[PATCH 01/11] drm/amdgpu: Optimize xxx_ras_late_init/xxx_ras_late_fini for each ras block

[PATCH 01/11] drm/amdgpu: Optimize xxx_ras_late_init/xxx_ras_late_fini for each ras block

[yipechai]

1. Define amdgpu_ras_block_late_init to create sysfs nodes
   and interrupt handles.
2. Define amdgpu_ras_block_late_fini to remove sysfs nodes
   and interrupt handles.
3. Replace ras block variable members in struct
   amdgpu_ras_block_object with struct ras_common_if, which
   can makes it easy to associate each ras block instance
   with each ras block functional interface.
4. Add .ras_cb to struct amdgpu_ras_block_object.
5. Change each ras block to fit for the changement of struct
   amdgpu_ras_block_object.

Signed-off-by: yipechai <YiPeng.Chai@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c  |  7 +++--
 drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c  | 35 +++++++++++++++++++++++-
 drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h  | 15 ++++++----
 drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c |  6 ++--
 drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c    |  4 +--
 drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c   |  4 +--
 drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c    |  8 +++---
 drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c    |  6 ++--
 drivers/gpu/drm/amd/amdgpu/mca_v3_0.c    | 28 +++++++++++--------
 drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c   |  6 ++--
 drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c   |  4 +--
 11 files changed, 86 insertions(+), 37 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c
index 52a60c2316a2..ad057d6b2c77 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c
@@ -83,14 +83,15 @@ int amdgpu_mca_ras_late_init(struct amdgpu_device *adev,
 		.sysfs_name = sysfs_name,
 	};
 
-	snprintf(sysfs_name, sizeof(sysfs_name), "%s_err_count", mca_dev->ras->ras_block.name);
+	snprintf(sysfs_name, sizeof(sysfs_name), "%s_err_count",
+		mca_dev->ras->ras_block.ras_comm.name);
 
 	if (!mca_dev->ras_if) {
 		mca_dev->ras_if = kmalloc(sizeof(struct ras_common_if), GFP_KERNEL);
 		if (!mca_dev->ras_if)
 			return -ENOMEM;
-		mca_dev->ras_if->block = mca_dev->ras->ras_block.block;
-		mca_dev->ras_if->sub_block_index = mca_dev->ras->ras_block.sub_block_index;
+		mca_dev->ras_if->block = mca_dev->ras->ras_block.ras_comm.block;
+		mca_dev->ras_if->sub_block_index = mca_dev->ras->ras_block.ras_comm.sub_block_index;
 		mca_dev->ras_if->type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE;
 	}
 	ih_info.head = fs_info.head = *mca_dev->ras_if;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
index 5934326b9db3..b7aed19db7e9 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
@@ -877,7 +877,7 @@ static int amdgpu_ras_block_match_default(struct amdgpu_ras_block_object *block_
 	if (!block_obj)
 		return -EINVAL;
 
-	if (block_obj->block == block)
+	if (block_obj->ras_comm.block == block)
 		return 0;
 
 	return -EINVAL;
@@ -2457,6 +2457,23 @@ int amdgpu_ras_late_init(struct amdgpu_device *adev,
 	return r;
 }
 
+int amdgpu_ras_block_late_init(struct amdgpu_device *adev,
+			struct ras_common_if *ras_block)
+{
+	char sysfs_name[32];
+	struct ras_ih_if ih_info;
+	struct ras_fs_if fs_info;
+	struct amdgpu_ras_block_object *obj;
+
+	obj = container_of(ras_block, struct amdgpu_ras_block_object, ras_comm);
+	ih_info.cb = obj->ras_cb;
+	ih_info.head = *ras_block;
+	snprintf(sysfs_name, sizeof(sysfs_name), "%s_err_count", ras_block->name);
+	fs_info.sysfs_name = (const char *)sysfs_name;
+	fs_info.head = *ras_block;
+	return amdgpu_ras_late_init(adev, ras_block, &fs_info, &ih_info);
+}
+
 /* helper function to remove ras fs node and interrupt handler */
 void amdgpu_ras_late_fini(struct amdgpu_device *adev,
 			  struct ras_common_if *ras_block,
@@ -2470,6 +2487,22 @@ void amdgpu_ras_late_fini(struct amdgpu_device *adev,
 		amdgpu_ras_interrupt_remove_handler(adev, ih_info);
 }
 
+void amdgpu_ras_block_late_fini(struct amdgpu_device *adev,
+			  struct ras_common_if *ras_block)
+{
+	struct ras_ih_if ih_info;
+	struct amdgpu_ras_block_object *obj;
+
+	if (!ras_block)
+		return;
+
+	obj = container_of(ras_block, struct amdgpu_ras_block_object, ras_comm);
+	ih_info.head = *ras_block;
+	ih_info.cb = obj->ras_cb;
+
+	amdgpu_ras_late_fini(adev, ras_block, &ih_info);
+}
+
 /* do some init work after IP late init as dependence.
  * and it runs in resume/gpu reset/booting up cases.
  */
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h
index a55743b12d57..8b94b556baf6 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h
@@ -486,17 +486,13 @@ struct ras_debug_if {
 };
 
 struct amdgpu_ras_block_object {
-	/* block name */
-	char name[32];
-
-	enum amdgpu_ras_block block;
-
-	uint32_t sub_block_index;
+	struct ras_common_if  ras_comm;
 
 	int (*ras_block_match)(struct amdgpu_ras_block_object *block_obj,
 				enum amdgpu_ras_block block, uint32_t sub_block_index);
 	int (*ras_late_init)(struct amdgpu_device *adev, void *ras_info);
 	void (*ras_fini)(struct amdgpu_device *adev);
+	ras_ih_cb ras_cb;
 	const struct amdgpu_ras_block_hw_ops *hw_ops;
 };
 
@@ -605,10 +601,17 @@ int amdgpu_ras_late_init(struct amdgpu_device *adev,
 			 struct ras_common_if *ras_block,
 			 struct ras_fs_if *fs_info,
 			 struct ras_ih_if *ih_info);
+
+int amdgpu_ras_block_late_init(struct amdgpu_device *adev,
+			struct ras_common_if *ras_block);
+
 void amdgpu_ras_late_fini(struct amdgpu_device *adev,
 			  struct ras_common_if *ras_block,
 			  struct ras_ih_if *ih_info);
 
+void amdgpu_ras_block_late_fini(struct amdgpu_device *adev,
+			  struct ras_common_if *ras_block);
+
 int amdgpu_ras_feature_enable(struct amdgpu_device *adev,
 		struct ras_common_if *head, bool enable);
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c
index 5929d6f528c9..15707af89212 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c
@@ -981,8 +981,10 @@ struct amdgpu_ras_block_hw_ops  xgmi_ras_hw_ops = {
 
 struct amdgpu_xgmi_ras xgmi_ras = {
 	.ras_block = {
-		.name = "xgmi",
-		.block = AMDGPU_RAS_BLOCK__XGMI_WAFL,
+		.ras_comm = {
+			.name = "xgmi",
+			.block = AMDGPU_RAS_BLOCK__XGMI_WAFL,
+		},
 		.hw_ops = &xgmi_ras_hw_ops,
 		.ras_late_init = amdgpu_xgmi_ras_late_init,
 		.ras_fini = amdgpu_xgmi_ras_fini,
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
index ca7b886c6ce6..0a291d2e5f91 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
@@ -2198,8 +2198,8 @@ static int gfx_v9_0_gpu_early_init(struct amdgpu_device *adev)
 			return err;
 		}
 
-		strcpy(adev->gfx.ras->ras_block.name,"gfx");
-		adev->gfx.ras->ras_block.block = AMDGPU_RAS_BLOCK__GFX;
+		strcpy(adev->gfx.ras->ras_block.ras_comm.name, "gfx");
+		adev->gfx.ras->ras_block.ras_comm.block = AMDGPU_RAS_BLOCK__GFX;
 
 		/* If not define special ras_late_init function, use gfx default ras_late_init */
 		if (!adev->gfx.ras->ras_block.ras_late_init)
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c
index bddaf2417344..2a362c570346 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c
@@ -672,8 +672,8 @@ static void gmc_v10_0_set_umc_funcs(struct amdgpu_device *adev)
 	if (adev->umc.ras) {
 		amdgpu_ras_register_ras_block(adev, &adev->umc.ras->ras_block);
 
-		strcpy(adev->umc.ras->ras_block.name, "umc");
-		adev->umc.ras->ras_block.block = AMDGPU_RAS_BLOCK__UMC;
+		strcpy(adev->umc.ras->ras_block.ras_comm.name, "umc");
+		adev->umc.ras->ras_block.ras_comm.block = AMDGPU_RAS_BLOCK__UMC;
 
 		/* If don't define special ras_late_init function, use default ras_late_init */
 		if (!adev->umc.ras->ras_block.ras_late_init)
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
index 4595027a8c63..af873c99d5e4 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
@@ -1232,8 +1232,8 @@ static void gmc_v9_0_set_umc_funcs(struct amdgpu_device *adev)
 	if (adev->umc.ras) {
 		amdgpu_ras_register_ras_block(adev, &adev->umc.ras->ras_block);
 
-		strcpy(adev->umc.ras->ras_block.name, "umc");
-		adev->umc.ras->ras_block.block = AMDGPU_RAS_BLOCK__UMC;
+		strcpy(adev->umc.ras->ras_block.ras_comm.name, "umc");
+		adev->umc.ras->ras_block.ras_comm.block = AMDGPU_RAS_BLOCK__UMC;
 
 		/* If don't define special ras_late_init function, use default ras_late_init */
 		if (!adev->umc.ras->ras_block.ras_late_init)
@@ -1280,8 +1280,8 @@ static void gmc_v9_0_set_mmhub_ras_funcs(struct amdgpu_device *adev)
 	if (adev->mmhub.ras) {
 		amdgpu_ras_register_ras_block(adev, &adev->mmhub.ras->ras_block);
 
-		strcpy(adev->mmhub.ras->ras_block.name,"mmhub");
-		adev->mmhub.ras->ras_block.block = AMDGPU_RAS_BLOCK__MMHUB;
+		strcpy(adev->mmhub.ras->ras_block.ras_comm.name, "mmhub");
+		adev->mmhub.ras->ras_block.ras_comm.block = AMDGPU_RAS_BLOCK__MMHUB;
 
 		/* If don't define special ras_late_init function, use default ras_late_init */
 		if (!adev->mmhub.ras->ras_block.ras_late_init)
diff --git a/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c b/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c
index 6b41fcbf4875..503c292b321e 100644
--- a/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c
@@ -157,8 +157,10 @@ struct amdgpu_ras_block_hw_ops hdp_v4_0_ras_hw_ops = {
 
 struct amdgpu_hdp_ras hdp_v4_0_ras = {
 	.ras_block = {
-		.name = "hdp",
-		.block = AMDGPU_RAS_BLOCK__HDP,
+		.ras_comm = {
+			.name = "hdp",
+			.block = AMDGPU_RAS_BLOCK__HDP,
+		},
 		.hw_ops = &hdp_v4_0_ras_hw_ops,
 		.ras_late_init = amdgpu_hdp_ras_late_init,
 		.ras_fini = amdgpu_hdp_ras_fini,
diff --git a/drivers/gpu/drm/amd/amdgpu/mca_v3_0.c b/drivers/gpu/drm/amd/amdgpu/mca_v3_0.c
index 68565262af9c..386416378a82 100644
--- a/drivers/gpu/drm/amd/amdgpu/mca_v3_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/mca_v3_0.c
@@ -53,8 +53,8 @@ static int mca_v3_0_ras_block_match(struct amdgpu_ras_block_object *block_obj,
 	if (!block_obj)
 		return -EINVAL;
 
-	if ((block_obj->block == block) &&
-		(block_obj->sub_block_index == sub_block_index)) {
+	if ((block_obj->ras_comm.block == block) &&
+		(block_obj->ras_comm.sub_block_index == sub_block_index)) {
 		return 0;
 	}
 
