAND(i), cmd);
+		}
 	}
 	spin_unlock(&pfdev->js->job_lock);
 }
-- 
2.31.1

[PATCH v5 16/16] drm/panfrost: Increase the AS_ACTIVE polling timeout

[Boris Brezillon]

Experience has shown that 1ms is sometimes not enough, even when the GPU
is running at its maximum frequency, not to mention that an MMU operation
might take longer if the GPU is running at a lower frequency, which is
likely to be the case if devfreq is active.

Let's pick a significantly bigger timeout value (1ms -> 100ms) to be on
the safe side.

v5:
* New patch

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
---
 drivers/gpu/drm/panfrost/panfrost_mmu.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/panfrost/panfrost_mmu.c b/drivers/gpu/drm/panfrost/panfrost_mmu.c
index 5267c3a1f02f..a32a3df5358e 100644
--- a/drivers/gpu/drm/panfrost/panfrost_mmu.c
+++ b/drivers/gpu/drm/panfrost/panfrost_mmu.c
@@ -34,7 +34,7 @@ static int wait_ready(struct panfrost_device *pfdev, u32 as_nr)
 	/* Wait for the MMU status to indicate there is no active command, in
 	 * case one is pending. */
 	ret = readl_relaxed_poll_timeout_atomic(pfdev->iomem + AS_STATUS(as_nr),
-		val, !(val & AS_STATUS_AS_ACTIVE), 10, 1000);
+		val, !(val & AS_STATUS_AS_ACTIVE), 10, 100000);
 
 	if (ret) {
 		/* The GPU hung, let's trigger a reset */
-- 
2.31.1

Re: [PATCH v5 02/16] drm/sched: Allow using a dedicated workqueue for the timeout/fault tdr

[Daniel Vetter]

On Tue, Jun 29, 2021 at 09:34:56AM +0200, Boris Brezillon wrote:
> Mali Midgard/Bifrost GPUs have 3 hardware queues but only a global GPU
> reset. This leads to extra complexity when we need to synchronize timeout
> works with the reset work. One solution to address that is to have an
> ordered workqueue at the driver level that will be used by the different
> schedulers to queue their timeout work. Thanks to the serialization
> provided by the ordered workqueue we are guaranteed that timeout
> handlers are executed sequentially, and can thus easily reset the GPU
> from the timeout handler without extra synchronization.
> 
> v5:
> * Add a new paragraph to the timedout_job() method
> 
> v3:
> * New patch
> 
> v4:
> * Actually use the timeout_wq to queue the timeout work
> 
> Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
> Reviewed-by: Steven Price <steven.price@arm.com>
> Reviewed-by: Lucas Stach <l.stach@pengutronix.de>
> Cc: Qiang Yu <yuq825@gmail.com>
> Cc: Emma Anholt <emma@anholt.net>
> Cc: Alex Deucher <alexander.deucher@amd.com>
> Cc: "Christian König" <christian.koenig@amd.com>

Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>

Also since I'm occasionally blinded by my own pride, add suggested-by: me?
I did spend quite a bit pondering how to untangle your various lockdep
splats in the trd handler :-)

