Merge tag 'amd-drm-fixes-6.5-2023-07-12' of https://gitlab.freedesktop.org/agd5f/linux into drm-fixes

amd-drm-fixes-6.5-2023-07-12:

amdgpu:
- SMU i2c locking fix
- Fix a possible deadlock in process restoration for ROCm apps
- Disable PCIe lane/speed switching on Intel platforms (the platforms don't support it)

Signed-off-by: Dave Airlie <airlied@redhat.com>

From: Alex Deucher <alexander.deucher@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230712184009.7740-1-alexander.deucher@amd.com

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

@@ -1296,6 +1296,7 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev,
 void amdgpu_device_pci_config_reset(struct amdgpu_device *adev);
 int amdgpu_device_pci_reset(struct amdgpu_device *adev);
 bool amdgpu_device_need_post(struct amdgpu_device *adev);
+bool amdgpu_device_pcie_dynamic_switching_supported(void);
 bool amdgpu_device_should_use_aspm(struct amdgpu_device *adev);
 bool amdgpu_device_aspm_support_quirk(void);
 

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

@@ -2881,6 +2881,9 @@ int amdgpu_amdkfd_gpuvm_restore_process_bos(void *info, struct dma_fence **ef)
 			if (!attachment->is_mapped)
 				continue;
 
+			if (attachment->bo_va->base.bo->tbo.pin_count)
+				continue;
+
 			kfd_mem_dmaunmap_attachment(mem, attachment);
 			ret = update_gpuvm_pte(mem, attachment, &sync_obj);
 			if (ret) {

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

@@ -1458,6 +1458,25 @@ bool amdgpu_device_need_post(struct amdgpu_device *adev)
 	return true;
 }
 
+/*
+ * Intel hosts such as Raptor Lake and Sapphire Rapids don't support dynamic
+ * speed switching. Until we have confirmation from Intel that a specific host
+ * supports it, it's safer that we keep it disabled for all.
+ *
+ * https://edc.intel.com/content/www/us/en/design/products/platforms/details/raptor-lake-s/13th-generation-core-processors-datasheet-volume-1-of-2/005/pci-express-support/
+ * https://gitlab.freedesktop.org/drm/amd/-/issues/2663
+ */
+bool amdgpu_device_pcie_dynamic_switching_supported(void)
+{
+#if IS_ENABLED(CONFIG_X86)
+	struct cpuinfo_x86 *c = &cpu_data(0);
+
+	if (c->x86_vendor == X86_VENDOR_INTEL)
+		return false;
+#endif
+	return true;
+}
+
 /**
  * amdgpu_device_should_use_aspm - check if the device should program ASPM
  *

drivers/gpu/drm/amd/pm/swsmu/inc/smu_v13_0.h

@@ -295,5 +295,9 @@ int smu_v13_0_get_pptable_from_firmware(struct smu_context *smu,
 					uint32_t *size,
 					uint32_t pptable_id);
 
+int smu_v13_0_update_pcie_parameters(struct smu_context *smu,
+				     uint32_t pcie_gen_cap,
+				     uint32_t pcie_width_cap);
+
 #endif
 #endif

drivers/gpu/drm/amd/pm/swsmu/smu11/arcturus_ppt.c

@@ -2113,7 +2113,6 @@ static int arcturus_i2c_xfer(struct i2c_adapter *i2c_adap,
 	}
 	mutex_lock(&adev->pm.mutex);
 	r = smu_cmn_update_table(smu, SMU_TABLE_I2C_COMMANDS, 0, req, true);
-	mutex_unlock(&adev->pm.mutex);
 	if (r)
 		goto fail;
 