@@ -68,9 +68,11 @@ const struct amdgpu_ras_block_hw_ops mca_v3_0_mp0_hw_ops = {
 
 struct amdgpu_mca_ras_block mca_v3_0_mp0_ras = {
 	.ras_block = {
-		.block = AMDGPU_RAS_BLOCK__MCA,
-		.sub_block_index = AMDGPU_RAS_MCA_BLOCK__MP0,
-		.name = "mp0",
+		.ras_comm = {
+			.block = AMDGPU_RAS_BLOCK__MCA,
+			.sub_block_index = AMDGPU_RAS_MCA_BLOCK__MP0,
+			.name = "mp0",
+		},
 		.hw_ops = &mca_v3_0_mp0_hw_ops,
 		.ras_block_match = mca_v3_0_ras_block_match,
 		.ras_late_init = mca_v3_0_mp0_ras_late_init,
@@ -103,9 +105,11 @@ const struct amdgpu_ras_block_hw_ops mca_v3_0_mp1_hw_ops = {
 
 struct amdgpu_mca_ras_block mca_v3_0_mp1_ras = {
 	.ras_block = {
-		.block = AMDGPU_RAS_BLOCK__MCA,
-		.sub_block_index = AMDGPU_RAS_MCA_BLOCK__MP1,
-		.name = "mp1",
+		.ras_comm = {
+			.block = AMDGPU_RAS_BLOCK__MCA,
+			.sub_block_index = AMDGPU_RAS_MCA_BLOCK__MP1,
+			.name = "mp1",
+		},
 		.hw_ops = &mca_v3_0_mp1_hw_ops,
 		.ras_block_match = mca_v3_0_ras_block_match,
 		.ras_late_init = mca_v3_0_mp1_ras_late_init,
@@ -138,9 +142,11 @@ const struct amdgpu_ras_block_hw_ops mca_v3_0_mpio_hw_ops = {
 
 struct amdgpu_mca_ras_block mca_v3_0_mpio_ras = {
 	.ras_block = {
-		.block = AMDGPU_RAS_BLOCK__MCA,
-		.sub_block_index = AMDGPU_RAS_MCA_BLOCK__MPIO,
-		.name = "mpio",
+		.ras_comm = {
+			.block = AMDGPU_RAS_BLOCK__MCA,
+			.sub_block_index = AMDGPU_RAS_MCA_BLOCK__MPIO,
+			.name = "mpio",
+		},
 		.hw_ops = &mca_v3_0_mpio_hw_ops,
 		.ras_block_match = mca_v3_0_ras_block_match,
 		.ras_late_init = mca_v3_0_mpio_ras_late_init,
diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c b/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c
index 39974b449341..c7cca87f1647 100644
--- a/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c
+++ b/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c
@@ -664,8 +664,10 @@ const struct amdgpu_ras_block_hw_ops nbio_v7_4_ras_hw_ops = {
 
 struct amdgpu_nbio_ras nbio_v7_4_ras = {
 	.ras_block = {
-		.name = "pcie_bif",
-		.block = AMDGPU_RAS_BLOCK__PCIE_BIF,
+		.ras_comm = {
+			.name = "pcie_bif",
+			.block = AMDGPU_RAS_BLOCK__PCIE_BIF,
+		},
 		.hw_ops = &nbio_v7_4_ras_hw_ops,
 		.ras_late_init = amdgpu_nbio_ras_late_init,
 		.ras_fini = amdgpu_nbio_ras_fini,
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c
index 06a7ceda4c87..8b0a8587dd36 100644
--- a/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c
@@ -2814,8 +2814,8 @@ static void sdma_v4_0_set_ras_funcs(struct amdgpu_device *adev)
 	if (adev->sdma.ras) {
 		amdgpu_ras_register_ras_block(adev, &adev->sdma.ras->ras_block);
 
-		strcpy(adev->sdma.ras->ras_block.name, "sdma");
-		adev->sdma.ras->ras_block.block = AMDGPU_RAS_BLOCK__SDMA;
+		strcpy(adev->sdma.ras->ras_block.ras_comm.name, "sdma");
+		adev->sdma.ras->ras_block.ras_comm.block = AMDGPU_RAS_BLOCK__SDMA;
 
 		/* If don't define special ras_late_init function, use default ras_late_init */
 		if (!adev->sdma.ras->ras_block.ras_late_init)
-- 
2.25.1

[PATCH 02/11] drm/amdgpu: Optimize amdgpu_gfx_ras_late_init/amdgpu_gfx_ras_fini function code

[yipechai]

Optimize amdgpu_gfx_ras_late_init/amdgpu_gfx_ras_fini function code.

Signed-off-by: yipechai <YiPeng.Chai@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c | 42 +++----------------------
 drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c   |  6 ++++
 2 files changed, 11 insertions(+), 37 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
index 43004822ec6f..fe392108b5c2 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
@@ -625,26 +625,9 @@ int amdgpu_get_gfx_off_status(struct amdgpu_device *adev, uint32_t *value)
 int amdgpu_gfx_ras_late_init(struct amdgpu_device *adev, void *ras_info)
 {
 	int r;
-	struct ras_fs_if fs_info = {
-		.sysfs_name = "gfx_err_count",
-	};
-	struct ras_ih_if ih_info = {
-		.cb = amdgpu_gfx_process_ras_data_cb,
-	};
-
-	if (!adev->gfx.ras_if) {
-		adev->gfx.ras_if = kmalloc(sizeof(struct ras_common_if), GFP_KERNEL);
-		if (!adev->gfx.ras_if)
-			return -ENOMEM;
-		adev->gfx.ras_if->block = AMDGPU_RAS_BLOCK__GFX;
-		adev->gfx.ras_if->type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE;
-		adev->gfx.ras_if->sub_block_index = 0;
-	}
-	fs_info.head = ih_info.head = *adev->gfx.ras_if;
-	r = amdgpu_ras_late_init(adev, adev->gfx.ras_if,
-				 &fs_info, &ih_info);
+	r = amdgpu_ras_block_late_init(adev, adev->gfx.ras_if);
 	if (r)
-		goto free;
+		return r;
 
 	if (amdgpu_ras_is_supported(adev, adev->gfx.ras_if->block)) {
 		if (!amdgpu_persistent_edc_harvesting_supported(adev))
@@ -653,34 +636,19 @@ int amdgpu_gfx_ras_late_init(struct amdgpu_device *adev, void *ras_info)
 		r = amdgpu_irq_get(adev, &adev->gfx.cp_ecc_error_irq, 0);
 		if (r)
 			goto late_fini;
-	} else {
-		/* free gfx ras_if if ras is not supported */
-		r = 0;
-		goto free;
 	}
 
 	return 0;
 late_fini:
-	amdgpu_ras_late_fini(adev, adev->gfx.ras_if, &ih_info);
-free:
-	kfree(adev->gfx.ras_if);
-	adev->gfx.ras_if = NULL;
+	amdgpu_ras_block_late_fini(adev, adev->gfx.ras_if);
 	return r;
 }
 
 void amdgpu_gfx_ras_fini(struct amdgpu_device *adev)
 {
 	if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__GFX) &&
-			adev->gfx.ras_if) {
-		struct ras_common_if *ras_if = adev->gfx.ras_if;
-		struct ras_ih_if ih_info = {
-			.head = *ras_if,
-			.cb = amdgpu_gfx_process_ras_data_cb,
-		};
-
-		amdgpu_ras_late_fini(adev, ras_if, &ih_info);
-		kfree(ras_if);
-	}
+			adev->gfx.ras_if)
+		amdgpu_ras_block_late_fini(adev, adev->gfx.ras_if);
 }
 
 int amdgpu_gfx_process_ras_data_cb(struct amdgpu_device *adev,
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
index 0a291d2e5f91..0c3b3972c04e 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
@@ -2200,6 +2200,8 @@ static int gfx_v9_0_gpu_early_init(struct amdgpu_device *adev)
 
 		strcpy(adev->gfx.ras->ras_block.ras_comm.name, "gfx");
 		adev->gfx.ras->ras_block.ras_comm.block = AMDGPU_RAS_BLOCK__GFX;
+		adev->gfx.ras->ras_block.ras_comm.type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE;
+		adev->gfx.ras_if = &adev->gfx.ras->ras_block.ras_comm;
 
 		/* If not define special ras_late_init function, use gfx default ras_late_init */
 		if (!adev->gfx.ras->ras_block.ras_late_init)
@@ -2208,6 +2210,10 @@ static int gfx_v9_0_gpu_early_init(struct amdgpu_device *adev)
 		/* If not define special ras_fini function, use gfx default ras_fini */
 		if (!adev->gfx.ras->ras_block.ras_fini)
 			adev->gfx.ras->ras_block.ras_fini = amdgpu_gfx_ras_fini;
+
+		/* If not defined special ras_cb function, use default ras_cb */
+		if (!adev->gfx.ras->ras_block.ras_cb)
+			adev->gfx.ras->ras_block.ras_cb = amdgpu_gfx_process_ras_data_cb;
 	}
 
 	adev->gfx.config.gb_addr_config = gb_addr_config;
-- 
2.25.1

[PATCH 03/11] drm/amdgpu: Optimize amdgpu_hdp_ras_late_init/amdgpu_hdp_ras_fini function code

[yipechai]

Optimize amdgpu_hdp_ras_late_init/amdgpu_hdp_ras_fini function code.

Signed-off-by: yipechai <YiPeng.Chai@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_hdp.c | 37 ++-----------------------
 drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c   |  1 +
 drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c   |  1 +
 3 files changed, 5 insertions(+), 34 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_hdp.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_hdp.c
index 518966a26130..21a5f884dd2a 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_hdp.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_hdp.c
@@ -26,43 +26,12 @@
 
 int amdgpu_hdp_ras_late_init(struct amdgpu_device *adev, void *ras_info)
 {
-	int r;
-	struct ras_ih_if ih_info = {
-		.cb = NULL,
-	};
-	struct ras_fs_if fs_info = {
-		.sysfs_name = "hdp_err_count",
-	};
-
-	if (!adev->hdp.ras_if) {
-		adev->hdp.ras_if = kmalloc(sizeof(struct ras_common_if), GFP_KERNEL);
-		if (!adev->hdp.ras_if)
-			return -ENOMEM;
-		adev->hdp.ras_if->block = AMDGPU_RAS_BLOCK__HDP;
-		adev->hdp.ras_if->type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE;
-		adev->hdp.ras_if->sub_block_index = 0;
-	}
-	ih_info.head = fs_info.head = *adev->hdp.ras_if;
-	r = amdgpu_ras_late_init(adev, adev->hdp.ras_if,
-				 &fs_info, &ih_info);
-	if (r || !amdgpu_ras_is_supported(adev, adev->hdp.ras_if->block)) {
-		kfree(adev->hdp.ras_if);
-		adev->hdp.ras_if = NULL;
-	}
-
-	return r;
+	return amdgpu_ras_block_late_init(adev, adev->hdp.ras_if);
 }
 
 void amdgpu_hdp_ras_fini(struct amdgpu_device *adev)
 {
 	if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__HDP) &&
-	    adev->hdp.ras_if) {
-		struct ras_common_if *ras_if = adev->hdp.ras_if;
-		struct ras_ih_if ih_info = {
-			.cb = NULL,
-		};
-
-		amdgpu_ras_late_fini(adev, ras_if, &ih_info);
-		kfree(ras_if);
-	}
+	    adev->hdp.ras_if)
+		amdgpu_ras_block_late_fini(adev, adev->hdp.ras_if);
 }
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
index af873c99d5e4..b12fe6703f02 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
@@ -1302,6 +1302,7 @@ static void gmc_v9_0_set_hdp_ras_funcs(struct amdgpu_device *adev)
 {
 	adev->hdp.ras = &hdp_v4_0_ras;
 	amdgpu_ras_register_ras_block(adev, &adev->hdp.ras->ras_block);
+	adev->hdp.ras_if = &adev->hdp.ras->ras_block.ras_comm;
 }
 
 static void gmc_v9_0_set_mca_funcs(struct amdgpu_device *adev)
diff --git a/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c b/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c
index 503c292b321e..a9ed4232cdeb 100644
--- a/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c
@@ -160,6 +160,7 @@ struct amdgpu_hdp_ras hdp_v4_0_ras = {
 		.ras_comm = {
 			.name = "hdp",
 			.block = AMDGPU_RAS_BLOCK__HDP,
+			.type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE,
 		},
 		.hw_ops = &hdp_v4_0_ras_hw_ops,
 		.ras_late_init = amdgpu_hdp_ras_late_init,
-- 
2.25.1

[PATCH 04/11] drm/amdgpu: Optimize amdgpu_mca_ras_late_init/amdgpu_mca_ras_fini function code

[yipechai]

Optimize amdgpu_mca_ras_late_init/amdgpu_mca_ras_fini function code.

