drm/amd/pm: implement ras_smu_drv interface for smu v13.0.12

implement ras_smu_drv interface for smu v13.0.12

Signed-off-by: Gangliang Xie <ganglxie@amd.com>
Signed-off-by: Yang Wang <kevinyang.wang@amd.com>
Reviewed-by: Tao Zhou <tao.zhou1@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Author: Gangliang Xie

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

@@ -503,6 +503,32 @@ struct ras_critical_region {
 	uint64_t size;
 };
 
+struct ras_eeprom_table_version {
+	uint32_t minor    : 16;
+	uint32_t major    : 16;
+};
+
+struct ras_eeprom_smu_funcs {
+	int (*get_ras_table_version)(struct amdgpu_device *adev,
+							uint32_t *table_version);
+	int (*get_badpage_count)(struct amdgpu_device *adev, uint32_t *count, uint32_t timeout);
+	int (*get_badpage_mca_addr)(struct amdgpu_device *adev, uint16_t index, uint64_t *mca_addr);
+	int (*set_timestamp)(struct amdgpu_device *adev, uint64_t timestamp);
+	int (*get_timestamp)(struct amdgpu_device *adev,
+							uint16_t index, uint64_t *timestamp);
+	int (*get_badpage_ipid)(struct amdgpu_device *adev, uint16_t index, uint64_t *ipid);
+	int (*erase_ras_table)(struct amdgpu_device *adev, uint32_t *result);
+};
+
+enum ras_smu_feature_flags {
+	RAS_SMU_FEATURE_BIT__RAS_EEPROM = BIT_ULL(0),
+};
+
+struct ras_smu_drv {
+	const struct ras_eeprom_smu_funcs *smu_eeprom_funcs;
+	void (*ras_smu_feature_flags)(struct amdgpu_device *adev, uint64_t *flags);
+};
+
 struct amdgpu_ras {
 	void *ras_mgr;
 	/* ras infrastructure */

drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_12_ppt.c

@@ -34,6 +34,7 @@
 #include "amdgpu_fru_eeprom.h"
 #include <linux/pci.h>
 #include "smu_cmn.h"
+#include "amdgpu_ras.h"
 
