Skip to contentSkip to conversation

drm/amdgpu: move convert_error_address out of umc_ras

PATCH In review Hawking Zhang 3 participants 3 messages Oct 14, 2022

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 drm/amdgpu: move convert_error_address out of umc_ras

RAS error address translation algorithm is common across dGPU and A + A platform as along as the SOC integrates the same generation of UMC IP.

UMC RAS is managed by x86 MCA on A + A platform, umc_ras in GPU driver is not initialized at all on A + A platform. In such case, any umc_ras callback implemented for dGPU config shouldn't be invoked from A + A specific callback.

The change moves convert_error_address out of dGPU umc_ras structure and makes it share between A + A and dGPU config.

---
 drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 15 +++++++++++----
 drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h |  3 ---
 drivers/gpu/drm/amd/amdgpu/umc_v6_7.c   |  7 +++----
 drivers/gpu/drm/amd/amdgpu/umc_v6_7.h   |  4 +++-
 4 files changed, 17 insertions(+), 12 deletions(-)
drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +11 -4
index 75f1402101f4..ff92ea99d513 100644
@@ -36,6 +36,7 @@
 #include "ivsrcid/nbio/irqsrcs_nbif_7_4.h"
 #include "atom.h"
 #include "amdgpu_reset.h"
+#include "umc_v6_7.h"
 
 #ifdef CONFIG_X86_MCE_AMD
 #include <asm/mce.h>
@@ -2885,10 +2886,16 @@ static int amdgpu_bad_page_notifier(struct notifier_block *nb,
 	/*
 	 * Translate UMC channel address to Physical address
 	 */
-	if (adev->umc.ras &&
-	    adev->umc.ras->convert_ras_error_address)
-		adev->umc.ras->convert_ras_error_address(adev,
-			&err_data, m->addr, ch_inst, umc_inst);
+	switch (adev->ip_versions[UMC_HWIP][0]) {
+	case IP_VERSION(6, 7, 0):
+		umc_v6_7_convert_error_address(adev,
+				&err_data, m->addr, ch_inst, umc_inst);
+		break;
+	default:
+		dev_warn(adev->dev,
+			 "UMC address to Physical address translation is not supported\n");
+		return NOTIFY_DONE;
+	}
 
 	if (amdgpu_bad_page_threshold != 0) {
 		amdgpu_ras_add_bad_pages(adev, err_data.err_addr,
drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h +0 -3
index e46439274f3a..3629d8f292ef 100644
@@ -51,9 +51,6 @@ struct amdgpu_umc_ras {
 	struct amdgpu_ras_block_object ras_block;
 	void (*err_cnt_init)(struct amdgpu_device *adev);
 	bool (*query_ras_poison_mode)(struct amdgpu_device *adev);
-	void (*convert_ras_error_address)(struct amdgpu_device *adev,
-				struct ras_err_data *err_data, uint64_t err_addr,
-				uint32_t ch_inst, uint32_t umc_inst);
 	void (*ecc_info_query_ras_error_count)(struct amdgpu_device *adev,
 				      void *ras_error_status);
 	void (*ecc_info_query_ras_error_address)(struct amdgpu_device *adev,
drivers/gpu/drm/amd/amdgpu/umc_v6_7.c +3 -4
index 5d5d031c9e7d..72fd963f178b 100644
@@ -187,9 +187,9 @@ static void umc_v6_7_ecc_info_query_ras_error_count(struct amdgpu_device *adev,
 	}
 }
 