Cheers, Daniel

> ---
>  drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c |  2 +-
>  drivers/gpu/drm/etnaviv/etnaviv_sched.c   |  3 ++-
>  drivers/gpu/drm/lima/lima_sched.c         |  3 ++-
>  drivers/gpu/drm/panfrost/panfrost_job.c   |  3 ++-
>  drivers/gpu/drm/scheduler/sched_main.c    | 14 +++++++++-----
>  drivers/gpu/drm/v3d/v3d_sched.c           | 10 +++++-----
>  include/drm/gpu_scheduler.h               | 23 ++++++++++++++++++++++-
>  7 files changed, 43 insertions(+), 15 deletions(-)
> 
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c
> index 47ea46859618..532636ea20bc 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c
> @@ -488,7 +488,7 @@ int amdgpu_fence_driver_init_ring(struct amdgpu_ring *ring,
>  
>  	r = drm_sched_init(&ring->sched, &amdgpu_sched_ops,
>  			   num_hw_submission, amdgpu_job_hang_limit,
> -			   timeout, sched_score, ring->name);
> +			   timeout, NULL, sched_score, ring->name);
>  	if (r) {
>  		DRM_ERROR("Failed to create scheduler on ring %s.\n",
>  			  ring->name);
> diff --git a/drivers/gpu/drm/etnaviv/etnaviv_sched.c b/drivers/gpu/drm/etnaviv/etnaviv_sched.c
> index 19826e504efc..feb6da1b6ceb 100644
> --- a/drivers/gpu/drm/etnaviv/etnaviv_sched.c
> +++ b/drivers/gpu/drm/etnaviv/etnaviv_sched.c
> @@ -190,7 +190,8 @@ int etnaviv_sched_init(struct etnaviv_gpu *gpu)
>  
>  	ret = drm_sched_init(&gpu->sched, &etnaviv_sched_ops,
>  			     etnaviv_hw_jobs_limit, etnaviv_job_hang_limit,
> -			     msecs_to_jiffies(500), NULL, dev_name(gpu->dev));
> +			     msecs_to_jiffies(500), NULL, NULL,
> +			     dev_name(gpu->dev));
>  	if (ret)
>  		return ret;
>  
> diff --git a/drivers/gpu/drm/lima/lima_sched.c b/drivers/gpu/drm/lima/lima_sched.c
> index ecf3267334ff..dba8329937a3 100644
> --- a/drivers/gpu/drm/lima/lima_sched.c
> +++ b/drivers/gpu/drm/lima/lima_sched.c
> @@ -508,7 +508,8 @@ int lima_sched_pipe_init(struct lima_sched_pipe *pipe, const char *name)
>  	INIT_WORK(&pipe->recover_work, lima_sched_recover_work);
>  
>  	return drm_sched_init(&pipe->base, &lima_sched_ops, 1,
> -			      lima_job_hang_limit, msecs_to_jiffies(timeout),
> +			      lima_job_hang_limit,
> +			      msecs_to_jiffies(timeout), NULL,
>  			      NULL, name);
>  }
>  
> diff --git a/drivers/gpu/drm/panfrost/panfrost_job.c b/drivers/gpu/drm/panfrost/panfrost_job.c
> index 682f2161b999..8ff79fd49577 100644
> --- a/drivers/gpu/drm/panfrost/panfrost_job.c
> +++ b/drivers/gpu/drm/panfrost/panfrost_job.c
> @@ -626,7 +626,8 @@ int panfrost_job_init(struct panfrost_device *pfdev)
>  
>  		ret = drm_sched_init(&js->queue[j].sched,
>  				     &panfrost_sched_ops,
> -				     1, 0, msecs_to_jiffies(JOB_TIMEOUT_MS),
> +				     1, 0,
> +				     msecs_to_jiffies(JOB_TIMEOUT_MS), NULL,
>  				     NULL, "pan_js");
>  		if (ret) {
>  			dev_err(pfdev->dev, "Failed to create scheduler: %d.", ret);
> diff --git a/drivers/gpu/drm/scheduler/sched_main.c b/drivers/gpu/drm/scheduler/sched_main.c
> index c0a2f8f8d472..3e180f0d4305 100644
> --- a/drivers/gpu/drm/scheduler/sched_main.c
> +++ b/drivers/gpu/drm/scheduler/sched_main.c
> @@ -232,7 +232,7 @@ static void drm_sched_start_timeout(struct drm_gpu_scheduler *sched)
>  {
>  	if (sched->timeout != MAX_SCHEDULE_TIMEOUT &&
>  	    !list_empty(&sched->pending_list))
> -		schedule_delayed_work(&sched->work_tdr, sched->timeout);
> +		queue_delayed_work(sched->timeout_wq, &sched->work_tdr, sched->timeout);
>  }
>  
>  /**
> @@ -244,7 +244,7 @@ static void drm_sched_start_timeout(struct drm_gpu_scheduler *sched)
>   */
>  void drm_sched_fault(struct drm_gpu_scheduler *sched)
>  {
> -	mod_delayed_work(system_wq, &sched->work_tdr, 0);
> +	mod_delayed_work(sched->timeout_wq, &sched->work_tdr, 0);
>  }
>  EXPORT_SYMBOL(drm_sched_fault);
>  
> @@ -270,7 +270,7 @@ unsigned long drm_sched_suspend_timeout(struct drm_gpu_scheduler *sched)
>  	 * Modify the timeout to an arbitrarily large value. This also prevents
>  	 * the timeout to be restarted when new submissions arrive
>  	 */
> -	if (mod_delayed_work(system_wq, &sched->work_tdr, MAX_SCHEDULE_TIMEOUT)
> +	if (mod_delayed_work(sched->timeout_wq, &sched->work_tdr, MAX_SCHEDULE_TIMEOUT)
>  			&& time_after(sched_timeout, now))
>  		return sched_timeout - now;
>  	else
> @@ -294,7 +294,7 @@ void drm_sched_resume_timeout(struct drm_gpu_scheduler *sched,
>  	if (list_empty(&sched->pending_list))
>  		cancel_delayed_work(&sched->work_tdr);
>  	else
> -		mod_delayed_work(system_wq, &sched->work_tdr, remaining);
> +		mod_delayed_work(sched->timeout_wq, &sched->work_tdr, remaining);
>  
>  	spin_unlock(&sched->job_list_lock);
>  }
> @@ -837,6 +837,8 @@ static int drm_sched_main(void *param)
>   * @hw_submission: number of hw submissions that can be in flight
>   * @hang_limit: number of times to allow a job to hang before dropping it
>   * @timeout: timeout value in jiffies for the scheduler
> + * @timeout_wq: workqueue to use for timeout work. If NULL, the system_wq is
> + *		used
>   * @score: optional score atomic shared with other schedulers
>   * @name: name used for debugging
>   *
> @@ -844,7 +846,8 @@ static int drm_sched_main(void *param)
>   */
>  int drm_sched_init(struct drm_gpu_scheduler *sched,
>  		   const struct drm_sched_backend_ops *ops,
> -		   unsigned hw_submission, unsigned hang_limit, long timeout,
> +		   unsigned hw_submission, unsigned hang_limit,
> +		   long timeout, struct workqueue_struct *timeout_wq,
>  		   atomic_t *score, const char *name)
>  {
>  	int i, ret;
> @@ -852,6 +855,7 @@ int drm_sched_init(struct drm_gpu_scheduler *sched,
>  	sched->hw_submission_limit = hw_submission;
>  	sched->name = name;
>  	sched->timeout = timeout;
> +	sched->timeout_wq = timeout_wq ? : system_wq;
>  	sched->hang_limit = hang_limit;
>  	sched->score = score ? score : &sched->_score;
>  	for (i = DRM_SCHED_PRIORITY_MIN; i < DRM_SCHED_PRIORITY_COUNT; i++)
> diff --git a/drivers/gpu/drm/v3d/v3d_sched.c b/drivers/gpu/drm/v3d/v3d_sched.c
> index 8992480c88fa..a39bdd5cfc4f 100644
> --- a/drivers/gpu/drm/v3d/v3d_sched.c
> +++ b/drivers/gpu/drm/v3d/v3d_sched.c
> @@ -402,7 +402,7 @@ v3d_sched_init(struct v3d_dev *v3d)
>  	ret = drm_sched_init(&v3d->queue[V3D_BIN].sched,
>  			     &v3d_bin_sched_ops,
>  			     hw_jobs_limit, job_hang_limit,
> -			     msecs_to_jiffies(hang_limit_ms),
> +			     msecs_to_jiffies(hang_limit_ms), NULL,
>  			     NULL, "v3d_bin");
>  	if (ret) {
>  		dev_err(v3d->drm.dev, "Failed to create bin scheduler: %d.", ret);
> @@ -412,7 +412,7 @@ v3d_sched_init(struct v3d_dev *v3d)
>  	ret = drm_sched_init(&v3d->queue[V3D_RENDER].sched,
>  			     &v3d_render_sched_ops,
>  			     hw_jobs_limit, job_hang_limit,
> -			     msecs_to_jiffies(hang_limit_ms),
> +			     msecs_to_jiffies(hang_limit_ms), NULL,
>  			     NULL, "v3d_render");
>  	if (ret) {
>  		dev_err(v3d->drm.dev, "Failed to create render scheduler: %d.",
> @@ -424,7 +424,7 @@ v3d_sched_init(struct v3d_dev *v3d)
>  	ret = drm_sched_init(&v3d->queue[V3D_TFU].sched,
>  			     &v3d_tfu_sched_ops,
>  			     hw_jobs_limit, job_hang_limit,
> -			     msecs_to_jiffies(hang_limit_ms),
> +			     msecs_to_jiffies(hang_limit_ms), NULL,
>  			     NULL, "v3d_tfu");
>  	if (ret) {
>  		dev_err(v3d->drm.dev, "Failed to create TFU scheduler: %d.",
> @@ -437,7 +437,7 @@ v3d_sched_init(struct v3d_dev *v3d)
>  		ret = drm_sched_init(&v3d->queue[V3D_CSD].sched,
>  				     &v3d_csd_sched_ops,
>  				     hw_jobs_limit, job_hang_limit,
> -				     msecs_to_jiffies(hang_limit_ms),
> +				     msecs_to_jiffies(hang_limit_ms), NULL,
>  				     NULL, "v3d_csd");
>  		if (ret) {
>  			dev_err(v3d->drm.dev, "Failed to create CSD scheduler: %d.",
> @@ -449,7 +449,7 @@ v3d_sched_init(struct v3d_dev *v3d)
>  		ret = drm_sched_init(&v3d->queue[V3D_CACHE_CLEAN].sched,
>  				     &v3d_cache_clean_sched_ops,
>  				     hw_jobs_limit, job_hang_limit,
> -				     msecs_to_jiffies(hang_limit_ms),
> +				     msecs_to_jiffies(hang_limit_ms), NULL,
>  				     NULL, "v3d_cache_clean");
>  		if (ret) {
>  			dev_err(v3d->drm.dev, "Failed to create CACHE_CLEAN scheduler: %d.",
> diff --git a/include/drm/gpu_scheduler.h b/include/drm/gpu_scheduler.h
> index 65700511e074..9c41904ffd83 100644
> --- a/include/drm/gpu_scheduler.h
> +++ b/include/drm/gpu_scheduler.h
> @@ -253,6 +253,24 @@ struct drm_sched_backend_ops {
>  	 * 5. Restart the scheduler using drm_sched_start(). At that point, new
>  	 *    jobs can be queued, and the scheduler thread is unblocked
>  	 *
> +	 * Note that some GPUs have distinct hardware queues but need to reset
> +	 * the GPU globally, which requires extra synchronization between the
> +	 * timeout handler of the different &drm_gpu_scheduler. One way to
> +	 * achieve this synchronization is to create an ordered workqueue
> +	 * (using alloc_ordered_workqueue()) at the driver level, and pass this
> +	 * queue to drm_sched_init(), to guarantee that timeout handlers are
> +	 * executed sequentially. The above workflow needs to be slightly
> +	 * adjusted in that case:
> +	 *
> +	 * 1. Stop all schedulers impacted by the reset using drm_sched_stop()
> +	 * 2. Try to gracefully stop non-faulty jobs on all queues impacted by
> +	 *    the reset (optional)
> +	 * 3. Issue a GPU reset on all faulty queues (driver-specific)
> +	 * 4. Re-submit jobs on all schedulers impacted by the reset using
> +	 *    drm_sched_resubmit_jobs()
> +	 * 5. Restart all schedulers that were stopped in step #1 using
> +	 *    drm_sched_start()
> +	 *
>  	 * Return DRM_GPU_SCHED_STAT_NOMINAL, when all is normal,
>  	 * and the underlying driver has started or completed recovery.
>  	 *
> @@ -283,6 +301,7 @@ struct drm_sched_backend_ops {
>   *                 finished.
>   * @hw_rq_count: the number of jobs currently in the hardware queue.
>   * @job_id_count: used to assign unique id to the each job.
> + * @timeout_wq: workqueue used to queue @work_tdr
>   * @work_tdr: schedules a delayed call to @drm_sched_job_timedout after the
>   *            timeout interval is over.
>   * @thread: the kthread on which the scheduler which run.
> @@ -307,6 +326,7 @@ struct drm_gpu_scheduler {
>  	wait_queue_head_t		job_scheduled;
>  	atomic_t			hw_rq_count;
>  	atomic64_t			job_id_count;
> +	struct workqueue_struct		*timeout_wq;
>  	struct delayed_work		work_tdr;
>  	struct task_struct		*thread;
>  	struct list_head		pending_list;
> @@ -320,7 +340,8 @@ struct drm_gpu_scheduler {
>  
>  int drm_sched_init(struct drm_gpu_scheduler *sched,
>  		   const struct drm_sched_backend_ops *ops,
> -		   uint32_t hw_submission, unsigned hang_limit, long timeout,
> +		   uint32_t hw_submission, unsigned hang_limit,
> +		   long timeout, struct workqueue_struct *timeout_wq,
>  		   atomic_t *score, const char *name);
>  
>  void drm_sched_fini(struct drm_gpu_scheduler *sched);
> -- 
> 2.31.1
> 

