Skip to contentSkip to conversation

drm/amd/pm: add send bad channel info function

PATCH In review Stanley.Yang 2 participants 3 messages Mar 4, 2022

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH 1/2 drm/amd/pm: add send bad channel info function

support message SMU update bad channel info to update HBM bad channel info in OOB table

Change-Id I1e50ed8118f4c1aaefb04c040e59ae4918cdc295
---
 drivers/gpu/drm/amd/pm/amdgpu_dpm.c           | 12 ++++++
 drivers/gpu/drm/amd/pm/inc/amdgpu_dpm.h       |  1 +
 drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c     | 10 +++++
 drivers/gpu/drm/amd/pm/swsmu/inc/amdgpu_smu.h |  7 +++
 .../pm/swsmu/inc/pmfw_if/aldebaran_ppsmc.h    |  3 +-
 drivers/gpu/drm/amd/pm/swsmu/inc/smu_types.h  |  3 +-
 .../drm/amd/pm/swsmu/smu13/aldebaran_ppt.c    | 43 +++++++++++++++++++
 7 files changed, 77 insertions(+), 2 deletions(-)
drivers/gpu/drm/amd/pm/amdgpu_dpm.c +12 -0
index 1d63f1e8884c..9a892d6d1d7a 100644
@@ -507,6 +507,18 @@ int amdgpu_dpm_send_hbm_bad_pages_num(struct amdgpu_device *adev, uint32_t size)
 	return ret;
 }
 
+int amdgpu_dpm_send_hbm_bad_channel_flag(struct amdgpu_device *adev, uint32_t size)
+{
+	struct smu_context *smu = adev->powerplay.pp_handle;
+	int ret = 0;
+
+	mutex_lock(&adev->pm.mutex);
+	ret = smu_send_hbm_bad_channel_flag(smu, size);
+	mutex_unlock(&adev->pm.mutex);
+
+	return ret;
+}
+
 int amdgpu_dpm_get_dpm_freq_range(struct amdgpu_device *adev,
 				  enum pp_clock_type type,
 				  uint32_t *min,
drivers/gpu/drm/amd/pm/inc/amdgpu_dpm.h +1 -0
index ddfa55b59d02..3e78b3057277 100644
@@ -412,6 +412,7 @@ void amdgpu_dpm_enable_jpeg(struct amdgpu_device *adev, bool enable);
 int amdgpu_pm_load_smu_firmware(struct amdgpu_device *adev, uint32_t *smu_version);
 int amdgpu_dpm_handle_passthrough_sbr(struct amdgpu_device *adev, bool enable);
 int amdgpu_dpm_send_hbm_bad_pages_num(struct amdgpu_device *adev, uint32_t size);
+int amdgpu_dpm_send_hbm_bad_channel_flag(struct amdgpu_device *adev, uint32_t size);
 int amdgpu_dpm_get_dpm_freq_range(struct amdgpu_device *adev,
 				       enum pp_clock_type type,
 				       uint32_t *min,
drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c +10 -0
index 7e79a67bb8ef..f1544755d8b4 100644
@@ -3052,3 +3052,13 @@ int smu_send_hbm_bad_pages_num(struct smu_context *smu, uint32_t size)
 
 	return ret;
 }
+
+int smu_send_hbm_bad_channel_flag(struct smu_context *smu, uint32_t size)
+{
+	int ret = 0;
+
+	if (smu->ppt_funcs && smu->ppt_funcs->send_hbm_bad_channel_flag)
+		ret = smu->ppt_funcs->send_hbm_bad_channel_flag(smu, size);
+
+	return ret;
+}
drivers/gpu/drm/amd/pm/swsmu/inc/amdgpu_smu.h +7 -0
index fbef3ab8d487..ef57b6089c69 100644
@@ -1292,6 +1292,12 @@ struct pptable_funcs {
 	 * @set_config_table: Apply the input DriverSmuConfig table settings.
 	 */
 	int (*set_config_table)(struct smu_context *smu, struct config_table_setting *table);
+
+	/**
+	 * @sned_hbm_bad_channel_flag:  message SMU to update bad channel info
+	 *										of SMUBUS table.
+	 */
+	int (*send_hbm_bad_channel_flag)(struct smu_context *smu, uint32_t size);
 };
 
