drm/amdgpu: fix vf error handling

The error handling for virtual functions assumed a single
vf per VM and didn't properly account for bare metal.  Make
the error arrays per device and add locking.

Reviewed-by: Gavin Wan <gavin.wan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

1 files changed, 13 insertions, 0 deletions

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h
index afcfb8bcfb65..e5fd0ff6b29d 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h
@@ -36,6 +36,18 @@ struct amdgpu_mm_table {
         uint64_t                gpu_addr;
 };
 
+#define AMDGPU_VF_ERROR_ENTRY_SIZE    16
+
+/* struct error_entry - amdgpu VF error information. */
+struct amdgpu_vf_error_buffer {
+        struct mutex lock;
+        int read_count;
+        int write_count;
+        uint16_t code[AMDGPU_VF_ERROR_ENTRY_SIZE];
+        uint16_t flags[AMDGPU_VF_ERROR_ENTRY_SIZE];
+        uint64_t data[AMDGPU_VF_ERROR_ENTRY_SIZE];
+};
+
 /**
  * struct amdgpu_virt_ops - amdgpu device virt operations
  */
@@ -59,6 +71,7 @@ struct amdgpu_virt {
         struct work_struct              flr_work;
         struct amdgpu_mm_table          mm_table;
         const struct amdgpu_virt_ops    *ops;
+        struct amdgpu_vf_error_buffer   vf_errors;
 };
 
 #define AMDGPU_CSA_SIZE    (8 * 1024)