drm/amdgpu: split gfx callbacks into ras and non-ras ones

gfx ras is only available in cerntain ip generations.
Signed-off-by: Hawking Zhang <Hawking.Zhang@amd.com>
Reviewed-by: Dennis Li <Dennis.Li@amd.com>
Reviewed-by: John Clements <John.Clements@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c

@@ -677,8 +677,9 @@ int amdgpu_gfx_process_ras_data_cb(struct amdgpu_device *adev,
 	 */
 	if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__GFX)) {
 		kgd2kfd_set_sram_ecc_flag(adev->kfd.dev);
-		if (adev->gfx.funcs->query_ras_error_count)
-			adev->gfx.funcs->query_ras_error_count(adev, err_data);
+		if (adev->gfx.ras_funcs &&
+		    adev->gfx.ras_funcs->query_ras_error_count)
+			adev->gfx.ras_funcs->query_ras_error_count(adev, err_data);
 		amdgpu_ras_reset_gpu(adev);
 	}
 	return AMDGPU_RAS_SUCCESS;

drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h

@@ -205,6 +205,19 @@ struct amdgpu_cu_info {
 	uint32_t bitmap[4][4];
 };
 
+struct amdgpu_gfx_ras_funcs {
+	int (*ras_late_init)(struct amdgpu_device *adev);
+	void (*ras_fini)(struct amdgpu_device *adev);
+	int (*ras_error_inject)(struct amdgpu_device *adev,
+				void *inject_if);
+	int (*query_ras_error_count)(struct amdgpu_device *adev,
+				     void *ras_error_status);
+	void (*reset_ras_error_count)(struct amdgpu_device *adev);
+	void (*query_ras_error_status)(struct amdgpu_device *adev);
+	void (*reset_ras_error_status)(struct amdgpu_device *adev);
+	void (*enable_watchdog_timer)(struct amdgpu_device *adev);
+};
+
 struct amdgpu_gfx_funcs {
 	/* get the gpu clock counter */
 	uint64_t (*get_gpu_clock_counter)(struct amdgpu_device *adev);
@@ -220,14 +233,8 @@ struct amdgpu_gfx_funcs {
 				uint32_t *dst);
 	void (*select_me_pipe_q)(struct amdgpu_device *adev, u32 me, u32 pipe,
 				 u32 queue, u32 vmid);
-	int (*ras_error_inject)(struct amdgpu_device *adev, void *inject_if);
-	int (*query_ras_error_count) (struct amdgpu_device *adev, void *ras_error_status);
-	void (*reset_ras_error_count) (struct amdgpu_device *adev);
 	void (*init_spm_golden)(struct amdgpu_device *adev);
-	void (*query_ras_error_status) (struct amdgpu_device *adev);
-	void (*reset_ras_error_status) (struct amdgpu_device *adev);
 	void (*update_perfmon_mgcg)(struct amdgpu_device *adev, bool enable);
-	void (*enable_watchdog_timer)(struct amdgpu_device *adev);
 };
 
 struct sq_work {
@@ -330,7 +337,8 @@ struct amdgpu_gfx {
 	DECLARE_BITMAP			(pipe_reserve_bitmap, AMDGPU_MAX_COMPUTE_QUEUES);
 
 	/*ras */
-	struct ras_common_if		*ras_if;
+	struct ras_common_if			*ras_if;
+	const struct amdgpu_gfx_ras_funcs	*ras_funcs;
 };
 
 #define amdgpu_gfx_get_gpu_clock_counter(adev) (adev)->gfx.funcs->get_gpu_clock_counter((adev))

drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c

@@ -792,11 +792,13 @@ int amdgpu_ras_query_error_status(struct amdgpu_device *adev,
 		}
 		break;
 	case AMDGPU_RAS_BLOCK__GFX:
-		if (adev->gfx.funcs->query_ras_error_count)
-			adev->gfx.funcs->query_ras_error_count(adev, &err_data);
+		if (adev->gfx.ras_funcs &&
+		    adev->gfx.ras_funcs->query_ras_error_count)
+			adev->gfx.ras_funcs->query_ras_error_count(adev, &err_data);
 
