From 04e9101766dfe1f140e59090935552b2906c5425 Mon Sep 17 00:00:00 2001
From: Asad Kamal <asad.kamal@amd.com>
Date: Wed, 30 Oct 2024 19:02:01 +0800
Subject: [PATCH 01/16] drm/amdgpu: Add supported NPS modes node

Add sysfs node to show supported NPS mode for the
partition configuration selected using xcp_config

v2: Hide node if dynamic nps switch not supported

v3: Fix removal of files in case of error

Signed-off-by: Asad Kamal <asad.kamal@amd.com>
Reviewed-by: Lijo Lazar <lijo.lazar@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.c | 48 ++++++++++++++++++++++++-
 1 file changed, 47 insertions(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.c
index 83a16918ea76..e209b5e101df 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.c
@@ -471,6 +471,16 @@ static const char *xcp_desc[] = {
 	[AMDGPU_CPX_PARTITION_MODE] = "CPX",
 };
 
+static const char *nps_desc[] = {
+	[UNKNOWN_MEMORY_PARTITION_MODE] = "UNKNOWN",
+	[AMDGPU_NPS1_PARTITION_MODE] = "NPS1",
+	[AMDGPU_NPS2_PARTITION_MODE] = "NPS2",
+	[AMDGPU_NPS3_PARTITION_MODE] = "NPS3",
+	[AMDGPU_NPS4_PARTITION_MODE] = "NPS4",
+	[AMDGPU_NPS6_PARTITION_MODE] = "NPS6",
+	[AMDGPU_NPS8_PARTITION_MODE] = "NPS8",
+};
+
 ATTRIBUTE_GROUPS(xcp_cfg_res_sysfs);
 
 #define to_xcp_attr(x) \
@@ -540,6 +550,26 @@ static ssize_t supported_xcp_configs_show(struct kobject *kobj,
 	return size;
 }
 
+static ssize_t supported_nps_configs_show(struct kobject *kobj,
+					  struct kobj_attribute *attr, char *buf)
+{
+	struct amdgpu_xcp_cfg *xcp_cfg = to_xcp_cfg(kobj);
+	int size = 0, mode;
+	char *sep = "";
+
+	if (!xcp_cfg || !xcp_cfg->compatible_nps_modes)
+		return sysfs_emit(buf, "Not supported\n");
+
+	for_each_inst(mode, xcp_cfg->compatible_nps_modes) {
+		size += sysfs_emit_at(buf, size, "%s%s", sep, nps_desc[mode]);
+		sep = ", ";
+	}
+
+	size += sysfs_emit_at(buf, size, "\n");
+
+	return size;
+}
+
 static ssize_t xcp_config_show(struct kobject *kobj,
 			       struct kobj_attribute *attr, char *buf)
 {
@@ -596,6 +626,9 @@ static const struct kobj_type xcp_cfg_sysfs_ktype = {
 static struct kobj_attribute supp_part_sysfs_mode =
 	__ATTR_RO(supported_xcp_configs);
 
+static struct kobj_attribute supp_nps_sysfs_mode =
+	__ATTR_RO(supported_nps_configs);
+
 static const struct attribute *xcp_attrs[] = {
 	&supp_part_sysfs_mode.attr,
 	&xcp_cfg_sysfs_mode.attr,
@@ -625,13 +658,24 @@ void amdgpu_xcp_cfg_sysfs_init(struct amdgpu_device *adev)
 	if (r)
 		goto err1;
 
+	if (adev->gmc.supported_nps_modes != 0) {
+		r = sysfs_create_file(&xcp_cfg->kobj, &supp_nps_sysfs_mode.attr);
+		if (r) {
+			sysfs_remove_files(&xcp_cfg->kobj, xcp_attrs);
+			goto err1;
+		}
+	}
+
 	mode = (xcp_cfg->xcp_mgr->mode ==
 		AMDGPU_UNKNOWN_COMPUTE_PARTITION_MODE) ?
 		       AMDGPU_SPX_PARTITION_MODE :
 		       xcp_cfg->xcp_mgr->mode;
 	r = amdgpu_xcp_get_res_info(xcp_cfg->xcp_mgr, mode, xcp_cfg);
-	if (r)
+	if (r) {
+		sysfs_remove_file(&xcp_cfg->kobj, &supp_nps_sysfs_mode.attr);
+		sysfs_remove_files(&xcp_cfg->kobj, xcp_attrs);
 		goto err1;
+	}
 
 	xcp_cfg->mode = mode;
 	for (i = 0; i < xcp_cfg->num_res; i++) {
@@ -653,6 +697,7 @@ err:
 		kobject_put(&xcp_res->kobj);
 	}
 
+	sysfs_remove_file(&xcp_cfg->kobj, &supp_nps_sysfs_mode.attr);
 	sysfs_remove_files(&xcp_cfg->kobj, xcp_attrs);
 err1:
 	kobject_put(&xcp_cfg->kobj);
@@ -673,6 +718,7 @@ void amdgpu_xcp_cfg_sysfs_fini(struct amdgpu_device *adev)
 		kobject_put(&xcp_res->kobj);
 	}
 
+	sysfs_remove_file(&xcp_cfg->kobj, &supp_nps_sysfs_mode.attr);
 	sysfs_remove_files(&xcp_cfg->kobj, xcp_attrs);
 	kobject_put(&xcp_cfg->kobj);
 }
-- 
2.51.0


From cd82f29ec51b2e616289db7b258a936127c16efa Mon Sep 17 00:00:00 2001
From: Jonathan Kim <jonathan.kim@amd.com>
Date: Tue, 5 Nov 2024 12:38:30 -0500
Subject: [PATCH 02/16] drm/amdkfd: remove gfx 12 trap handler page size cap

GFX 12 does not require a page size cap for the trap handler because
it does not require a CWSR work around like GFX 11 did.

Signed-off-by: Jonathan Kim <jonathan.kim@amd.com>
Reviewed-by: David Belanger <david.belanger@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/amd/amdkfd/kfd_device.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device.c b/drivers/gpu/drm/amd/amdkfd/kfd_device.c
index 348925254bff..956198da7859 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_device.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_device.c
@@ -534,7 +534,8 @@ static void kfd_cwsr_init(struct kfd_dev *kfd)
 			kfd->cwsr_isa = cwsr_trap_gfx11_hex;
 			kfd->cwsr_isa_size = sizeof(cwsr_trap_gfx11_hex);
 		} else {
-			BUILD_BUG_ON(sizeof(cwsr_trap_gfx12_hex) > PAGE_SIZE);
+			BUILD_BUG_ON(sizeof(cwsr_trap_gfx12_hex)
+					     > KFD_CWSR_TMA_OFFSET);
 			kfd->cwsr_isa = cwsr_trap_gfx12_hex;
 			kfd->cwsr_isa_size = sizeof(cwsr_trap_gfx12_hex);
 		}
-- 
2.51.0


From f4a3246a2c7a595161f1ba11db53639b7f580104 Mon Sep 17 00:00:00 2001
From: chongli2 <chongli2@amd.com>
Date: Wed, 6 Nov 2024 11:43:09 +0800
Subject: [PATCH 03/16] drm/amdgpu: fix return random value when multiple
 threads read registers via mes.
MIME-Version: 1.0
Content-Type: text/plain; charset=utf8
Content-Transfer-Encoding: 8bit

The currect code use the address "adev->mes.read_val_ptr" to
store the value read from register via mes.
So when multiple threads read register,
multiple threads have to share the one address,
and overwrite the value each other.

Assign an address by "amdgpu_device_wb_get" to store register value.
each thread will has an address to store register value.

Signed-off-by: chongli2 <chongli2@amd.com>
Reviewed-by: Emily Deng <Emily.Deng@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c | 30 +++++++++++--------------
 drivers/gpu/drm/amd/amdgpu/amdgpu_mes.h |  3 ---
 2 files changed, 13 insertions(+), 20 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c
index b10383f83d73..f702b8b9c318 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c
@@ -192,17 +192,6 @@ int amdgpu_mes_init(struct amdgpu_device *adev)
 			(uint64_t *)&adev->wb.wb[adev->mes.query_status_fence_offs[i]];
 	}
 
-	r = amdgpu_device_wb_get(adev, &adev->mes.read_val_offs);
-	if (r) {
-		dev_err(adev->dev,
-			"(%d) read_val_offs alloc failed\n", r);
-		goto error;
-	}
-	adev->mes.read_val_gpu_addr =
-		adev->wb.gpu_addr + (adev->mes.read_val_offs * 4);
-	adev->mes.read_val_ptr =
-		(uint32_t *)&adev->wb.wb[adev->mes.read_val_offs];
-
 	r = amdgpu_mes_doorbell_init(adev);
 	if (r)
 		goto error;
@@ -223,8 +212,6 @@ error:
 			amdgpu_device_wb_free(adev,
 				      adev->mes.query_status_fence_offs[i]);
 	}
-	if (adev->mes.read_val_ptr)
-		amdgpu_device_wb_free(adev, adev->mes.read_val_offs);
 
 	idr_destroy(&adev->mes.pasid_idr);
 	idr_destroy(&adev->mes.gang_id_idr);
@@ -249,8 +236,6 @@ void amdgpu_mes_fini(struct amdgpu_device *adev)
 			amdgpu_device_wb_free(adev,
 				      adev->mes.query_status_fence_offs[i]);
 	}
-	if (adev->mes.read_val_ptr)
-		amdgpu_device_wb_free(adev, adev->mes.read_val_offs);
 
 	amdgpu_mes_doorbell_free(adev);
 
@@ -921,10 +906,19 @@ uint32_t amdgpu_mes_rreg(struct amdgpu_device *adev, uint32_t reg)
 {
 	struct mes_misc_op_input op_input;
 	int r, val = 0;
+	uint32_t addr_offset = 0;
+	uint64_t read_val_gpu_addr;
+	uint32_t *read_val_ptr;
 