-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch

Re: [PATCH v5 02/16] drm/sched: Allow using a dedicated workqueue for the timeout/fault tdr

[Boris Brezillon]

On Tue, 29 Jun 2021 10:50:36 +0200
Daniel Vetter <daniel@ffwll.ch> wrote:

> On Tue, Jun 29, 2021 at 09:34:56AM +0200, Boris Brezillon wrote:
> > Mali Midgard/Bifrost GPUs have 3 hardware queues but only a global GPU
> > reset. This leads to extra complexity when we need to synchronize timeout
> > works with the reset work. One solution to address that is to have an
> > ordered workqueue at the driver level that will be used by the different
> > schedulers to queue their timeout work. Thanks to the serialization
> > provided by the ordered workqueue we are guaranteed that timeout
> > handlers are executed sequentially, and can thus easily reset the GPU
> > from the timeout handler without extra synchronization.
> > 
> > v5:
> > * Add a new paragraph to the timedout_job() method
> > 
> > v3:
> > * New patch
> > 
> > v4:
> > * Actually use the timeout_wq to queue the timeout work
> > 
> > Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
> > Reviewed-by: Steven Price <steven.price@arm.com>
> > Reviewed-by: Lucas Stach <l.stach@pengutronix.de>
> > Cc: Qiang Yu <yuq825@gmail.com>
> > Cc: Emma Anholt <emma@anholt.net>
> > Cc: Alex Deucher <alexander.deucher@amd.com>
> > Cc: "Christian König" <christian.koenig@amd.com>  
> 
> Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
> 
> Also since I'm occasionally blinded by my own pride, add suggested-by: me?

Duh, it's an oversight (I thought I had that 'Suggested-by: Daniel
Vetter ...' already).

> I did spend quite a bit pondering how to untangle your various lockdep
> splats in the trd handler :-)

And I'm grateful for your help ;-).

Re: [PATCH v5 01/16] drm/sched: Document what the timedout_job method should do

[Daniel Vetter]

On Tue, Jun 29, 2021 at 09:34:55AM +0200, Boris Brezillon wrote:
> The documentation is a bit vague and doesn't really describe what the
> ->timedout_job() is expected to do. Let's add a few more details.
> 
> v5:
> * New patch
> 
> Suggested-by: Daniel Vetter <daniel.vetter@ffwll.ch>
> Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>

Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>

> ---
>  include/drm/gpu_scheduler.h | 14 ++++++++++++++
>  1 file changed, 14 insertions(+)
> 
> diff --git a/include/drm/gpu_scheduler.h b/include/drm/gpu_scheduler.h
> index 10225a0a35d0..65700511e074 100644
> --- a/include/drm/gpu_scheduler.h
> +++ b/include/drm/gpu_scheduler.h
> @@ -239,6 +239,20 @@ struct drm_sched_backend_ops {
>  	 * @timedout_job: Called when a job has taken too long to execute,
>  	 * to trigger GPU recovery.
>  	 *
> +	 * This method is called in a workqueue context.
> +	 *
> +	 * Drivers typically issue a reset to recover from GPU hangs, and this
> +	 * procedure usually follows the following workflow:
> +	 *
> +	 * 1. Stop the scheduler using drm_sched_stop(). This will park the
> +	 *    scheduler thread and cancel the timeout work, guaranteeing that
> +	 *    nothing is queued while we reset the hardware queue
> +	 * 2. Try to gracefully stop non-faulty jobs (optional)
> +	 * 3. Issue a GPU reset (driver-specific)
> +	 * 4. Re-submit jobs using drm_sched_resubmit_jobs()
> +	 * 5. Restart the scheduler using drm_sched_start(). At that point, new
> +	 *    jobs can be queued, and the scheduler thread is unblocked
> +	 *
>  	 * Return DRM_GPU_SCHED_STAT_NOMINAL, when all is normal,
>  	 * and the underlying driver has started or completed recovery.
>  	 *
> -- 
> 2.31.1
> 

-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch

Re: [PATCH v5 02/16] drm/sched: Allow using a dedicated workqueue for the timeout/fault tdr

[Christian König]

Am 29.06.21 um 09:34 schrieb Boris Brezillon:
> Mali Midgard/Bifrost GPUs have 3 hardware queues but only a global GPU
> reset. This leads to extra complexity when we need to synchronize timeout
> works with the reset work. One solution to address that is to have an
> ordered workqueue at the driver level that will be used by the different
> schedulers to queue their timeout work. Thanks to the serialization
> provided by the ordered workqueue we are guaranteed that timeout
> handlers are executed sequentially, and can thus easily reset the GPU
> from the timeout handler without extra synchronization.