-		if (adev->gfx.funcs->query_ras_error_status)
-			adev->gfx.funcs->query_ras_error_status(adev);
+		if (adev->gfx.ras_funcs &&
+		    adev->gfx.ras_funcs->query_ras_error_status)
+			adev->gfx.ras_funcs->query_ras_error_status(adev);
 		break;
 	case AMDGPU_RAS_BLOCK__MMHUB:
 		if (adev->mmhub.ras_funcs &&
@@ -852,11 +854,13 @@ int amdgpu_ras_reset_error_status(struct amdgpu_device *adev,
 
 	switch (block) {
 	case AMDGPU_RAS_BLOCK__GFX:
-		if (adev->gfx.funcs->reset_ras_error_count)
-			adev->gfx.funcs->reset_ras_error_count(adev);
+		if (adev->gfx.ras_funcs &&
+		    adev->gfx.ras_funcs->reset_ras_error_count)
+			adev->gfx.ras_funcs->reset_ras_error_count(adev);
 
-		if (adev->gfx.funcs->reset_ras_error_status)
-			adev->gfx.funcs->reset_ras_error_status(adev);
+		if (adev->gfx.ras_funcs &&
+		    adev->gfx.ras_funcs->reset_ras_error_status)
+			adev->gfx.ras_funcs->reset_ras_error_status(adev);
 		break;
 	case AMDGPU_RAS_BLOCK__MMHUB:
 		if (adev->mmhub.ras_funcs &&
@@ -926,8 +930,9 @@ int amdgpu_ras_error_inject(struct amdgpu_device *adev,
 
 	switch (info->head.block) {
 	case AMDGPU_RAS_BLOCK__GFX:
-		if (adev->gfx.funcs->ras_error_inject)
-			ret = adev->gfx.funcs->ras_error_inject(adev, info);
+		if (adev->gfx.ras_funcs &&
+		    adev->gfx.ras_funcs->ras_error_inject)
+			ret = adev->gfx.ras_funcs->ras_error_inject(adev, info);
 		else
 			ret = -EINVAL;
 		break;
@@ -1514,8 +1519,9 @@ static void amdgpu_ras_error_status_query(struct amdgpu_device *adev,
 	 */
 	switch (info->head.block) {
 	case AMDGPU_RAS_BLOCK__GFX:
-		if (adev->gfx.funcs->query_ras_error_status)
-			adev->gfx.funcs->query_ras_error_status(adev);
+		if (adev->gfx.ras_funcs &&
+		    adev->gfx.ras_funcs->query_ras_error_status)
+			adev->gfx.ras_funcs->query_ras_error_status(adev);
 		break;
 	case AMDGPU_RAS_BLOCK__MMHUB:
 		if (adev->mmhub.ras_funcs &&

drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c

@@ -2103,45 +2103,22 @@ static void gfx_v9_0_select_me_pipe_q(struct amdgpu_device *adev,
 }
 
 static const struct amdgpu_gfx_funcs gfx_v9_0_gfx_funcs = {
-	.get_gpu_clock_counter = &gfx_v9_0_get_gpu_clock_counter,
-	.select_se_sh = &gfx_v9_0_select_se_sh,
-	.read_wave_data = &gfx_v9_0_read_wave_data,
-	.read_wave_sgprs = &gfx_v9_0_read_wave_sgprs,
-	.read_wave_vgprs = &gfx_v9_0_read_wave_vgprs,
-	.select_me_pipe_q = &gfx_v9_0_select_me_pipe_q,
+        .get_gpu_clock_counter = &gfx_v9_0_get_gpu_clock_counter,
+        .select_se_sh = &gfx_v9_0_select_se_sh,
+        .read_wave_data = &gfx_v9_0_read_wave_data,
+        .read_wave_sgprs = &gfx_v9_0_read_wave_sgprs,
+        .read_wave_vgprs = &gfx_v9_0_read_wave_vgprs,
+        .select_me_pipe_q = &gfx_v9_0_select_me_pipe_q,
+};
+
+static const struct amdgpu_gfx_ras_funcs gfx_v9_0_ras_funcs = {
+	.ras_late_init = amdgpu_gfx_ras_late_init,
+	.ras_fini = amdgpu_gfx_ras_fini,
 	.ras_error_inject = &gfx_v9_0_ras_error_inject,
 	.query_ras_error_count = &gfx_v9_0_query_ras_error_count,
 	.reset_ras_error_count = &gfx_v9_0_reset_ras_error_count,
 };
 