+	if (amdgpu_device_wb_get(adev, &addr_offset)) {
+		DRM_ERROR("critical bug! too many mes readers\n");
+		goto error;
+	}
+	read_val_gpu_addr = adev->wb.gpu_addr + (addr_offset * 4);
+	read_val_ptr = (uint32_t *)&adev->wb.wb[addr_offset];
 	op_input.op = MES_MISC_OP_READ_REG;
 	op_input.read_reg.reg_offset = reg;
-	op_input.read_reg.buffer_addr = adev->mes.read_val_gpu_addr;
+	op_input.read_reg.buffer_addr = read_val_gpu_addr;
 
 	if (!adev->mes.funcs->misc_op) {
 		DRM_ERROR("mes rreg is not supported!\n");
@@ -935,9 +929,11 @@ uint32_t amdgpu_mes_rreg(struct amdgpu_device *adev, uint32_t reg)
 	if (r)
 		DRM_ERROR("failed to read reg (0x%x)\n", reg);
 	else
-		val = *(adev->mes.read_val_ptr);
+		val = *(read_val_ptr);
 
 error:
+	if (addr_offset)
+		amdgpu_device_wb_free(adev, addr_offset);
 	return val;
 }
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.h
index 0684e482a204..129b8d1cacef 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.h
@@ -120,9 +120,6 @@ struct amdgpu_mes {
 	uint32_t			query_status_fence_offs[AMDGPU_MAX_MES_PIPES];
 	uint64_t			query_status_fence_gpu_addr[AMDGPU_MAX_MES_PIPES];
 	uint64_t			*query_status_fence_ptr[AMDGPU_MAX_MES_PIPES];
-	uint32_t                        read_val_offs;
-	uint64_t			read_val_gpu_addr;
-	uint32_t			*read_val_ptr;
 
 	uint32_t			saved_flags;
 
-- 
2.51.0


From 6c8d1f4b042e706ccd7575beb0397a75d545d71b Mon Sep 17 00:00:00 2001
From: "Jesse.zhang@amd.com" <Jesse.zhang@amd.com>
Date: Tue, 5 Nov 2024 15:22:56 +0800
Subject: [PATCH 04/16] drm/amdgpu: Add sysfs interface for gc reset mask

Add two sysfs interfaces for gfx and compute:
gfx_reset_mask
compute_reset_mask

These interfaces are read-only and show the resets supported by the IP.
For example, full adapter reset (mode1/mode2/BACO/etc),
soft reset, queue reset, and pipe reset.

V2: the sysfs node returns a text string instead of some flags (Christian)
v3: add a generic helper which takes the ring as parameter
    and print the strings in the order they are applied (Christian)

    check amdgpu_gpu_recovery  before creating sysfs file itself,
    and initialize supported_reset_types in IP version files (Lijo)
v4: Fixing uninitialized variables (Tim)

Signed-off-by: Jesse Zhang <Jesse.Zhang@amd.com>
Suggested-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Tim Huang <tim.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu.h        |  8 +++
 drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 44 ++++++++++++++
 drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c    | 70 ++++++++++++++++++++++
 drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h    |  2 +
 drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c     |  5 ++
 drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c     | 18 ++++++
 drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c     |  6 ++
 drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c      |  6 ++
 drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c    | 13 ++++
 9 files changed, 172 insertions(+)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
index 7645e498faa4..d8bc6da50016 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
@@ -299,6 +299,12 @@ extern int amdgpu_wbrf;
 #define AMDGPU_RESET_VCE			(1 << 13)
 #define AMDGPU_RESET_VCE1			(1 << 14)
 
+/* reset mask */
+#define AMDGPU_RESET_TYPE_FULL (1 << 0) /* full adapter reset, mode1/mode2/BACO/etc. */
+#define AMDGPU_RESET_TYPE_SOFT_RESET (1 << 1) /* IP level soft reset */
+#define AMDGPU_RESET_TYPE_PER_QUEUE (1 << 2) /* per queue */
+#define AMDGPU_RESET_TYPE_PER_PIPE (1 << 3) /* per pipe */
+
 /* max cursor sizes (in pixels) */
 #define CIK_CURSOR_WIDTH 128
 #define CIK_CURSOR_HEIGHT 128
@@ -1464,6 +1470,8 @@ struct dma_fence *amdgpu_device_get_gang(struct amdgpu_device *adev);
 struct dma_fence *amdgpu_device_switch_gang(struct amdgpu_device *adev,
 					    struct dma_fence *gang);
 bool amdgpu_device_has_display_hardware(struct amdgpu_device *adev);
+ssize_t amdgpu_get_soft_full_reset_mask(struct amdgpu_ring *ring);
+ssize_t amdgpu_show_reset_mask(char *buf, uint32_t supported_reset);
 
 /* atpx handler */
 #if defined(CONFIG_VGA_SWITCHEROO)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
index 0450eab6ade7..38a17571da75 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
@@ -6715,3 +6715,47 @@ uint32_t amdgpu_device_wait_on_rreg(struct amdgpu_device *adev,
 	}
 	return ret;
 }
+
+ssize_t amdgpu_get_soft_full_reset_mask(struct amdgpu_ring *ring)
+{
+	ssize_t size = 0;
+
+	if (!ring || !ring->adev)
+		return size;
+
+	if (amdgpu_device_should_recover_gpu(ring->adev))
+		size |= AMDGPU_RESET_TYPE_FULL;
+
+	if (unlikely(!ring->adev->debug_disable_soft_recovery) &&
+	    !amdgpu_sriov_vf(ring->adev) && ring->funcs->soft_recovery)
+		size |= AMDGPU_RESET_TYPE_SOFT_RESET;
+
+	return size;
+}
+
+ssize_t amdgpu_show_reset_mask(char *buf, uint32_t supported_reset)
+{
+	ssize_t size = 0;
+
+	if (supported_reset == 0) {
+		size += sysfs_emit_at(buf, size, "unsupported");
+		size += sysfs_emit_at(buf, size, "\n");
+		return size;
+
+	}
+
+	if (supported_reset & AMDGPU_RESET_TYPE_SOFT_RESET)
+		size += sysfs_emit_at(buf, size, "soft ");
+
+	if (supported_reset & AMDGPU_RESET_TYPE_PER_QUEUE)
+		size += sysfs_emit_at(buf, size, "queue ");
+
+	if (supported_reset & AMDGPU_RESET_TYPE_PER_PIPE)
+		size += sysfs_emit_at(buf, size, "pipe ");
+
+	if (supported_reset & AMDGPU_RESET_TYPE_FULL)
+		size += sysfs_emit_at(buf, size, "full ");
+
+	size += sysfs_emit_at(buf, size, "\n");
+	return size;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
index 2f3f09dfb1fd..6cc6484bde06 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
@@ -1588,6 +1588,32 @@ static ssize_t amdgpu_gfx_set_enforce_isolation(struct device *dev,
 	return count;
 }
 
+static ssize_t amdgpu_gfx_get_gfx_reset_mask(struct device *dev,
+						struct device_attribute *attr,
+						char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+
+	if (!adev)
+		return -ENODEV;
+
+	return amdgpu_show_reset_mask(buf, adev->gfx.gfx_supported_reset);
+}
+
+static ssize_t amdgpu_gfx_get_compute_reset_mask(struct device *dev,
+						struct device_attribute *attr,
+						char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+
+	if (!adev)
+		return -ENODEV;
+
+	return amdgpu_show_reset_mask(buf, adev->gfx.compute_supported_reset);
+}
+
 static DEVICE_ATTR(run_cleaner_shader, 0200,
 		   NULL, amdgpu_gfx_set_run_cleaner_shader);
 
@@ -1601,6 +1627,11 @@ static DEVICE_ATTR(current_compute_partition, 0644,
 
 static DEVICE_ATTR(available_compute_partition, 0444,
 		   amdgpu_gfx_get_available_compute_partition, NULL);
+static DEVICE_ATTR(gfx_reset_mask, 0444,
+		   amdgpu_gfx_get_gfx_reset_mask, NULL);
+
+static DEVICE_ATTR(compute_reset_mask, 0444,
+		   amdgpu_gfx_get_compute_reset_mask, NULL);
 
 static int amdgpu_gfx_sysfs_xcp_init(struct amdgpu_device *adev)
 {
@@ -1666,6 +1697,40 @@ static void amdgpu_gfx_sysfs_isolation_shader_fini(struct amdgpu_device *adev)
 		device_remove_file(adev->dev, &dev_attr_run_cleaner_shader);
 }
 
+static int amdgpu_gfx_sysfs_reset_mask_init(struct amdgpu_device *adev)
+{
+	int r = 0;
+
+	if (!amdgpu_gpu_recovery)
+		return r;
+
+	if (adev->gfx.num_gfx_rings) {
+		r = device_create_file(adev->dev, &dev_attr_gfx_reset_mask);
+		if (r)
+			return r;
+	}
+
+	if (adev->gfx.num_compute_rings) {
+		r = device_create_file(adev->dev, &dev_attr_compute_reset_mask);
+		if (r)
+			return r;
+	}
+
+	return r;
+}
+
+static void amdgpu_gfx_sysfs_reset_mask_fini(struct amdgpu_device *adev)
+{
+	if (!amdgpu_gpu_recovery)
+		return;
+
+	if (adev->gfx.num_gfx_rings)
+		device_remove_file(adev->dev, &dev_attr_gfx_reset_mask);
+
+	if (adev->gfx.num_compute_rings)
+		device_remove_file(adev->dev, &dev_attr_compute_reset_mask);
+}
+
 int amdgpu_gfx_sysfs_init(struct amdgpu_device *adev)
 {
 	int r;
@@ -1680,6 +1745,10 @@ int amdgpu_gfx_sysfs_init(struct amdgpu_device *adev)
 	if (r)
 		dev_err(adev->dev, "failed to create isolation sysfs files");
 