@@ -2130,6 +2129,7 @@ static int arcturus_i2c_xfer(struct i2c_adapter *i2c_adap,
 	}
 	r = num_msgs;
 fail:
+	mutex_unlock(&adev->pm.mutex);
 	kfree(req);
 	return r;
 }

drivers/gpu/drm/amd/pm/swsmu/smu11/navi10_ppt.c

@@ -3021,7 +3021,6 @@ static int navi10_i2c_xfer(struct i2c_adapter *i2c_adap,
 	}
 	mutex_lock(&adev->pm.mutex);
 	r = smu_cmn_update_table(smu, SMU_TABLE_I2C_COMMANDS, 0, req, true);
-	mutex_unlock(&adev->pm.mutex);
 	if (r)
 		goto fail;
 
@@ -3038,6 +3037,7 @@ static int navi10_i2c_xfer(struct i2c_adapter *i2c_adap,
 	}
 	r = num_msgs;
 fail:
+	mutex_unlock(&adev->pm.mutex);
 	kfree(req);
 	return r;
 }

drivers/gpu/drm/amd/pm/swsmu/smu11/sienna_cichlid_ppt.c

@@ -2077,89 +2077,36 @@ static int sienna_cichlid_display_disable_memory_clock_switch(struct smu_context
 	return ret;
 }
 
-static void sienna_cichlid_get_override_pcie_settings(struct smu_context *smu,
-						      uint32_t *gen_speed_override,
-						      uint32_t *lane_width_override)
-{
-	struct amdgpu_device *adev = smu->adev;
-
-	*gen_speed_override = 0xff;
-	*lane_width_override = 0xff;
-
-	switch (adev->pdev->device) {
-	case 0x73A0:
-	case 0x73A1:
-	case 0x73A2:
-	case 0x73A3:
-	case 0x73AB:
-	case 0x73AE:
-		/* Bit 7:0: PCIE lane width, 1 to 7 corresponds is x1 to x32 */
-		*lane_width_override = 6;
-		break;
-	case 0x73E0:
-	case 0x73E1:
-	case 0x73E3:
-		*lane_width_override = 4;
-		break;
-	case 0x7420:
-	case 0x7421:
-	case 0x7422:
-	case 0x7423:
-	case 0x7424:
-		*lane_width_override = 3;
-		break;
-	default:
-		break;
-	}
-}
-
-#define MAX(a, b)	((a) > (b) ? (a) : (b))
-
 static int sienna_cichlid_update_pcie_parameters(struct smu_context *smu,
 					 uint32_t pcie_gen_cap,
 					 uint32_t pcie_width_cap)
 {
 	struct smu_11_0_dpm_context *dpm_context = smu->smu_dpm.dpm_context;
 	struct smu_11_0_pcie_table *pcie_table = &dpm_context->dpm_tables.pcie_table;
-	uint32_t gen_speed_override, lane_width_override;
+	u32 smu_pcie_arg;
-	uint8_t *table_member1, *table_member2;
-	uint32_t min_gen_speed, max_gen_speed;
-	uint32_t min_lane_width, max_lane_width;
-	uint32_t smu_pcie_arg;
 	int ret, i;
 
-	GET_PPTABLE_MEMBER(PcieGenSpeed, &table_member1);
+	/* PCIE gen speed and lane width override */
-	GET_PPTABLE_MEMBER(PcieLaneCount, &table_member2);
+	if (!amdgpu_device_pcie_dynamic_switching_supported()) {
+		if (pcie_table->pcie_gen[NUM_LINK_LEVELS - 1] < pcie_gen_cap)
+			pcie_gen_cap = pcie_table->pcie_gen[NUM_LINK_LEVELS - 1];
 
-	sienna_cichlid_get_override_pcie_settings(smu,
+		if (pcie_table->pcie_lane[NUM_LINK_LEVELS - 1] < pcie_width_cap)
-						  &gen_speed_override,
+			pcie_width_cap = pcie_table->pcie_lane[NUM_LINK_LEVELS - 1];
-						  &lane_width_override);
 
