drm/panfrost: Hold runtime PM reference until jobs complete
authorRob Herring <robh@kernel.org>
Mon, 26 Aug 2019 22:33:11 +0000 (17:33 -0500)
committerRob Herring <robh@kernel.org>
Fri, 30 Aug 2019 14:51:51 +0000 (09:51 -0500)
Doing a pm_runtime_put as soon as a job is submitted is wrong as it should
not happen until the job completes. It works currently because we are
relying on the autosuspend timeout to keep the h/w enabled.

Fixes: f3ba91228e8e ("drm/panfrost: Add initial panfrost driver")
Cc: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Cc: Steven Price <steven.price@arm.com>
Cc: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Cc: David Airlie <airlied@linux.ie>
Cc: Daniel Vetter <daniel@ffwll.ch>
Signed-off-by: Rob Herring <robh@kernel.org>
Acked-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Steven Price <steven.price@arm.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190826223317.28509-3-robh@kernel.org
drivers/gpu/drm/panfrost/panfrost_job.c

index 05c85f45a0dea9eb660c2fd506a4c5e44af109b7..18bcc9bac6d2242ec6291e67ebf0fed74819e789 100644 (file)
@@ -150,8 +150,10 @@ static void panfrost_job_hw_submit(struct panfrost_job *job, int js)
        if (ret < 0)
                return;
 
-       if (WARN_ON(job_read(pfdev, JS_COMMAND_NEXT(js))))
-               goto end;
+       if (WARN_ON(job_read(pfdev, JS_COMMAND_NEXT(js)))) {
+               pm_runtime_put_sync_autosuspend(pfdev->dev);
+               return;
+       }
 
        cfg = panfrost_mmu_as_get(pfdev, &job->file_priv->mmu);
 
@@ -187,10 +189,6 @@ static void panfrost_job_hw_submit(struct panfrost_job *job, int js)
        job_write(pfdev, JS_COMMAND_NEXT(js), JS_COMMAND_START);
 
        spin_unlock_irqrestore(&pfdev->hwaccess_lock, flags);
-
-end:
-       pm_runtime_mark_last_busy(pfdev->dev);
-       pm_runtime_put_autosuspend(pfdev->dev);
 }
 
 static void panfrost_acquire_object_fences(struct drm_gem_object **bos,
@@ -369,6 +367,7 @@ static void panfrost_job_timedout(struct drm_sched_job *sched_job)
        struct panfrost_job *job = to_panfrost_job(sched_job);
        struct panfrost_device *pfdev = job->pfdev;
        int js = panfrost_job_get_slot(job);
+       unsigned long flags;
        int i;
 
        /*
@@ -394,6 +393,15 @@ static void panfrost_job_timedout(struct drm_sched_job *sched_job)
        if (sched_job)
                drm_sched_increase_karma(sched_job);
 
+       spin_lock_irqsave(&pfdev->js->job_lock, flags);
+       for (i = 0; i < NUM_JOB_SLOTS; i++) {
+               if (pfdev->jobs[i]) {
+                       pm_runtime_put_noidle(pfdev->dev);
+                       pfdev->jobs[i] = NULL;
+               }
+       }
+       spin_unlock_irqrestore(&pfdev->js->job_lock, flags);
+
        /* panfrost_core_dump(pfdev); */
 
        panfrost_devfreq_record_transition(pfdev, js);
@@ -450,12 +458,21 @@ static irqreturn_t panfrost_job_irq_handler(int irq, void *data)
                }
 
                if (status & JOB_INT_MASK_DONE(j)) {
-                       struct panfrost_job *job = pfdev->jobs[j];
+                       struct panfrost_job *job;
+
+                       spin_lock(&pfdev->js->job_lock);
+                       job = pfdev->jobs[j];
+                       /* Only NULL if job timeout occurred */
+                       if (job) {
+                               pfdev->jobs[j] = NULL;
+
+                               panfrost_mmu_as_put(pfdev, &job->file_priv->mmu);
+                               panfrost_devfreq_record_transition(pfdev, j);
 
-                       pfdev->jobs[j] = NULL;
-                       panfrost_mmu_as_put(pfdev, &job->file_priv->mmu);
-                       panfrost_devfreq_record_transition(pfdev, j);
-                       dma_fence_signal(job->done_fence);
+                               dma_fence_signal_locked(job->done_fence);
+                               pm_runtime_put_autosuspend(pfdev->dev);
+                       }
+                       spin_unlock(&pfdev->js->job_lock);
                }
 
                status &= ~mask;