+	r = amdgpu_gfx_sysfs_reset_mask_init(adev);
+	if (r)
+		dev_err(adev->dev, "failed to create reset mask sysfs files");
+
 	return r;
 }
 
@@ -1687,6 +1756,7 @@ void amdgpu_gfx_sysfs_fini(struct amdgpu_device *adev)
 {
 	amdgpu_gfx_sysfs_xcp_fini(adev);
 	amdgpu_gfx_sysfs_isolation_shader_fini(adev);
+	amdgpu_gfx_sysfs_reset_mask_fini(adev);
 }
 
 int amdgpu_gfx_cleaner_shader_sw_init(struct amdgpu_device *adev,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h
index fd73e527f446..8b5bd63b5773 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h
@@ -424,6 +424,8 @@ struct amdgpu_gfx {
 	/* reset mask */
 	uint32_t                        grbm_soft_reset;
 	uint32_t                        srbm_soft_reset;
+	uint32_t 			gfx_supported_reset;
+	uint32_t 			compute_supported_reset;
 
 	/* gfx off */
 	bool                            gfx_off_state;      /* true: enabled, false: disabled */
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c
index d1a18ca584dd..24dce803a829 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c
@@ -4825,6 +4825,11 @@ static int gfx_v10_0_sw_init(struct amdgpu_ip_block *ip_block)
 			}
 		}
 	}
+	/* TODO: Add queue reset mask when FW fully supports it */
+	adev->gfx.gfx_supported_reset =
+		amdgpu_get_soft_full_reset_mask(&adev->gfx.gfx_ring[0]);
+	adev->gfx.compute_supported_reset =
+		amdgpu_get_soft_full_reset_mask(&adev->gfx.compute_ring[0]);
 
 	r = amdgpu_gfx_kiq_init(adev, GFX10_MEC_HPD_SIZE, 0);
 	if (r) {
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c
index 62e4c446793d..28fe8d23c91f 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c
@@ -1691,6 +1691,24 @@ static int gfx_v11_0_sw_init(struct amdgpu_ip_block *ip_block)
 		}
 	}
 
+	adev->gfx.gfx_supported_reset =
+		amdgpu_get_soft_full_reset_mask(&adev->gfx.gfx_ring[0]);
+	adev->gfx.compute_supported_reset =
+		amdgpu_get_soft_full_reset_mask(&adev->gfx.compute_ring[0]);
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(11, 0, 0):
+	case IP_VERSION(11, 0, 2):
+	case IP_VERSION(11, 0, 3):
+		if ((adev->gfx.me_fw_version >= 2280) &&
+			    (adev->gfx.mec_fw_version >= 2410)) {
+				adev->gfx.compute_supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+				adev->gfx.gfx_supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+		}
+		break;
+	default:
+		break;
+	}
+
 	if (!adev->enable_mes_kiq) {
 		r = amdgpu_gfx_kiq_init(adev, GFX11_MEC_HPD_SIZE, 0);
 		if (r) {
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c
index 1b99f90cd193..fe7c48f2fb2a 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c
@@ -1437,6 +1437,12 @@ static int gfx_v12_0_sw_init(struct amdgpu_ip_block *ip_block)
 		}
 	}
 
+	/* TODO: Add queue reset mask when FW fully supports it */
+	adev->gfx.gfx_supported_reset =
+		amdgpu_get_soft_full_reset_mask(&adev->gfx.gfx_ring[0]);
+	adev->gfx.compute_supported_reset =
+		amdgpu_get_soft_full_reset_mask(&adev->gfx.compute_ring[0]);
+
 	if (!adev->enable_mes_kiq) {
 		r = amdgpu_gfx_kiq_init(adev, GFX12_MEC_HPD_SIZE, 0);
 		if (r) {
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
index a880dce16ae2..0b6f09f2cc9b 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
@@ -2374,6 +2374,12 @@ static int gfx_v9_0_sw_init(struct amdgpu_ip_block *ip_block)
 		}
 	}
 
+	/* TODO: Add queue reset mask when FW fully supports it */
+	adev->gfx.gfx_supported_reset =
+		amdgpu_get_soft_full_reset_mask(&adev->gfx.gfx_ring[0]);
+	adev->gfx.compute_supported_reset =
+		amdgpu_get_soft_full_reset_mask(&adev->gfx.compute_ring[0]);
+
 	r = amdgpu_gfx_kiq_init(adev, GFX9_MEC_HPD_SIZE, 0);
 	if (r) {
 		DRM_ERROR("Failed to init KIQ BOs!\n");
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c
index 983088805c3a..e2b3dda57030 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c
@@ -1157,6 +1157,19 @@ static int gfx_v9_4_3_sw_init(struct amdgpu_ip_block *ip_block)
 			return r;
 	}
 
+	adev->gfx.compute_supported_reset =
+		amdgpu_get_soft_full_reset_mask(&adev->gfx.compute_ring[0]);
+	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
+	case IP_VERSION(9, 4, 3):
+	case IP_VERSION(9, 4, 4):
+		if (adev->gfx.mec_fw_version >= 155) {
+			adev->gfx.compute_supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+			adev->gfx.compute_supported_reset |= AMDGPU_RESET_TYPE_PER_PIPE;
+		}
+		break;
+	default:
+		break;
+	}
 	r = gfx_v9_4_3_gpu_early_init(adev);
 	if (r)
 		return r;
-- 
2.51.0


From 2bb7dced1c2f8c0e705cc74840f776406db492c3 Mon Sep 17 00:00:00 2001
From: Yang Wang <kevinyang.wang@amd.com>
Date: Wed, 6 Nov 2024 14:49:56 +0800
Subject: [PATCH 05/16] drm/amdgpu: fix ACA bank count boundary check error

fix ACA bank count boundary check error.

Fixes: f5e4cc8461c4 ("drm/amdgpu: implement RAS ACA driver framework")
Signed-off-by: Yang Wang <kevinyang.wang@amd.com>
Reviewed-by: Tao Zhou <tao.zhou1@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_aca.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.c
index 2ca127173135..9d6345146495 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.c
@@ -158,7 +158,7 @@ static int aca_smu_get_valid_aca_banks(struct amdgpu_device *adev, enum aca_smu_
 		return -EINVAL;
 	}
 
-	if (start + count >= max_count)
+	if (start + count > max_count)
 		return -EINVAL;
 
 	count = min_t(int, count, max_count);
-- 
2.51.0


From fa31798582882740f2b13d19e1bd43b4ef918e2f Mon Sep 17 00:00:00 2001
From: Lijo Lazar <lijo.lazar@amd.com>
Date: Tue, 5 Nov 2024 10:30:20 +0530
Subject: [PATCH 06/16] drm/amdgpu: Fix map/unmap queue logic

In current logic, it calls ring_alloc followed by a ring_test. ring_test
in turn will call another ring_alloc. This is illegal usage as a
ring_alloc is expected to be closed properly with a ring_commit. Change
to commit the map/unmap queue packet first followed by a ring_test. Add a
comment about the usage of ring_test.

Also, reorder the current pre-condition checks of job hang or kiq ring
scheduler not ready. Without them being met, it is not useful to attempt
ring or memory allocations.

Fixes tag refers to the original patch which introduced this issue which
then got carried over into newer code.

Signed-off-by: Lijo Lazar <lijo.lazar@amd.com>
Reviewed-by: Le Ma <le.ma@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Fixes: 6c10b5cc4eaa ("drm/amdgpu: Remove duplicate code in gfx_v8_0.c")
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c | 13 ++++-
 drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c    | 63 +++++++++++++++-------
 drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c      |  7 +++
 3 files changed, 63 insertions(+), 20 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
index 24343c312480..3afcd1e8aa54 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
@@ -834,6 +834,9 @@ int amdgpu_amdkfd_unmap_hiq(struct amdgpu_device *adev, u32 doorbell_off,
 	if (!kiq->pmf || !kiq->pmf->kiq_unmap_queues)
 		return -EINVAL;
 
+	if (!kiq_ring->sched.ready || adev->job_hang)
+		return 0;
+
 	ring_funcs = kzalloc(sizeof(*ring_funcs), GFP_KERNEL);
 	if (!ring_funcs)
 		return -ENOMEM;
@@ -858,8 +861,14 @@ int amdgpu_amdkfd_unmap_hiq(struct amdgpu_device *adev, u32 doorbell_off,
 
 	kiq->pmf->kiq_unmap_queues(kiq_ring, ring, RESET_QUEUES, 0, 0);
 
-	if (kiq_ring->sched.ready && !adev->job_hang)
-		r = amdgpu_ring_test_helper(kiq_ring);
+	/* Submit unmap queue packet */
+	amdgpu_ring_commit(kiq_ring);
+	/*
+	 * Ring test will do a basic scratch register change check. Just run
+	 * this to ensure that unmap queues that is submitted before got
+	 * processed successfully before returning.
+	 */
+	r = amdgpu_ring_test_helper(kiq_ring);
 
 	spin_unlock(&kiq->ring_lock);
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
index 6cc6484bde06..5136cb3bdd4b 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
@@ -515,6 +515,17 @@ int amdgpu_gfx_disable_kcq(struct amdgpu_device *adev, int xcc_id)
 	if (!kiq->pmf || !kiq->pmf->kiq_unmap_queues)
 		return -EINVAL;
 