Well, we had already tried this and it didn't worked the way it is expected.

The major problem is that you not only want to serialize the queue, but 
rather have a single reset for all queues.

Otherwise you schedule multiple resets for each hardware queue. E.g. for 
your 3 hardware queues you would reset the GPU 3 times if all of them 
time out at the same time (which is rather likely).

Using a single delayed work item doesn't work either because you then 
only have one timeout.

What could be done is to cancel all delayed work items from all stopped 
schedulers.

Regards,
Christian.

>
> v5:
> * Add a new paragraph to the timedout_job() method
>
> v3:
> * New patch
>
> v4:
> * Actually use the timeout_wq to queue the timeout work
>
> Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
> Reviewed-by: Steven Price <steven.price@arm.com>
> Reviewed-by: Lucas Stach <l.stach@pengutronix.de>
> Cc: Qiang Yu <yuq825@gmail.com>
> Cc: Emma Anholt <emma@anholt.net>
> Cc: Alex Deucher <alexander.deucher@amd.com>
> Cc: "Christian König" <christian.koenig@amd.com>
> ---
>   drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c |  2 +-
>   drivers/gpu/drm/etnaviv/etnaviv_sched.c   |  3 ++-
>   drivers/gpu/drm/lima/lima_sched.c         |  3 ++-
>   drivers/gpu/drm/panfrost/panfrost_job.c   |  3 ++-
>   drivers/gpu/drm/scheduler/sched_main.c    | 14 +++++++++-----
>   drivers/gpu/drm/v3d/v3d_sched.c           | 10 +++++-----
>   include/drm/gpu_scheduler.h               | 23 ++++++++++++++++++++++-
>   7 files changed, 43 insertions(+), 15 deletions(-)
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c
> index 47ea46859618..532636ea20bc 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c
> @@ -488,7 +488,7 @@ int amdgpu_fence_driver_init_ring(struct amdgpu_ring *ring,
>   
>   	r = drm_sched_init(&ring->sched, &amdgpu_sched_ops,
>   			   num_hw_submission, amdgpu_job_hang_limit,
> -			   timeout, sched_score, ring->name);
> +			   timeout, NULL, sched_score, ring->name);
>   	if (r) {
>   		DRM_ERROR("Failed to create scheduler on ring %s.\n",
>   			  ring->name);
> diff --git a/drivers/gpu/drm/etnaviv/etnaviv_sched.c b/drivers/gpu/drm/etnaviv/etnaviv_sched.c
> index 19826e504efc..feb6da1b6ceb 100644
> --- a/drivers/gpu/drm/etnaviv/etnaviv_sched.c
> +++ b/drivers/gpu/drm/etnaviv/etnaviv_sched.c
> @@ -190,7 +190,8 @@ int etnaviv_sched_init(struct etnaviv_gpu *gpu)
>   
>   	ret = drm_sched_init(&gpu->sched, &etnaviv_sched_ops,
>   			     etnaviv_hw_jobs_limit, etnaviv_job_hang_limit,
> -			     msecs_to_jiffies(500), NULL, dev_name(gpu->dev));
> +			     msecs_to_jiffies(500), NULL, NULL,
> +			     dev_name(gpu->dev));
>   	if (ret)
>   		return ret;
>   
> diff --git a/drivers/gpu/drm/lima/lima_sched.c b/drivers/gpu/drm/lima/lima_sched.c
> index ecf3267334ff..dba8329937a3 100644
> --- a/drivers/gpu/drm/lima/lima_sched.c
> +++ b/drivers/gpu/drm/lima/lima_sched.c
> @@ -508,7 +508,8 @@ int lima_sched_pipe_init(struct lima_sched_pipe *pipe, const char *name)
>   	INIT_WORK(&pipe->recover_work, lima_sched_recover_work);
>   
>   	return drm_sched_init(&pipe->base, &lima_sched_ops, 1,
> -			      lima_job_hang_limit, msecs_to_jiffies(timeout),
> +			      lima_job_hang_limit,
> +			      msecs_to_jiffies(timeout), NULL,
>   			      NULL, name);
>   }
>   
> diff --git a/drivers/gpu/drm/panfrost/panfrost_job.c b/drivers/gpu/drm/panfrost/panfrost_job.c
> index 682f2161b999..8ff79fd49577 100644
> --- a/drivers/gpu/drm/panfrost/panfrost_job.c
> +++ b/drivers/gpu/drm/panfrost/panfrost_job.c
> @@ -626,7 +626,8 @@ int panfrost_job_init(struct panfrost_device *pfdev)
>   
>   		ret = drm_sched_init(&js->queue[j].sched,
>   				     &panfrost_sched_ops,
> -				     1, 0, msecs_to_jiffies(JOB_TIMEOUT_MS),
> +				     1, 0,
> +				     msecs_to_jiffies(JOB_TIMEOUT_MS), NULL,
>   				     NULL, "pan_js");
>   		if (ret) {
>   			dev_err(pfdev->dev, "Failed to create scheduler: %d.", ret);
> diff --git a/drivers/gpu/drm/scheduler/sched_main.c b/drivers/gpu/drm/scheduler/sched_main.c
> index c0a2f8f8d472..3e180f0d4305 100644
> --- a/drivers/gpu/drm/scheduler/sched_main.c
> +++ b/drivers/gpu/drm/scheduler/sched_main.c
> @@ -232,7 +232,7 @@ static void drm_sched_start_timeout(struct drm_gpu_scheduler *sched)
>   {
>   	if (sched->timeout != MAX_SCHEDULE_TIMEOUT &&
>   	    !list_empty(&sched->pending_list))
> -		schedule_delayed_work(&sched->work_tdr, sched->timeout);
> +		queue_delayed_work(sched->timeout_wq, &sched->work_tdr, sched->timeout);
>   }
>   
>   /**
> @@ -244,7 +244,7 @@ static void drm_sched_start_timeout(struct drm_gpu_scheduler *sched)
>    */
>   void drm_sched_fault(struct drm_gpu_scheduler *sched)
>   {
> -	mod_delayed_work(system_wq, &sched->work_tdr, 0);
> +	mod_delayed_work(sched->timeout_wq, &sched->work_tdr, 0);
>   }
>   EXPORT_SYMBOL(drm_sched_fault);
>   
> @@ -270,7 +270,7 @@ unsigned long drm_sched_suspend_timeout(struct drm_gpu_scheduler *sched)
>   	 * Modify the timeout to an arbitrarily large value. This also prevents
>   	 * the timeout to be restarted when new submissions arrive
>   	 */
> -	if (mod_delayed_work(system_wq, &sched->work_tdr, MAX_SCHEDULE_TIMEOUT)
> +	if (mod_delayed_work(sched->timeout_wq, &sched->work_tdr, MAX_SCHEDULE_TIMEOUT)
>   			&& time_after(sched_timeout, now))
>   		return sched_timeout - now;
>   	else
> @@ -294,7 +294,7 @@ void drm_sched_resume_timeout(struct drm_gpu_scheduler *sched,
>   	if (list_empty(&sched->pending_list))
>   		cancel_delayed_work(&sched->work_tdr);
>   	else
> -		mod_delayed_work(system_wq, &sched->work_tdr, remaining);
> +		mod_delayed_work(sched->timeout_wq, &sched->work_tdr, remaining);
>   
>   	spin_unlock(&sched->job_list_lock);
>   }
> @@ -837,6 +837,8 @@ static int drm_sched_main(void *param)
>    * @hw_submission: number of hw submissions that can be in flight
>    * @hang_limit: number of times to allow a job to hang before dropping it
>    * @timeout: timeout value in jiffies for the scheduler