Signed-off-by: yipechai <YiPeng.Chai@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c | 41 ++-----------------------
 drivers/gpu/drm/amd/amdgpu/mca_v3_0.c   |  6 ++++
 2 files changed, 8 insertions(+), 39 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c
index ad057d6b2c77..1c77fe7e9e68 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c
@@ -74,48 +74,11 @@ void amdgpu_mca_query_ras_error_count(struct amdgpu_device *adev,
 int amdgpu_mca_ras_late_init(struct amdgpu_device *adev,
 			     struct amdgpu_mca_ras *mca_dev)
 {
-	char sysfs_name[32] = {0};
-	int r;
-	struct ras_ih_if ih_info = {
-		.cb = NULL,
-	};
-	struct ras_fs_if fs_info= {
-		.sysfs_name = sysfs_name,
-	};
-
-	snprintf(sysfs_name, sizeof(sysfs_name), "%s_err_count",
-		mca_dev->ras->ras_block.ras_comm.name);
-
-	if (!mca_dev->ras_if) {
-		mca_dev->ras_if = kmalloc(sizeof(struct ras_common_if), GFP_KERNEL);
-		if (!mca_dev->ras_if)
-			return -ENOMEM;
-		mca_dev->ras_if->block = mca_dev->ras->ras_block.ras_comm.block;
-		mca_dev->ras_if->sub_block_index = mca_dev->ras->ras_block.ras_comm.sub_block_index;
-		mca_dev->ras_if->type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE;
-	}
-	ih_info.head = fs_info.head = *mca_dev->ras_if;
-	r = amdgpu_ras_late_init(adev, mca_dev->ras_if,
-				 &fs_info, &ih_info);
-	if (r || !amdgpu_ras_is_supported(adev, mca_dev->ras_if->block)) {
-		kfree(mca_dev->ras_if);
-		mca_dev->ras_if = NULL;
-	}
-
-	return r;
+	return amdgpu_ras_block_late_init(adev, mca_dev->ras_if);
 }
 
 void amdgpu_mca_ras_fini(struct amdgpu_device *adev,
 			 struct amdgpu_mca_ras *mca_dev)
 {
-	struct ras_ih_if ih_info = {
-		.cb = NULL,
-	};
-
-	if (!mca_dev->ras_if)
-		return;
-
-	amdgpu_ras_late_fini(adev, mca_dev->ras_if, &ih_info);
-	kfree(mca_dev->ras_if);
-	mca_dev->ras_if = NULL;
+	amdgpu_ras_block_late_fini(adev, mca_dev->ras_if);
 }
\ No newline at end of file
diff --git a/drivers/gpu/drm/amd/amdgpu/mca_v3_0.c b/drivers/gpu/drm/amd/amdgpu/mca_v3_0.c
index 386416378a82..a307f336f7ab 100644
--- a/drivers/gpu/drm/amd/amdgpu/mca_v3_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/mca_v3_0.c
@@ -71,6 +71,7 @@ struct amdgpu_mca_ras_block mca_v3_0_mp0_ras = {
 		.ras_comm = {
 			.block = AMDGPU_RAS_BLOCK__MCA,
 			.sub_block_index = AMDGPU_RAS_MCA_BLOCK__MP0,
+			.type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE,
 			.name = "mp0",
 		},
 		.hw_ops = &mca_v3_0_mp0_hw_ops,
@@ -108,6 +109,7 @@ struct amdgpu_mca_ras_block mca_v3_0_mp1_ras = {
 		.ras_comm = {
 			.block = AMDGPU_RAS_BLOCK__MCA,
 			.sub_block_index = AMDGPU_RAS_MCA_BLOCK__MP1,
+			.type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE,
 			.name = "mp1",
 		},
 		.hw_ops = &mca_v3_0_mp1_hw_ops,
@@ -145,6 +147,7 @@ struct amdgpu_mca_ras_block mca_v3_0_mpio_ras = {
 		.ras_comm = {
 			.block = AMDGPU_RAS_BLOCK__MCA,
 			.sub_block_index = AMDGPU_RAS_MCA_BLOCK__MPIO,
+			.type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE,
 			.name = "mpio",
 		},
 		.hw_ops = &mca_v3_0_mpio_hw_ops,
@@ -165,6 +168,9 @@ static void mca_v3_0_init(struct amdgpu_device *adev)
 	amdgpu_ras_register_ras_block(adev, &mca->mp0.ras->ras_block);
 	amdgpu_ras_register_ras_block(adev, &mca->mp1.ras->ras_block);
 	amdgpu_ras_register_ras_block(adev, &mca->mpio.ras->ras_block);
+	mca->mp0.ras_if = &mca_v3_0_mp0_ras.ras_block.ras_comm;
+	mca->mp1.ras_if = &mca_v3_0_mp1_ras.ras_block.ras_comm;
+	mca->mpio.ras_if = &mca_v3_0_mpio_ras.ras_block.ras_comm;
 }
 
 const struct amdgpu_mca_funcs mca_v3_0_funcs = {
-- 
2.25.1

[PATCH 05/11] drm/amdgpu: Optimize amdgpu_mmhub_ras_late_init/amdgpu_mmhub_ras_fini function code

[yipechai]

Optimize amdgpu_mmhub_ras_late_init/amdgpu_mmhub_ras_fini function code.

Signed-off-by: yipechai <YiPeng.Chai@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_mmhub.c | 37 ++---------------------
 drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c     |  2 ++
 2 files changed, 5 insertions(+), 34 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mmhub.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_mmhub.c
index f9b5472a75d7..2bdb4d8b7955 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mmhub.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mmhub.c
@@ -26,43 +26,12 @@
 
 int amdgpu_mmhub_ras_late_init(struct amdgpu_device *adev, void *ras_info)
 {
-	int r;
-	struct ras_ih_if ih_info = {
-		.cb = NULL,
-	};
-	struct ras_fs_if fs_info = {
-		.sysfs_name = "mmhub_err_count",
-	};
-
-	if (!adev->mmhub.ras_if) {
-		adev->mmhub.ras_if = kmalloc(sizeof(struct ras_common_if), GFP_KERNEL);
-		if (!adev->mmhub.ras_if)
-			return -ENOMEM;
-		adev->mmhub.ras_if->block = AMDGPU_RAS_BLOCK__MMHUB;
-		adev->mmhub.ras_if->type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE;
-		adev->mmhub.ras_if->sub_block_index = 0;
-	}
-	ih_info.head = fs_info.head = *adev->mmhub.ras_if;
-	r = amdgpu_ras_late_init(adev, adev->mmhub.ras_if,
-				 &fs_info, &ih_info);
-	if (r || !amdgpu_ras_is_supported(adev, adev->mmhub.ras_if->block)) {
-		kfree(adev->mmhub.ras_if);
-		adev->mmhub.ras_if = NULL;
-	}
-
-	return r;
+	return amdgpu_ras_block_late_init(adev, adev->mmhub.ras_if);
 }
 
 void amdgpu_mmhub_ras_fini(struct amdgpu_device *adev)
 {
 	if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__MMHUB) &&
-			adev->mmhub.ras_if) {
-		struct ras_common_if *ras_if = adev->mmhub.ras_if;
-		struct ras_ih_if ih_info = {
-			.cb = NULL,
-		};
-
-		amdgpu_ras_late_fini(adev, ras_if, &ih_info);
-		kfree(ras_if);
-	}
+			adev->mmhub.ras_if)
+		amdgpu_ras_block_late_fini(adev, adev->mmhub.ras_if);
 }
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
index b12fe6703f02..15958fd45f64 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
@@ -1282,6 +1282,8 @@ static void gmc_v9_0_set_mmhub_ras_funcs(struct amdgpu_device *adev)
 
 		strcpy(adev->mmhub.ras->ras_block.ras_comm.name, "mmhub");
 		adev->mmhub.ras->ras_block.ras_comm.block = AMDGPU_RAS_BLOCK__MMHUB;
+		adev->mmhub.ras->ras_block.ras_comm.type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE;
+		adev->mmhub.ras_if = &adev->mmhub.ras->ras_block.ras_comm;
 
 		/* If don't define special ras_late_init function, use default ras_late_init */
 		if (!adev->mmhub.ras->ras_block.ras_late_init)
-- 
2.25.1

[PATCH 06/11] drm/amdgpu: Optimize amdgpu_nbio_ras_late_init/amdgpu_nbio_ras_fini function code

[yipechai]

Optimize amdgpu_nbio_ras_late_init/amdgpu_nbio_ras_fini function code.

Signed-off-by: yipechai <YiPeng.Chai@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_nbio.c | 40 +++---------------------
 drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c  |  1 +
 drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c   |  1 +
 3 files changed, 7 insertions(+), 35 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_nbio.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_nbio.c
index 6ace2e390e77..89e61fdd3580 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_nbio.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_nbio.c
@@ -25,26 +25,9 @@
 int amdgpu_nbio_ras_late_init(struct amdgpu_device *adev, void *ras_info)
 {
 	int r;
-	struct ras_ih_if ih_info = {
-		.cb = NULL,
-	};
-	struct ras_fs_if fs_info = {
-		.sysfs_name = "pcie_bif_err_count",
-	};
-
-	if (!adev->nbio.ras_if) {
-		adev->nbio.ras_if = kmalloc(sizeof(struct ras_common_if), GFP_KERNEL);
-		if (!adev->nbio.ras_if)
-			return -ENOMEM;
-		adev->nbio.ras_if->block = AMDGPU_RAS_BLOCK__PCIE_BIF;
-		adev->nbio.ras_if->type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE;
-		adev->nbio.ras_if->sub_block_index = 0;
-	}
-	ih_info.head = fs_info.head = *adev->nbio.ras_if;
-	r = amdgpu_ras_late_init(adev, adev->nbio.ras_if,
-				 &fs_info, &ih_info);
+	r = amdgpu_ras_block_late_init(adev, adev->nbio.ras_if);
 	if (r)
-		goto free;
+		return r;
 
 	if (amdgpu_ras_is_supported(adev, adev->nbio.ras_if->block)) {
 		r = amdgpu_irq_get(adev, &adev->nbio.ras_controller_irq, 0);
@@ -53,30 +36,17 @@ int amdgpu_nbio_ras_late_init(struct amdgpu_device *adev, void *ras_info)
 		r = amdgpu_irq_get(adev, &adev->nbio.ras_err_event_athub_irq, 0);
 		if (r)
 			goto late_fini;
-	} else {
-		r = 0;
-		goto free;
 	}
 
 	return 0;
 late_fini:
-	amdgpu_ras_late_fini(adev, adev->nbio.ras_if, &ih_info);
-free:
-	kfree(adev->nbio.ras_if);
-	adev->nbio.ras_if = NULL;
+	amdgpu_ras_block_late_fini(adev, adev->nbio.ras_if);
 	return r;
 }
 
 void amdgpu_nbio_ras_fini(struct amdgpu_device *adev)
 {
 	if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__PCIE_BIF) &&
-			adev->nbio.ras_if) {
-		struct ras_common_if *ras_if = adev->nbio.ras_if;
-		struct ras_ih_if ih_info = {
-			.cb = NULL,
-		};
-
-		amdgpu_ras_late_fini(adev, ras_if, &ih_info);
-		kfree(ras_if);
-	}
+			adev->nbio.ras_if)
+		amdgpu_ras_block_late_fini(adev, adev->nbio.ras_if);
 }
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
index b7aed19db7e9..8a76a4e07659 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
@@ -2301,6 +2301,7 @@ int amdgpu_ras_init(struct amdgpu_device *adev)
 		if (!adev->gmc.xgmi.connected_to_cpu) {
 			adev->nbio.ras = &nbio_v7_4_ras;
 			amdgpu_ras_register_ras_block(adev, &adev->nbio.ras->ras_block);
+			adev->nbio.ras_if = &adev->nbio.ras->ras_block.ras_comm;
 		}
 		break;
 	default:
diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c b/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c
index c7cca87f1647..14768570c298 100644
--- a/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c
+++ b/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c
@@ -667,6 +667,7 @@ struct amdgpu_nbio_ras nbio_v7_4_ras = {
 		.ras_comm = {
 			.name = "pcie_bif",
 			.block = AMDGPU_RAS_BLOCK__PCIE_BIF,
+			.type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE,
 		},
 		.hw_ops = &nbio_v7_4_ras_hw_ops,
 		.ras_late_init = amdgpu_nbio_ras_late_init,
-- 
2.25.1

[PATCH 07/11] drm/amdgpu: Optimize amdgpu_sdma_ras_late_init/amdgpu_sdma_ras_fini function code

[yipechai]

Optimize amdgpu_sdma_ras_late_init/amdgpu_sdma_ras_fini function code.