-	/* PCIE gen speed override */
+		/* Force all levels to use the same settings */
-	if (gen_speed_override != 0xff) {
+		for (i = 0; i < NUM_LINK_LEVELS; i++) {
-		min_gen_speed = MIN(pcie_gen_cap, gen_speed_override);
+			pcie_table->pcie_gen[i] = pcie_gen_cap;
-		max_gen_speed = MIN(pcie_gen_cap, gen_speed_override);
+			pcie_table->pcie_lane[i] = pcie_width_cap;
+		}
 	} else {
-		min_gen_speed = MAX(0, table_member1[0]);
+		for (i = 0; i < NUM_LINK_LEVELS; i++) {
-		max_gen_speed = MIN(pcie_gen_cap, table_member1[1]);
+			if (pcie_table->pcie_gen[i] > pcie_gen_cap)
-		min_gen_speed = min_gen_speed > max_gen_speed ?
+				pcie_table->pcie_gen[i] = pcie_gen_cap;
-				max_gen_speed : min_gen_speed;
+			if (pcie_table->pcie_lane[i] > pcie_width_cap)
+				pcie_table->pcie_lane[i] = pcie_width_cap;
+		}
 	}
-	pcie_table->pcie_gen[0] = min_gen_speed;
-	pcie_table->pcie_gen[1] = max_gen_speed;
-
-	/* PCIE lane width override */
-	if (lane_width_override != 0xff) {
-		min_lane_width = MIN(pcie_width_cap, lane_width_override);
-		max_lane_width = MIN(pcie_width_cap, lane_width_override);
-	} else {
-		min_lane_width = MAX(1, table_member2[0]);
-		max_lane_width = MIN(pcie_width_cap, table_member2[1]);
-		min_lane_width = min_lane_width > max_lane_width ?
-				 max_lane_width : min_lane_width;
-	}
-	pcie_table->pcie_lane[0] = min_lane_width;
-	pcie_table->pcie_lane[1] = max_lane_width;
 
 	for (i = 0; i < NUM_LINK_LEVELS; i++) {
 		smu_pcie_arg = (i << 16 |
@@ -3842,7 +3789,6 @@ static int sienna_cichlid_i2c_xfer(struct i2c_adapter *i2c_adap,
 	}
 	mutex_lock(&adev->pm.mutex);
 	r = smu_cmn_update_table(smu, SMU_TABLE_I2C_COMMANDS, 0, req, true);
-	mutex_unlock(&adev->pm.mutex);
 	if (r)
 		goto fail;
 
@@ -3859,6 +3805,7 @@ static int sienna_cichlid_i2c_xfer(struct i2c_adapter *i2c_adap,
 	}
 	r = num_msgs;
 fail:
+	mutex_unlock(&adev->pm.mutex);
 	kfree(req);
 	return r;
 }

drivers/gpu/drm/amd/pm/swsmu/smu13/aldebaran_ppt.c

@@ -1525,7 +1525,6 @@ static int aldebaran_i2c_xfer(struct i2c_adapter *i2c_adap,
 	}
 	mutex_lock(&adev->pm.mutex);
 	r = smu_cmn_update_table(smu, SMU_TABLE_I2C_COMMANDS, 0, req, true);
-	mutex_unlock(&adev->pm.mutex);
 	if (r)
 		goto fail;
 
@@ -1542,6 +1541,7 @@ static int aldebaran_i2c_xfer(struct i2c_adapter *i2c_adap,
 	}
 	r = num_msgs;
 fail:
+	mutex_unlock(&adev->pm.mutex);
 	kfree(req);
 	return r;
 }

drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0.c

@@ -2424,3 +2424,51 @@ int smu_v13_0_mode1_reset(struct smu_context *smu)
 
