drm/amdgpu: fix context switch

Properly protect the state and also handle submission failures.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Jammy Zhou <Jammy.Zhou@amd.com>
Reviewed-by: Monk Liu <monk.liu@amd.com>

1 files changed, 7 insertions, 1 deletions

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c
index 74ed94ee7102..560c5fd347be 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c
@@ -140,6 +140,7 @@ int amdgpu_ib_schedule(struct amdgpu_device *adev, unsigned num_ibs,
 {
         struct amdgpu_ib *ib = &ibs[0];
         struct amdgpu_ring *ring;
+        struct amdgpu_ctx *ctx, *old_ctx;
         struct amdgpu_vm *vm;
         unsigned i;
         int r = 0;
@@ -148,6 +149,7 @@ int amdgpu_ib_schedule(struct amdgpu_device *adev, unsigned num_ibs,
                 return -EINVAL;
 
         ring = ibs->ring;
+        ctx = ibs->ctx;
         vm = ibs->vm;
 
         if (!ring->ready) {
@@ -189,19 +191,23 @@ int amdgpu_ib_schedule(struct amdgpu_device *adev, unsigned num_ibs,
         if (ring->funcs->emit_hdp_flush)
                 amdgpu_ring_emit_hdp_flush(ring);
 
+        old_ctx = ring->current_ctx;
         for (i = 0; i < num_ibs; ++i) {
                 ib = &ibs[i];
 
-                if (ib->ring != ring) {
+                if (ib->ring != ring || ib->ctx != ctx || ib->vm != vm) {
+                        ring->current_ctx = old_ctx;
                         amdgpu_ring_unlock_undo(ring);
                         return -EINVAL;
                 }
                 amdgpu_ring_emit_ib(ring, ib);
+                ring->current_ctx = ctx;
         }
 
         r = amdgpu_fence_emit(ring, owner, &ib->fence);
         if (r) {
                 dev_err(adev->dev, "failed to emit fence (%d)\n", r);
+                ring->current_ctx = old_ctx;
                 amdgpu_ring_unlock_undo(ring);
                 return r;
         }