 typedef enum {
@@ -1428,5 +1434,6 @@ int smu_get_ecc_info(struct smu_context *smu, void *umc_ecc);
 int smu_stb_collect_info(struct smu_context *smu, void *buff, uint32_t size);
 void amdgpu_smu_stb_debug_fs_init(struct amdgpu_device *adev);
 int smu_send_hbm_bad_pages_num(struct smu_context *smu, uint32_t size);
+int smu_send_hbm_bad_channel_flag(struct smu_context *smu, uint32_t size);
 #endif
 #endif
drivers/gpu/drm/amd/pm/swsmu/inc/pmfw_if/aldebaran_ppsmc.h +2 -1
index ab66a4b9e438..0f498baf6838 100644
@@ -103,7 +103,8 @@
 #define PPSMC_MSG_GfxDriverResetRecovery	0x42
 #define PPSMC_MSG_BoardPowerCalibration 	0x43
 #define PPSMC_MSG_HeavySBR                      0x45
-#define PPSMC_Message_Count			0x46
+#define PPSMC_MSG_SetBadHBMPagesRetiredFlagsPerChannel	0x46
+#define PPSMC_Message_Count			0x47
 
 
 //PPSMC Reset Types
drivers/gpu/drm/amd/pm/swsmu/inc/smu_types.h +2 -1
index d787c3b9fc52..9f6f306eeca0 100644
@@ -232,7 +232,8 @@
 	__SMU_DUMMY_MAP(ForceGfxVid),             \
 	__SMU_DUMMY_MAP(Spare0),                  \
 	__SMU_DUMMY_MAP(UnforceGfxVid),           \
-	__SMU_DUMMY_MAP(HeavySBR),
+	__SMU_DUMMY_MAP(HeavySBR),			\
+	__SMU_DUMMY_MAP(SetBadHBMPagesRetiredFlagsPerChannel),
 
 #undef __SMU_DUMMY_MAP
 #define __SMU_DUMMY_MAP(type)	SMU_MSG_##type
drivers/gpu/drm/amd/pm/swsmu/smu13/aldebaran_ppt.c +43 -0

67 lines not on this page · raw message

Signature · 1 line
2.17.1
#1 lore raw

PATCH 2/2 drm/amdgpu: message smu to update bad channel info

It should notice SMU to update bad channel info when detected uncorrectable error in UMC block

Change-Id I2dc8848affdb53e52891013953ae9383fff5f20f
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c       |  7 ++++++
 drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h       |  3 +++
 .../gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c    | 25 +++++++++++++++++--
 .../gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.h    |  4 +++
 drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c       |  5 ++++
 5 files changed, 42 insertions(+), 2 deletions(-)
drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +7 -0
index d3875618ebf5..f9104f99eb9c 100644
@@ -2068,6 +2068,7 @@ int amdgpu_ras_recovery_init(struct amdgpu_device *adev)
 	mutex_init(&con->recovery_lock);
 	INIT_WORK(&con->recovery_work, amdgpu_ras_do_recovery);
 	atomic_set(&con->in_recovery, 0);
+	con->eeprom_control.bad_channel_bitmap = 0;
 
 	max_eeprom_records_count = amdgpu_ras_eeprom_max_record_count();
 	amdgpu_ras_validate_threshold(adev, max_eeprom_records_count);
@@ -2092,6 +2093,11 @@ int amdgpu_ras_recovery_init(struct amdgpu_device *adev)
 			goto free;
 
 		amdgpu_dpm_send_hbm_bad_pages_num(adev, con->eeprom_control.ras_num_recs);
+
+		if (con->update_channel_flag == true) {
+			amdgpu_dpm_send_hbm_bad_channel_flag(adev, con->eeprom_control.bad_channel_bitmap);
+			con->update_channel_flag = false;
+		}
 	}
 
 #ifdef CONFIG_X86_MCE_AMD
@@ -2285,6 +2291,7 @@ int amdgpu_ras_init(struct amdgpu_device *adev)
 		goto release_con;
 	}
 
+	con->update_channel_flag = false;
 	con->features = 0;
 	INIT_LIST_HEAD(&con->head);
 	/* Might need get this flag from vbios. */
drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h +3 -0
index 7cddaad90d6d..9314fde81e68 100644
@@ -374,6 +374,9 @@ struct amdgpu_ras {
 