> + * @timeout_wq: workqueue to use for timeout work. If NULL, the system_wq is
> + *		used
>    * @score: optional score atomic shared with other schedulers
>    * @name: name used for debugging
>    *
> @@ -844,7 +846,8 @@ static int drm_sched_main(void *param)
>    */
>   int drm_sched_init(struct drm_gpu_scheduler *sched,
>   		   const struct drm_sched_backend_ops *ops,
> -		   unsigned hw_submission, unsigned hang_limit, long timeout,
> +		   unsigned hw_submission, unsigned hang_limit,
> +		   long timeout, struct workqueue_struct *timeout_wq,
>   		   atomic_t *score, const char *name)
>   {
>   	int i, ret;
> @@ -852,6 +855,7 @@ int drm_sched_init(struct drm_gpu_scheduler *sched,
>   	sched->hw_submission_limit = hw_submission;
>   	sched->name = name;
>   	sched->timeout = timeout;
> +	sched->timeout_wq = timeout_wq ? : system_wq;
>   	sched->hang_limit = hang_limit;
>   	sched->score = score ? score : &sched->_score;
>   	for (i = DRM_SCHED_PRIORITY_MIN; i < DRM_SCHED_PRIORITY_COUNT; i++)
> diff --git a/drivers/gpu/drm/v3d/v3d_sched.c b/drivers/gpu/drm/v3d/v3d_sched.c
> index 8992480c88fa..a39bdd5cfc4f 100644
> --- a/drivers/gpu/drm/v3d/v3d_sched.c
> +++ b/drivers/gpu/drm/v3d/v3d_sched.c
> @@ -402,7 +402,7 @@ v3d_sched_init(struct v3d_dev *v3d)
>   	ret = drm_sched_init(&v3d->queue[V3D_BIN].sched,
>   			     &v3d_bin_sched_ops,
>   			     hw_jobs_limit, job_hang_limit,
> -			     msecs_to_jiffies(hang_limit_ms),
> +			     msecs_to_jiffies(hang_limit_ms), NULL,
>   			     NULL, "v3d_bin");
>   	if (ret) {
>   		dev_err(v3d->drm.dev, "Failed to create bin scheduler: %d.", ret);
> @@ -412,7 +412,7 @@ v3d_sched_init(struct v3d_dev *v3d)
>   	ret = drm_sched_init(&v3d->queue[V3D_RENDER].sched,
>   			     &v3d_render_sched_ops,
>   			     hw_jobs_limit, job_hang_limit,
> -			     msecs_to_jiffies(hang_limit_ms),
> +			     msecs_to_jiffies(hang_limit_ms), NULL,
>   			     NULL, "v3d_render");
>   	if (ret) {
>   		dev_err(v3d->drm.dev, "Failed to create render scheduler: %d.",
> @@ -424,7 +424,7 @@ v3d_sched_init(struct v3d_dev *v3d)
>   	ret = drm_sched_init(&v3d->queue[V3D_TFU].sched,
>   			     &v3d_tfu_sched_ops,
>   			     hw_jobs_limit, job_hang_limit,
> -			     msecs_to_jiffies(hang_limit_ms),
> +			     msecs_to_jiffies(hang_limit_ms), NULL,
>   			     NULL, "v3d_tfu");
>   	if (ret) {
>   		dev_err(v3d->drm.dev, "Failed to create TFU scheduler: %d.",
> @@ -437,7 +437,7 @@ v3d_sched_init(struct v3d_dev *v3d)
>   		ret = drm_sched_init(&v3d->queue[V3D_CSD].sched,
>   				     &v3d_csd_sched_ops,
>   				     hw_jobs_limit, job_hang_limit,
> -				     msecs_to_jiffies(hang_limit_ms),
> +				     msecs_to_jiffies(hang_limit_ms), NULL,
>   				     NULL, "v3d_csd");
>   		if (ret) {
>   			dev_err(v3d->drm.dev, "Failed to create CSD scheduler: %d.",
> @@ -449,7 +449,7 @@ v3d_sched_init(struct v3d_dev *v3d)
>   		ret = drm_sched_init(&v3d->queue[V3D_CACHE_CLEAN].sched,
>   				     &v3d_cache_clean_sched_ops,
>   				     hw_jobs_limit, job_hang_limit,
> -				     msecs_to_jiffies(hang_limit_ms),
> +				     msecs_to_jiffies(hang_limit_ms), NULL,
>   				     NULL, "v3d_cache_clean");
>   		if (ret) {
>   			dev_err(v3d->drm.dev, "Failed to create CACHE_CLEAN scheduler: %d.",
> diff --git a/include/drm/gpu_scheduler.h b/include/drm/gpu_scheduler.h
> index 65700511e074..9c41904ffd83 100644
> --- a/include/drm/gpu_scheduler.h
> +++ b/include/drm/gpu_scheduler.h
> @@ -253,6 +253,24 @@ struct drm_sched_backend_ops {
>   	 * 5. Restart the scheduler using drm_sched_start(). At that point, new
>   	 *    jobs can be queued, and the scheduler thread is unblocked
>   	 *
> +	 * Note that some GPUs have distinct hardware queues but need to reset
> +	 * the GPU globally, which requires extra synchronization between the
> +	 * timeout handler of the different &drm_gpu_scheduler. One way to
> +	 * achieve this synchronization is to create an ordered workqueue
> +	 * (using alloc_ordered_workqueue()) at the driver level, and pass this
> +	 * queue to drm_sched_init(), to guarantee that timeout handlers are
> +	 * executed sequentially. The above workflow needs to be slightly
> +	 * adjusted in that case:
> +	 *
> +	 * 1. Stop all schedulers impacted by the reset using drm_sched_stop()
> +	 * 2. Try to gracefully stop non-faulty jobs on all queues impacted by
> +	 *    the reset (optional)
> +	 * 3. Issue a GPU reset on all faulty queues (driver-specific)
> +	 * 4. Re-submit jobs on all schedulers impacted by the reset using
> +	 *    drm_sched_resubmit_jobs()
> +	 * 5. Restart all schedulers that were stopped in step #1 using
> +	 *    drm_sched_start()
> +	 *
>   	 * Return DRM_GPU_SCHED_STAT_NOMINAL, when all is normal,
>   	 * and the underlying driver has started or completed recovery.
>   	 *
> @@ -283,6 +301,7 @@ struct drm_sched_backend_ops {
>    *                 finished.
>    * @hw_rq_count: the number of jobs currently in the hardware queue.
>    * @job_id_count: used to assign unique id to the each job.
> + * @timeout_wq: workqueue used to queue @work_tdr
>    * @work_tdr: schedules a delayed call to @drm_sched_job_timedout after the
>    *            timeout interval is over.
>    * @thread: the kthread on which the scheduler which run.
> @@ -307,6 +326,7 @@ struct drm_gpu_scheduler {
>   	wait_queue_head_t		job_scheduled;
>   	atomic_t			hw_rq_count;
>   	atomic64_t			job_id_count;
> +	struct workqueue_struct		*timeout_wq;
>   	struct delayed_work		work_tdr;
>   	struct task_struct		*thread;
>   	struct list_head		pending_list;
> @@ -320,7 +340,8 @@ struct drm_gpu_scheduler {
>   
>   int drm_sched_init(struct drm_gpu_scheduler *sched,
>   		   const struct drm_sched_backend_ops *ops,
> -		   uint32_t hw_submission, unsigned hang_limit, long timeout,
> +		   uint32_t hw_submission, unsigned hang_limit,
> +		   long timeout, struct workqueue_struct *timeout_wq,
>   		   atomic_t *score, const char *name);
>   
>   void drm_sched_fini(struct drm_gpu_scheduler *sched);