Signed-off-by: yipechai <YiPeng.Chai@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.c | 46 +++---------------------
 drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c   | 12 ++++---
 2 files changed, 13 insertions(+), 45 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.c
index 65debb65a5df..242a7b4dcad9 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.c
@@ -90,28 +90,10 @@ int amdgpu_sdma_ras_late_init(struct amdgpu_device *adev,
 			      void *ras_ih_info)
 {
 	int r, i;
-	struct ras_ih_if *ih_info = (struct ras_ih_if *)ras_ih_info;
-	struct ras_fs_if fs_info = {
-		.sysfs_name = "sdma_err_count",
-	};
-
-	if (!ih_info)
-		return -EINVAL;
 
-	if (!adev->sdma.ras_if) {
-		adev->sdma.ras_if = kmalloc(sizeof(struct ras_common_if), GFP_KERNEL);
-		if (!adev->sdma.ras_if)
-			return -ENOMEM;
-		adev->sdma.ras_if->block = AMDGPU_RAS_BLOCK__SDMA;
-		adev->sdma.ras_if->type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE;
-		adev->sdma.ras_if->sub_block_index = 0;
-	}
-	fs_info.head = ih_info->head = *adev->sdma.ras_if;
-
-	r = amdgpu_ras_late_init(adev, adev->sdma.ras_if,
-				 &fs_info, ih_info);
+	r = amdgpu_ras_block_late_init(adev, adev->sdma.ras_if);
 	if (r)
-		goto free;
+		return r;
 
 	if (amdgpu_ras_is_supported(adev, adev->sdma.ras_if->block)) {
 		for (i = 0; i < adev->sdma.num_instances; i++) {
@@ -120,38 +102,20 @@ int amdgpu_sdma_ras_late_init(struct amdgpu_device *adev,
 			if (r)
 				goto late_fini;
 		}
-	} else {
-		r = 0;
-		goto free;
 	}
 
 	return 0;
 
 late_fini:
-	amdgpu_ras_late_fini(adev, adev->sdma.ras_if, ih_info);
-free:
-	kfree(adev->sdma.ras_if);
-	adev->sdma.ras_if = NULL;
+	amdgpu_ras_block_late_fini(adev, adev->sdma.ras_if);
 	return r;
 }
 
 void amdgpu_sdma_ras_fini(struct amdgpu_device *adev)
 {
 	if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__SDMA) &&
-			adev->sdma.ras_if) {
-		struct ras_common_if *ras_if = adev->sdma.ras_if;
-		struct ras_ih_if ih_info = {
-			.head = *ras_if,
-			/* the cb member will not be used by
-			 * amdgpu_ras_interrupt_remove_handler, init it only
-			 * to cheat the check in ras_late_fini
-			 */
-			.cb = amdgpu_sdma_process_ras_data_cb,
-		};
-
-		amdgpu_ras_late_fini(adev, ras_if, &ih_info);
-		kfree(ras_if);
-	}
+			adev->sdma.ras_if)
+		amdgpu_ras_block_late_fini(adev, adev->sdma.ras_if);
 }
 
 int amdgpu_sdma_process_ras_data_cb(struct amdgpu_device *adev,
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c
index 8b0a8587dd36..82a31dfa8c21 100644
--- a/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c
@@ -1885,9 +1885,6 @@ static int sdma_v4_0_process_ras_data_cb(struct amdgpu_device *adev,
 static int sdma_v4_0_late_init(void *handle)
 {
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-	struct ras_ih_if ih_info = {
-		.cb = sdma_v4_0_process_ras_data_cb,
-	};
 
 	sdma_v4_0_setup_ulv(adev);
 
@@ -1898,7 +1895,7 @@ static int sdma_v4_0_late_init(void *handle)
 	}
 
 	if (adev->sdma.ras && adev->sdma.ras->ras_block.ras_late_init)
-		return adev->sdma.ras->ras_block.ras_late_init(adev, &ih_info);
+		return adev->sdma.ras->ras_block.ras_late_init(adev, NULL);
 	else
 		return 0;
 }
@@ -2794,6 +2791,7 @@ const struct amdgpu_ras_block_hw_ops sdma_v4_0_ras_hw_ops = {
 static struct amdgpu_sdma_ras sdma_v4_0_ras = {
 	.ras_block = {
 		.hw_ops = &sdma_v4_0_ras_hw_ops,
+		.ras_cb = sdma_v4_0_process_ras_data_cb,
 	},
 };
 
@@ -2816,6 +2814,8 @@ static void sdma_v4_0_set_ras_funcs(struct amdgpu_device *adev)
 
 		strcpy(adev->sdma.ras->ras_block.ras_comm.name, "sdma");
 		adev->sdma.ras->ras_block.ras_comm.block = AMDGPU_RAS_BLOCK__SDMA;
+		adev->sdma.ras->ras_block.ras_comm.type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE;
+		adev->sdma.ras_if = &adev->sdma.ras->ras_block.ras_comm;
 
 		/* If don't define special ras_late_init function, use default ras_late_init */
 		if (!adev->sdma.ras->ras_block.ras_late_init)
@@ -2824,6 +2824,10 @@ static void sdma_v4_0_set_ras_funcs(struct amdgpu_device *adev)
 		/* If don't define special ras_fini function, use default ras_fini */
 		if (!adev->sdma.ras->ras_block.ras_fini)
 			adev->sdma.ras->ras_block.ras_fini = amdgpu_sdma_ras_fini;
+
+		/* If not defined special ras_cb function, use default ras_cb */
+		if (!adev->sdma.ras->ras_block.ras_cb)
+			adev->sdma.ras->ras_block.ras_cb = amdgpu_sdma_process_ras_data_cb;
 	}
 }
 
-- 
2.25.1

[PATCH 08/11] drm/amdgpu: Optimize amdgpu_umc_ras_late_init/amdgpu_umc_ras_fini function code

[yipechai]

Optimize amdgpu_umc_ras_late_init/amdgpu_umc_ras_fini function code.

Signed-off-by: yipechai <YiPeng.Chai@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c | 44 ++++---------------------
 drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h |  4 +++
 drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c   |  6 ++++
 3 files changed, 16 insertions(+), 38 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c
index ff7805beda38..9f1406e1a48a 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c
@@ -129,7 +129,7 @@ int amdgpu_umc_poison_handler(struct amdgpu_device *adev,
 	return ret;
 }
 
-static int amdgpu_umc_process_ras_data_cb(struct amdgpu_device *adev,
+int amdgpu_umc_process_ras_data_cb(struct amdgpu_device *adev,
 		void *ras_error_status,
 		struct amdgpu_iv_entry *entry)
 {
@@ -139,36 +139,15 @@ static int amdgpu_umc_process_ras_data_cb(struct amdgpu_device *adev,
 int amdgpu_umc_ras_late_init(struct amdgpu_device *adev, void *ras_info)
 {
 	int r;
-	struct ras_fs_if fs_info = {
-		.sysfs_name = "umc_err_count",
-	};
-	struct ras_ih_if ih_info = {
-		.cb = amdgpu_umc_process_ras_data_cb,
-	};
 
-	if (!adev->umc.ras_if) {
-		adev->umc.ras_if =
-			kmalloc(sizeof(struct ras_common_if), GFP_KERNEL);
-		if (!adev->umc.ras_if)
-			return -ENOMEM;
-		adev->umc.ras_if->block = AMDGPU_RAS_BLOCK__UMC;
-		adev->umc.ras_if->type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE;
-		adev->umc.ras_if->sub_block_index = 0;
-	}
-	ih_info.head = fs_info.head = *adev->umc.ras_if;
-
-	r = amdgpu_ras_late_init(adev, adev->umc.ras_if,
-				 &fs_info, &ih_info);
+	r = amdgpu_ras_block_late_init(adev, adev->umc.ras_if);
 	if (r)
-		goto free;
+		return r;
 
 	if (amdgpu_ras_is_supported(adev, adev->umc.ras_if->block)) {
 		r = amdgpu_irq_get(adev, &adev->gmc.ecc_irq, 0);
 		if (r)
 			goto late_fini;
-	} else {
-		r = 0;
-		goto free;
 	}
 
 	/* ras init of specific umc version */
@@ -179,26 +158,15 @@ int amdgpu_umc_ras_late_init(struct amdgpu_device *adev, void *ras_info)
 	return 0;
 
 late_fini:
-	amdgpu_ras_late_fini(adev, adev->umc.ras_if, &ih_info);
-free:
-	kfree(adev->umc.ras_if);
-	adev->umc.ras_if = NULL;
+	amdgpu_ras_block_late_fini(adev, adev->umc.ras_if);
 	return r;
 }
 
 void amdgpu_umc_ras_fini(struct amdgpu_device *adev)
 {
 	if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__UMC) &&
-			adev->umc.ras_if) {
-		struct ras_common_if *ras_if = adev->umc.ras_if;
-		struct ras_ih_if ih_info = {
-			.head = *ras_if,
-			.cb = amdgpu_umc_process_ras_data_cb,
-		};
-
-		amdgpu_ras_late_fini(adev, ras_if, &ih_info);
-		kfree(ras_if);
-	}
+			adev->umc.ras_if)
+		amdgpu_ras_block_late_fini(adev, adev->umc.ras_if);
 }
 
 int amdgpu_umc_process_ecc_irq(struct amdgpu_device *adev,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h
index 4db0526d0be4..ec15b3640399 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h
@@ -85,4 +85,8 @@ void amdgpu_umc_fill_error_record(struct ras_err_data *err_data,
 		uint64_t retired_page,
 		uint32_t channel_index,
 		uint32_t umc_inst);
+
+int amdgpu_umc_process_ras_data_cb(struct amdgpu_device *adev,
+		void *ras_error_status,
+		struct amdgpu_iv_entry *entry);
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
index 15958fd45f64..94095b965e2c 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
@@ -1234,6 +1234,8 @@ static void gmc_v9_0_set_umc_funcs(struct amdgpu_device *adev)
 
 		strcpy(adev->umc.ras->ras_block.ras_comm.name, "umc");
 		adev->umc.ras->ras_block.ras_comm.block = AMDGPU_RAS_BLOCK__UMC;
+		adev->umc.ras->ras_block.ras_comm.type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE;
+		adev->umc.ras_if = &adev->umc.ras->ras_block.ras_comm;
 
 		/* If don't define special ras_late_init function, use default ras_late_init */
 		if (!adev->umc.ras->ras_block.ras_late_init)
@@ -1242,6 +1244,10 @@ static void gmc_v9_0_set_umc_funcs(struct amdgpu_device *adev)
 		/* If don't define special ras_fini function, use default ras_fini */
 		if (!adev->umc.ras->ras_block.ras_fini)
 				adev->umc.ras->ras_block.ras_fini = amdgpu_umc_ras_fini;
+
+		/* If not defined special ras_cb function, use default ras_cb */
+		if (!adev->umc.ras->ras_block.ras_cb)
+			adev->umc.ras->ras_block.ras_cb = amdgpu_umc_process_ras_data_cb;
 	}
 }
 
-- 
2.25.1

[PATCH 09/11] drm/amdgpu: Optimize amdgpu_xgmi_ras_late_init/amdgpu_xgmi_ras_fini function code

[yipechai]

Optimize amdgpu_xgmi_ras_late_init/amdgpu_xgmi_ras_fini function code.