-static const struct amdgpu_gfx_funcs gfx_v9_4_gfx_funcs = {
-	.get_gpu_clock_counter = &gfx_v9_0_get_gpu_clock_counter,
-	.select_se_sh = &gfx_v9_0_select_se_sh,
-	.read_wave_data = &gfx_v9_0_read_wave_data,
-	.read_wave_sgprs = &gfx_v9_0_read_wave_sgprs,
-	.read_wave_vgprs = &gfx_v9_0_read_wave_vgprs,
-	.select_me_pipe_q = &gfx_v9_0_select_me_pipe_q,
-	.ras_error_inject = &gfx_v9_4_ras_error_inject,
-	.query_ras_error_count = &gfx_v9_4_query_ras_error_count,
-	.reset_ras_error_count = &gfx_v9_4_reset_ras_error_count,
-	.query_ras_error_status = &gfx_v9_4_query_ras_error_status,
-};
-
-static const struct amdgpu_gfx_funcs gfx_v9_4_2_gfx_funcs = {
-	.get_gpu_clock_counter = &gfx_v9_0_get_gpu_clock_counter,
-	.select_se_sh = &gfx_v9_0_select_se_sh,
-	.read_wave_data = &gfx_v9_0_read_wave_data,
-	.read_wave_sgprs = &gfx_v9_0_read_wave_sgprs,
-	.read_wave_vgprs = &gfx_v9_0_read_wave_vgprs,
-	.select_me_pipe_q = &gfx_v9_0_select_me_pipe_q,
-	.ras_error_inject = &gfx_v9_4_2_ras_error_inject,
-	.query_ras_error_count = &gfx_v9_4_2_query_ras_error_count,
-	.reset_ras_error_count = &gfx_v9_4_2_reset_ras_error_count,
-	.query_ras_error_status = &gfx_v9_4_2_query_ras_error_status,
-	.reset_ras_error_status = &gfx_v9_4_2_reset_ras_error_status,
-	.enable_watchdog_timer = &gfx_v9_4_2_enable_watchdog_timer,
-};
-
 static int gfx_v9_0_gpu_early_init(struct amdgpu_device *adev)
 {
 	u32 gb_addr_config;
@@ -2168,6 +2145,7 @@ static int gfx_v9_0_gpu_early_init(struct amdgpu_device *adev)
 		DRM_INFO("fix gfx.config for vega12\n");
 		break;
 	case CHIP_VEGA20:
+		adev->gfx.ras_funcs = &gfx_v9_0_ras_funcs;
 		adev->gfx.config.max_hw_contexts = 8;
 		adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
 		adev->gfx.config.sc_prim_fifo_size_backend = 0x100;
@@ -2193,7 +2171,7 @@ static int gfx_v9_0_gpu_early_init(struct amdgpu_device *adev)
 			gb_addr_config = RAVEN_GB_ADDR_CONFIG_GOLDEN;
 		break;
 	case CHIP_ARCTURUS:
-		adev->gfx.funcs = &gfx_v9_4_gfx_funcs;
+		adev->gfx.ras_funcs = &gfx_v9_4_ras_funcs;
 		adev->gfx.config.max_hw_contexts = 8;
 		adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
 		adev->gfx.config.sc_prim_fifo_size_backend = 0x100;
@@ -2214,7 +2192,7 @@ static int gfx_v9_0_gpu_early_init(struct amdgpu_device *adev)
 		gb_addr_config |= 0x22010042;
 		break;
 	case CHIP_ALDEBARAN:
-		adev->gfx.funcs = &gfx_v9_4_2_gfx_funcs;
+		adev->gfx.ras_funcs = &gfx_v9_4_2_ras_funcs;
 		adev->gfx.config.max_hw_contexts = 8;
 		adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
 		adev->gfx.config.sc_prim_fifo_size_backend = 0x100;
@@ -2447,7 +2425,9 @@ static int gfx_v9_0_sw_fini(void *handle)
 	int i;
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
-	amdgpu_gfx_ras_fini(adev);
+	if (adev->gfx.ras_funcs &&
+	    adev->gfx.ras_funcs->ras_fini)
+		adev->gfx.ras_funcs->ras_fini(adev);
 