+	if (!kiq_ring->sched.ready || adev->job_hang)
+		return 0;
+	/**
+	 * This is workaround: only skip kiq_ring test
+	 * during ras recovery in suspend stage for gfx9.4.3
+	 */
+	if ((amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 3) ||
+	     amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 4)) &&
+	    amdgpu_ras_in_recovery(adev))
+		return 0;
+
 	spin_lock(&kiq->ring_lock);
 	if (amdgpu_ring_alloc(kiq_ring, kiq->pmf->unmap_queues_size *
 					adev->gfx.num_compute_rings)) {
@@ -528,20 +539,15 @@ int amdgpu_gfx_disable_kcq(struct amdgpu_device *adev, int xcc_id)
 					   &adev->gfx.compute_ring[j],
 					   RESET_QUEUES, 0, 0);
 	}
-
-	/**
-	 * This is workaround: only skip kiq_ring test
-	 * during ras recovery in suspend stage for gfx9.4.3
+	/* Submit unmap queue packet */
+	amdgpu_ring_commit(kiq_ring);
+	/*
+	 * Ring test will do a basic scratch register change check. Just run
+	 * this to ensure that unmap queues that is submitted before got
+	 * processed successfully before returning.
 	 */
-	if ((amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 3) ||
-	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 4)) &&
-	    amdgpu_ras_in_recovery(adev)) {
-		spin_unlock(&kiq->ring_lock);
-		return 0;
-	}
+	r = amdgpu_ring_test_helper(kiq_ring);
 
-	if (kiq_ring->sched.ready && !adev->job_hang)
-		r = amdgpu_ring_test_helper(kiq_ring);
 	spin_unlock(&kiq->ring_lock);
 
 	return r;
@@ -569,8 +575,11 @@ int amdgpu_gfx_disable_kgq(struct amdgpu_device *adev, int xcc_id)
 	if (!kiq->pmf || !kiq->pmf->kiq_unmap_queues)
 		return -EINVAL;
 
-	spin_lock(&kiq->ring_lock);
+	if (!adev->gfx.kiq[0].ring.sched.ready || adev->job_hang)
+		return 0;
+
 	if (amdgpu_gfx_is_master_xcc(adev, xcc_id)) {
+		spin_lock(&kiq->ring_lock);
 		if (amdgpu_ring_alloc(kiq_ring, kiq->pmf->unmap_queues_size *
 						adev->gfx.num_gfx_rings)) {
 			spin_unlock(&kiq->ring_lock);
@@ -583,11 +592,17 @@ int amdgpu_gfx_disable_kgq(struct amdgpu_device *adev, int xcc_id)
 						   &adev->gfx.gfx_ring[j],
 						   PREEMPT_QUEUES, 0, 0);
 		}
-	}
+		/* Submit unmap queue packet */
+		amdgpu_ring_commit(kiq_ring);
 
-	if (adev->gfx.kiq[0].ring.sched.ready && !adev->job_hang)
+		/*
+		 * Ring test will do a basic scratch register change check.
+		 * Just run this to ensure that unmap queues that is submitted
+		 * before got processed successfully before returning.
+		 */
 		r = amdgpu_ring_test_helper(kiq_ring);
-	spin_unlock(&kiq->ring_lock);
+		spin_unlock(&kiq->ring_lock);
+	}
 
 	return r;
 }
@@ -692,7 +707,13 @@ int amdgpu_gfx_enable_kcq(struct amdgpu_device *adev, int xcc_id)
 		kiq->pmf->kiq_map_queues(kiq_ring,
 					 &adev->gfx.compute_ring[j]);
 	}
-
+	/* Submit map queue packet */
+	amdgpu_ring_commit(kiq_ring);
+	/*
+	 * Ring test will do a basic scratch register change check. Just run
+	 * this to ensure that map queues that is submitted before got
+	 * processed successfully before returning.
+	 */
 	r = amdgpu_ring_test_helper(kiq_ring);
 	spin_unlock(&kiq->ring_lock);
 	if (r)
@@ -743,7 +764,13 @@ int amdgpu_gfx_enable_kgq(struct amdgpu_device *adev, int xcc_id)
 						 &adev->gfx.gfx_ring[j]);
 		}
 	}
-
+	/* Submit map queue packet */
+	amdgpu_ring_commit(kiq_ring);
+	/*
+	 * Ring test will do a basic scratch register change check. Just run
+	 * this to ensure that map queues that is submitted before got
+	 * processed successfully before returning.
+	 */
 	r = amdgpu_ring_test_helper(kiq_ring);
 	spin_unlock(&kiq->ring_lock);
 	if (r)
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
index 480c41ee947e..b7006c41e270 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
@@ -4823,6 +4823,13 @@ static int gfx_v8_0_kcq_disable(struct amdgpu_device *adev)
 		amdgpu_ring_write(kiq_ring, 0);
 		amdgpu_ring_write(kiq_ring, 0);
 	}
+	/* Submit unmap queue packet */
+	amdgpu_ring_commit(kiq_ring);
+	/*
+	 * Ring test will do a basic scratch register change check. Just run
+	 * this to ensure that unmap queues that is submitted before got
+	 * processed successfully before returning.
+	 */
 	r = amdgpu_ring_test_helper(kiq_ring);
 	if (r)
 		DRM_ERROR("KCQ disable failed\n");
-- 
2.51.0


From 7b1ebbe856fcb3d870017c0682d97e3d3376bf82 Mon Sep 17 00:00:00 2001
From: Lijo Lazar <lijo.lazar@amd.com>
Date: Tue, 5 Nov 2024 11:04:09 +0530
Subject: [PATCH 07/16] drm/amdgpu: Avoid kcq disable during reset

Reset sequence indicates that hardware already ran into a bad state.
Avoid sending unmap queue request to reset KCQ. This will also cover RAS
error scenarios which need a reset to recover, hence remove the check.

Signed-off-by: Lijo Lazar <lijo.lazar@amd.com>
Reviewed-by: Le Ma <le.ma@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c | 10 +---------
 1 file changed, 1 insertion(+), 9 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
index 5136cb3bdd4b..826a41bd10df 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
@@ -515,15 +515,7 @@ int amdgpu_gfx_disable_kcq(struct amdgpu_device *adev, int xcc_id)
 	if (!kiq->pmf || !kiq->pmf->kiq_unmap_queues)
 		return -EINVAL;
 
-	if (!kiq_ring->sched.ready || adev->job_hang)
-		return 0;
-	/**
-	 * This is workaround: only skip kiq_ring test
-	 * during ras recovery in suspend stage for gfx9.4.3
-	 */
-	if ((amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 3) ||
-	     amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 4)) &&
-	    amdgpu_ras_in_recovery(adev))
+	if (!kiq_ring->sched.ready || adev->job_hang || amdgpu_in_reset(adev))
 		return 0;
 
 	spin_lock(&kiq->ring_lock);
-- 
2.51.0


From edd345f7ef799848a9e2be9de82bbfcb98fdcc43 Mon Sep 17 00:00:00 2001
From: Sathishkumar S <sathishkumar.sundararaju@amd.com>
Date: Tue, 5 Nov 2024 10:53:01 +0530
Subject: [PATCH 08/16] drm/amdgpu: Normalize reg offsets on VCN v4.0.3

Remote access to external AIDs isn't possible with VCN RRMT disabled
and it is disabled on SoCs with GC 9.4.4, so use only local offsets.

Signed-off-by: Sathishkumar S <sathishkumar.sundararaju@amd.com>
Reviewed-by: Lijo Lazar <lijo.lazar@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/vcn_v4_0_3.c | 15 +++++++++++----
 1 file changed, 11 insertions(+), 4 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_3.c b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_3.c
index 0d5c94bfc0ef..cf808a153fce 100644
--- a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_3.c
+++ b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_3.c
@@ -95,6 +95,13 @@ static void vcn_v4_0_3_unified_ring_set_wptr(struct amdgpu_ring *ring);
 static void vcn_v4_0_3_set_ras_funcs(struct amdgpu_device *adev);
 static void vcn_v4_0_3_enable_ras(struct amdgpu_device *adev,
 				  int inst_idx, bool indirect);
+
+static inline bool vcn_v4_0_3_normalizn_reqd(struct amdgpu_device *adev)
+{
+	return (amdgpu_sriov_vf(adev) ||
+		(amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 4)));
+}
+
 /**
  * vcn_v4_0_3_early_init - set function pointers
  *
@@ -1428,8 +1435,8 @@ static uint64_t vcn_v4_0_3_unified_ring_get_wptr(struct amdgpu_ring *ring)
 static void vcn_v4_0_3_enc_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg,
 				uint32_t val, uint32_t mask)
 {
-	/* For VF, only local offsets should be used */
-	if (amdgpu_sriov_vf(ring->adev))
+	/* Use normalized offsets when required */
+	if (vcn_v4_0_3_normalizn_reqd(ring->adev))
 		reg = NORMALIZE_VCN_REG_OFFSET(reg);
 
 	amdgpu_ring_write(ring, VCN_ENC_CMD_REG_WAIT);
@@ -1440,8 +1447,8 @@ static void vcn_v4_0_3_enc_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t
 
 static void vcn_v4_0_3_enc_ring_emit_wreg(struct amdgpu_ring *ring, uint32_t reg, uint32_t val)
 {
-	/* For VF, only local offsets should be used */
-	if (amdgpu_sriov_vf(ring->adev))
+	/* Use normalized offsets when required */
+	if (vcn_v4_0_3_normalizn_reqd(ring->adev))
 		reg = NORMALIZE_VCN_REG_OFFSET(reg);
 
 	amdgpu_ring_write(ring, VCN_ENC_CMD_REG_WRITE);
-- 
2.51.0


From 59fd50b8663b4e703b44f9b51a2e715dc6e344c1 Mon Sep 17 00:00:00 2001
From: "Jesse.zhang@amd.com" <Jesse.zhang@amd.com>
Date: Mon, 4 Nov 2024 13:15:49 +0800
Subject: [PATCH 09/16] drm/amdgpu: Add sysfs interface for sdma reset mask

Add the sysfs interface for sdma:
sdma_reset_mask

The interface is read-only and show the resets supported by the IP.
For example, full adapter reset (mode1/mode2/BACO/etc),
soft reset, queue reset, and pipe reset.

