drm/msm/gpu: Track global faults per address-space

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>
Link: https://lore.kernel.org/r/20220201161618.778455-3-robdclark@gmail.com
Reviewed-by: Emma Anholt <emma@anholt.net>
Signed-off-by: Rob Clark <robdclark@chromium.org>

diff --git a/drivers/gpu/drm/msm/adreno/adreno_gpu.c b/drivers/gpu/drm/msm/adreno/adreno_gpu.c
index caa9076197deb2dfaa34c5fb59e506ddc43a3c30..58dfb23cf2afacb239f176aa692c2168702563aa 100644 (file)
--- 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 54ca0817d8071793f551fb3694fa6315a4ac58a3..af612add52647ba8f33cf1e53a22414e04445c5b 100644 (file)
--- 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 998e705c15efae7f2106c2195beb6532cd2efe8f..173ebd449f2fd8330f929e7c353ea8c27b16be21 100644 (file)
--- 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 ba840723134002b910119079cb50b82b46b655cd..c99627fc99ddb8f22c2a812ce5932260df7a3d3b 100644 (file)
--- 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;