 	/* record umc error info queried from smu */
 	struct umc_ecc_info umc_ecc;
+
+	/* Indicates smu whether need update bad channel info */
+	bool update_channel_flag;
 };
 
 struct ras_fs_data {
drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c +23 -2
index 2b844a5aafdb..ad5d8667756d 100644
@@ -265,6 +265,7 @@ int amdgpu_ras_eeprom_reset_table(struct amdgpu_ras_eeprom_control *control)
 {
 	struct amdgpu_device *adev = to_amdgpu_device(control);
 	struct amdgpu_ras_eeprom_table_header *hdr = &control->tbl_hdr;
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
 	u8 csum;
 	int res;
 
@@ -285,6 +286,10 @@ int amdgpu_ras_eeprom_reset_table(struct amdgpu_ras_eeprom_control *control)
 
 	amdgpu_dpm_send_hbm_bad_pages_num(adev, control->ras_num_recs);
 
+	control->bad_channel_bitmap = 0;
+	amdgpu_dpm_send_hbm_bad_channel_flag(adev, control->bad_channel_bitmap);
+	con->update_channel_flag = false;
+
 	amdgpu_ras_debugfs_set_ret_size(control);
 
 	mutex_unlock(&control->ras_tbl_mutex);
@@ -418,6 +423,7 @@ amdgpu_ras_eeprom_append_table(struct amdgpu_ras_eeprom_control *control,
 			       struct eeprom_table_record *record,
 			       const u32 num)
 {
+	struct amdgpu_ras *con = amdgpu_ras_get_context(to_amdgpu_device(control));
 	u32 a, b, i;
 	u8 *buf, *pp;
 	int res;
@@ -429,9 +435,16 @@ amdgpu_ras_eeprom_append_table(struct amdgpu_ras_eeprom_control *control,
 	/* Encode all of them in one go.
 	 */
 	pp = buf;
-	for (i = 0; i < num; i++, pp += RAS_TABLE_RECORD_SIZE)
+	for (i = 0; i < num; i++, pp += RAS_TABLE_RECORD_SIZE) {
 		__encode_table_record_to_buf(control, &record[i], pp);
 
+		/* update bad channel bitmap */
+		if (!(control->bad_channel_bitmap & (1 << record[i].mem_channel))) {
+			control->bad_channel_bitmap |= 1 << record[i].mem_channel;
+			con->update_channel_flag = true;
+		}
+	}
+
 	/* a, first record index to write into.
 	 * b, last record index to write into.
 	 * a = first index to read (fri) + number of records in the table,
@@ -684,6 +697,7 @@ int amdgpu_ras_eeprom_read(struct amdgpu_ras_eeprom_control *control,
 			   const u32 num)
 {
 	struct amdgpu_device *adev = to_amdgpu_device(control);
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
 	int i, res;
 	u8 *buf, *pp;
 	u32 g0, g1;
@@ -751,8 +765,15 @@ int amdgpu_ras_eeprom_read(struct amdgpu_ras_eeprom_control *control,
 	/* Read up everything? Then transform.
 	 */
 	pp = buf;
-	for (i = 0; i < num; i++, pp += RAS_TABLE_RECORD_SIZE)
+	for (i = 0; i < num; i++, pp += RAS_TABLE_RECORD_SIZE) {
 		__decode_table_record_from_buf(control, &record[i], pp);
+
+		/* update bad channel bitmap */
+		if (!(control->bad_channel_bitmap & (1 << record[i].mem_channel))) {
+			control->bad_channel_bitmap |= 1 << record[i].mem_channel;
+			con->update_channel_flag = true;
+		}
+	}
 Out:
 	kfree(buf);
 	mutex_unlock(&control->ras_tbl_mutex);
drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.h +4 -0
index 6bb00578bfbb..54d9bfe0881d 100644
@@ -80,6 +80,10 @@ struct amdgpu_ras_eeprom_control {
 	/* Protect table access via this mutex.
 	 */
 	struct mutex ras_tbl_mutex;
+
+	/* Record channel info which occurred bad pages
+	 */
+	u32 bad_channel_bitmap;
 };
 