Signed-off-by: yipechai <YiPeng.Chai@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c  |  1 +
 drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c | 40 +++---------------------
 2 files changed, 6 insertions(+), 35 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c
index d426de48d299..956cc994ca7d 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c
@@ -441,6 +441,7 @@ int amdgpu_gmc_ras_early_init(struct amdgpu_device *adev)
 	if (!adev->gmc.xgmi.connected_to_cpu) {
 		adev->gmc.xgmi.ras = &xgmi_ras;
 		amdgpu_ras_register_ras_block(adev, &adev->gmc.xgmi.ras->ras_block);
+		adev->gmc.xgmi.ras_if = &adev->gmc.xgmi.ras->ras_block.ras_comm;
 	}
 
 	return 0;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c
index 15707af89212..a785b1e088cd 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c
@@ -734,51 +734,20 @@ int amdgpu_xgmi_remove_device(struct amdgpu_device *adev)
 
 static int amdgpu_xgmi_ras_late_init(struct amdgpu_device *adev, void *ras_info)
 {
-	int r;
-	struct ras_ih_if ih_info = {
-		.cb = NULL,
-	};
-	struct ras_fs_if fs_info = {
-		.sysfs_name = "xgmi_wafl_err_count",
-	};
-
 	if (!adev->gmc.xgmi.supported ||
 	    adev->gmc.xgmi.num_physical_nodes == 0)
 		return 0;
 
 	adev->gmc.xgmi.ras->ras_block.hw_ops->reset_ras_error_count(adev);
 
-	if (!adev->gmc.xgmi.ras_if) {
-		adev->gmc.xgmi.ras_if = kmalloc(sizeof(struct ras_common_if), GFP_KERNEL);
-		if (!adev->gmc.xgmi.ras_if)
-			return -ENOMEM;
-		adev->gmc.xgmi.ras_if->block = AMDGPU_RAS_BLOCK__XGMI_WAFL;
-		adev->gmc.xgmi.ras_if->type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE;
-		adev->gmc.xgmi.ras_if->sub_block_index = 0;
-	}
-	ih_info.head = fs_info.head = *adev->gmc.xgmi.ras_if;
-	r = amdgpu_ras_late_init(adev, adev->gmc.xgmi.ras_if,
-				 &fs_info, &ih_info);
-	if (r || !amdgpu_ras_is_supported(adev, adev->gmc.xgmi.ras_if->block)) {
-		kfree(adev->gmc.xgmi.ras_if);
-		adev->gmc.xgmi.ras_if = NULL;
-	}
-
-	return r;
+	return amdgpu_ras_block_late_init(adev, adev->gmc.xgmi.ras_if);
 }
 
 static void amdgpu_xgmi_ras_fini(struct amdgpu_device *adev)
 {
 	if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__XGMI_WAFL) &&
-			adev->gmc.xgmi.ras_if) {
-		struct ras_common_if *ras_if = adev->gmc.xgmi.ras_if;
-		struct ras_ih_if ih_info = {
-			.cb = NULL,
-		};
-
-		amdgpu_ras_late_fini(adev, ras_if, &ih_info);
-		kfree(ras_if);
-	}
+			adev->gmc.xgmi.ras_if)
+		amdgpu_ras_block_late_fini(adev, adev->gmc.xgmi.ras_if);
 }
 
 uint64_t amdgpu_xgmi_get_relative_phy_addr(struct amdgpu_device *adev,
@@ -982,8 +951,9 @@ struct amdgpu_ras_block_hw_ops  xgmi_ras_hw_ops = {
 struct amdgpu_xgmi_ras xgmi_ras = {
 	.ras_block = {
 		.ras_comm = {
-			.name = "xgmi",
+			.name = "xgmi_wafl",
 			.block = AMDGPU_RAS_BLOCK__XGMI_WAFL,
+			.type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE,
 		},
 		.hw_ops = &xgmi_ras_hw_ops,
 		.ras_late_init = amdgpu_xgmi_ras_late_init,
-- 
2.25.1

[PATCH 10/11] drm/amdgpu: Optimize operating sysfs and interrupt function interface in amdgpu_ras.c

[yipechai]

In order to reduce redundant struct conversion, modify
operating sysfs and interrupt function interface parameters.

Signed-off-by: yipechai <YiPeng.Chai@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 37 ++++++++++++-------------
 drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h |  6 ++--
 2 files changed, 21 insertions(+), 22 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
index 8a76a4e07659..9f8f596663ab 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
@@ -1276,18 +1276,17 @@ static int amdgpu_ras_sysfs_remove_feature_node(struct amdgpu_device *adev)
 }
 
 int amdgpu_ras_sysfs_create(struct amdgpu_device *adev,
-		struct ras_fs_if *head)
+		struct ras_common_if *head)
 {
-	struct ras_manager *obj = amdgpu_ras_find_obj(adev, &head->head);
+	struct ras_manager *obj = amdgpu_ras_find_obj(adev, head);
 
 	if (!obj || obj->attr_inuse)
 		return -EINVAL;
 
 	get_obj(obj);
 
-	memcpy(obj->fs_data.sysfs_name,
-			head->sysfs_name,
-			sizeof(obj->fs_data.sysfs_name));
+	snprintf(obj->fs_data.sysfs_name, sizeof(obj->fs_data.sysfs_name),
+		"%s_err_count", head->name);
 
 	obj->sysfs_attr = (struct device_attribute){
 		.attr = {
@@ -1594,9 +1593,9 @@ int amdgpu_ras_interrupt_dispatch(struct amdgpu_device *adev,
 }
 
 int amdgpu_ras_interrupt_remove_handler(struct amdgpu_device *adev,
-		struct ras_ih_if *info)
+		struct ras_common_if *head)
 {
-	struct ras_manager *obj = amdgpu_ras_find_obj(adev, &info->head);
+	struct ras_manager *obj = amdgpu_ras_find_obj(adev, head);
 	struct ras_ih_data *data;
 
 	if (!obj)
@@ -1616,24 +1615,27 @@ int amdgpu_ras_interrupt_remove_handler(struct amdgpu_device *adev,
 }
 
 int amdgpu_ras_interrupt_add_handler(struct amdgpu_device *adev,
-		struct ras_ih_if *info)
+		struct ras_common_if *head)
 {
-	struct ras_manager *obj = amdgpu_ras_find_obj(adev, &info->head);
+	struct ras_manager *obj = amdgpu_ras_find_obj(adev, head);
 	struct ras_ih_data *data;
+	struct amdgpu_ras_block_object *ras_obj;
 
 	if (!obj) {
 		/* in case we registe the IH before enable ras feature */
-		obj = amdgpu_ras_create_obj(adev, &info->head);
+		obj = amdgpu_ras_create_obj(adev, head);
 		if (!obj)
 			return -EINVAL;
 	} else
 		get_obj(obj);
 
+	ras_obj = container_of(head, struct amdgpu_ras_block_object, ras_comm);
+
 	data = &obj->ih_data;
 	/* add the callback.etc */
 	*data = (struct ras_ih_data) {
 		.inuse = 0,
-		.cb = info->cb,
+		.cb = ras_obj->ras_cb,
 		.element_size = sizeof(struct amdgpu_iv_entry),
 		.rptr = 0,
 		.wptr = 0,
@@ -1662,10 +1664,7 @@ static int amdgpu_ras_interrupt_remove_all(struct amdgpu_device *adev)
 	struct ras_manager *obj, *tmp;
 
 	list_for_each_entry_safe(obj, tmp, &con->head, node) {
-		struct ras_ih_if info = {
-			.head = obj->head,
-		};
-		amdgpu_ras_interrupt_remove_handler(adev, &info);
+		amdgpu_ras_interrupt_remove_handler(adev, &obj->head);
 	}
 
 	return 0;
@@ -2431,12 +2430,12 @@ int amdgpu_ras_late_init(struct amdgpu_device *adev,
 		return 0;
 
 	if (ih_info->cb) {
-		r = amdgpu_ras_interrupt_add_handler(adev, ih_info);
+		r = amdgpu_ras_interrupt_add_handler(adev, ras_block);
 		if (r)
 			goto interrupt;
 	}
 
-	r = amdgpu_ras_sysfs_create(adev, fs_info);
+	r = amdgpu_ras_sysfs_create(adev, ras_block);
 	if (r)
 		goto sysfs;
 
@@ -2452,7 +2451,7 @@ int amdgpu_ras_late_init(struct amdgpu_device *adev,
 	amdgpu_ras_sysfs_remove(adev, ras_block);
 sysfs:
 	if (ih_info->cb)
-		amdgpu_ras_interrupt_remove_handler(adev, ih_info);
+		amdgpu_ras_interrupt_remove_handler(adev, ras_block);
 interrupt:
 	amdgpu_ras_feature_enable(adev, ras_block, 0);
 	return r;
@@ -2485,7 +2484,7 @@ void amdgpu_ras_late_fini(struct amdgpu_device *adev,
 
 	amdgpu_ras_sysfs_remove(adev, ras_block);
 	if (ih_info->cb)
-		amdgpu_ras_interrupt_remove_handler(adev, ih_info);
+		amdgpu_ras_interrupt_remove_handler(adev, &ih_info->head);
 }
 
 void amdgpu_ras_block_late_fini(struct amdgpu_device *adev,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h
index 8b94b556baf6..ae8741ac526f 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h
@@ -619,7 +619,7 @@ int amdgpu_ras_feature_enable_on_boot(struct amdgpu_device *adev,
 		struct ras_common_if *head, bool enable);
 
 int amdgpu_ras_sysfs_create(struct amdgpu_device *adev,
-		struct ras_fs_if *head);
+		struct ras_common_if *head);
 
 int amdgpu_ras_sysfs_remove(struct amdgpu_device *adev,
 		struct ras_common_if *head);
@@ -636,10 +636,10 @@ int amdgpu_ras_error_inject(struct amdgpu_device *adev,
 		struct ras_inject_if *info);
 
 int amdgpu_ras_interrupt_add_handler(struct amdgpu_device *adev,
-		struct ras_ih_if *info);
+		struct ras_common_if *head);
 
 int amdgpu_ras_interrupt_remove_handler(struct amdgpu_device *adev,
-		struct ras_ih_if *info);
+		struct ras_common_if *head);
 
 int amdgpu_ras_interrupt_dispatch(struct amdgpu_device *adev,
 		struct ras_dispatch_if *info);
-- 
2.25.1

[PATCH 11/11] drm/amdgpu: Merge amdgpu_ras_late_init/amdgpu_ras_late_fini to amdgpu_ras_block_late_init/amdgpu_ras_block_late_fini

[yipechai]

1. Merge amdgpu_ras_late_init to
   amdgpu_ras_block_late_init.
2. Remove amdgpu_ras_late_init since no ras block
   calls amdgpu_ras_late_init.
3. Merge amdgpu_ras_late_fini to
   amdgpu_ras_block_late_fini.
4. Remove amdgpu_ras_late_fini since no ras block
   calls amdgpu_ras_late_fini.

Signed-off-by: yipechai <YiPeng.Chai@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 53 +++++--------------------
 drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h |  8 ----
 2 files changed, 11 insertions(+), 50 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
index 9f8f596663ab..1aff88fcea76 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
@@ -2397,11 +2397,10 @@ bool amdgpu_ras_is_poison_mode_supported(struct amdgpu_device *adev)
 }
 
 /* helper function to handle common stuff in ip late init phase */
-int amdgpu_ras_late_init(struct amdgpu_device *adev,
-			 struct ras_common_if *ras_block,
-			 struct ras_fs_if *fs_info,
-			 struct ras_ih_if *ih_info)
+int amdgpu_ras_block_late_init(struct amdgpu_device *adev,
+			 struct ras_common_if *ras_block)
 {
+	struct amdgpu_ras_block_object *ras_obj;
 	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
 	unsigned long ue_count, ce_count;
 	int r;
@@ -2429,7 +2428,8 @@ int amdgpu_ras_late_init(struct amdgpu_device *adev,
 	if (adev->in_suspend || amdgpu_in_reset(adev))
 		return 0;
 
-	if (ih_info->cb) {
+	ras_obj = container_of(ras_block, struct amdgpu_ras_block_object, ras_comm);
+	if (ras_obj->ras_cb) {
 		r = amdgpu_ras_interrupt_add_handler(adev, ras_block);
 		if (r)
 			goto interrupt;
@@ -2450,57 +2450,26 @@ int amdgpu_ras_late_init(struct amdgpu_device *adev,
 cleanup:
 	amdgpu_ras_sysfs_remove(adev, ras_block);
 sysfs:
-	if (ih_info->cb)
+	if (ras_obj->ras_cb)
 		amdgpu_ras_interrupt_remove_handler(adev, ras_block);
 interrupt:
 	amdgpu_ras_feature_enable(adev, ras_block, 0);
 	return r;
 }
 
-int amdgpu_ras_block_late_init(struct amdgpu_device *adev,
-			struct ras_common_if *ras_block)
-{
-	char sysfs_name[32];
-	struct ras_ih_if ih_info;
-	struct ras_fs_if fs_info;
-	struct amdgpu_ras_block_object *obj;
-
-	obj = container_of(ras_block, struct amdgpu_ras_block_object, ras_comm);
-	ih_info.cb = obj->ras_cb;
-	ih_info.head = *ras_block;
-	snprintf(sysfs_name, sizeof(sysfs_name), "%s_err_count", ras_block->name);
-	fs_info.sysfs_name = (const char *)sysfs_name;
-	fs_info.head = *ras_block;
-	return amdgpu_ras_late_init(adev, ras_block, &fs_info, &ih_info);
-}
-
 /* helper function to remove ras fs node and interrupt handler */
-void amdgpu_ras_late_fini(struct amdgpu_device *adev,
-			  struct ras_common_if *ras_block,
-			  struct ras_ih_if *ih_info)
-{
-	if (!ras_block || !ih_info)
-		return;
-
-	amdgpu_ras_sysfs_remove(adev, ras_block);
-	if (ih_info->cb)
-		amdgpu_ras_interrupt_remove_handler(adev, &ih_info->head);
-}
-
 void amdgpu_ras_block_late_fini(struct amdgpu_device *adev,
 			  struct ras_common_if *ras_block)
 {
-	struct ras_ih_if ih_info;
-	struct amdgpu_ras_block_object *obj;
-
+	struct amdgpu_ras_block_object *ras_obj;
 	if (!ras_block)
 		return;
 