V2: the sysfs node returns a text string instead of some flags (Christian)
v3: add a generic helper which takes the ring as parameter
   and print the strings in the order they are applied (Christian)

   check amdgpu_gpu_recovery  before creating sysfs file itself,
   and initialize supported_reset_types in IP version files (Lijo)

Signed-off-by: Jesse Zhang <Jesse.Zhang@amd.com>
Suggested-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Tim Huang <tim.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.c | 41 ++++++++++++++++++++++++
 drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.h |  3 ++
 drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c |  9 ++++++
 drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c   | 18 +++++++++++
 drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c   | 23 +++++++++++++
 drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c   | 18 +++++++++++
 6 files changed, 112 insertions(+)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.c
index 5868b4a32ea6..8c89b69edc20 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.c
@@ -413,3 +413,44 @@ void amdgpu_debugfs_sdma_sched_mask_init(struct amdgpu_device *adev)
 			    &amdgpu_debugfs_sdma_sched_mask_fops);
 #endif
 }
+
+static ssize_t amdgpu_get_sdma_reset_mask(struct device *dev,
+						struct device_attribute *attr,
+						char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+
+	if (!adev)
+		return -ENODEV;
+
+	return amdgpu_show_reset_mask(buf, adev->sdma.supported_reset);
+}
+
+static DEVICE_ATTR(sdma_reset_mask, 0444,
+		   amdgpu_get_sdma_reset_mask, NULL);
+
+int amdgpu_sdma_sysfs_reset_mask_init(struct amdgpu_device *adev)
+{
+	int r = 0;
+
+	if (!amdgpu_gpu_recovery)
+		return r;
+
+	if (adev->sdma.num_instances) {
+		r = device_create_file(adev->dev, &dev_attr_sdma_reset_mask);
+		if (r)
+			return r;
+	}
+
+	return r;
+}
+
+void amdgpu_sdma_sysfs_reset_mask_fini(struct amdgpu_device *adev)
+{
+	if (!amdgpu_gpu_recovery)
+		return;
+
+	if (adev->sdma.num_instances)
+		device_remove_file(adev->dev, &dev_attr_sdma_reset_mask);
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.h
index a37fcd9bb981..2db58b5812a8 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.h
@@ -116,6 +116,7 @@ struct amdgpu_sdma {
 	struct ras_common_if	*ras_if;
 	struct amdgpu_sdma_ras	*ras;
 	uint32_t		*ip_dump;
+	uint32_t 		supported_reset;
 };
 
 /*
@@ -176,4 +177,6 @@ void amdgpu_sdma_destroy_inst_ctx(struct amdgpu_device *adev,
         bool duplicate);
 int amdgpu_sdma_ras_sw_init(struct amdgpu_device *adev);
 void amdgpu_debugfs_sdma_sched_mask_init(struct amdgpu_device *adev);
+int amdgpu_sdma_sysfs_reset_mask_init(struct amdgpu_device *adev);
+void amdgpu_sdma_sysfs_reset_mask_fini(struct amdgpu_device *adev);
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c b/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c
index 9c7cea0890c9..a38553f38fdc 100644
--- a/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c
@@ -1430,6 +1430,10 @@ static int sdma_v4_4_2_sw_init(struct amdgpu_ip_block *ip_block)
 		}
 	}
 
+	/* TODO: Add queue reset mask when FW fully supports it */
+	adev->sdma.supported_reset =
+		amdgpu_get_soft_full_reset_mask(&adev->sdma.instance[0].ring);
+
 	if (amdgpu_sdma_ras_sw_init(adev)) {
 		dev_err(adev->dev, "fail to initialize sdma ras block\n");
 		return -EINVAL;
@@ -1442,6 +1446,10 @@ static int sdma_v4_4_2_sw_init(struct amdgpu_ip_block *ip_block)
 	else
 		DRM_ERROR("Failed to allocated memory for SDMA IP Dump\n");
 
+	r = amdgpu_sdma_sysfs_reset_mask_init(adev);
+	if (r)
+		return r;
+
 	return r;
 }
 
@@ -1456,6 +1464,7 @@ static int sdma_v4_4_2_sw_fini(struct amdgpu_ip_block *ip_block)
 			amdgpu_ring_fini(&adev->sdma.instance[i].page);
 	}
 
+	amdgpu_sdma_sysfs_reset_mask_fini(adev);
 	if (amdgpu_ip_version(adev, SDMA0_HWIP, 0) == IP_VERSION(4, 4, 2) ||
 	    amdgpu_ip_version(adev, SDMA0_HWIP, 0) == IP_VERSION(4, 4, 5))
 		amdgpu_sdma_destroy_inst_ctx(adev, true);
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c
index d31c4860933f..fa9b40934957 100644
--- a/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c
@@ -1452,6 +1452,19 @@ static int sdma_v5_0_sw_init(struct amdgpu_ip_block *ip_block)
 			return r;
 	}
 
+	adev->sdma.supported_reset =
+		amdgpu_get_soft_full_reset_mask(&adev->sdma.instance[0].ring);
+	switch (amdgpu_ip_version(adev, SDMA0_HWIP, 0)) {
+	case IP_VERSION(5, 0, 0):
+	case IP_VERSION(5, 0, 2):
+	case IP_VERSION(5, 0, 5):
+		if (adev->sdma.instance[0].fw_version >= 35)
+			adev->sdma.supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+		break;
+	default:
+		break;
+	}
+
 	/* Allocate memory for SDMA IP Dump buffer */
 	ptr = kcalloc(adev->sdma.num_instances * reg_count, sizeof(uint32_t), GFP_KERNEL);
 	if (ptr)
@@ -1459,6 +1472,10 @@ static int sdma_v5_0_sw_init(struct amdgpu_ip_block *ip_block)
 	else
 		DRM_ERROR("Failed to allocated memory for SDMA IP Dump\n");
 
+	r = amdgpu_sdma_sysfs_reset_mask_init(adev);
+	if (r)
+		return r;
+
 	return r;
 }
 
@@ -1470,6 +1487,7 @@ static int sdma_v5_0_sw_fini(struct amdgpu_ip_block *ip_block)
 	for (i = 0; i < adev->sdma.num_instances; i++)
 		amdgpu_ring_fini(&adev->sdma.instance[i].ring);
 
+	amdgpu_sdma_sysfs_reset_mask_fini(adev);
 	amdgpu_sdma_destroy_inst_ctx(adev, false);
 
 	kfree(adev->sdma.ip_dump);
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c b/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c
index ffa8c62ac101..ba5160399ab2 100644
--- a/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c
@@ -1357,6 +1357,24 @@ static int sdma_v5_2_sw_init(struct amdgpu_ip_block *ip_block)
 			return r;
 	}
 
+	adev->sdma.supported_reset =
+		amdgpu_get_soft_full_reset_mask(&adev->sdma.instance[0].ring);
+	switch (amdgpu_ip_version(adev, SDMA0_HWIP, 0)) {
+	case IP_VERSION(5, 2, 0):
+	case IP_VERSION(5, 2, 2):
+	case IP_VERSION(5, 2, 3):
+	case IP_VERSION(5, 2, 4):
+		if (adev->sdma.instance[0].fw_version >= 76)
+			adev->sdma.supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+		break;
+	case IP_VERSION(5, 2, 5):
+		if (adev->sdma.instance[0].fw_version >= 34)
+			adev->sdma.supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+		break;
+	default:
+		break;
+	}
+
 	/* Allocate memory for SDMA IP Dump buffer */
 	ptr = kcalloc(adev->sdma.num_instances * reg_count, sizeof(uint32_t), GFP_KERNEL);
 	if (ptr)
@@ -1364,6 +1382,10 @@ static int sdma_v5_2_sw_init(struct amdgpu_ip_block *ip_block)
 	else
 		DRM_ERROR("Failed to allocated memory for SDMA IP Dump\n");
 
+	r = amdgpu_sdma_sysfs_reset_mask_init(adev);
+	if (r)
+		return r;
+
 	return r;
 }
 
@@ -1375,6 +1397,7 @@ static int sdma_v5_2_sw_fini(struct amdgpu_ip_block *ip_block)
 	for (i = 0; i < adev->sdma.num_instances; i++)
 		amdgpu_ring_fini(&adev->sdma.instance[i].ring);
 
+	amdgpu_sdma_sysfs_reset_mask_fini(adev);
 	amdgpu_sdma_destroy_inst_ctx(adev, true);
 
 	kfree(adev->sdma.ip_dump);
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c
index 5635f2d84090..d46128b0ec92 100644
--- a/drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c
@@ -1350,6 +1350,19 @@ static int sdma_v6_0_sw_init(struct amdgpu_ip_block *ip_block)
 			return r;
 	}
 
+	adev->sdma.supported_reset =
+		amdgpu_get_soft_full_reset_mask(&adev->sdma.instance[0].ring);
+	switch (amdgpu_ip_version(adev, SDMA0_HWIP, 0)) {
+	case IP_VERSION(6, 0, 0):
+	case IP_VERSION(6, 0, 2):
+	case IP_VERSION(6, 0, 3):
+		if (adev->sdma.instance[0].fw_version >= 21)
+			adev->sdma.supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE;
+		break;
+	default:
+		break;
+	}
+
 	if (amdgpu_sdma_ras_sw_init(adev)) {
 		dev_err(adev->dev, "Failed to initialize sdma ras block!\n");
 		return -EINVAL;
@@ -1362,6 +1375,10 @@ static int sdma_v6_0_sw_init(struct amdgpu_ip_block *ip_block)
 	else
 		DRM_ERROR("Failed to allocated memory for SDMA IP Dump\n");
 
+	r = amdgpu_sdma_sysfs_reset_mask_init(adev);
+	if (r)
+		return r;
+
 	return r;
 }
 
@@ -1373,6 +1390,7 @@ static int sdma_v6_0_sw_fini(struct amdgpu_ip_block *ip_block)
 	for (i = 0; i < adev->sdma.num_instances; i++)
 		amdgpu_ring_fini(&adev->sdma.instance[i].ring);
 
+	amdgpu_sdma_sysfs_reset_mask_fini(adev);
 	amdgpu_sdma_destroy_inst_ctx(adev, true);
 
 	kfree(adev->sdma.ip_dump);
-- 
2.51.0


From ea02ea9437deebb3d997e9662022159953ecf7e0 Mon Sep 17 00:00:00 2001
From: "Jesse.zhang@amd.com" <Jesse.zhang@amd.com>
Date: Tue, 29 Oct 2024 14:33:05 +0800
Subject: [PATCH 10/16] drm/amdgpu: Add sysfs interface for vpe reset mask

Add the sysfs interface for vpe:
    vpe_reset_mask

The interface is read-only and show the resets supported by the IP.
For example, full adapter reset (mode1/mode2/BACO/etc),
soft reset, queue reset, and pipe reset.