 /*
drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c +5 -0
index 85da6cbaf3b7..aad3c8b4c810 100644
@@ -97,6 +97,11 @@ static int amdgpu_umc_do_page_retirement(struct amdgpu_device *adev,
 			amdgpu_ras_save_bad_pages(adev);
 
 			amdgpu_dpm_send_hbm_bad_pages_num(adev, con->eeprom_control.ras_num_recs);
+
+			if (con->update_channel_flag == true) {
+				amdgpu_dpm_send_hbm_bad_channel_flag(adev, con->eeprom_control.bad_channel_bitmap);
+				con->update_channel_flag = false;
+			}
 		}
 
 		if (reset)
Signature · 1 line
2.17.1
Reviewed-by on v1 2/2
#2 lore raw

RE: [PATCH Review 2/2] drm/amdgpu: message smu to update bad channel info

[AMD Official Use Only]

The series is:

Quoted 202 lines

-----Original Message-----
From: Stanley.Yang <[email protected]>
Sent: Friday, March 4, 2022 2:51 PM
To: [email protected]; Zhang, Hawking <[email protected]>; Zhou1, Tao <[email protected]>; Joo, Maria <[email protected]>
Cc: Yang, Stanley <[email protected]>
Subject: [PATCH Review 2/2] drm/amdgpu: message smu to update bad channel info

It should notice SMU to update bad channel info when detected uncorrectable error in UMC block

Change-Id I2dc8848affdb53e52891013953ae9383fff5f20f
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c       |  7 ++++++
 drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h       |  3 +++
 .../gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c    | 25 +++++++++++++++++--
 .../gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.h    |  4 +++
 drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c       |  5 ++++
 5 files changed, 42 insertions(+), 2 deletions(-)
drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +1 -0
index d3875618ebf5..f9104f99eb9c 100644
@@ -2068,6 +2068,7 @@ int amdgpu_ras_recovery_init(struct amdgpu_device
 *adev)
 	mutex_init(&con->recovery_lock);
 	INIT_WORK(&con->recovery_work, amdgpu_ras_do_recovery);
 	atomic_set(&con->in_recovery, 0);
+	con->eeprom_control.bad_channel_bitmap = 0;
 
 	max_eeprom_records_count =

amdgpu_ras_eeprom_max_record_count();

drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +5 -0
 	amdgpu_ras_validate_threshold(adev, max_eeprom_records_count);
@@ -2092,6 +2093,11 @@ int amdgpu_ras_recovery_init(struct amdgpu_device
 *adev)
 			goto free;
 
 		amdgpu_dpm_send_hbm_bad_pages_num(adev, con-
 >eeprom_control.ras_num_recs);
+
+		if (con->update_channel_flag == true) {
+			amdgpu_dpm_send_hbm_bad_channel_flag(adev, con-
 >eeprom_control.bad_channel_bitmap);
+			con->update_channel_flag = false;
+		}
drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +1 -0
 	}
 
 #ifdef CONFIG_X86_MCE_AMD
@@ -2285,6 +2291,7 @@ int amdgpu_ras_init(struct amdgpu_device *adev)
 		goto release_con;
 	}
 
+	con->update_channel_flag = false;
 	con->features = 0;
 	INIT_LIST_HEAD(&con->head);
 	/* Might need get this flag from vbios. */ diff --git

a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h
b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h
index 7cddaad90d6d..9314fde81e68 100644

drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h +3 -0
@@ -374,6 +374,9 @@ struct amdgpu_ras {
 
 	/* record umc error info queried from smu */
 	struct umc_ecc_info umc_ecc;
+
+	/* Indicates smu whether need update bad channel info */
+	bool update_channel_flag;
 };
 
 struct ras_fs_data {
drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c +5 -0
index 2b844a5aafdb..ad5d8667756d 100644
@@ -265,6 +265,7 @@ int amdgpu_ras_eeprom_reset_table(struct
 amdgpu_ras_eeprom_control *control)  {
 	struct amdgpu_device *adev = to_amdgpu_device(control);
 	struct amdgpu_ras_eeprom_table_header *hdr = &control->tbl_hdr;
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
 	u8 csum;
 	int res;
 