-	obj = container_of(ras_block, struct amdgpu_ras_block_object, ras_comm);
-	ih_info.head = *ras_block;
-	ih_info.cb = obj->ras_cb;
+	amdgpu_ras_sysfs_remove(adev, ras_block);
 
-	amdgpu_ras_late_fini(adev, ras_block, &ih_info);
+	ras_obj = container_of(ras_block, struct amdgpu_ras_block_object, ras_comm);
+	if (ras_obj->ras_cb)
+		amdgpu_ras_interrupt_remove_handler(adev, ras_block);
 }
 
 /* do some init work after IP late init as dependence.
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h
index ae8741ac526f..5de567c6a8f7 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h
@@ -597,18 +597,10 @@ amdgpu_ras_error_to_ta(enum amdgpu_ras_error_type error) {
 int amdgpu_ras_init(struct amdgpu_device *adev);
 int amdgpu_ras_fini(struct amdgpu_device *adev);
 int amdgpu_ras_pre_fini(struct amdgpu_device *adev);
-int amdgpu_ras_late_init(struct amdgpu_device *adev,
-			 struct ras_common_if *ras_block,
-			 struct ras_fs_if *fs_info,
-			 struct ras_ih_if *ih_info);
 
 int amdgpu_ras_block_late_init(struct amdgpu_device *adev,
 			struct ras_common_if *ras_block);
 
-void amdgpu_ras_late_fini(struct amdgpu_device *adev,
-			  struct ras_common_if *ras_block,
-			  struct ras_ih_if *ih_info);
-
 void amdgpu_ras_block_late_fini(struct amdgpu_device *adev,
 			  struct ras_common_if *ras_block);
 
-- 
2.25.1

RE: [PATCH 03/11] drm/amdgpu: Optimize amdgpu_hdp_ras_late_init/amdgpu_hdp_ras_fini function code

[Zhou1, Tao]

[AMD Official Use Only]



> -----Original Message-----
> From: Chai, Thomas <YiPeng.Chai@amd.com>
> Sent: Wednesday, February 9, 2022 1:57 PM
> To: amd-gfx@lists.freedesktop.org
> Cc: Chai, Thomas <YiPeng.Chai@amd.com>; Zhang, Hawking
> <Hawking.Zhang@amd.com>; Zhou1, Tao <Tao.Zhou1@amd.com>; Clements,
> John <John.Clements@amd.com>; Chai, Thomas <YiPeng.Chai@amd.com>
> Subject: [PATCH 03/11] drm/amdgpu: Optimize
> amdgpu_hdp_ras_late_init/amdgpu_hdp_ras_fini function code
> 
> Optimize amdgpu_hdp_ras_late_init/amdgpu_hdp_ras_fini function code.
> 
> Signed-off-by: yipechai <YiPeng.Chai@amd.com>
> ---
>  drivers/gpu/drm/amd/amdgpu/amdgpu_hdp.c | 37 ++-----------------------
>  drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c   |  1 +
>  drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c   |  1 +
>  3 files changed, 5 insertions(+), 34 deletions(-)
> 
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_hdp.c
> b/drivers/gpu/drm/amd/amdgpu/amdgpu_hdp.c
> index 518966a26130..21a5f884dd2a 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_hdp.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_hdp.c
> @@ -26,43 +26,12 @@
> 
>  int amdgpu_hdp_ras_late_init(struct amdgpu_device *adev, void *ras_info)  {
> -	int r;
> -	struct ras_ih_if ih_info = {
> -		.cb = NULL,
> -	};
> -	struct ras_fs_if fs_info = {
> -		.sysfs_name = "hdp_err_count",
> -	};
> -
> -	if (!adev->hdp.ras_if) {
> -		adev->hdp.ras_if = kmalloc(sizeof(struct ras_common_if),
> GFP_KERNEL);
> -		if (!adev->hdp.ras_if)
> -			return -ENOMEM;
> -		adev->hdp.ras_if->block = AMDGPU_RAS_BLOCK__HDP;
> -		adev->hdp.ras_if->type =
> AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE;
> -		adev->hdp.ras_if->sub_block_index = 0;
> -	}
> -	ih_info.head = fs_info.head = *adev->hdp.ras_if;
> -	r = amdgpu_ras_late_init(adev, adev->hdp.ras_if,
> -				 &fs_info, &ih_info);
> -	if (r || !amdgpu_ras_is_supported(adev, adev->hdp.ras_if->block)) {
> -		kfree(adev->hdp.ras_if);
> -		adev->hdp.ras_if = NULL;
> -	}
> -
> -	return r;
> +	return amdgpu_ras_block_late_init(adev, adev->hdp.ras_if);
>  }
> 
>  void amdgpu_hdp_ras_fini(struct amdgpu_device *adev)  {
>  	if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__HDP) &&
> -	    adev->hdp.ras_if) {
> -		struct ras_common_if *ras_if = adev->hdp.ras_if;
> -		struct ras_ih_if ih_info = {
> -			.cb = NULL,
> -		};
> -
> -		amdgpu_ras_late_fini(adev, ras_if, &ih_info);
> -		kfree(ras_if);
> -	}
> +	    adev->hdp.ras_if)
> +		amdgpu_ras_block_late_fini(adev, adev->hdp.ras_if);
>  }
[Tao]: Since hdp_ras_late_init/fini are simple wrapper, can we remove them and call amdgpu_ras_block_late_init/fini directly?
The same comment to other blocks.

> diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
> b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
> index af873c99d5e4..b12fe6703f02 100644
> --- a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
> +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
> @@ -1302,6 +1302,7 @@ static void gmc_v9_0_set_hdp_ras_funcs(struct
> amdgpu_device *adev)  {
>  	adev->hdp.ras = &hdp_v4_0_ras;
>  	amdgpu_ras_register_ras_block(adev, &adev->hdp.ras->ras_block);
> +	adev->hdp.ras_if = &adev->hdp.ras->ras_block.ras_comm;
>  }
> 
>  static void gmc_v9_0_set_mca_funcs(struct amdgpu_device *adev) diff --git
> a/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c
> b/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c
> index 503c292b321e..a9ed4232cdeb 100644
> --- a/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c
> +++ b/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c
> @@ -160,6 +160,7 @@ struct amdgpu_hdp_ras hdp_v4_0_ras = {
>  		.ras_comm = {
>  			.name = "hdp",
>  			.block = AMDGPU_RAS_BLOCK__HDP,
> +			.type =
> AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE,
>  		},
>  		.hw_ops = &hdp_v4_0_ras_hw_ops,
>  		.ras_late_init = amdgpu_hdp_ras_late_init,
> --
> 2.25.1

RE: [PATCH 01/11] drm/amdgpu: Optimize xxx_ras_late_init/xxx_ras_late_fini for each ras block

[Zhou1, Tao]

[AMD Official Use Only]



> -----Original Message-----
> From: Chai, Thomas <YiPeng.Chai@amd.com>
> Sent: Wednesday, February 9, 2022 1:57 PM
> To: amd-gfx@lists.freedesktop.org
> Cc: Chai, Thomas <YiPeng.Chai@amd.com>; Zhang, Hawking
> <Hawking.Zhang@amd.com>; Zhou1, Tao <Tao.Zhou1@amd.com>; Clements,
> John <John.Clements@amd.com>; Chai, Thomas <YiPeng.Chai@amd.com>
> Subject: [PATCH 01/11] drm/amdgpu: Optimize
> xxx_ras_late_init/xxx_ras_late_fini for each ras block
> 
> 1. Define amdgpu_ras_block_late_init to create sysfs nodes
>    and interrupt handles.
> 2. Define amdgpu_ras_block_late_fini to remove sysfs nodes
>    and interrupt handles.
> 3. Replace ras block variable members in struct
>    amdgpu_ras_block_object with struct ras_common_if, which
>    can makes it easy to associate each ras block instance
[Tao]: makes -> make