-static void umc_v6_7_convert_error_address(struct amdgpu_device *adev,
-					struct ras_err_data *err_data, uint64_t err_addr,
-					uint32_t ch_inst, uint32_t umc_inst)
+void umc_v6_7_convert_error_address(struct amdgpu_device *adev,
+				    struct ras_err_data *err_data, uint64_t err_addr,
+				    uint32_t ch_inst, uint32_t umc_inst)
 {
 	uint32_t channel_index;
 	uint64_t soc_pa, retired_page, column;
@@ -553,5 +553,4 @@ struct amdgpu_umc_ras umc_v6_7_ras = {
 	.query_ras_poison_mode = umc_v6_7_query_ras_poison_mode,
 	.ecc_info_query_ras_error_count = umc_v6_7_ecc_info_query_ras_error_count,
 	.ecc_info_query_ras_error_address = umc_v6_7_ecc_info_query_ras_error_address,
-	.convert_ras_error_address = umc_v6_7_convert_error_address,
 };
drivers/gpu/drm/amd/amdgpu/umc_v6_7.h +3 -1
index fe41ed2f5945..105245d5b6e5 100644
@@ -71,5 +71,7 @@ extern const uint32_t
 	umc_v6_7_channel_idx_tbl_second[UMC_V6_7_UMC_INSTANCE_NUM][UMC_V6_7_CHANNEL_INSTANCE_NUM];
 extern const uint32_t
 	umc_v6_7_channel_idx_tbl_first[UMC_V6_7_UMC_INSTANCE_NUM][UMC_V6_7_CHANNEL_INSTANCE_NUM];
-
+void umc_v6_7_convert_error_address(struct amdgpu_device *adev,
+                                    struct ras_err_data *err_data, uint64_t err_addr,
+                                    uint32_t ch_inst, uint32_t umc_inst);
 #endif
Signature · 1 line
2.17.1
Reviewed-by on v1 1/1
#1 lore raw

[AMD Official Use Only - General]

Regards,
Stanley

Quoted 132 lines

-----Original Message-----
From: amd-gfx <[email protected]> On Behalf Of Hawking Zhang
Sent: Friday, October 14, 2022 2:19 PM
To: [email protected]; Zhou1, Tao <[email protected]>; Yang, Stanley <[email protected]>
Cc: Russell, Kent <[email protected]>; Zhang, Hawking <[email protected]>
Subject: [PATCH] drm/amdgpu: move convert_error_address out of umc_ras

RAS error address translation algorithm is common across dGPU and A + A platform as along as the SOC integrates the same generation of UMC IP.

UMC RAS is managed by x86 MCA on A + A platform, umc_ras in GPU driver is not initialized at all on A + A platform. In such case, any umc_ras callback implemented for dGPU config shouldn't be invoked from A + A specific callback.

The change moves convert_error_address out of dGPU umc_ras structure and makes it share between A + A and dGPU config.

---
 drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 15 +++++++++++----
drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h |  3 ---
 drivers/gpu/drm/amd/amdgpu/umc_v6_7.c   |  7 +++----
 drivers/gpu/drm/amd/amdgpu/umc_v6_7.h   |  4 +++-
 4 files changed, 17 insertions(+), 12 deletions(-)
drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +11 -4
index 75f1402101f4..ff92ea99d513 100644
@@ -36,6 +36,7 @@
 #include "ivsrcid/nbio/irqsrcs_nbif_7_4.h"
 #include "atom.h"
 #include "amdgpu_reset.h"
+#include "umc_v6_7.h"
 
 #ifdef CONFIG_X86_MCE_AMD
 #include <asm/mce.h>
@@ -2885,10 +2886,16 @@ static int amdgpu_bad_page_notifier(struct
 notifier_block *nb,
 	/*
 	 * Translate UMC channel address to Physical address
 	 */
-	if (adev->umc.ras &&
-	    adev->umc.ras->convert_ras_error_address)
-		adev->umc.ras->convert_ras_error_address(adev,
-			&err_data, m->addr, ch_inst, umc_inst);
+	switch (adev->ip_versions[UMC_HWIP][0]) {
+	case IP_VERSION(6, 7, 0):
+		umc_v6_7_convert_error_address(adev,
+				&err_data, m->addr, ch_inst, umc_inst);
+		break;
+	default:
+		dev_warn(adev->dev,
+			 "UMC address to Physical address translation is not
 supported\n");
+		return NOTIFY_DONE;
+	}
 
drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +0 -0
 	if (amdgpu_bad_page_threshold != 0) {
 		amdgpu_ras_add_bad_pages(adev, err_data.err_addr, diff --

git a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h
b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h
index e46439274f3a..3629d8f292ef 100644

drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h +0 -3
@@ -51,9 +51,6 @@ struct amdgpu_umc_ras {
 	struct amdgpu_ras_block_object ras_block;
 	void (*err_cnt_init)(struct amdgpu_device *adev);
 	bool (*query_ras_poison_mode)(struct amdgpu_device *adev);
-	void (*convert_ras_error_address)(struct amdgpu_device *adev,
-				struct ras_err_data *err_data, uint64_t
 err_addr,
-				uint32_t ch_inst, uint32_t umc_inst);
 	void (*ecc_info_query_ras_error_count)(struct amdgpu_device
 *adev,
drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h +0 -0
 				      void *ras_error_status);
 	void (*ecc_info_query_ras_error_address)(struct amdgpu_device