V2: the sysfs node returns a text string instead of some flags (Christian)
v3: add a generic helper which takes the ring as parameter
    and print the strings in the order they are applied (Christian)

    check amdgpu_gpu_recovery  before creating sysfs file itself,
    and initialize supported_reset_types in IP version files (Lijo)

Signed-off-by: Jesse Zhang <Jesse.Zhang@amd.com>
Suggested-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Tim Huang <tim.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_vpe.c | 43 +++++++++++++++++++++++++
 drivers/gpu/drm/amd/amdgpu/amdgpu_vpe.h |  3 ++
 2 files changed, 46 insertions(+)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vpe.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vpe.c
index 46713a158d90..3e6f9dfb61bb 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vpe.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vpe.c
@@ -377,6 +377,13 @@ static int vpe_sw_init(struct amdgpu_ip_block *ip_block)
 	ret = vpe_init_microcode(vpe);
 	if (ret)
 		goto out;
+
+	/* TODO: Add queue reset mask when FW fully supports it */
+	adev->vpe.supported_reset =
+		 amdgpu_get_soft_full_reset_mask(&adev->vpe.ring);
+	ret = amdgpu_vpe_sysfs_reset_mask_init(adev);
+	if (ret)
+		goto out;
 out:
 	return ret;
 }
@@ -389,6 +396,7 @@ static int vpe_sw_fini(struct amdgpu_ip_block *ip_block)
 	release_firmware(vpe->fw);
 	vpe->fw = NULL;
 
+	amdgpu_vpe_sysfs_reset_mask_fini(adev);
 	vpe_ring_fini(vpe);
 
 	amdgpu_bo_free_kernel(&adev->vpe.cmdbuf_obj,
@@ -865,6 +873,41 @@ static void vpe_ring_end_use(struct amdgpu_ring *ring)
 	schedule_delayed_work(&adev->vpe.idle_work, VPE_IDLE_TIMEOUT);
 }
 
+static ssize_t amdgpu_get_vpe_reset_mask(struct device *dev,
+						struct device_attribute *attr,
+						char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+
+	if (!adev)
+		return -ENODEV;
+
+	return amdgpu_show_reset_mask(buf, adev->vpe.supported_reset);
+}
+
+static DEVICE_ATTR(vpe_reset_mask, 0444,
+		   amdgpu_get_vpe_reset_mask, NULL);
+
+int amdgpu_vpe_sysfs_reset_mask_init(struct amdgpu_device *adev)
+{
+	int r = 0;
+
+	if (adev->vpe.num_instances) {
+		r = device_create_file(adev->dev, &dev_attr_vpe_reset_mask);
+		if (r)
+			return r;
+	}
+
+	return r;
+}
+
+void amdgpu_vpe_sysfs_reset_mask_fini(struct amdgpu_device *adev)
+{
+	if (adev->vpe.num_instances)
+		device_remove_file(adev->dev, &dev_attr_vpe_reset_mask);
+}
+
 static const struct amdgpu_ring_funcs vpe_ring_funcs = {
 	.type = AMDGPU_RING_TYPE_VPE,
 	.align_mask = 0xf,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vpe.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vpe.h
index 231d86d0953e..695da740a97e 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vpe.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vpe.h
@@ -79,6 +79,7 @@ struct amdgpu_vpe {
 
 	uint32_t			num_instances;
 	bool				collaborate_mode;
+	uint32_t			supported_reset;
 };
 
 int amdgpu_vpe_psp_update_sram(struct amdgpu_device *adev);
@@ -86,6 +87,8 @@ int amdgpu_vpe_init_microcode(struct amdgpu_vpe *vpe);
 int amdgpu_vpe_ring_init(struct amdgpu_vpe *vpe);
 int amdgpu_vpe_ring_fini(struct amdgpu_vpe *vpe);
 int amdgpu_vpe_configure_dpm(struct amdgpu_vpe *vpe);
+void amdgpu_vpe_sysfs_reset_mask_fini(struct amdgpu_device *adev);
+int amdgpu_vpe_sysfs_reset_mask_init(struct amdgpu_device *adev);
 
 #define vpe_ring_init(vpe) ((vpe)->funcs->ring_init ? (vpe)->funcs->ring_init((vpe)) : 0)
 #define vpe_ring_start(vpe) ((vpe)->funcs->ring_start ? (vpe)->funcs->ring_start((vpe)) : 0)
-- 
2.51.0


From 96f0b56c34d257e4e7532eb99a071ca6c8497467 Mon Sep 17 00:00:00 2001
From: "Jesse.zhang@amd.com" <Jesse.zhang@amd.com>
Date: Tue, 29 Oct 2024 14:36:35 +0800
Subject: [PATCH 11/16] drm/amdgpu: Add sysfs interface for jpeg reset mask

Add the sysfs interface for jpeg:
jpeg_reset_mask

The interface is read-only and show the resets supported by the IP.
For example, full adapter reset (mode1/mode2/BACO/etc),
soft reset, queue reset, and pipe reset.

V2: the sysfs node returns a text string instead of some flags (Christian)
v3: add a generic helper which takes the ring as parameter
    and print the strings in the order they are applied (Christian)

    check amdgpu_gpu_recovery  before creating sysfs file itself,
    and initialize supported_reset_types in IP version files (Lijo)

Signed-off-by: Jesse Zhang <Jesse.Zhang@amd.com>
Suggested-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Tim Huang <tim.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.c | 35 ++++++++++++++++++++++++
 drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.h |  3 ++
 drivers/gpu/drm/amd/amdgpu/jpeg_v4_0.c   |  7 +++++
 drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.c |  8 ++++++
 drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_5.c |  8 ++++++
 drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_0.c |  7 +++++
 6 files changed, 68 insertions(+)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.c
index 95e2796919fc..f971ffdffce9 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.c
@@ -415,3 +415,38 @@ void amdgpu_debugfs_jpeg_sched_mask_init(struct amdgpu_device *adev)
 			    &amdgpu_debugfs_jpeg_sched_mask_fops);
 #endif
 }
+
+static ssize_t amdgpu_get_jpeg_reset_mask(struct device *dev,
+						struct device_attribute *attr,
+						char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+
+	if (!adev)
+		return -ENODEV;
+
+	return amdgpu_show_reset_mask(buf, adev->jpeg.supported_reset);
+}
+
+static DEVICE_ATTR(jpeg_reset_mask, 0444,
+		   amdgpu_get_jpeg_reset_mask, NULL);
+
+int amdgpu_jpeg_sysfs_reset_mask_init(struct amdgpu_device *adev)
+{
+	int r = 0;
+
+	if (adev->jpeg.num_jpeg_inst) {
+		r = device_create_file(adev->dev, &dev_attr_jpeg_reset_mask);
+		if (r)
+			return r;
+	}
+
+	return r;
+}
+
+void amdgpu_jpeg_sysfs_reset_mask_fini(struct amdgpu_device *adev)
+{
+	if (adev->jpeg.num_jpeg_inst)
+		device_remove_file(adev->dev, &dev_attr_jpeg_reset_mask);
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.h
index 819dc7a0af99..3eb4a4653fce 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.h
@@ -128,6 +128,7 @@ struct amdgpu_jpeg {
 	uint16_t inst_mask;
 	uint8_t num_inst_per_aid;
 	bool	indirect_sram;
+	uint32_t supported_reset;
 };
 
 int amdgpu_jpeg_sw_init(struct amdgpu_device *adev);
@@ -150,5 +151,7 @@ int amdgpu_jpeg_ras_sw_init(struct amdgpu_device *adev);
 int amdgpu_jpeg_psp_update_sram(struct amdgpu_device *adev, int inst_idx,
 			       enum AMDGPU_UCODE_ID ucode_id);
 void amdgpu_debugfs_jpeg_sched_mask_init(struct amdgpu_device *adev);
+int amdgpu_jpeg_sysfs_reset_mask_init(struct amdgpu_device *adev);
+void amdgpu_jpeg_sysfs_reset_mask_fini(struct amdgpu_device *adev);
 
 #endif /*__AMDGPU_JPEG_H__*/
diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0.c
index 89953c0f5f1f..193dfac5dc76 100644
--- a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0.c
@@ -121,6 +121,12 @@ static int jpeg_v4_0_sw_init(struct amdgpu_ip_block *ip_block)
 	adev->jpeg.inst->external.jpeg_pitch[0] = SOC15_REG_OFFSET(JPEG, 0, regUVD_JPEG_PITCH);
 
 	r = amdgpu_jpeg_ras_sw_init(adev);
+	if (r)
+		return r;
+	/* TODO: Add queue reset mask when FW fully supports it */
+	adev->jpeg.supported_reset =
+		amdgpu_get_soft_full_reset_mask(&adev->jpeg.inst[0].ring_dec[0]);
+	r = amdgpu_jpeg_sysfs_reset_mask_init(adev);
 	if (r)
 		return r;
 