 	return ret;
 }
+
+int smu_v13_0_update_pcie_parameters(struct smu_context *smu,
+				     uint32_t pcie_gen_cap,
+				     uint32_t pcie_width_cap)
+{
+	struct smu_13_0_dpm_context *dpm_context = smu->smu_dpm.dpm_context;
+	struct smu_13_0_pcie_table *pcie_table =
+				&dpm_context->dpm_tables.pcie_table;
+	int num_of_levels = pcie_table->num_of_link_levels;
+	uint32_t smu_pcie_arg;
+	int ret, i;
+
+	if (!amdgpu_device_pcie_dynamic_switching_supported()) {
+		if (pcie_table->pcie_gen[num_of_levels - 1] < pcie_gen_cap)
+			pcie_gen_cap = pcie_table->pcie_gen[num_of_levels - 1];
+
+		if (pcie_table->pcie_lane[num_of_levels - 1] < pcie_width_cap)
+			pcie_width_cap = pcie_table->pcie_lane[num_of_levels - 1];
+
+		/* Force all levels to use the same settings */
+		for (i = 0; i < num_of_levels; i++) {
+			pcie_table->pcie_gen[i] = pcie_gen_cap;
+			pcie_table->pcie_lane[i] = pcie_width_cap;
+		}
+	} else {
+		for (i = 0; i < num_of_levels; i++) {
+			if (pcie_table->pcie_gen[i] > pcie_gen_cap)
+				pcie_table->pcie_gen[i] = pcie_gen_cap;
+			if (pcie_table->pcie_lane[i] > pcie_width_cap)
+				pcie_table->pcie_lane[i] = pcie_width_cap;
+		}
+	}
+
+	for (i = 0; i < num_of_levels; i++) {
+		smu_pcie_arg = i << 16;
+		smu_pcie_arg |= pcie_table->pcie_gen[i] << 8;
+		smu_pcie_arg |= pcie_table->pcie_lane[i];
+
+		ret = smu_cmn_send_smc_msg_with_param(smu,
+						      SMU_MSG_OverridePcieParameters,
+						      smu_pcie_arg,
+						      NULL);
+		if (ret)
+			return ret;
+	}
+
+	return 0;
+}

drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_0_ppt.c

@@ -1645,37 +1645,6 @@ static int smu_v13_0_0_force_clk_levels(struct smu_context *smu,
 	return ret;
 }
 
-static int smu_v13_0_0_update_pcie_parameters(struct smu_context *smu,
-					      uint32_t pcie_gen_cap,
-					      uint32_t pcie_width_cap)
-{
-	struct smu_13_0_dpm_context *dpm_context = smu->smu_dpm.dpm_context;
-	struct smu_13_0_pcie_table *pcie_table =
-				&dpm_context->dpm_tables.pcie_table;
-	uint32_t smu_pcie_arg;
-	int ret, i;
-
-	for (i = 0; i < pcie_table->num_of_link_levels; i++) {
-		if (pcie_table->pcie_gen[i] > pcie_gen_cap)
-			pcie_table->pcie_gen[i] = pcie_gen_cap;
-		if (pcie_table->pcie_lane[i] > pcie_width_cap)
-			pcie_table->pcie_lane[i] = pcie_width_cap;
-
-		smu_pcie_arg = i << 16;
-		smu_pcie_arg |= pcie_table->pcie_gen[i] << 8;
-		smu_pcie_arg |= pcie_table->pcie_lane[i];
-
-		ret = smu_cmn_send_smc_msg_with_param(smu,
-						      SMU_MSG_OverridePcieParameters,
-						      smu_pcie_arg,
-						      NULL);
-		if (ret)
-			return ret;
-	}
-
-	return 0;
-}
-
 static const struct smu_temperature_range smu13_thermal_policy[] = {
 	{-273150,  99000, 99000, -273150, 99000, 99000, -273150, 99000, 99000},
 	{ 120000, 120000, 120000, 120000, 120000, 120000, 120000, 120000, 120000},
@@ -2320,7 +2289,6 @@ static int smu_v13_0_0_i2c_xfer(struct i2c_adapter *i2c_adap,
 	}
 	mutex_lock(&adev->pm.mutex);
 	r = smu_cmn_update_table(smu, SMU_TABLE_I2C_COMMANDS, 0, req, true);
-	mutex_unlock(&adev->pm.mutex);
 	if (r)
 		goto fail;
 