 	for (i = 0; i < adev->gfx.num_gfx_rings; i++)
 		amdgpu_ring_fini(&adev->gfx.gfx_ring[i]);
@@ -4766,12 +4746,16 @@ static int gfx_v9_0_ecc_late_init(void *handle)
 	if (r)
 		return r;
 
-	r = amdgpu_gfx_ras_late_init(adev);
-	if (r)
-		return r;
+	if (adev->gfx.ras_funcs &&
+	    adev->gfx.ras_funcs->ras_late_init) {
+		r = adev->gfx.ras_funcs->ras_late_init(adev);
+		if (r)
+			return r;
+	}
 
-	if (adev->gfx.funcs->enable_watchdog_timer)
-		adev->gfx.funcs->enable_watchdog_timer(adev);
+	if (adev->gfx.ras_funcs &&
+	    adev->gfx.ras_funcs->enable_watchdog_timer)
+		adev->gfx.ras_funcs->enable_watchdog_timer(adev);
 
 	return 0;
 }

drivers/gpu/drm/amd/amdgpu/gfx_v9_4.c

@@ -863,8 +863,8 @@ static int gfx_v9_4_ras_error_count(struct amdgpu_device *adev,
 	return 0;
 }
 
-int gfx_v9_4_query_ras_error_count(struct amdgpu_device *adev,
-				   void *ras_error_status)
+static int gfx_v9_4_query_ras_error_count(struct amdgpu_device *adev,
+					  void *ras_error_status)
 {
 	struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status;
 	uint32_t sec_count = 0, ded_count = 0;
@@ -906,7 +906,7 @@ int gfx_v9_4_query_ras_error_count(struct amdgpu_device *adev,
 	return 0;
 }
 
-void gfx_v9_4_reset_ras_error_count(struct amdgpu_device *adev)
+static void gfx_v9_4_reset_ras_error_count(struct amdgpu_device *adev)
 {
 	int i, j, k;
 
@@ -971,7 +971,8 @@ void gfx_v9_4_reset_ras_error_count(struct amdgpu_device *adev)
 	WREG32_SOC15(GC, 0, mmATC_L2_CACHE_4K_DSM_INDEX, 255);
 }
 
-int gfx_v9_4_ras_error_inject(struct amdgpu_device *adev, void *inject_if)
+static int gfx_v9_4_ras_error_inject(struct amdgpu_device *adev,
+				     void *inject_if)
 {
 	struct ras_inject_if *info = (struct ras_inject_if *)inject_if;
 	int ret;
@@ -996,7 +997,7 @@ int gfx_v9_4_ras_error_inject(struct amdgpu_device *adev, void *inject_if)
 static const struct soc15_reg_entry gfx_v9_4_rdrsp_status_regs =
 	{ SOC15_REG_ENTRY(GC, 0, mmGCEA_ERR_STATUS), 0, 1, 32 };
 