@@ -143,6 +149,7 @@ static int jpeg_v4_0_sw_fini(struct amdgpu_ip_block *ip_block)
 	if (r)
 		return r;
 
+	amdgpu_jpeg_sysfs_reset_mask_fini(adev);
 	r = amdgpu_jpeg_sw_fini(adev);
 
 	return r;
diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.c
index 6917e4a8e96a..67b51bcbacd1 100644
--- a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.c
+++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.c
@@ -159,6 +159,13 @@ static int jpeg_v4_0_3_sw_init(struct amdgpu_ip_block *ip_block)
 		}
 	}
 
+	/* TODO: Add queue reset mask when FW fully supports it */
+	adev->jpeg.supported_reset =
+		amdgpu_get_soft_full_reset_mask(&adev->jpeg.inst[0].ring_dec[0]);
+	r = amdgpu_jpeg_sysfs_reset_mask_init(adev);
+	if (r)
+		return r;
+
 	return 0;
 }
 
@@ -178,6 +185,7 @@ static int jpeg_v4_0_3_sw_fini(struct amdgpu_ip_block *ip_block)
 	if (r)
 		return r;
 
+	amdgpu_jpeg_sysfs_reset_mask_fini(adev);
 	r = amdgpu_jpeg_sw_fini(adev);
 
 	return r;
diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_5.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_5.c
index f3cce523f3cb..b48e2412e6cc 100644
--- a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_5.c
+++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_5.c
@@ -153,6 +153,13 @@ static int jpeg_v4_0_5_sw_init(struct amdgpu_ip_block *ip_block)
 		adev->jpeg.inst[i].external.jpeg_pitch[0] = SOC15_REG_OFFSET(JPEG, i, regUVD_JPEG_PITCH);
 	}
 
+	/* TODO: Add queue reset mask when FW fully supports it */
+	adev->jpeg.supported_reset =
+		amdgpu_get_soft_full_reset_mask(&adev->jpeg.inst[0].ring_dec[0]);
+	r = amdgpu_jpeg_sysfs_reset_mask_init(adev);
+	if (r)
+		return r;
+
 	return 0;
 }
 
@@ -172,6 +179,7 @@ static int jpeg_v4_0_5_sw_fini(struct amdgpu_ip_block *ip_block)
 	if (r)
 		return r;
 
+	amdgpu_jpeg_sysfs_reset_mask_fini(adev);
 	r = amdgpu_jpeg_sw_fini(adev);
 
 	return r;
diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_0.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_0.c
index 06840d1dae79..686f9605239d 100644
--- a/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_0.c
@@ -100,6 +100,12 @@ static int jpeg_v5_0_0_sw_init(struct amdgpu_ip_block *ip_block)
 	adev->jpeg.internal.jpeg_pitch[0] = regUVD_JPEG_PITCH_INTERNAL_OFFSET;
 	adev->jpeg.inst->external.jpeg_pitch[0] = SOC15_REG_OFFSET(JPEG, 0, regUVD_JPEG_PITCH);
 
+	/* TODO: Add queue reset mask when FW fully supports it */
+	adev->jpeg.supported_reset =
+		amdgpu_get_soft_full_reset_mask(&adev->jpeg.inst[0].ring_dec[0]);
+	r = amdgpu_jpeg_sysfs_reset_mask_init(adev);
+	if (r)
+		return r;
 	return 0;
 }
 
@@ -119,6 +125,7 @@ static int jpeg_v5_0_0_sw_fini(struct amdgpu_ip_block *ip_block)
 	if (r)
 		return r;
 
+	amdgpu_jpeg_sysfs_reset_mask_fini(adev);
 	r = amdgpu_jpeg_sw_fini(adev);
 
 	return r;
-- 
2.51.0


From 69e9a9e65b1ea542d07e3fdd4222b46e9f5a3a29 Mon Sep 17 00:00:00 2001
From: David Rosca <david.rosca@amd.com>
Date: Mon, 21 Oct 2024 09:36:11 +0200
Subject: [PATCH 12/16] drm/amdgpu: Fix video caps for H264 and HEVC encode
 maximum size

H264 supports 4096x4096 starting from Polaris.
HEVC also supports 4096x4096, with VCN 3 and newer 8192x4352
is supported.

Signed-off-by: David Rosca <david.rosca@amd.com>
Reviewed-by: Leo Liu <leo.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/nv.c    | 12 ++++++------
 drivers/gpu/drm/amd/amdgpu/soc15.c |  4 ++--
 drivers/gpu/drm/amd/amdgpu/soc21.c | 12 ++++++------
 drivers/gpu/drm/amd/amdgpu/soc24.c |  2 +-
 drivers/gpu/drm/amd/amdgpu/vi.c    |  8 ++++----
 5 files changed, 19 insertions(+), 19 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/nv.c b/drivers/gpu/drm/amd/amdgpu/nv.c
index 6b72169be8f8..3bad565ded73 100644
--- a/drivers/gpu/drm/amd/amdgpu/nv.c
+++ b/drivers/gpu/drm/amd/amdgpu/nv.c
@@ -67,8 +67,8 @@ static const struct amd_ip_funcs nv_common_ip_funcs;
 
 /* Navi */
 static const struct amdgpu_video_codec_info nv_video_codecs_encode_array[] = {
-	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 2304, 0)},
-	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 4096, 2304, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 4096, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 4096, 4096, 0)},
 };
 
 static const struct amdgpu_video_codecs nv_video_codecs_encode = {
@@ -94,8 +94,8 @@ static const struct amdgpu_video_codecs nv_video_codecs_decode = {
 
 /* Sienna Cichlid */
 static const struct amdgpu_video_codec_info sc_video_codecs_encode_array[] = {
-	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 2160, 0)},
-	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 7680, 4352, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 4096, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 8192, 4352, 0)},
 };
 
 static const struct amdgpu_video_codecs sc_video_codecs_encode = {
@@ -136,8 +136,8 @@ static const struct amdgpu_video_codecs sc_video_codecs_decode_vcn1 = {
 
 /* SRIOV Sienna Cichlid, not const since data is controlled by host */
 static struct amdgpu_video_codec_info sriov_sc_video_codecs_encode_array[] = {
-	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 2160, 0)},
-	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 7680, 4352, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 4096, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 8192, 4352, 0)},
 };
 
 static struct amdgpu_video_codec_info sriov_sc_video_codecs_decode_array_vcn0[] = {
diff --git a/drivers/gpu/drm/amd/amdgpu/soc15.c b/drivers/gpu/drm/amd/amdgpu/soc15.c
index 533b4b2b432d..ede072758dab 100644
--- a/drivers/gpu/drm/amd/amdgpu/soc15.c
+++ b/drivers/gpu/drm/amd/amdgpu/soc15.c
@@ -90,8 +90,8 @@ static const struct amd_ip_funcs soc15_common_ip_funcs;
 /* Vega, Raven, Arcturus */
 static const struct amdgpu_video_codec_info vega_video_codecs_encode_array[] =
 {
-	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 2304, 0)},
-	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 4096, 2304, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 4096, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 4096, 4096, 0)},
 };
 
 static const struct amdgpu_video_codecs vega_video_codecs_encode =
diff --git a/drivers/gpu/drm/amd/amdgpu/soc21.c b/drivers/gpu/drm/amd/amdgpu/soc21.c
index 93fbb3354720..d6999835918f 100644
--- a/drivers/gpu/drm/amd/amdgpu/soc21.c
+++ b/drivers/gpu/drm/amd/amdgpu/soc21.c
@@ -49,13 +49,13 @@ static const struct amd_ip_funcs soc21_common_ip_funcs;
 
 /* SOC21 */
 static const struct amdgpu_video_codec_info vcn_4_0_0_video_codecs_encode_array_vcn0[] = {
-	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 2304, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 4096, 0)},
 	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 8192, 4352, 0)},
 	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_AV1, 8192, 4352, 0)},
 };
 
 static const struct amdgpu_video_codec_info vcn_4_0_0_video_codecs_encode_array_vcn1[] = {
-	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 2304, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 4096, 0)},
 	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 8192, 4352, 0)},
 };
 
@@ -96,14 +96,14 @@ static const struct amdgpu_video_codecs vcn_4_0_0_video_codecs_decode_vcn1 = {
 
 /* SRIOV SOC21, not const since data is controlled by host */
 static struct amdgpu_video_codec_info sriov_vcn_4_0_0_video_codecs_encode_array_vcn0[] = {
-	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 2304, 0)},
-	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 4096, 2304, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 4096, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 8192, 4352, 0)},
 	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_AV1, 8192, 4352, 0)},
 };
 
 static struct amdgpu_video_codec_info sriov_vcn_4_0_0_video_codecs_encode_array_vcn1[] = {
-	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 2304, 0)},
-	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 4096, 2304, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 4096, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 8192, 4352, 0)},
 };
 
 static struct amdgpu_video_codecs sriov_vcn_4_0_0_video_codecs_encode_vcn0 = {
diff --git a/drivers/gpu/drm/amd/amdgpu/soc24.c b/drivers/gpu/drm/amd/amdgpu/soc24.c
index 3af10ef4b793..be96de92b2f5 100644
--- a/drivers/gpu/drm/amd/amdgpu/soc24.c
+++ b/drivers/gpu/drm/amd/amdgpu/soc24.c
@@ -48,7 +48,7 @@
 static const struct amd_ip_funcs soc24_common_ip_funcs;
 
 static const struct amdgpu_video_codec_info vcn_5_0_0_video_codecs_encode_array_vcn0[] = {
-	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 2304, 0)},
+	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 4096, 0)},
 	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 8192, 4352, 0)},
 	{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_AV1, 8192, 4352, 0)},
 };
diff --git a/drivers/gpu/drm/amd/amdgpu/vi.c b/drivers/gpu/drm/amd/amdgpu/vi.c
index b3fa54c0514e..a83505815d39 100644
--- a/drivers/gpu/drm/amd/amdgpu/vi.c
+++ b/drivers/gpu/drm/amd/amdgpu/vi.c
@@ -136,15 +136,15 @@ static const struct amdgpu_video_codec_info polaris_video_codecs_encode_array[]
 	{
 		.codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC,
 		.max_width = 4096,
-		.max_height = 2304,
-		.max_pixels_per_frame = 4096 * 2304,
+		.max_height = 4096,
+		.max_pixels_per_frame = 4096 * 4096,
 		.max_level = 0,
 	},
 	{
 		.codec_type = AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC,
 		.max_width = 4096,
-		.max_height = 2304,
-		.max_pixels_per_frame = 4096 * 2304,
+		.max_height = 4096,
+		.max_pixels_per_frame = 4096 * 4096,
 		.max_level = 0,
 	},
 };
-- 
2.51.0