Re: [PATCH v5 02/16] drm/sched: Allow using a dedicated workqueue for the timeout/fault tdr

[Boris Brezillon]

Hi Christian,

On Tue, 29 Jun 2021 13:03:58 +0200
Christian König <christian.koenig@amd.com> wrote:

> Am 29.06.21 um 09:34 schrieb Boris Brezillon:
> > Mali Midgard/Bifrost GPUs have 3 hardware queues but only a global GPU
> > reset. This leads to extra complexity when we need to synchronize timeout
> > works with the reset work. One solution to address that is to have an
> > ordered workqueue at the driver level that will be used by the different
> > schedulers to queue their timeout work. Thanks to the serialization
> > provided by the ordered workqueue we are guaranteed that timeout
> > handlers are executed sequentially, and can thus easily reset the GPU
> > from the timeout handler without extra synchronization.  
> 
> Well, we had already tried this and it didn't worked the way it is expected.
> 
> The major problem is that you not only want to serialize the queue, but 
> rather have a single reset for all queues.
> 
> Otherwise you schedule multiple resets for each hardware queue. E.g. for 
> your 3 hardware queues you would reset the GPU 3 times if all of them 
> time out at the same time (which is rather likely).
> 
> Using a single delayed work item doesn't work either because you then 
> only have one timeout.
> 
> What could be done is to cancel all delayed work items from all stopped 
> schedulers.

drm_sched_stop() does that already, and since we call drm_sched_stop()
on all queues in the timeout handler, we end up with only one global
reset happening even if several queues report a timeout at the same
time.

Regards,

Boris

Re: [PATCH v5 02/16] drm/sched: Allow using a dedicated workqueue for the timeout/fault tdr

[Christian König]

Am 29.06.21 um 13:18 schrieb Boris Brezillon:
> Hi Christian,
>
> On Tue, 29 Jun 2021 13:03:58 +0200
> Christian König <christian.koenig@amd.com> wrote:
>
>> Am 29.06.21 um 09:34 schrieb Boris Brezillon:
>>> Mali Midgard/Bifrost GPUs have 3 hardware queues but only a global GPU
>>> reset. This leads to extra complexity when we need to synchronize timeout
>>> works with the reset work. One solution to address that is to have an
>>> ordered workqueue at the driver level that will be used by the different
>>> schedulers to queue their timeout work. Thanks to the serialization
>>> provided by the ordered workqueue we are guaranteed that timeout
>>> handlers are executed sequentially, and can thus easily reset the GPU
>>> from the timeout handler without extra synchronization.
>> Well, we had already tried this and it didn't worked the way it is expected.
>>
>> The major problem is that you not only want to serialize the queue, but
>> rather have a single reset for all queues.
>>
>> Otherwise you schedule multiple resets for each hardware queue. E.g. for
>> your 3 hardware queues you would reset the GPU 3 times if all of them
>> time out at the same time (which is rather likely).
>>
>> Using a single delayed work item doesn't work either because you then
>> only have one timeout.
>>
>> What could be done is to cancel all delayed work items from all stopped
>> schedulers.
> drm_sched_stop() does that already, and since we call drm_sched_stop()
> on all queues in the timeout handler, we end up with only one global
> reset happening even if several queues report a timeout at the same
> time.

Ah, nice. Yeah, in this case it should indeed work as expected.

Feel free to add an Acked-by: Christian König <christian.koenig@amd.com> 
to it.

Regards,
Christian.

>
> Regards,
>
> Boris

Re: [PATCH v5 09/16] drm/panfrost: Simplify the reset serialization logic

[Boris Brezillon]

On Tue, 29 Jun 2021 09:35:03 +0200
Boris Brezillon <boris.brezillon@collabora.com> wrote:

> @@ -379,57 +370,72 @@ void panfrost_job_enable_interrupts(struct panfrost_device *pfdev)
>  	job_write(pfdev, JOB_INT_MASK, irq_mask);
>  }
>  
> -static bool panfrost_scheduler_stop(struct panfrost_queue_state *queue,
> -				    struct drm_sched_job *bad)
> +static void panfrost_reset(struct panfrost_device *pfdev,
> +			   struct drm_sched_job *bad)
>  {
> -	enum panfrost_queue_status old_status;
> -	bool stopped = false;
> +	unsigned int i;
> +	bool cookie;
>  
> -	mutex_lock(&queue->lock);
> -	old_status = atomic_xchg(&queue->status,
> -				 PANFROST_QUEUE_STATUS_STOPPED);
> -	if (old_status == PANFROST_QUEUE_STATUS_STOPPED)
> -		goto out;
> +	if (!atomic_read(&pfdev->reset.pending))
> +		return;
> +
> +	/* Stop the schedulers.
> +	 *
> +	 * FIXME: We temporarily get out of the dma_fence_signalling section
> +	 * because the cleanup path generate lockdep splats when taking locks
> +	 * to release job resources. We should rework the code to follow this
> +	 * pattern:
> +	 *
> +	 *	try_lock
> +	 *	if (locked)
> +	 *		release
> +	 *	else
> +	 *		schedule_work_to_release_later
> +	 */
> +	for (i = 0; i < NUM_JOB_SLOTS; i++)
> +		drm_sched_stop(&pfdev->js->queue[i].sched, bad);
> +
> +	cookie = dma_fence_begin_signalling();
>  
> -	WARN_ON(old_status != PANFROST_QUEUE_STATUS_ACTIVE);
> -	drm_sched_stop(&queue->sched, bad);
>  	if (bad)
>  		drm_sched_increase_karma(bad);
>  
> -	stopped = true;
> +	spin_lock(&pfdev->js->job_lock);
> +	for (i = 0; i < NUM_JOB_SLOTS; i++) {
> +		if (pfdev->jobs[i]) {
> +			pm_runtime_put_noidle(pfdev->dev);
> +			panfrost_devfreq_record_idle(&pfdev->pfdevfreq);
> +			pfdev->jobs[i] = NULL;
> +		}
> +	}
> +	spin_unlock(&pfdev->js->job_lock);
>  
> -	/*
> -	 * Set the timeout to max so the timer doesn't get started
> -	 * when we return from the timeout handler (restored in
> -	 * panfrost_scheduler_start()).
> +	panfrost_device_reset(pfdev);
> +
> +	/* GPU has been reset, we can cancel timeout/fault work that may have
> +	 * been queued in the meantime and clear the reset pending bit.
>  	 */
> -	queue->sched.timeout = MAX_SCHEDULE_TIMEOUT;
> +	atomic_set(&pfdev->reset.pending, 0);
> +	for (i = 0; i < NUM_JOB_SLOTS; i++)
> +		cancel_delayed_work(&pfdev->js->queue[i].sched.work_tdr);
>  

Those cancel_delayed_work() calls are useless, drm_sched_stop()
canceled those works already. I'll get rid of them in v6.

Re: [PATCH v5 02/16] drm/sched: Allow using a dedicated workqueue for the timeout/fault tdr

[Daniel Vetter]

On Tue, Jun 29, 2021 at 01:24:53PM +0200, Christian König wrote:
> Am 29.06.21 um 13:18 schrieb Boris Brezillon:
> > Hi Christian,
> > 
> > On Tue, 29 Jun 2021 13:03:58 +0200
> > Christian König <christian.koenig@amd.com> wrote:
> > 
> > > Am 29.06.21 um 09:34 schrieb Boris Brezillon:
> > > > Mali Midgard/Bifrost GPUs have 3 hardware queues but only a global GPU
> > > > reset. This leads to extra complexity when we need to synchronize timeout
> > > > works with the reset work. One solution to address that is to have an
> > > > ordered workqueue at the driver level that will be used by the different
> > > > schedulers to queue their timeout work. Thanks to the serialization
> > > > provided by the ordered workqueue we are guaranteed that timeout
> > > > handlers are executed sequentially, and can thus easily reset the GPU
> > > > from the timeout handler without extra synchronization.
> > > Well, we had already tried this and it didn't worked the way it is expected.
> > > 
> > > The major problem is that you not only want to serialize the queue, but
> > > rather have a single reset for all queues.
> > > 
> > > Otherwise you schedule multiple resets for each hardware queue. E.g. for
> > > your 3 hardware queues you would reset the GPU 3 times if all of them
> > > time out at the same time (which is rather likely).
> > > 
> > > Using a single delayed work item doesn't work either because you then
> > > only have one timeout.
> > > 
> > > What could be done is to cancel all delayed work items from all stopped
> > > schedulers.
> > drm_sched_stop() does that already, and since we call drm_sched_stop()
> > on all queues in the timeout handler, we end up with only one global
> > reset happening even if several queues report a timeout at the same
> > time.
> 
> Ah, nice. Yeah, in this case it should indeed work as expected.
> 
> Feel free to add an Acked-by: Christian König <christian.koenig@amd.com> to
> it.

Yeah this is also what we want for the case where you have cascading
reset. Like if you have per-engine reset, but if that fails, have to reset
the entire chip. That's similar but not quite to the soft recover, then
chip reset amdgpu does since for the engine reset you do have to stop the
scheduler. For full context the i915 reset cascade:

- try to evict the offending work with preemption (similar in it's impact
  to amdgpu's soft recovery I think), other requests on the same engine
  are unaffected here.

- reset the affected engine

- reset the entire chip

So my plan at least is that we'd only call drm_sched_stop as we escalate.
Also on earlier chips first and second might not be available.

Lucky for us chip reset only resets the render side of things, display
survives on gen4+, so we don't have the nasty locking issue amdgpu has.
-Daniel
-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch

Re: [PATCH v5 02/16] drm/sched: Allow using a dedicated workqueue for the timeout/fault tdr

[Andrey Grodzovsky]

On 2021-06-29 7:24 a.m., Christian König wrote:

> Am 29.06.21 um 13:18 schrieb Boris Brezillon:
>> Hi Christian,
>>
>> On Tue, 29 Jun 2021 13:03:58 +0200
>> Christian König <christian.koenig@amd.com> wrote:
>>
>>> Am 29.06.21 um 09:34 schrieb Boris Brezillon:
>>>> Mali Midgard/Bifrost GPUs have 3 hardware queues but only a global GPU
>>>> reset. This leads to extra complexity when we need to synchronize 
>>>> timeout
>>>> works with the reset work. One solution to address that is to have an
>>>> ordered workqueue at the driver level that will be used by the 
>>>> different
>>>> schedulers to queue their timeout work. Thanks to the serialization
>>>> provided by the ordered workqueue we are guaranteed that timeout
>>>> handlers are executed sequentially, and can thus easily reset the GPU
>>>> from the timeout handler without extra synchronization.
>>> Well, we had already tried this and it didn't worked the way it is 
>>> expected.
>>>
>>> The major problem is that you not only want to serialize the queue, but
>>> rather have a single reset for all queues.
>>>
>>> Otherwise you schedule multiple resets for each hardware queue. E.g. 
>>> for
>>> your 3 hardware queues you would reset the GPU 3 times if all of them
>>> time out at the same time (which is rather likely).
>>>
>>> Using a single delayed work item doesn't work either because you then
>>> only have one timeout.
>>>
>>> What could be done is to cancel all delayed work items from all stopped
>>> schedulers.
>> drm_sched_stop() does that already, and since we call drm_sched_stop()
>> on all queues in the timeout handler, we end up with only one global
>> reset happening even if several queues report a timeout at the same
>> time.
>
> Ah, nice. Yeah, in this case it should indeed work as expected.
>
> Feel free to add an Acked-by: Christian König 
> <christian.koenig@amd.com> to it.
>
> Regards,
> Christian.


Seems to me that for this to work we need to change cancel_delayed_work 
to cancel_delayed_work_sync
so not only pending TO handlers  are cancelled but also any in progress 
are waited for and to to prevent rearming.
Also move it right after kthread_park - before we start touching pending 
list.