>    with each ras block functional interface.
> 4. Add .ras_cb to struct amdgpu_ras_block_object.
> 5. Change each ras block to fit for the changement of struct
>    amdgpu_ras_block_object.
> 
> Signed-off-by: yipechai <YiPeng.Chai@amd.com>
> ---
>  drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c  |  7 +++--
> drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c  | 35
> +++++++++++++++++++++++-  drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h  |
> 15 ++++++----  drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c |  6 ++--
>  drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c    |  4 +--
>  drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c   |  4 +--
>  drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c    |  8 +++---
>  drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c    |  6 ++--
>  drivers/gpu/drm/amd/amdgpu/mca_v3_0.c    | 28 +++++++++++--------
>  drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c   |  6 ++--
>  drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c   |  4 +--
>  11 files changed, 86 insertions(+), 37 deletions(-)
> 
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c
> b/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c
> index 52a60c2316a2..ad057d6b2c77 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c
> @@ -83,14 +83,15 @@ int amdgpu_mca_ras_late_init(struct amdgpu_device
> *adev,
>  		.sysfs_name = sysfs_name,
>  	};
> 
> -	snprintf(sysfs_name, sizeof(sysfs_name), "%s_err_count", mca_dev-
> >ras->ras_block.name);
> +	snprintf(sysfs_name, sizeof(sysfs_name), "%s_err_count",
> +		mca_dev->ras->ras_block.ras_comm.name);
> 
>  	if (!mca_dev->ras_if) {
>  		mca_dev->ras_if = kmalloc(sizeof(struct ras_common_if),
> GFP_KERNEL);
>  		if (!mca_dev->ras_if)
>  			return -ENOMEM;
> -		mca_dev->ras_if->block = mca_dev->ras->ras_block.block;
> -		mca_dev->ras_if->sub_block_index = mca_dev->ras-
> >ras_block.sub_block_index;
> +		mca_dev->ras_if->block = mca_dev->ras-
> >ras_block.ras_comm.block;
> +		mca_dev->ras_if->sub_block_index =
> +mca_dev->ras->ras_block.ras_comm.sub_block_index;
>  		mca_dev->ras_if->type =
> AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE;
>  	}
>  	ih_info.head = fs_info.head = *mca_dev->ras_if; diff --git
> a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
> b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
> index 5934326b9db3..b7aed19db7e9 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
> @@ -877,7 +877,7 @@ static int amdgpu_ras_block_match_default(struct
> amdgpu_ras_block_object *block_
>  	if (!block_obj)
>  		return -EINVAL;
> 
> -	if (block_obj->block == block)
> +	if (block_obj->ras_comm.block == block)
>  		return 0;
> 
>  	return -EINVAL;
> @@ -2457,6 +2457,23 @@ int amdgpu_ras_late_init(struct amdgpu_device
> *adev,
>  	return r;
>  }
> 
> +int amdgpu_ras_block_late_init(struct amdgpu_device *adev,
> +			struct ras_common_if *ras_block)
> +{
> +	char sysfs_name[32];
> +	struct ras_ih_if ih_info;
> +	struct ras_fs_if fs_info;
> +	struct amdgpu_ras_block_object *obj;
> +
> +	obj = container_of(ras_block, struct amdgpu_ras_block_object,
> ras_comm);
> +	ih_info.cb = obj->ras_cb;
> +	ih_info.head = *ras_block;
> +	snprintf(sysfs_name, sizeof(sysfs_name), "%s_err_count", ras_block-
> >name);
> +	fs_info.sysfs_name = (const char *)sysfs_name;
> +	fs_info.head = *ras_block;
> +	return amdgpu_ras_late_init(adev, ras_block, &fs_info, &ih_info); }
> +
>  /* helper function to remove ras fs node and interrupt handler */  void
> amdgpu_ras_late_fini(struct amdgpu_device *adev,
>  			  struct ras_common_if *ras_block,
> @@ -2470,6 +2487,22 @@ void amdgpu_ras_late_fini(struct amdgpu_device
> *adev,
>  		amdgpu_ras_interrupt_remove_handler(adev, ih_info);  }
> 
> +void amdgpu_ras_block_late_fini(struct amdgpu_device *adev,
> +			  struct ras_common_if *ras_block)
> +{
> +	struct ras_ih_if ih_info;
> +	struct amdgpu_ras_block_object *obj;
> +
> +	if (!ras_block)
> +		return;
> +
> +	obj = container_of(ras_block, struct amdgpu_ras_block_object,
> ras_comm);
> +	ih_info.head = *ras_block;
> +	ih_info.cb = obj->ras_cb;
> +
> +	amdgpu_ras_late_fini(adev, ras_block, &ih_info); }
> +
>  /* do some init work after IP late init as dependence.
>   * and it runs in resume/gpu reset/booting up cases.
>   */
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h
> b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h
> index a55743b12d57..8b94b556baf6 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h
> @@ -486,17 +486,13 @@ struct ras_debug_if {  };
> 
>  struct amdgpu_ras_block_object {
> -	/* block name */
> -	char name[32];
> -
> -	enum amdgpu_ras_block block;
> -
> -	uint32_t sub_block_index;
> +	struct ras_common_if  ras_comm;
> 
>  	int (*ras_block_match)(struct amdgpu_ras_block_object *block_obj,
>  				enum amdgpu_ras_block block, uint32_t
> sub_block_index);
>  	int (*ras_late_init)(struct amdgpu_device *adev, void *ras_info);
>  	void (*ras_fini)(struct amdgpu_device *adev);
> +	ras_ih_cb ras_cb;
>  	const struct amdgpu_ras_block_hw_ops *hw_ops;  };
> 
> @@ -605,10 +601,17 @@ int amdgpu_ras_late_init(struct amdgpu_device
> *adev,
>  			 struct ras_common_if *ras_block,
>  			 struct ras_fs_if *fs_info,
>  			 struct ras_ih_if *ih_info);
> +
> +int amdgpu_ras_block_late_init(struct amdgpu_device *adev,
> +			struct ras_common_if *ras_block);
> +
>  void amdgpu_ras_late_fini(struct amdgpu_device *adev,
>  			  struct ras_common_if *ras_block,
>  			  struct ras_ih_if *ih_info);
> 
> +void amdgpu_ras_block_late_fini(struct amdgpu_device *adev,
> +			  struct ras_common_if *ras_block);
> +
>  int amdgpu_ras_feature_enable(struct amdgpu_device *adev,
>  		struct ras_common_if *head, bool enable);
> 
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c
> b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c
> index 5929d6f528c9..15707af89212 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c
> @@ -981,8 +981,10 @@ struct amdgpu_ras_block_hw_ops  xgmi_ras_hw_ops
> = {
> 
>  struct amdgpu_xgmi_ras xgmi_ras = {
>  	.ras_block = {
> -		.name = "xgmi",
> -		.block = AMDGPU_RAS_BLOCK__XGMI_WAFL,
> +		.ras_comm = {
> +			.name = "xgmi",
> +			.block = AMDGPU_RAS_BLOCK__XGMI_WAFL,
> +		},
>  		.hw_ops = &xgmi_ras_hw_ops,
>  		.ras_late_init = amdgpu_xgmi_ras_late_init,
>  		.ras_fini = amdgpu_xgmi_ras_fini,
> diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
> b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
> index ca7b886c6ce6..0a291d2e5f91 100644
> --- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
> +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
> @@ -2198,8 +2198,8 @@ static int gfx_v9_0_gpu_early_init(struct
> amdgpu_device *adev)
>  			return err;
>  		}
> 
> -		strcpy(adev->gfx.ras->ras_block.name,"gfx");
> -		adev->gfx.ras->ras_block.block = AMDGPU_RAS_BLOCK__GFX;
> +		strcpy(adev->gfx.ras->ras_block.ras_comm.name, "gfx");
> +		adev->gfx.ras->ras_block.ras_comm.block =
> AMDGPU_RAS_BLOCK__GFX;
> 
>  		/* If not define special ras_late_init function, use gfx default
> ras_late_init */
>  		if (!adev->gfx.ras->ras_block.ras_late_init)
> diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c
> b/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c
> index bddaf2417344..2a362c570346 100644
> --- a/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c
> +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c
> @@ -672,8 +672,8 @@ static void gmc_v10_0_set_umc_funcs(struct
> amdgpu_device *adev)
>  	if (adev->umc.ras) {
>  		amdgpu_ras_register_ras_block(adev, &adev->umc.ras-
> >ras_block);
> 
> -		strcpy(adev->umc.ras->ras_block.name, "umc");
> -		adev->umc.ras->ras_block.block =
> AMDGPU_RAS_BLOCK__UMC;
> +		strcpy(adev->umc.ras->ras_block.ras_comm.name, "umc");
> +		adev->umc.ras->ras_block.ras_comm.block =
> AMDGPU_RAS_BLOCK__UMC;
> 
>  		/* If don't define special ras_late_init function, use default
> ras_late_init */
>  		if (!adev->umc.ras->ras_block.ras_late_init)
> diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
> b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
> index 4595027a8c63..af873c99d5e4 100644
> --- a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
> +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
> @@ -1232,8 +1232,8 @@ static void gmc_v9_0_set_umc_funcs(struct
> amdgpu_device *adev)
>  	if (adev->umc.ras) {
>  		amdgpu_ras_register_ras_block(adev, &adev->umc.ras-
> >ras_block);
> 
> -		strcpy(adev->umc.ras->ras_block.name, "umc");
> -		adev->umc.ras->ras_block.block =
> AMDGPU_RAS_BLOCK__UMC;
> +		strcpy(adev->umc.ras->ras_block.ras_comm.name, "umc");
> +		adev->umc.ras->ras_block.ras_comm.block =
> AMDGPU_RAS_BLOCK__UMC;
> 
>  		/* If don't define special ras_late_init function, use default
> ras_late_init */
>  		if (!adev->umc.ras->ras_block.ras_late_init)
> @@ -1280,8 +1280,8 @@ static void gmc_v9_0_set_mmhub_ras_funcs(struct
> amdgpu_device *adev)
>  	if (adev->mmhub.ras) {
>  		amdgpu_ras_register_ras_block(adev, &adev->mmhub.ras-
> >ras_block);
> 
> -		strcpy(adev->mmhub.ras->ras_block.name,"mmhub");
> -		adev->mmhub.ras->ras_block.block =
> AMDGPU_RAS_BLOCK__MMHUB;
> +		strcpy(adev->mmhub.ras->ras_block.ras_comm.name,
> "mmhub");
> +		adev->mmhub.ras->ras_block.ras_comm.block =
> AMDGPU_RAS_BLOCK__MMHUB;
> 
>  		/* If don't define special ras_late_init function, use default
> ras_late_init */
>  		if (!adev->mmhub.ras->ras_block.ras_late_init)
> diff --git a/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c
> b/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c
> index 6b41fcbf4875..503c292b321e 100644
> --- a/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c
> +++ b/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c
> @@ -157,8 +157,10 @@ struct amdgpu_ras_block_hw_ops
> hdp_v4_0_ras_hw_ops = {
> 
>  struct amdgpu_hdp_ras hdp_v4_0_ras = {
>  	.ras_block = {
> -		.name = "hdp",
> -		.block = AMDGPU_RAS_BLOCK__HDP,
> +		.ras_comm = {
> +			.name = "hdp",
> +			.block = AMDGPU_RAS_BLOCK__HDP,
> +		},
>  		.hw_ops = &hdp_v4_0_ras_hw_ops,
>  		.ras_late_init = amdgpu_hdp_ras_late_init,
>  		.ras_fini = amdgpu_hdp_ras_fini,
> diff --git a/drivers/gpu/drm/amd/amdgpu/mca_v3_0.c
> b/drivers/gpu/drm/amd/amdgpu/mca_v3_0.c
> index 68565262af9c..386416378a82 100644
> --- a/drivers/gpu/drm/amd/amdgpu/mca_v3_0.c
> +++ b/drivers/gpu/drm/amd/amdgpu/mca_v3_0.c
> @@ -53,8 +53,8 @@ static int mca_v3_0_ras_block_match(struct
> amdgpu_ras_block_object *block_obj,
>  	if (!block_obj)
>  		return -EINVAL;
> 
> -	if ((block_obj->block == block) &&
> -		(block_obj->sub_block_index == sub_block_index)) {
> +	if ((block_obj->ras_comm.block == block) &&
> +		(block_obj->ras_comm.sub_block_index == sub_block_index)) {
>  		return 0;
>  	}
> 
> @@ -68,9 +68,11 @@ const struct amdgpu_ras_block_hw_ops
> mca_v3_0_mp0_hw_ops = {
> 
>  struct amdgpu_mca_ras_block mca_v3_0_mp0_ras = {
>  	.ras_block = {
> -		.block = AMDGPU_RAS_BLOCK__MCA,
> -		.sub_block_index = AMDGPU_RAS_MCA_BLOCK__MP0,
> -		.name = "mp0",
> +		.ras_comm = {
> +			.block = AMDGPU_RAS_BLOCK__MCA,
> +			.sub_block_index = AMDGPU_RAS_MCA_BLOCK__MP0,
> +			.name = "mp0",
> +		},
>  		.hw_ops = &mca_v3_0_mp0_hw_ops,
>  		.ras_block_match = mca_v3_0_ras_block_match,
>  		.ras_late_init = mca_v3_0_mp0_ras_late_init, @@ -103,9
> +105,11 @@ const struct amdgpu_ras_block_hw_ops mca_v3_0_mp1_hw_ops
> = {
> 
>  struct amdgpu_mca_ras_block mca_v3_0_mp1_ras = {
>  	.ras_block = {
> -		.block = AMDGPU_RAS_BLOCK__MCA,
> -		.sub_block_index = AMDGPU_RAS_MCA_BLOCK__MP1,
> -		.name = "mp1",
> +		.ras_comm = {
> +			.block = AMDGPU_RAS_BLOCK__MCA,
> +			.sub_block_index = AMDGPU_RAS_MCA_BLOCK__MP1,
> +			.name = "mp1",
> +		},
>  		.hw_ops = &mca_v3_0_mp1_hw_ops,
>  		.ras_block_match = mca_v3_0_ras_block_match,
>  		.ras_late_init = mca_v3_0_mp1_ras_late_init, @@ -138,9
> +142,11 @@ const struct amdgpu_ras_block_hw_ops mca_v3_0_mpio_hw_ops
> = {
> 
>  struct amdgpu_mca_ras_block mca_v3_0_mpio_ras = {
>  	.ras_block = {
> -		.block = AMDGPU_RAS_BLOCK__MCA,
> -		.sub_block_index = AMDGPU_RAS_MCA_BLOCK__MPIO,
> -		.name = "mpio",
> +		.ras_comm = {
> +			.block = AMDGPU_RAS_BLOCK__MCA,
> +			.sub_block_index =
> AMDGPU_RAS_MCA_BLOCK__MPIO,
> +			.name = "mpio",
> +		},
>  		.hw_ops = &mca_v3_0_mpio_hw_ops,
>  		.ras_block_match = mca_v3_0_ras_block_match,
>  		.ras_late_init = mca_v3_0_mpio_ras_late_init, diff --git
> a/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c
> b/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c
> index 39974b449341..c7cca87f1647 100644
> --- a/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c
> +++ b/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c
> @@ -664,8 +664,10 @@ const struct amdgpu_ras_block_hw_ops
> nbio_v7_4_ras_hw_ops = {
> 
>  struct amdgpu_nbio_ras nbio_v7_4_ras = {
>  	.ras_block = {
> -		.name = "pcie_bif",
> -		.block = AMDGPU_RAS_BLOCK__PCIE_BIF,
> +		.ras_comm = {
> +			.name = "pcie_bif",
> +			.block = AMDGPU_RAS_BLOCK__PCIE_BIF,
> +		},
>  		.hw_ops = &nbio_v7_4_ras_hw_ops,
>  		.ras_late_init = amdgpu_nbio_ras_late_init,
>  		.ras_fini = amdgpu_nbio_ras_fini,
> diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c
> b/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c
> index 06a7ceda4c87..8b0a8587dd36 100644
> --- a/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c
> +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c
> @@ -2814,8 +2814,8 @@ static void sdma_v4_0_set_ras_funcs(struct
> amdgpu_device *adev)
>  	if (adev->sdma.ras) {
>  		amdgpu_ras_register_ras_block(adev, &adev->sdma.ras-
> >ras_block);
> 
> -		strcpy(adev->sdma.ras->ras_block.name, "sdma");
> -		adev->sdma.ras->ras_block.block =
> AMDGPU_RAS_BLOCK__SDMA;
> +		strcpy(adev->sdma.ras->ras_block.ras_comm.name, "sdma");
> +		adev->sdma.ras->ras_block.ras_comm.block =
> AMDGPU_RAS_BLOCK__SDMA;
> 
>  		/* If don't define special ras_late_init function, use default
> ras_late_init */
>  		if (!adev->sdma.ras->ras_block.ras_late_init)
> --
> 2.25.1