*adev, diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v6_7.c b/drivers/gpu/drm/amd/amdgpu/umc_v6_7.c
index 5d5d031c9e7d..72fd963f178b 100644

drivers/gpu/drm/amd/amdgpu/umc_v6_7.c +3 -3
@@ -187,9 +187,9 @@ static void
 umc_v6_7_ecc_info_query_ras_error_count(struct amdgpu_device *adev,
 	}
 }
 
-static void umc_v6_7_convert_error_address(struct amdgpu_device *adev,
-					struct ras_err_data *err_data,
 uint64_t err_addr,
-					uint32_t ch_inst, uint32_t umc_inst)
+void umc_v6_7_convert_error_address(struct amdgpu_device *adev,
+				    struct ras_err_data *err_data, uint64_t
 err_addr,
+				    uint32_t ch_inst, uint32_t umc_inst)

{

drivers/gpu/drm/amd/amdgpu/umc_v6_7.c +0 -0
 	uint32_t channel_index;
 	uint64_t soc_pa, retired_page, column; @@ -553,5 +553,4 @@ struct

amdgpu_umc_ras umc_v6_7_ras = {

drivers/gpu/drm/amd/amdgpu/umc_v6_7.c +0 -0
 	.query_ras_poison_mode = umc_v6_7_query_ras_poison_mode,
 	.ecc_info_query_ras_error_count =

umc_v6_7_ecc_info_query_ras_error_count,
.ecc_info_query_ras_error_address =
umc_v6_7_ecc_info_query_ras_error_address,

drivers/gpu/drm/amd/amdgpu/umc_v6_7.c +0 -1
-	.convert_ras_error_address = umc_v6_7_convert_error_address,
 };
drivers/gpu/drm/amd/amdgpu/umc_v6_7.h +0 -0
index fe41ed2f5945..105245d5b6e5 100644
@@ -71,5 +71,7 @@ extern const uint32_t
 
 	umc_v6_7_channel_idx_tbl_second[UMC_V6_7_UMC_INSTANCE_NU
 M][UMC_V6_7_CHANNEL_INSTANCE_NUM];
 extern const uint32_t
 
	umc_v6_7_channel_idx_tbl_first[UMC_V6_7_UMC_INSTANCE_NUM]

[UMC_V6_7_CHANNEL_INSTANCE_NUM];

drivers/gpu/drm/amd/amdgpu/umc_v6_7.h +4 -2
-
+void umc_v6_7_convert_error_address(struct amdgpu_device *adev,
+                                    struct ras_err_data *err_data, uint64_t err_addr,
+                                    uint32_t ch_inst, uint32_t
+umc_inst);
 #endif
--

2.17.1

#2 lore raw
In reply to Hawking Zhang

default:
+ dev_warn(adev->dev,
+ "UMC address to Physical address translation is not supported\n");
+ return NOTIFY_DONE;

Before returning, maybe it's necessary to free err_data.err_addr?

Regards,
Guchun

-----Original Message-----
From: amd-gfx <[email protected]> On Behalf Of Hawking Zhang
Sent: Friday, October 14, 2022 2:19 PM
To: [email protected]; Zhou1, Tao <[email protected]>; Yang, Stanley <[email protected]>
Cc: Russell, Kent <[email protected]>; Zhang, Hawking <[email protected]>
Subject: [PATCH] drm/amdgpu: move convert_error_address out of umc_ras

RAS error address translation algorithm is common across dGPU and A + A platform as along as the SOC integrates the same generation of UMC IP.

UMC RAS is managed by x86 MCA on A + A platform, umc_ras in GPU driver is not initialized at all on A + A platform. In such case, any umc_ras callback implemented for dGPU config shouldn't be invoked from A + A specific callback.

The change moves convert_error_address out of dGPU umc_ras structure and makes it share between A + A and dGPU config.