@@ -285,6 +286,10 @@ int amdgpu_ras_eeprom_reset_table(struct
 amdgpu_ras_eeprom_control *control)
 
 	amdgpu_dpm_send_hbm_bad_pages_num(adev, control-
 >ras_num_recs);
 
+	control->bad_channel_bitmap = 0;
+	amdgpu_dpm_send_hbm_bad_channel_flag(adev, control-
 >bad_channel_bitmap);
+	con->update_channel_flag = false;
+
drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c +10 -1
 	amdgpu_ras_debugfs_set_ret_size(control);
 
 	mutex_unlock(&control->ras_tbl_mutex);
@@ -418,6 +423,7 @@ amdgpu_ras_eeprom_append_table(struct
 amdgpu_ras_eeprom_control *control,
 			       struct eeprom_table_record *record,
 			       const u32 num)
 {
+	struct amdgpu_ras *con =
+amdgpu_ras_get_context(to_amdgpu_device(control));
 	u32 a, b, i;
 	u8 *buf, *pp;
 	int res;
@@ -429,9 +435,16 @@ amdgpu_ras_eeprom_append_table(struct
 amdgpu_ras_eeprom_control *control,
 	/* Encode all of them in one go.
 	 */
 	pp = buf;
-	for (i = 0; i < num; i++, pp += RAS_TABLE_RECORD_SIZE)
+	for (i = 0; i < num; i++, pp += RAS_TABLE_RECORD_SIZE) {
 		__encode_table_record_to_buf(control, &record[i], pp);
 
+		/* update bad channel bitmap */
+		if (!(control->bad_channel_bitmap & (1 <<
 record[i].mem_channel))) {
+			control->bad_channel_bitmap |= 1 <<
 record[i].mem_channel;
+			con->update_channel_flag = true;
+		}
+	}
+
drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c +0 -0
 	/* a, first record index to write into.
 	 * b, last record index to write into.
 	 * a = first index to read (fri) + number of records in the table, @@ -

684,6 +697,7 @@ int amdgpu_ras_eeprom_read(struct
amdgpu_ras_eeprom_control *control,

drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c +9 -1
 			   const u32 num)
 {
 	struct amdgpu_device *adev = to_amdgpu_device(control);
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
 	int i, res;
 	u8 *buf, *pp;
 	u32 g0, g1;
@@ -751,8 +765,15 @@ int amdgpu_ras_eeprom_read(struct
 amdgpu_ras_eeprom_control *control,
 	/* Read up everything? Then transform.
 	 */
 	pp = buf;
-	for (i = 0; i < num; i++, pp += RAS_TABLE_RECORD_SIZE)
+	for (i = 0; i < num; i++, pp += RAS_TABLE_RECORD_SIZE) {
 		__decode_table_record_from_buf(control, &record[i], pp);
+
+		/* update bad channel bitmap */
+		if (!(control->bad_channel_bitmap & (1 <<
 record[i].mem_channel))) {
+			control->bad_channel_bitmap |= 1 <<
 record[i].mem_channel;
+			con->update_channel_flag = true;
+		}
+	}

Out:

drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c +0 -0
 	kfree(buf);
 	mutex_unlock(&control->ras_tbl_mutex);
drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.h +4 -0
index 6bb00578bfbb..54d9bfe0881d 100644
@@ -80,6 +80,10 @@ struct amdgpu_ras_eeprom_control {
 	/* Protect table access via this mutex.
 	 */
 	struct mutex ras_tbl_mutex;
+
+	/* Record channel info which occurred bad pages
+	 */
+	u32 bad_channel_bitmap;
 };
 
 /*
drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c +3 -0
index 85da6cbaf3b7..aad3c8b4c810 100644
@@ -97,6 +97,11 @@ static int amdgpu_umc_do_page_retirement(struct
 amdgpu_device *adev,
 			amdgpu_ras_save_bad_pages(adev);
 
 			amdgpu_dpm_send_hbm_bad_pages_num(adev, con-
 >eeprom_control.ras_num_recs);
+
+			if (con->update_channel_flag == true) {
+
 	amdgpu_dpm_send_hbm_bad_channel_flag(adev, con-
Quoted 1 line

eeprom_control.bad_channel_bitmap);

drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c +2 -1
+				con->update_channel_flag = false;
+			}
 		}
 
 		if (reset)
--

2.17.1