Andrey


>
>>
>> Regards,
>>
>> Boris
>

Re: [PATCH v5 02/16] drm/sched: Allow using a dedicated workqueue for the timeout/fault tdr

[Boris Brezillon]

On Tue, 7 Sep 2021 14:53:58 -0400
Andrey Grodzovsky <andrey.grodzovsky@amd.com> wrote:

> On 2021-06-29 7:24 a.m., Christian König wrote:
> 
> > Am 29.06.21 um 13:18 schrieb Boris Brezillon:  
> >> Hi Christian,
> >>
> >> On Tue, 29 Jun 2021 13:03:58 +0200
> >> Christian König <christian.koenig@amd.com> wrote:
> >>  
> >>> Am 29.06.21 um 09:34 schrieb Boris Brezillon:  
> >>>> Mali Midgard/Bifrost GPUs have 3 hardware queues but only a global GPU
> >>>> reset. This leads to extra complexity when we need to synchronize 
> >>>> timeout
> >>>> works with the reset work. One solution to address that is to have an
> >>>> ordered workqueue at the driver level that will be used by the 
> >>>> different
> >>>> schedulers to queue their timeout work. Thanks to the serialization
> >>>> provided by the ordered workqueue we are guaranteed that timeout
> >>>> handlers are executed sequentially, and can thus easily reset the GPU
> >>>> from the timeout handler without extra synchronization.  
> >>> Well, we had already tried this and it didn't worked the way it is 
> >>> expected.
> >>>
> >>> The major problem is that you not only want to serialize the queue, but
> >>> rather have a single reset for all queues.
> >>>
> >>> Otherwise you schedule multiple resets for each hardware queue. E.g. 
> >>> for
> >>> your 3 hardware queues you would reset the GPU 3 times if all of them
> >>> time out at the same time (which is rather likely).
> >>>
> >>> Using a single delayed work item doesn't work either because you then
> >>> only have one timeout.
> >>>
> >>> What could be done is to cancel all delayed work items from all stopped
> >>> schedulers.  
> >> drm_sched_stop() does that already, and since we call drm_sched_stop()
> >> on all queues in the timeout handler, we end up with only one global
> >> reset happening even if several queues report a timeout at the same
> >> time.  
> >
> > Ah, nice. Yeah, in this case it should indeed work as expected.
> >
> > Feel free to add an Acked-by: Christian König 
> > <christian.koenig@amd.com> to it.
> >
> > Regards,
> > Christian.  
> 
> 
> Seems to me that for this to work we need to change cancel_delayed_work 
> to cancel_delayed_work_sync
> so not only pending TO handlers  are cancelled but also any in progress 
> are waited for and to to prevent rearming.
> Also move it right after kthread_park - before we start touching pending 
> list.

I'm probably missing something, but I don't really see why this
specific change would require replacing cancel_delayed_work() calls by
the sync variant. I mean, if there's a situation where we need to wait
for in-flight timeout handler to return, it was already the case before
that patch. Note that we need to be careful to not call the sync
variant in helpers that are called from the interrupt handler itself
to avoid deadlocks (i.e. drm_sched_stop()).

Re: [PATCH v5 02/16] drm/sched: Allow using a dedicated workqueue for the timeout/fault tdr

[Andrey Grodzovsky]

On 2021-09-08 2:50 a.m., Boris Brezillon wrote:
> On Tue, 7 Sep 2021 14:53:58 -0400
> Andrey Grodzovsky <andrey.grodzovsky@amd.com> wrote:
>
>> On 2021-06-29 7:24 a.m., Christian König wrote:
>>
>>> Am 29.06.21 um 13:18 schrieb Boris Brezillon:
>>>> Hi Christian,
>>>>
>>>> On Tue, 29 Jun 2021 13:03:58 +0200
>>>> Christian König <christian.koenig@amd.com> wrote:
>>>>   
>>>>> Am 29.06.21 um 09:34 schrieb Boris Brezillon:
>>>>>> Mali Midgard/Bifrost GPUs have 3 hardware queues but only a global GPU
>>>>>> reset. This leads to extra complexity when we need to synchronize
>>>>>> timeout
>>>>>> works with the reset work. One solution to address that is to have an
>>>>>> ordered workqueue at the driver level that will be used by the
>>>>>> different
>>>>>> schedulers to queue their timeout work. Thanks to the serialization
>>>>>> provided by the ordered workqueue we are guaranteed that timeout
>>>>>> handlers are executed sequentially, and can thus easily reset the GPU
>>>>>> from the timeout handler without extra synchronization.
>>>>> Well, we had already tried this and it didn't worked the way it is
>>>>> expected.
>>>>>
>>>>> The major problem is that you not only want to serialize the queue, but
>>>>> rather have a single reset for all queues.
>>>>>
>>>>> Otherwise you schedule multiple resets for each hardware queue. E.g.
>>>>> for
>>>>> your 3 hardware queues you would reset the GPU 3 times if all of them
>>>>> time out at the same time (which is rather likely).
>>>>>
>>>>> Using a single delayed work item doesn't work either because you then
>>>>> only have one timeout.
>>>>>
>>>>> What could be done is to cancel all delayed work items from all stopped
>>>>> schedulers.
>>>> drm_sched_stop() does that already, and since we call drm_sched_stop()
>>>> on all queues in the timeout handler, we end up with only one global
>>>> reset happening even if several queues report a timeout at the same
>>>> time.
>>> Ah, nice. Yeah, in this case it should indeed work as expected.
>>>
>>> Feel free to add an Acked-by: Christian König
>>> <christian.koenig@amd.com> to it.
>>>
>>> Regards,
>>> Christian.
>>
>> Seems to me that for this to work we need to change cancel_delayed_work
>> to cancel_delayed_work_sync
>> so not only pending TO handlers  are cancelled but also any in progress
>> are waited for and to to prevent rearming.
>> Also move it right after kthread_park - before we start touching pending
>> list.
> I'm probably missing something, but I don't really see why this
> specific change would require replacing cancel_delayed_work() calls by
> the sync variant.


I see, I missed the point that since now we have a single threaded 
processing and
only one TDR handler runs at given time there is no need to wait for 
other parallel in flight TDR handlers.


> I mean, if there's a situation where we need to wait
> for in-flight timeout handler to return, it was already the case before
> that patch.


In amdgpu case we avoided that by trylock on a common lock
and returning early in case it was already taken by another TDR handler


> Note that we need to be careful to not call the sync
> variant in helpers that are called from the interrupt handler itself
> to avoid deadlocks (i.e. drm_sched_stop()).


I am not clear here - which interrupt handler is drm_sched_stop
called from ? It's called from TDR work as far as I see in the code.

Andrey

Re: [PATCH v5 02/16] drm/sched: Allow using a dedicated workqueue for the timeout/fault tdr

[Boris Brezillon]

On Wed, 8 Sep 2021 10:53:21 -0400
Andrey Grodzovsky <andrey.grodzovsky@amd.com> wrote:

> > Note that we need to be careful to not call the sync
> > variant in helpers that are called from the interrupt handler itself
> > to avoid deadlocks (i.e. drm_sched_stop()).  
> 
> 
> I am not clear here - which interrupt handler is drm_sched_stop
> called from ? It's called from TDR work as far as I see in the code.

My bad, I meant the timeout handler, not the interrupt handler.