drm/amdgpu: add amdgpu interface to query cu info

Signed-off-by: Flora Cui <flora.cui@amd.com>
Signed-off-by: Harish Kasiviswanathan <Harish.Kasiviswanathan@amd.com>
Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
Reviewed-by: Oded Gabbay <oded.gabbay@gmail.com>
Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>

7 files changed, 39 insertions, 0 deletions

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
index bbe06e04dcb6..e40c5df55f95 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
@@ -959,6 +959,7 @@ struct amdgpu_gfx_config {
 };
 
 struct amdgpu_cu_info {
+	uint32_t simd_per_cu;
 	uint32_t max_waves_per_simd;
 	uint32_t wave_front_size;
 	uint32_t max_scratch_slots_per_cu;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
index 01aa851938f8..48c1675aa342 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
@@ -304,3 +304,26 @@ uint32_t get_max_engine_clock_in_mhz(struct kgd_dev *kgd)
 
 	return amdgpu_dpm_get_sclk(adev, false) / 100;
 }
+
+void get_cu_info(struct kgd_dev *kgd, struct kfd_cu_info *cu_info)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
+	struct amdgpu_cu_info acu_info = adev->gfx.cu_info;
+
+	memset(cu_info, 0, sizeof(*cu_info));
+	if (sizeof(cu_info->cu_bitmap) != sizeof(acu_info.bitmap))
+		return;
+
+	cu_info->cu_active_number = acu_info.number;
+	cu_info->cu_ao_mask = acu_info.ao_cu_mask;
+	memcpy(&cu_info->cu_bitmap[0], &acu_info.bitmap[0],
+	       sizeof(acu_info.bitmap));
+	cu_info->num_shader_engines = adev->gfx.config.max_shader_engines;
+	cu_info->num_shader_arrays_per_engine = adev->gfx.config.max_sh_per_se;
+	cu_info->num_cu_per_sh = adev->gfx.config.max_cu_per_sh;
+	cu_info->simd_per_cu = acu_info.simd_per_cu;
+	cu_info->max_waves_per_simd = acu_info.max_waves_per_simd;
+	cu_info->wave_front_size = acu_info.wave_front_size;
+	cu_info->max_scratch_slots_per_cu = acu_info.max_scratch_slots_per_cu;
+	cu_info->lds_size = acu_info.lds_size;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h
index 8d689ab7e429..a8fa2253817b 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h
@@ -60,6 +60,7 @@ uint64_t get_vmem_size(struct kgd_dev *kgd);
 uint64_t get_gpu_clock_counter(struct kgd_dev *kgd);
 
 uint32_t get_max_engine_clock_in_mhz(struct kgd_dev *kgd);
+void get_cu_info(struct kgd_dev *kgd, struct kfd_cu_info *cu_info);
 
 #define read_user_wptr(mmptr, wptr, dst)				\
 	({								\
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c
index 12feba8091df..c9b98d09a09b 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c
@@ -200,6 +200,7 @@ static const struct kfd2kgd_calls kfd2kgd = {
 	.get_fw_version = get_fw_version,
 	.set_scratch_backing_va = set_scratch_backing_va,
 	.get_tile_config = get_tile_config,
+	.get_cu_info = get_cu_info
 };
 
 struct kfd2kgd_calls *amdgpu_amdkfd_gfx_7_get_functions(void)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c
index b38049569264..c538e3029f6f 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c
@@ -161,6 +161,7 @@ static const struct kfd2kgd_calls kfd2kgd = {
 	.get_fw_version = get_fw_version,
 	.set_scratch_backing_va = set_scratch_backing_va,
 	.get_tile_config = get_tile_config,
+	.get_cu_info = get_cu_info
 };
 
 struct kfd2kgd_calls *amdgpu_amdkfd_gfx_8_0_get_functions(void)
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c
index 83d94c23aa78..9c62ebd5a19c 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c
@@ -48,6 +48,8 @@
 #include "oss/oss_2_0_d.h"
 #include "oss/oss_2_0_sh_mask.h"
 
+#define NUM_SIMD_PER_CU 0x4 /* missing from the gfx_7 IP headers */
+
 #define GFX7_NUM_GFX_RINGS     1
 #define GFX7_MEC_HPD_SIZE      2048
 
@@ -5277,6 +5279,11 @@ static void gfx_v7_0_get_cu_info(struct amdgpu_device *adev)
 
 	cu_info->number = active_cu_number;
 	cu_info->ao_cu_mask = ao_cu_mask;
+	cu_info->simd_per_cu = NUM_SIMD_PER_CU;
+	cu_info->max_waves_per_simd = 10;
+	cu_info->max_scratch_slots_per_cu = 32;
+	cu_info->wave_front_size = 64;
+	cu_info->lds_size = 64;
 }
 
 const struct amdgpu_ip_block_version gfx_v7_0_ip_block =
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
index 46550b588982..e18c2e62a20b 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
@@ -7116,6 +7116,11 @@ static void gfx_v8_0_get_cu_info(struct amdgpu_device *adev)
 
 	cu_info->number = active_cu_number;
 	cu_info->ao_cu_mask = ao_cu_mask;
+	cu_info->simd_per_cu = NUM_SIMD_PER_CU;
+	cu_info->max_waves_per_simd = 10;
+	cu_info->max_scratch_slots_per_cu = 32;
+	cu_info->wave_front_size = 64;
+	cu_info->lds_size = 64;
 }
 
 const struct amdgpu_ip_block_version gfx_v8_0_ip_block =