From 8e29057eecb83e45898a31285ff8b82dff188dd1 Mon Sep 17 00:00:00 2001
From: Ramesh Errabolu <Ramesh.Errabolu@amd.com>
Date: Tue, 5 Nov 2024 18:50:02 -0600
Subject: [PATCH 13/16] drm/amdgpu: Inform if PCIe based P2P links are not
 available

Raise an info message in kernel log if PCIe root complex
determines that a AMD GPU device D<i> cannot have P2P
communication with another AMD GPU device D<j>

Signed-off-by: Ramesh Errabolu <Ramesh.Errabolu@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
index 38a17571da75..a72c90da3d1d 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
@@ -6200,6 +6200,9 @@ bool amdgpu_device_is_peer_accessible(struct amdgpu_device *adev,
 	bool p2p_access =
 		!adev->gmc.xgmi.connected_to_cpu &&
 		!(pci_p2pdma_distance(adev->pdev, peer_adev->dev, false) < 0);
+	if (!p2p_access)
+		dev_info(adev->dev, "PCIe P2P access from peer device %s is not supported by the chipset\n",
+			pci_name(peer_adev->pdev));
 
 	bool is_large_bar = adev->gmc.visible_vram_size &&
 		adev->gmc.real_vram_size == adev->gmc.visible_vram_size;
-- 
2.51.0


From d4be16ccfd5bf822176740a51ff2306679a2247e Mon Sep 17 00:00:00 2001
From: Tim Huang <tim.huang@amd.com>
Date: Mon, 28 Oct 2024 13:51:50 +0800
Subject: [PATCH 14/16] drm/amd/pm: print pp_dpm_mclk in ascending order on SMU
 v14.0.0

Currently, the pp_dpm_mclk values are reported in descending order
on SMU IP v14.0.0/1/4. Adjust to ascending order for consistency
with other clock interfaces.

Signed-off-by: Tim Huang <tim.huang@amd.com>
Reviewed-by: Yifan Zhang <yifan1.zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/amd/pm/swsmu/smu14/smu_v14_0_0_ppt.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu14/smu_v14_0_0_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu14/smu_v14_0_0_ppt.c
index 8798ebfcea83..84f9b007b59f 100644
--- a/drivers/gpu/drm/amd/pm/swsmu/smu14/smu_v14_0_0_ppt.c
+++ b/drivers/gpu/drm/amd/pm/swsmu/smu14/smu_v14_0_0_ppt.c
@@ -1132,7 +1132,7 @@ static int smu_v14_0_common_get_dpm_level_count(struct smu_context *smu,
 static int smu_v14_0_0_print_clk_levels(struct smu_context *smu,
 					enum smu_clk_type clk_type, char *buf)
 {
-	int i, size = 0, ret = 0;
+	int i, idx, ret = 0, size = 0;
 	uint32_t cur_value = 0, value = 0, count = 0;
 	uint32_t min, max;
 
@@ -1168,7 +1168,8 @@ static int smu_v14_0_0_print_clk_levels(struct smu_context *smu,
 			break;
 
 		for (i = 0; i < count; i++) {
-			ret = smu_v14_0_common_get_dpm_freq_by_index(smu, clk_type, i, &value);
+			idx = (clk_type == SMU_MCLK) ? (count - i - 1) : i;
+			ret = smu_v14_0_common_get_dpm_freq_by_index(smu, clk_type, idx, &value);
 			if (ret)
 				break;
 
-- 
2.51.0


From 1b4ca8546f5b5c482717bedb8e031227b1541539 Mon Sep 17 00:00:00 2001
From: =?utf8?q?Christian=20K=C3=B6nig?= <christian.koenig@amd.com>
Date: Thu, 31 Oct 2024 10:04:17 +0100
Subject: [PATCH 15/16] drm/amdgpu: fix check in gmc_v9_0_get_vm_pte()
MIME-Version: 1.0
Content-Type: text/plain; charset=utf8
Content-Transfer-Encoding: 8bit

The coherency flags can only be determined when the BO is locked and that
in turn is only guaranteed when the mapping is validated.

Fix the check, move the resource check into the function and add an assert
that the BO is locked.

Signed-off-by: Christian König <christian.koenig@amd.com>
Fixes: d1a372af1c3d ("drm/amdgpu: Set MTYPE in PTE based on BO flags")
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c | 13 ++++++++-----
 1 file changed, 8 insertions(+), 5 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
index f43ded8a0aab..50c5da3020cb 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
@@ -1130,8 +1130,10 @@ static void gmc_v9_0_get_coherence_flags(struct amdgpu_device *adev,
 					 uint64_t *flags)
 {
 	struct amdgpu_device *bo_adev = amdgpu_ttm_adev(bo->tbo.bdev);
-	bool is_vram = bo->tbo.resource->mem_type == TTM_PL_VRAM;
-	bool coherent = bo->flags & (AMDGPU_GEM_CREATE_COHERENT | AMDGPU_GEM_CREATE_EXT_COHERENT);
+	bool is_vram = bo->tbo.resource &&
+		bo->tbo.resource->mem_type == TTM_PL_VRAM;
+	bool coherent = bo->flags & (AMDGPU_GEM_CREATE_COHERENT |
+				     AMDGPU_GEM_CREATE_EXT_COHERENT);
 	bool ext_coherent = bo->flags & AMDGPU_GEM_CREATE_EXT_COHERENT;
 	bool uncached = bo->flags & AMDGPU_GEM_CREATE_UNCACHED;
 	struct amdgpu_vm *vm = mapping->bo_va->base.vm;
@@ -1139,6 +1141,8 @@ static void gmc_v9_0_get_coherence_flags(struct amdgpu_device *adev,
 	bool snoop = false;
 	bool is_local;
 
+	dma_resv_assert_held(bo->tbo.base.resv);
+
 	switch (amdgpu_ip_version(adev, GC_HWIP, 0)) {
 	case IP_VERSION(9, 4, 1):
 	case IP_VERSION(9, 4, 2):
@@ -1257,9 +1261,8 @@ static void gmc_v9_0_get_vm_pte(struct amdgpu_device *adev,
 		*flags &= ~AMDGPU_PTE_VALID;
 	}
 
-	if (bo && bo->tbo.resource)
-		gmc_v9_0_get_coherence_flags(adev, mapping->bo_va->base.bo,
-					     mapping, flags);
+	if ((*flags & AMDGPU_PTE_VALID) && bo)
+		gmc_v9_0_get_coherence_flags(adev, bo, mapping, flags);
 }
 
 static void gmc_v9_0_override_vm_pte_flags(struct amdgpu_device *adev,
-- 
2.51.0


From 21cae8debc6a1d243f64fa82cd1b41cb612b5c61 Mon Sep 17 00:00:00 2001
From: Yuan Can <yuancan@huawei.com>
Date: Wed, 6 Nov 2024 09:35:41 +0800
Subject: [PATCH 16/16] drm/amdkfd: Fix wrong usage of INIT_WORK()

In kfd_procfs_show(), the sdma_activity_work_handler is a local variable
and the sdma_activity_work_handler.sdma_activity_work should initialize
with INIT_WORK_ONSTACK() instead of INIT_WORK().

Fixes: 32cb59f31362 ("drm/amdkfd: Track SDMA utilization per process")
Signed-off-by: Yuan Can <yuancan@huawei.com>
Signed-off-by: Felix Kuehling <felix.kuehling@amd.com>
Reviewed-by: Felix Kuehling <felix.kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/amd/amdkfd/kfd_process.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_process.c b/drivers/gpu/drm/amd/amdkfd/kfd_process.c
index 6bab6fc6a35d..ff34bb1ac9db 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_process.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_process.c
@@ -341,8 +341,8 @@ static ssize_t kfd_procfs_show(struct kobject *kobj, struct attribute *attr,
 							      attr_sdma);
 		struct kfd_sdma_activity_handler_workarea sdma_activity_work_handler;
 
-		INIT_WORK(&sdma_activity_work_handler.sdma_activity_work,
-					kfd_sdma_activity_worker);
+		INIT_WORK_ONSTACK(&sdma_activity_work_handler.sdma_activity_work,
+				  kfd_sdma_activity_worker);
 
 		sdma_activity_work_handler.pdd = pdd;
 		sdma_activity_work_handler.sdma_activity_counter = 0;
@@ -350,6 +350,7 @@ static ssize_t kfd_procfs_show(struct kobject *kobj, struct attribute *attr,
 		schedule_work(&sdma_activity_work_handler.sdma_activity_work);
 
 		flush_work(&sdma_activity_work_handler.sdma_activity_work);
+		destroy_work_on_stack(&sdma_activity_work_handler.sdma_activity_work);
 
 		return snprintf(buffer, PAGE_SIZE, "%llu\n",
 				(sdma_activity_work_handler.sdma_activity_counter)/
-- 
2.51.0