gpu: nvgpu: fix deinit of GR

Existing implementation of GR de-init doesn't account for multiple
instances of struct nvgpu_gr. As a fix, below changes are added.

1) nvgpu_gr_free is unified for VGPU as well as native.
2) All the GR instances are freed.
3) Appropriate NULL checks are added when freeing GR memories.
4) 2D, 3D, I2M and ZBC etc are explicitely disabled when MIG is set.
5) In ioctl_ctrl, checks are added to not return error when zbc is NULL
   for VGPU as requests are rerouted to RMserver.

Jira NVGPU-6920

Change-Id: Icaa40f88f523c2cdbfe3a4fd6a55681ea7a83d12
Signed-off-by: Debarshi Dutta <ddutta@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nvgpu/+/2578500
Reviewed-by: svc_kernel_abi <svc_kernel_abi@nvidia.com>
Reviewed-by: Vijayakumar Subbu <vsubbu@nvidia.com>
Reviewed-by: Dinesh T <dt@nvidia.com>
Reviewed-by: Seshendra Gadagottu <sgadagottu@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: Antony Clince Alex <aalex@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
GVS: Gerrit_Virtual_Submit

drivers/gpu/nvgpu/common/gr/global_ctx.c

@@ -67,7 +67,9 @@ nvgpu_gr_global_ctx_desc_alloc(struct gk20a *g)
 void nvgpu_gr_global_ctx_desc_free(struct gk20a *g,
 	struct nvgpu_gr_global_ctx_buffer_desc *desc)
 {
+	if (desc != NULL) {
 		nvgpu_kfree(g, desc);
+	}
 }
 
 

drivers/gpu/nvgpu/common/gr/gr.c

@@ -305,29 +305,31 @@ out:
 
 static void gr_remove_support(struct gk20a *g)
 {
-	struct nvgpu_gr *gr = g->gr;
+	struct nvgpu_gr *gr = NULL;
+	u32 i;
 
 	nvgpu_log_fn(g, " ");
 
-	nvgpu_gr_global_ctx_buffer_free(g, gr->global_ctx_buffer);
-	nvgpu_gr_global_ctx_desc_free(g, gr->global_ctx_buffer);
-
-	nvgpu_gr_ctx_desc_free(g, gr->gr_ctx_desc);
-
-	nvgpu_gr_config_deinit(g, gr->config);
-
 	nvgpu_netlist_deinit_ctx_vars(g);
 
-#ifdef CONFIG_NVGPU_DEBUGGER
+	for (i = 0U; i < g->num_gr_instances; i++) {
-	nvgpu_gr_hwpm_map_deinit(g, gr->hwpm_map);
+		gr = &g->gr[i];
-#endif
 
-#ifdef CONFIG_NVGPU_GRAPHICS
+		nvgpu_gr_global_ctx_buffer_free(g, gr->global_ctx_buffer);
-	nvgpu_gr_zbc_deinit(g, gr->zbc);
+		nvgpu_gr_global_ctx_desc_free(g, gr->global_ctx_buffer);
-	nvgpu_gr_zcull_deinit(g, gr->zcull);
+		gr->global_ctx_buffer = NULL;
-#endif /* CONFIG_NVGPU_GRAPHICS */
+
+		nvgpu_gr_ctx_desc_free(g, gr->gr_ctx_desc);
+		gr->gr_ctx_desc = NULL;
+
+	#ifdef CONFIG_NVGPU_DEBUGGER
+		nvgpu_gr_hwpm_map_deinit(g, gr->hwpm_map);
+		gr->hwpm_map = NULL;
+	#endif
 
 		nvgpu_gr_obj_ctx_deinit(g, gr->golden_image);
+		gr->golden_image = NULL;
+	}
 
 	nvgpu_gr_free(g);
 }
@@ -522,6 +524,9 @@ static int gr_init_setup_sw(struct gk20a *g, struct nvgpu_gr *gr)
 		if (err != 0) {
 			goto clean_up;
 		}
+	} else {
+		gr->zbc = NULL;
+		gr->zcull = NULL;
 	}
 #endif /* CONFIG_NVGPU_GRAPHICS */
 
@@ -1026,11 +1031,21 @@ void nvgpu_gr_free(struct gk20a *g)
 	for (i = 0U; i < g->num_gr_instances; i++) {
 		gr = &g->gr[i];
 
+		nvgpu_gr_config_deinit(g, gr->config);
+		gr->config = NULL;
+
 		nvgpu_gr_falcon_remove_support(g, gr->falcon);
 		gr->falcon = NULL;
 
 		nvgpu_gr_intr_remove_support(g, gr->intr);
 		gr->intr = NULL;
+
+#ifdef CONFIG_NVGPU_GRAPHICS
+		nvgpu_gr_zbc_deinit(g, gr->zbc);
+		nvgpu_gr_zcull_deinit(g, gr->zcull);
+		gr->zbc = NULL;
+		gr->zcull = NULL;
+#endif /* CONFIG_NVGPU_GRAPHICS */
 	}
 
 	nvgpu_kfree(g, g->gr);