---
 drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 15 +++++++++++----  drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h |  3 ---
 drivers/gpu/drm/amd/amdgpu/umc_v6_7.c   |  7 +++----
 drivers/gpu/drm/amd/amdgpu/umc_v6_7.h   |  4 +++-
 4 files changed, 17 insertions(+), 12 deletions(-)
drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +11 -4
index 75f1402101f4..ff92ea99d513 100644
@@ -36,6 +36,7 @@
 #include "ivsrcid/nbio/irqsrcs_nbif_7_4.h"
 #include "atom.h"
 #include "amdgpu_reset.h"
+#include "umc_v6_7.h"
 
 #ifdef CONFIG_X86_MCE_AMD
 #include <asm/mce.h>
@@ -2885,10 +2886,16 @@ static int amdgpu_bad_page_notifier(struct notifier_block *nb,
 	/*
 	 * Translate UMC channel address to Physical address
 	 */
-	if (adev->umc.ras &&
-	    adev->umc.ras->convert_ras_error_address)
-		adev->umc.ras->convert_ras_error_address(adev,
-			&err_data, m->addr, ch_inst, umc_inst);
+	switch (adev->ip_versions[UMC_HWIP][0]) {
+	case IP_VERSION(6, 7, 0):
+		umc_v6_7_convert_error_address(adev,
+				&err_data, m->addr, ch_inst, umc_inst);
+		break;
+	default:
+		dev_warn(adev->dev,
+			 "UMC address to Physical address translation is not supported\n");
+		return NOTIFY_DONE;
+	}
 
 	if (amdgpu_bad_page_threshold != 0) {
 		amdgpu_ras_add_bad_pages(adev, err_data.err_addr, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h

index e46439274f3a..3629d8f292ef 100644

drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h +0 -3
@@ -51,9 +51,6 @@ struct amdgpu_umc_ras {
 	struct amdgpu_ras_block_object ras_block;
 	void (*err_cnt_init)(struct amdgpu_device *adev);
 	bool (*query_ras_poison_mode)(struct amdgpu_device *adev);
-	void (*convert_ras_error_address)(struct amdgpu_device *adev,
-				struct ras_err_data *err_data, uint64_t err_addr,
-				uint32_t ch_inst, uint32_t umc_inst);
 	void (*ecc_info_query_ras_error_count)(struct amdgpu_device *adev,
 				      void *ras_error_status);
 	void (*ecc_info_query_ras_error_address)(struct amdgpu_device *adev, diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v6_7.c b/drivers/gpu/drm/amd/amdgpu/umc_v6_7.c

index 5d5d031c9e7d..72fd963f178b 100644

drivers/gpu/drm/amd/amdgpu/umc_v6_7.c +3 -3
@@ -187,9 +187,9 @@ static void umc_v6_7_ecc_info_query_ras_error_count(struct amdgpu_device *adev,
 	}
 }
 
-static void umc_v6_7_convert_error_address(struct amdgpu_device *adev,
-					struct ras_err_data *err_data, uint64_t err_addr,
-					uint32_t ch_inst, uint32_t umc_inst)
+void umc_v6_7_convert_error_address(struct amdgpu_device *adev,
+				    struct ras_err_data *err_data, uint64_t err_addr,
+				    uint32_t ch_inst, uint32_t umc_inst)
 {
 	uint32_t channel_index;
 	uint64_t soc_pa, retired_page, column; @@ -553,5 +553,4 @@ struct amdgpu_umc_ras umc_v6_7_ras = {

.query_ras_poison_mode = umc_v6_7_query_ras_poison_mode, .ecc_info_query_ras_error_count = umc_v6_7_ecc_info_query_ras_error_count, .ecc_info_query_ras_error_address = umc_v6_7_ecc_info_query_ras_error_address,
- .convert_ras_error_address = umc_v6_7_convert_error_address,
};

drivers/gpu/drm/amd/amdgpu/umc_v6_7.h +4 -1
index fe41ed2f5945..105245d5b6e5 100644
@@ -71,5 +71,7 @@ extern const uint32_t
 	umc_v6_7_channel_idx_tbl_second[UMC_V6_7_UMC_INSTANCE_NUM][UMC_V6_7_CHANNEL_INSTANCE_NUM];
 extern const uint32_t
 	umc_v6_7_channel_idx_tbl_first[UMC_V6_7_UMC_INSTANCE_NUM][UMC_V6_7_CHANNEL_INSTANCE_NUM];
-
+void umc_v6_7_convert_error_address(struct amdgpu_device *adev,
+                                    struct ras_err_data *err_data, uint64_t err_addr,
+                                    uint32_t ch_inst, uint32_t 
+umc_inst);

#endif
--
2.17.1