[PATCH v2 2/2] drm/msm/gpu: Track global faults per address-space

[Rob Clark <robdclark@gmail.com>]

From: Rob Clark <robdclark@chromium.org>

Other processes don't need to know about faults that they are isolated
from by virtue of address space isolation.  They are only interested in
whether some of their state might have been corrupted.

But to be safe, also track unattributed faults.  This case should really
never happen unless there is a kernel bug (and that would never happen,
right?)

v2: Instead of adding a new param, just change the behavior of the
    existing param to match what userspace actually wants [anholt]

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/5934
Signed-off-by: Rob Clark <robdclark@chromium.org>
---
 drivers/gpu/drm/msm/adreno/adreno_gpu.c | 2 +-
 drivers/gpu/drm/msm/msm_gem.h           | 3 +++
 drivers/gpu/drm/msm/msm_gpu.c           | 8 +++++++-
 drivers/gpu/drm/msm/msm_gpu.h           | 5 ++++-
 4 files changed, 15 insertions(+), 3 deletions(-)

diff --git a/drivers/gpu/drm/msm/adreno/adreno_gpu.c b/drivers/gpu/drm/msm/adreno/adreno_gpu.c
index caa9076197de..58dfb23cf2af 100644
--- a/drivers/gpu/drm/msm/adreno/adreno_gpu.c
+++ b/drivers/gpu/drm/msm/adreno/adreno_gpu.c
@@ -269,7 +269,7 @@ int adreno_get_param(struct msm_gpu *gpu, struct msm_file_private *ctx,
 		*value = 0;
 		return 0;
 	case MSM_PARAM_FAULTS:
-		*value = gpu->global_faults;
+		*value = gpu->global_faults + ctx->aspace->faults;
 		return 0;
 	case MSM_PARAM_SUSPENDS:
 		*value = gpu->suspend_count;
diff --git a/drivers/gpu/drm/msm/msm_gem.h b/drivers/gpu/drm/msm/msm_gem.h
index 54ca0817d807..af612add5264 100644
--- a/drivers/gpu/drm/msm/msm_gem.h
+++ b/drivers/gpu/drm/msm/msm_gem.h
@@ -35,6 +35,9 @@ struct msm_gem_address_space {
 	 * will be non-NULL:
 	 */
 	struct pid *pid;
+
+	/* @faults: the number of GPU hangs associated with this address space */
+	int faults;
 };
 
 struct msm_gem_vma {
diff --git a/drivers/gpu/drm/msm/msm_gpu.c b/drivers/gpu/drm/msm/msm_gpu.c
index 2c1049c0ea14..942bf41403ff 100644
--- a/drivers/gpu/drm/msm/msm_gpu.c
+++ b/drivers/gpu/drm/msm/msm_gpu.c
@@ -370,8 +370,8 @@ static void recover_worker(struct kthread_work *work)
 		struct task_struct *task;
 
 		/* Increment the fault counts */
-		gpu->global_faults++;
 		submit->queue->faults++;
+		submit->aspace->faults++;
 
 		task = get_pid_task(submit->pid, PIDTYPE_PID);
 		if (task) {
@@ -389,6 +389,12 @@ static void recover_worker(struct kthread_work *work)
 		} else {
 			msm_rd_dump_submit(priv->hangrd, submit, NULL);
 		}
+	} else {
+		/*
+		 * We couldn't attribute this fault to any particular context,
+		 * so increment the global fault count instead.
+		 */
+		gpu->global_faults++;
 	}
 
 	/* Record the crash state */
diff --git a/drivers/gpu/drm/msm/msm_gpu.h b/drivers/gpu/drm/msm/msm_gpu.h
index ba8407231340..c99627fc99dd 100644
--- a/drivers/gpu/drm/msm/msm_gpu.h
+++ b/drivers/gpu/drm/msm/msm_gpu.h
@@ -200,7 +200,10 @@ struct msm_gpu {
 	/* does gpu need hw_init? */
 	bool needs_hw_init;
 
-	/* number of GPU hangs (for all contexts) */
+	/**
+	 * global_faults: number of GPU hangs not attributed to a particular
+	 * address space
+	 */
 	int global_faults;
 
 	void __iomem *mmio;
-- 
2.34.1