@@ -2337,6 +2305,7 @@ static int smu_v13_0_0_i2c_xfer(struct i2c_adapter *i2c_adap,
 	}
 	r = num_msgs;
 fail:
+	mutex_unlock(&adev->pm.mutex);
 	kfree(req);
 	return r;
 }
@@ -2654,7 +2623,7 @@ static const struct pptable_funcs smu_v13_0_0_ppt_funcs = {
 	.feature_is_enabled = smu_cmn_feature_is_enabled,
 	.print_clk_levels = smu_v13_0_0_print_clk_levels,
 	.force_clk_levels = smu_v13_0_0_force_clk_levels,
-	.update_pcie_parameters = smu_v13_0_0_update_pcie_parameters,
+	.update_pcie_parameters = smu_v13_0_update_pcie_parameters,
 	.get_thermal_temperature_range = smu_v13_0_0_get_thermal_temperature_range,
 	.register_irq_handler = smu_v13_0_register_irq_handler,
 	.enable_thermal_alert = smu_v13_0_enable_thermal_alert,

drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.c

@@ -1763,7 +1763,6 @@ static int smu_v13_0_6_i2c_xfer(struct i2c_adapter *i2c_adap,
 	}
 	mutex_lock(&adev->pm.mutex);
 	r = smu_v13_0_6_request_i2c_xfer(smu, req);
-	mutex_unlock(&adev->pm.mutex);
 	if (r)
 		goto fail;
 
@@ -1780,6 +1779,7 @@ static int smu_v13_0_6_i2c_xfer(struct i2c_adapter *i2c_adap,
 	}
 	r = num_msgs;
 fail:
+	mutex_unlock(&adev->pm.mutex);
 	kfree(req);
 	return r;
 }

drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_7_ppt.c

@@ -1635,37 +1635,6 @@ static int smu_v13_0_7_force_clk_levels(struct smu_context *smu,
 	return ret;
 }
 
-static int smu_v13_0_7_update_pcie_parameters(struct smu_context *smu,
-					      uint32_t pcie_gen_cap,
-					      uint32_t pcie_width_cap)
-{
-	struct smu_13_0_dpm_context *dpm_context = smu->smu_dpm.dpm_context;
-	struct smu_13_0_pcie_table *pcie_table =
-				&dpm_context->dpm_tables.pcie_table;
-	uint32_t smu_pcie_arg;
-	int ret, i;
-
-	for (i = 0; i < pcie_table->num_of_link_levels; i++) {
-		if (pcie_table->pcie_gen[i] > pcie_gen_cap)
-			pcie_table->pcie_gen[i] = pcie_gen_cap;
-		if (pcie_table->pcie_lane[i] > pcie_width_cap)
-			pcie_table->pcie_lane[i] = pcie_width_cap;
-
-		smu_pcie_arg = i << 16;
-		smu_pcie_arg |= pcie_table->pcie_gen[i] << 8;
-		smu_pcie_arg |= pcie_table->pcie_lane[i];
-
-		ret = smu_cmn_send_smc_msg_with_param(smu,
-						      SMU_MSG_OverridePcieParameters,
-						      smu_pcie_arg,
-						      NULL);
-		if (ret)
-			return ret;
-	}
-
-	return 0;
-}
-
 static const struct smu_temperature_range smu13_thermal_policy[] =
 {
 	{-273150,  99000, 99000, -273150, 99000, 99000, -273150, 99000, 99000},
@@ -2234,7 +2203,7 @@ static const struct pptable_funcs smu_v13_0_7_ppt_funcs = {
 	.feature_is_enabled = smu_cmn_feature_is_enabled,
 	.print_clk_levels = smu_v13_0_7_print_clk_levels,
 	.force_clk_levels = smu_v13_0_7_force_clk_levels,
-	.update_pcie_parameters = smu_v13_0_7_update_pcie_parameters,
+	.update_pcie_parameters = smu_v13_0_update_pcie_parameters,
 	.get_thermal_temperature_range = smu_v13_0_7_get_thermal_temperature_range,
 	.register_irq_handler = smu_v13_0_register_irq_handler,
 	.enable_thermal_alert = smu_v13_0_enable_thermal_alert,