 #undef MP1_Public
 #undef smnMP1_FIRMWARE_FLAGS
@@ -925,3 +926,131 @@ const struct smu_temp_funcs smu_v13_0_12_temp_funcs = {
 	.temp_metrics_is_supported = smu_v13_0_12_is_temp_metrics_supported,
 	.get_temp_metrics = smu_v13_0_12_get_temp_metrics,
 };
+
+static int smu_v13_0_12_get_ras_table_version(struct amdgpu_device *adev,
+					      uint32_t *table_version)
+{
+	struct smu_context *smu = adev->powerplay.pp_handle;
+
+	return smu_cmn_send_smc_msg_with_param(smu,
+			SMU_MSG_GetRASTableVersion, 0, table_version);
+}
+
+static int smu_v13_0_12_get_badpage_count(struct amdgpu_device *adev, uint32_t *count,
+					  uint32_t timeout)
+{
+	struct smu_context *smu = adev->powerplay.pp_handle;
+	uint64_t end, now;
+	int ret = 0;
+
+	now = (uint64_t)ktime_to_ms(ktime_get());
+	end = now + timeout;
+	do {
+		ret = smu_cmn_send_smc_msg_with_param(smu,
+			SMU_MSG_GetBadPageCount, 0, count);
+		/* eeprom is not ready */
+		if (ret != -EBUSY)
+			return ret;
+		mdelay(10);
+		now = (uint64_t)ktime_to_ms(ktime_get());
+	} while (now < end);
+
+	return ret;
+}
+
+static int smu_v13_0_12_set_timestamp(struct amdgpu_device *adev, uint64_t timestamp)
+{
+	struct smu_context *smu = adev->powerplay.pp_handle;
+
+	return smu_cmn_send_smc_msg_with_param(smu,
+			SMU_MSG_SetTimestamp, (uint32_t)timestamp, 0);
+}
+
+static int smu_v13_0_12_get_timestamp(struct amdgpu_device *adev,
+				      uint16_t index, uint64_t *timestamp)
+{
+	struct smu_context *smu = adev->powerplay.pp_handle;
+	uint32_t temp;
+	int ret;
+
+	ret = smu_cmn_send_smc_msg_with_param(smu,
+			SMU_MSG_GetTimestamp, index, &temp);
+	if (!ret)
+		*timestamp = temp;
+
+	return ret;
+}
+
+static int smu_v13_0_12_get_badpage_ipid(struct amdgpu_device *adev,
+					 uint16_t index, uint64_t *ipid)
+{
+	struct smu_context *smu = adev->powerplay.pp_handle;
+	uint32_t temp_arg, temp_ipid_lo, temp_ipid_high;
+	int ret;
+
+	temp_arg = index | (1 << 16);
+	ret = smu_cmn_send_smc_msg_with_param(smu,
+			SMU_MSG_GetBadPageIpid, temp_arg, &temp_ipid_lo);
+	if (ret)
+		return ret;
+
+	temp_arg = index | (2 << 16);
+	ret = smu_cmn_send_smc_msg_with_param(smu,
+			SMU_MSG_GetBadPageIpid, temp_arg, &temp_ipid_high);
+	if (!ret)
+		*ipid = (uint64_t)temp_ipid_high << 32 | temp_ipid_lo;
+	return ret;
+}
+
+static int smu_v13_0_12_erase_ras_table(struct amdgpu_device *adev,
+					uint32_t *result)
+{
+	struct smu_context *smu = adev->powerplay.pp_handle;
+
+	return smu_cmn_send_smc_msg_with_param(smu,
+			SMU_MSG_EraseRasTable, 0, result);
+}
+
+static int smu_v13_0_12_get_badpage_mca_addr(struct amdgpu_device *adev,
+					     uint16_t index, uint64_t *mca_addr)
+{
+	struct smu_context *smu = adev->powerplay.pp_handle;
+	uint32_t temp_arg, temp_addr_lo, temp_addr_high;
+	int ret;
+
+	temp_arg = index | (1 << 16);
+	ret = smu_cmn_send_smc_msg_with_param(smu,
+			SMU_MSG_GetBadPageMcaAddr, temp_arg, &temp_addr_lo);
+	if (ret)
+		return ret;
+
+	temp_arg = index | (2 << 16);
+	ret = smu_cmn_send_smc_msg_with_param(smu,
+			SMU_MSG_GetBadPageMcaAddr, temp_arg, &temp_addr_high);
+	if (!ret)
+		*mca_addr = (uint64_t)temp_addr_high << 32 | temp_addr_lo;
+	return ret;
+}
+
+static const struct ras_eeprom_smu_funcs smu_v13_0_12_eeprom_smu_funcs = {
+	.get_ras_table_version = smu_v13_0_12_get_ras_table_version,
+	.get_badpage_count = smu_v13_0_12_get_badpage_count,
+	.get_badpage_mca_addr = smu_v13_0_12_get_badpage_mca_addr,
+	.set_timestamp = smu_v13_0_12_set_timestamp,
+	.get_timestamp = smu_v13_0_12_get_timestamp,
+	.get_badpage_ipid = smu_v13_0_12_get_badpage_ipid,
+	.erase_ras_table = smu_v13_0_12_erase_ras_table,
+};
+
+static void smu_v13_0_12_ras_smu_feature_flags(struct amdgpu_device *adev, uint64_t *flags)
+{
+	if (!flags)
+		return;
+
+	*flags = 0ULL;
+}
+
+const struct ras_smu_drv smu_v13_0_12_ras_smu_drv = {
+	.smu_eeprom_funcs = &smu_v13_0_12_eeprom_smu_funcs,
+	.ras_smu_feature_flags = smu_v13_0_12_ras_smu_feature_flags,
+};

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

@@ -105,6 +105,7 @@ int smu_v13_0_12_get_npm_data(struct smu_context *smu,
 extern const struct cmn2asic_mapping smu_v13_0_12_feature_mask_map[];
 extern const struct cmn2asic_msg_mapping smu_v13_0_12_message_map[];
 extern const struct smu_temp_funcs smu_v13_0_12_temp_funcs;
+extern const struct ras_smu_drv smu_v13_0_12_ras_smu_drv;
 
 #if defined(SWSMU_CODE_LAYER_L2)
 #include "smu_cmn.h"