RE: [PATCH 03/11] drm/amdgpu: Optimize amdgpu_hdp_ras_late_init/amdgpu_hdp_ras_fini function code

[Chai, Thomas]

[AMD Official Use Only]



-----Original Message-----
From: Zhou1, Tao <Tao.Zhou1@amd.com> 
Sent: Wednesday, February 9, 2022 4:54 PM
To: Chai, Thomas <YiPeng.Chai@amd.com>; amd-gfx@lists.freedesktop.org
Cc: Zhang, Hawking <Hawking.Zhang@amd.com>; Clements, John <John.Clements@amd.com>
Subject: RE: [PATCH 03/11] drm/amdgpu: Optimize amdgpu_hdp_ras_late_init/amdgpu_hdp_ras_fini function code

[AMD Official Use Only]



> -----Original Message-----
> From: Chai, Thomas <YiPeng.Chai@amd.com>
> Sent: Wednesday, February 9, 2022 1:57 PM
> To: amd-gfx@lists.freedesktop.org
> Cc: Chai, Thomas <YiPeng.Chai@amd.com>; Zhang, Hawking 
> <Hawking.Zhang@amd.com>; Zhou1, Tao <Tao.Zhou1@amd.com>; Clements, 
> John <John.Clements@amd.com>; Chai, Thomas <YiPeng.Chai@amd.com>
> Subject: [PATCH 03/11] drm/amdgpu: Optimize 
> amdgpu_hdp_ras_late_init/amdgpu_hdp_ras_fini function code
> 
> Optimize amdgpu_hdp_ras_late_init/amdgpu_hdp_ras_fini function code.
> 
> Signed-off-by: yipechai <YiPeng.Chai@amd.com>
> ---
>  drivers/gpu/drm/amd/amdgpu/amdgpu_hdp.c | 37 ++-----------------------
>  drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c   |  1 +
>  drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c   |  1 +
>  3 files changed, 5 insertions(+), 34 deletions(-)
> 
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_hdp.c
> b/drivers/gpu/drm/amd/amdgpu/amdgpu_hdp.c
> index 518966a26130..21a5f884dd2a 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_hdp.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_hdp.c
> @@ -26,43 +26,12 @@
> 
>  int amdgpu_hdp_ras_late_init(struct amdgpu_device *adev, void *ras_info)  {
> -	int r;
> -	struct ras_ih_if ih_info = {
> -		.cb = NULL,
> -	};
> -	struct ras_fs_if fs_info = {
> -		.sysfs_name = "hdp_err_count",
> -	};
> -
> -	if (!adev->hdp.ras_if) {
> -		adev->hdp.ras_if = kmalloc(sizeof(struct ras_common_if),
> GFP_KERNEL);
> -		if (!adev->hdp.ras_if)
> -			return -ENOMEM;
> -		adev->hdp.ras_if->block = AMDGPU_RAS_BLOCK__HDP;
> -		adev->hdp.ras_if->type =
> AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE;
> -		adev->hdp.ras_if->sub_block_index = 0;
> -	}
> -	ih_info.head = fs_info.head = *adev->hdp.ras_if;
> -	r = amdgpu_ras_late_init(adev, adev->hdp.ras_if,
> -				 &fs_info, &ih_info);
> -	if (r || !amdgpu_ras_is_supported(adev, adev->hdp.ras_if->block)) {
> -		kfree(adev->hdp.ras_if);
> -		adev->hdp.ras_if = NULL;
> -	}
> -
> -	return r;
> +	return amdgpu_ras_block_late_init(adev, adev->hdp.ras_if);
>  }
> 
>  void amdgpu_hdp_ras_fini(struct amdgpu_device *adev)  {
>  	if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__HDP) &&
> -	    adev->hdp.ras_if) {
> -		struct ras_common_if *ras_if = adev->hdp.ras_if;
> -		struct ras_ih_if ih_info = {
> -			.cb = NULL,
> -		};
> -
> -		amdgpu_ras_late_fini(adev, ras_if, &ih_info);
> -		kfree(ras_if);
> -	}
> +	    adev->hdp.ras_if)
> +		amdgpu_ras_block_late_fini(adev, adev->hdp.ras_if);
>  }
>[Tao]: Since hdp_ras_late_init/fini are simple wrapper, can we remove them and call amdgpu_ras_block_late_init/fini directly?
>The same comment to other blocks.

[Thomas] Compared with amdgpu_ras_block_late_init/fin,  hdp_ras_late_init/fini have different function interface parameters.
                 But can do it as a new ticket later.

> diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
> b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
> index af873c99d5e4..b12fe6703f02 100644
> --- a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
> +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
> @@ -1302,6 +1302,7 @@ static void gmc_v9_0_set_hdp_ras_funcs(struct 
> amdgpu_device *adev)  {
>  	adev->hdp.ras = &hdp_v4_0_ras;
>  	amdgpu_ras_register_ras_block(adev, &adev->hdp.ras->ras_block);
> +	adev->hdp.ras_if = &adev->hdp.ras->ras_block.ras_comm;
>  }
> 
>  static void gmc_v9_0_set_mca_funcs(struct amdgpu_device *adev) diff 
> --git a/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c
> b/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c
> index 503c292b321e..a9ed4232cdeb 100644
> --- a/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c
> +++ b/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c
> @@ -160,6 +160,7 @@ struct amdgpu_hdp_ras hdp_v4_0_ras = {
>  		.ras_comm = {
>  			.name = "hdp",
>  			.block = AMDGPU_RAS_BLOCK__HDP,
> +			.type =
> AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE,
>  		},
>  		.hw_ops = &hdp_v4_0_ras_hw_ops,
>  		.ras_late_init = amdgpu_hdp_ras_late_init,
> --
> 2.25.1

RE: [PATCH 03/11] drm/amdgpu: Optimize amdgpu_hdp_ras_late_init/amdgpu_hdp_ras_fini function code

[Zhou1, Tao]

[AMD Official Use Only]

OK, if there is further refinement, the series is:

Reviewed-by: Tao Zhou <tao.zhou1@amd.com>

> -----Original Message-----
> From: Chai, Thomas <YiPeng.Chai@amd.com>
> Sent: Thursday, February 10, 2022 10:59 AM
> To: Zhou1, Tao <Tao.Zhou1@amd.com>; amd-gfx@lists.freedesktop.org
> Cc: Zhang, Hawking <Hawking.Zhang@amd.com>; Clements, John
> <John.Clements@amd.com>
> Subject: RE: [PATCH 03/11] drm/amdgpu: Optimize
> amdgpu_hdp_ras_late_init/amdgpu_hdp_ras_fini function code
> 
> [AMD Official Use Only]
> 
> 
> 
> -----Original Message-----
> From: Zhou1, Tao <Tao.Zhou1@amd.com>
> Sent: Wednesday, February 9, 2022 4:54 PM
> To: Chai, Thomas <YiPeng.Chai@amd.com>; amd-gfx@lists.freedesktop.org
> Cc: Zhang, Hawking <Hawking.Zhang@amd.com>; Clements, John
> <John.Clements@amd.com>
> Subject: RE: [PATCH 03/11] drm/amdgpu: Optimize
> amdgpu_hdp_ras_late_init/amdgpu_hdp_ras_fini function code
> 
> [AMD Official Use Only]
> 
> 
> 
> > -----Original Message-----
> > From: Chai, Thomas <YiPeng.Chai@amd.com>
> > Sent: Wednesday, February 9, 2022 1:57 PM
> > To: amd-gfx@lists.freedesktop.org
> > Cc: Chai, Thomas <YiPeng.Chai@amd.com>; Zhang, Hawking
> > <Hawking.Zhang@amd.com>; Zhou1, Tao <Tao.Zhou1@amd.com>; Clements,
> > John <John.Clements@amd.com>; Chai, Thomas <YiPeng.Chai@amd.com>
> > Subject: [PATCH 03/11] drm/amdgpu: Optimize
> > amdgpu_hdp_ras_late_init/amdgpu_hdp_ras_fini function code
> >
> > Optimize amdgpu_hdp_ras_late_init/amdgpu_hdp_ras_fini function code.
> >
> > Signed-off-by: yipechai <YiPeng.Chai@amd.com>
> > ---
> >  drivers/gpu/drm/amd/amdgpu/amdgpu_hdp.c | 37 ++-----------------------
> >  drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c   |  1 +
> >  drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c   |  1 +
> >  3 files changed, 5 insertions(+), 34 deletions(-)
> >
> > diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_hdp.c
> > b/drivers/gpu/drm/amd/amdgpu/amdgpu_hdp.c
> > index 518966a26130..21a5f884dd2a 100644
> > --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_hdp.c
> > +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_hdp.c
> > @@ -26,43 +26,12 @@
> >
> >  int amdgpu_hdp_ras_late_init(struct amdgpu_device *adev, void *ras_info)  {
> > -	int r;
> > -	struct ras_ih_if ih_info = {
> > -		.cb = NULL,
> > -	};
> > -	struct ras_fs_if fs_info = {
> > -		.sysfs_name = "hdp_err_count",
> > -	};
> > -
> > -	if (!adev->hdp.ras_if) {
> > -		adev->hdp.ras_if = kmalloc(sizeof(struct ras_common_if),
> > GFP_KERNEL);
> > -		if (!adev->hdp.ras_if)
> > -			return -ENOMEM;
> > -		adev->hdp.ras_if->block = AMDGPU_RAS_BLOCK__HDP;
> > -		adev->hdp.ras_if->type =
> > AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE;
> > -		adev->hdp.ras_if->sub_block_index = 0;
> > -	}
> > -	ih_info.head = fs_info.head = *adev->hdp.ras_if;
> > -	r = amdgpu_ras_late_init(adev, adev->hdp.ras_if,
> > -				 &fs_info, &ih_info);
> > -	if (r || !amdgpu_ras_is_supported(adev, adev->hdp.ras_if->block)) {
> > -		kfree(adev->hdp.ras_if);
> > -		adev->hdp.ras_if = NULL;
> > -	}
> > -
> > -	return r;
> > +	return amdgpu_ras_block_late_init(adev, adev->hdp.ras_if);
> >  }
> >
> >  void amdgpu_hdp_ras_fini(struct amdgpu_device *adev)  {
> >  	if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__HDP) &&
> > -	    adev->hdp.ras_if) {
> > -		struct ras_common_if *ras_if = adev->hdp.ras_if;
> > -		struct ras_ih_if ih_info = {
> > -			.cb = NULL,
> > -		};
> > -
> > -		amdgpu_ras_late_fini(adev, ras_if, &ih_info);
> > -		kfree(ras_if);
> > -	}
> > +	    adev->hdp.ras_if)
> > +		amdgpu_ras_block_late_fini(adev, adev->hdp.ras_if);
> >  }
> >[Tao]: Since hdp_ras_late_init/fini are simple wrapper, can we remove them
> and call amdgpu_ras_block_late_init/fini directly?
> >The same comment to other blocks.
> 
> [Thomas] Compared with amdgpu_ras_block_late_init/fin,
> hdp_ras_late_init/fini have different function interface parameters.
>                  But can do it as a new ticket later.
> 
> > diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
> > b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
> > index af873c99d5e4..b12fe6703f02 100644
> > --- a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
> > +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
> > @@ -1302,6 +1302,7 @@ static void gmc_v9_0_set_hdp_ras_funcs(struct
> > amdgpu_device *adev)  {
> >  	adev->hdp.ras = &hdp_v4_0_ras;
> >  	amdgpu_ras_register_ras_block(adev, &adev->hdp.ras->ras_block);
> > +	adev->hdp.ras_if = &adev->hdp.ras->ras_block.ras_comm;
> >  }
> >
> >  static void gmc_v9_0_set_mca_funcs(struct amdgpu_device *adev) diff
> > --git a/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c
> > b/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c
> > index 503c292b321e..a9ed4232cdeb 100644
> > --- a/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c
> > +++ b/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c
> > @@ -160,6 +160,7 @@ struct amdgpu_hdp_ras hdp_v4_0_ras = {
> >  		.ras_comm = {
> >  			.name = "hdp",
> >  			.block = AMDGPU_RAS_BLOCK__HDP,
> > +			.type =
> > AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE,
> >  		},
> >  		.hw_ops = &hdp_v4_0_ras_hw_ops,
> >  		.ras_late_init = amdgpu_hdp_ras_late_init,
> > --
> > 2.25.1