drivers/gpu/nvgpu/common/gr/gr_config.c

@@ -692,12 +692,15 @@ void nvgpu_gr_config_deinit(struct gk20a *g, struct nvgpu_gr_config *config)
 	nvgpu_kfree(g, config->map_tiles);
 #endif
 	nvgpu_kfree(g, config->sm_to_cluster);
+	config->sm_to_cluster = NULL;
 #ifdef CONFIG_NVGPU_SM_DIVERSITY
 	if (config->sm_to_cluster_redex_config != NULL) {
 		nvgpu_kfree(g, config->sm_to_cluster_redex_config);
 		config->sm_to_cluster_redex_config = NULL;
 	}
 #endif
+
+	nvgpu_kfree(g, config);
 }
 
 u32 nvgpu_gr_config_get_max_gpc_count(struct nvgpu_gr_config *config)

drivers/gpu/nvgpu/common/init/nvgpu_init.c

@@ -1030,6 +1030,10 @@ int nvgpu_init_gpu_characteristics(struct gk20a *g)
 		nvgpu_set_enabled(g, NVGPU_SUPPORT_I2M, true);
 		nvgpu_set_enabled(g, NVGPU_SUPPORT_ZBC, true);
 	} else {
+		nvgpu_set_enabled(g, NVGPU_SUPPORT_2D, false);
+		nvgpu_set_enabled(g, NVGPU_SUPPORT_3D, false);
+		nvgpu_set_enabled(g, NVGPU_SUPPORT_I2M, false);
+		nvgpu_set_enabled(g, NVGPU_SUPPORT_ZBC, false);
 		nvgpu_set_enabled(g, NVGPU_SUPPORT_ZBC_STENCIL, false);
 		nvgpu_set_enabled(g, NVGPU_SUPPORT_PREEMPTION_GFXP, false);
 	}

drivers/gpu/nvgpu/common/vgpu/gr/gr_vgpu.c

@@ -597,22 +597,6 @@ static void vgpu_remove_gr_support(struct gk20a *g)
 
 	nvgpu_log_fn(gr->g, " ");
 
-	nvgpu_kfree(gr->g, gr->config->sm_to_cluster);
-	gr->config->sm_to_cluster = NULL;
-
-#ifdef CONFIG_NVGPU_SM_DIVERSITY
-	if (gr->config->sm_to_cluster_redex_config != NULL) {
-		nvgpu_kfree(g, gr->config->sm_to_cluster_redex_config);
-		gr->config->sm_to_cluster_redex_config = NULL;
-	}
-#endif
-
-	nvgpu_gr_config_deinit(gr->g, gr->config);
-
-#ifdef CONFIG_NVGPU_GRAPHICS
-	nvgpu_gr_zcull_deinit(gr->g, gr->zcull);
-#endif
-
 	nvgpu_gr_free(g);
 }
 

drivers/gpu/nvgpu/include/nvgpu/gk20a.h

@@ -477,7 +477,7 @@ struct gk20a {
 	/** @cond DOXYGEN_SHOULD_SKIP_THIS */
 	struct nvgpu_nvlink_dev nvlink;
 	/** @endcond */
-	/** Pointer to struct maintaining GR unit's software state. */
+	/** Pointer to struct maintaining multiple GR instance's software state. */
 	struct nvgpu_gr *gr;
 	u32 num_gr_instances;
 	/** Pointer to struct maintaining fbp unit's software state. */

drivers/gpu/nvgpu/os/linux/ioctl_ctrl.c

@@ -1951,12 +1951,18 @@ long gk20a_ctrl_dev_ioctl(struct file *filp, unsigned int cmd, unsigned long arg
 	switch (cmd) {
 #ifdef CONFIG_NVGPU_GRAPHICS
 	case NVGPU_GPU_IOCTL_ZCULL_GET_CTX_SIZE:
+		if (gr_zcull == NULL)
+			return -ENODEV;
+
 		get_ctx_size_args = (struct nvgpu_gpu_zcull_get_ctx_size_args *)buf;
 
 		get_ctx_size_args->size = nvgpu_gr_get_ctxsw_zcull_size(g, gr_zcull);
 
 		break;
 	case NVGPU_GPU_IOCTL_ZCULL_GET_INFO:
+		if (gr_zcull == NULL)
+			return -ENODEV;
+
 		get_info_args = (struct nvgpu_gpu_zcull_get_info_args *)buf;
 
 		(void) memset(get_info_args, 0,
@@ -1987,6 +1993,9 @@ long gk20a_ctrl_dev_ioctl(struct file *filp, unsigned int cmd, unsigned long arg
 		nvgpu_kfree(g, zcull_info);
 		break;
 	case NVGPU_GPU_IOCTL_ZBC_SET_TABLE:
+		if (!nvgpu_is_enabled(g, NVGPU_SUPPORT_ZBC))
+			return -ENODEV;
+
 		set_table_args = (struct nvgpu_gpu_zbc_set_table_args *)buf;
 
 		zbc_val = nvgpu_gr_zbc_entry_alloc(g);
@@ -2027,6 +2036,9 @@ long gk20a_ctrl_dev_ioctl(struct file *filp, unsigned int cmd, unsigned long arg
 			nvgpu_gr_zbc_entry_free(g, zbc_val);
 		break;
 	case NVGPU_GPU_IOCTL_ZBC_QUERY_TABLE:
+		if (!nvgpu_is_enabled(g, NVGPU_SUPPORT_ZBC))
+			return -ENODEV;
+
 		query_table_args = (struct nvgpu_gpu_zbc_query_table_args *)buf;
 
 		zbc_tbl = nvgpu_kzalloc(g, sizeof(struct nvgpu_gr_zbc_query_params));