-void gfx_v9_4_query_ras_error_status(struct amdgpu_device *adev)
+static void gfx_v9_4_query_ras_error_status(struct amdgpu_device *adev)
 {
 	uint32_t i, j;
 	uint32_t reg_value;
@@ -1021,3 +1022,12 @@ void gfx_v9_4_query_ras_error_status(struct amdgpu_device *adev)
 	gfx_v9_4_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
 	mutex_unlock(&adev->grbm_idx_mutex);
 }
+
+const struct amdgpu_gfx_ras_funcs gfx_v9_4_ras_funcs = {
+        .ras_late_init = amdgpu_gfx_ras_late_init,
+        .ras_fini = amdgpu_gfx_ras_fini,
+        .ras_error_inject = &gfx_v9_4_ras_error_inject,
+        .query_ras_error_count = &gfx_v9_4_query_ras_error_count,
+        .reset_ras_error_count = &gfx_v9_4_reset_ras_error_count,
+        .query_ras_error_status = &gfx_v9_4_query_ras_error_status,
+};

drivers/gpu/drm/amd/amdgpu/gfx_v9_4.h

@@ -24,16 +24,6 @@
 #ifndef __GFX_V9_4_H__
 #define __GFX_V9_4_H__
 
-void gfx_v9_4_clear_ras_edc_counter(struct amdgpu_device *adev);
-
-int gfx_v9_4_query_ras_error_count(struct amdgpu_device *adev,
-				   void *ras_error_status);
-
-int gfx_v9_4_ras_error_inject(struct amdgpu_device *adev,
-				     void *inject_if);
-
-void gfx_v9_4_reset_ras_error_count(struct amdgpu_device *adev);
-
-void gfx_v9_4_query_ras_error_status(struct amdgpu_device *adev);
+extern const struct amdgpu_gfx_ras_funcs gfx_v9_4_ras_funcs;
 
 #endif /* __GFX_V9_4_H__ */

drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2.c

@@ -1283,4 +1283,15 @@ static void gfx_v9_4_2_reset_sq_timeout_status(struct amdgpu_device *adev)
 	}
 	gfx_v9_4_2_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
 	mutex_unlock(&adev->grbm_idx_mutex);
-}
+}
\ No newline at end of file
+
+const struct amdgpu_gfx_ras_funcs gfx_v9_4_2_ras_funcs = {
+	.ras_late_init = amdgpu_gfx_ras_late_init,
+	.ras_fini = amdgpu_gfx_ras_fini,
+	.ras_error_inject = &gfx_v9_4_2_ras_error_inject,
+	.query_ras_error_count = &gfx_v9_4_2_query_ras_error_count,
+	.reset_ras_error_count = &gfx_v9_4_2_reset_ras_error_count,
+	.query_ras_error_status = &gfx_v9_4_2_query_ras_error_status,
+	.reset_ras_error_status = &gfx_v9_4_2_reset_ras_error_status,
+	.enable_watchdog_timer = &gfx_v9_4_2_enable_watchdog_timer,
+};

drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2.h

@@ -30,11 +30,6 @@ void gfx_v9_4_2_init_golden_registers(struct amdgpu_device *adev,
 				      uint32_t die_id);
 void gfx_v9_4_2_set_power_brake_sequence(struct amdgpu_device *adev);
 
-void gfx_v9_4_2_reset_ras_error_count(struct amdgpu_device *adev);
-int gfx_v9_4_2_ras_error_inject(struct amdgpu_device *adev, void *inject_if);
-void gfx_v9_4_2_query_ras_error_status(struct amdgpu_device *adev);
-int gfx_v9_4_2_query_ras_error_count(struct amdgpu_device *adev,
-				   void *ras_error_status);
-void gfx_v9_4_2_reset_ras_error_status(struct amdgpu_device *adev);
-void gfx_v9_4_2_enable_watchdog_timer(struct amdgpu_device *adev);
+extern const struct amdgpu_gfx_ras_funcs gfx_v9_4_2_ras_funcs;
+
 #endif /* __GFX_V9_4_2_H__ */