[PATCH 1/3] accel/amdxdna: Add hardware specific attributes

[PATCH 1/3] accel/amdxdna: Add hardware specific attributes

[Lizhi Hou]

Add three hardware specific attributes to describe device capabilities:
  hwctx_limit: The maximum number of hardware context supported.
  max_tops: The maximum TOPS supported.
  curr_tops: The TOPS achievable with the current power and frequency
             configuration.

Signed-off-by: Lizhi Hou <lizhi.hou@amd.com>
---
 drivers/accel/amdxdna/aie2_pci.h  |  3 +++
 drivers/accel/amdxdna/aie2_smu.c  | 11 +++++++++++
 drivers/accel/amdxdna/npu1_regs.c |  1 +
 drivers/accel/amdxdna/npu2_regs.c |  1 +
 drivers/accel/amdxdna/npu4_regs.c |  1 +
 drivers/accel/amdxdna/npu5_regs.c |  1 +
 drivers/accel/amdxdna/npu6_regs.c |  1 +
 7 files changed, 19 insertions(+)

diff --git a/drivers/accel/amdxdna/aie2_pci.h b/drivers/accel/amdxdna/aie2_pci.h
index 6cc24641d3db..a79f4f71ff6b 100644
--- a/drivers/accel/amdxdna/aie2_pci.h
+++ b/drivers/accel/amdxdna/aie2_pci.h
@@ -195,6 +195,8 @@ struct amdxdna_dev_hdl {
 	u32				clk_gating;
 	u32				npuclk_freq;
 	u32				hclk_freq;
+	u32				max_tops;
+	u32				curr_tops;
 
 	/* Mailbox and the management channel */
 	struct mailbox			*mbox;
@@ -246,6 +248,7 @@ struct amdxdna_dev_priv {
 	u32				mbox_dev_addr;
 	/* If mbox_size is 0, use BAR size. See MBOX_SIZE macro */
 	u32				mbox_size;
+	u32				hwctx_limit;
 	u32				sram_dev_addr;
 	struct aie2_bar_off_pair	sram_offs[SRAM_MAX_INDEX];
 	struct aie2_bar_off_pair	psp_regs_off[PSP_MAX_REGS];
diff --git a/drivers/accel/amdxdna/aie2_smu.c b/drivers/accel/amdxdna/aie2_smu.c
index 7f292a615ed8..11c0e9e7b03a 100644
--- a/drivers/accel/amdxdna/aie2_smu.c
+++ b/drivers/accel/amdxdna/aie2_smu.c
@@ -23,6 +23,13 @@
 #define AIE2_SMU_SET_SOFT_DPMLEVEL	0x7
 #define AIE2_SMU_SET_HARD_DPMLEVEL	0x8
 
+#define NPU4_DPM_TOPS(ndev, dpm_level) \
+({ \
+	typeof(ndev) _ndev = ndev; \
+	(4096 * (_ndev)->total_col * \
+	 (_ndev)->priv->dpm_clk_tbl[dpm_level].hclk / 1000000); \
+})
+
 static int aie2_smu_exec(struct amdxdna_dev_hdl *ndev, u32 reg_cmd,
 			 u32 reg_arg, u32 *out)
 {
@@ -84,6 +91,8 @@ int npu1_set_dpm(struct amdxdna_dev_hdl *ndev, u32 dpm_level)
 	amdxdna_pm_suspend_put(ndev->xdna);
 	ndev->hclk_freq = freq;
 	ndev->dpm_level = dpm_level;
+	ndev->max_tops = 2 * ndev->total_col;
+	ndev->curr_tops = ndev->max_tops * freq / 1028;
 
 	XDNA_DBG(ndev->xdna, "MP-NPU clock %d, H clock %d\n",
 		 ndev->npuclk_freq, ndev->hclk_freq);
@@ -121,6 +130,8 @@ int npu4_set_dpm(struct amdxdna_dev_hdl *ndev, u32 dpm_level)
 	ndev->npuclk_freq = ndev->priv->dpm_clk_tbl[dpm_level].npuclk;
 	ndev->hclk_freq = ndev->priv->dpm_clk_tbl[dpm_level].hclk;
 	ndev->dpm_level = dpm_level;
+	ndev->max_tops = NPU4_DPM_TOPS(ndev, ndev->max_dpm_level);
+	ndev->curr_tops = NPU4_DPM_TOPS(ndev, dpm_level);
 
 	XDNA_DBG(ndev->xdna, "MP-NPU clock %d, H clock %d\n",
 		 ndev->npuclk_freq, ndev->hclk_freq);
diff --git a/drivers/accel/amdxdna/npu1_regs.c b/drivers/accel/amdxdna/npu1_regs.c
index 4a43d02404d1..ec407f3b48fc 100644
--- a/drivers/accel/amdxdna/npu1_regs.c
+++ b/drivers/accel/amdxdna/npu1_regs.c
@@ -79,6 +79,7 @@ static const struct amdxdna_dev_priv npu1_dev_priv = {
 	.mbox_dev_addr  = NPU1_MBOX_BAR_BASE,
 	.mbox_size      = 0, /* Use BAR size */
 	.sram_dev_addr  = NPU1_SRAM_BAR_BASE,
+	.hwctx_limit    = 6,
 	.sram_offs      = {
 		DEFINE_BAR_OFFSET(MBOX_CHANN_OFF, NPU1_SRAM, MPNPU_SRAM_X2I_MAILBOX_0),
 		DEFINE_BAR_OFFSET(FW_ALIVE_OFF,   NPU1_SRAM, MPNPU_SRAM_I2X_MAILBOX_15),
diff --git a/drivers/accel/amdxdna/npu2_regs.c b/drivers/accel/amdxdna/npu2_regs.c
index 97df2f09356a..86f87d0d1354 100644
--- a/drivers/accel/amdxdna/npu2_regs.c
+++ b/drivers/accel/amdxdna/npu2_regs.c
@@ -72,6 +72,7 @@ static const struct amdxdna_dev_priv npu2_dev_priv = {
 	.mbox_dev_addr  = NPU2_MBOX_BAR_BASE,
 	.mbox_size      = 0, /* Use BAR size */
 	.sram_dev_addr  = NPU2_SRAM_BAR_BASE,
+	.hwctx_limit    = 16,
 	.sram_offs      = {
 		DEFINE_BAR_OFFSET(MBOX_CHANN_OFF, NPU2_SRAM, MPNPU_SRAM_X2I_MAILBOX_0),
 		DEFINE_BAR_OFFSET(FW_ALIVE_OFF,   NPU2_SRAM, MPNPU_SRAM_X2I_MAILBOX_15),
diff --git a/drivers/accel/amdxdna/npu4_regs.c b/drivers/accel/amdxdna/npu4_regs.c
index 5a4ed0c363f8..d90777275a9f 100644
--- a/drivers/accel/amdxdna/npu4_regs.c
+++ b/drivers/accel/amdxdna/npu4_regs.c
@@ -99,6 +99,7 @@ static const struct amdxdna_dev_priv npu4_dev_priv = {
 	.mbox_dev_addr  = NPU4_MBOX_BAR_BASE,
 	.mbox_size      = 0, /* Use BAR size */
 	.sram_dev_addr  = NPU4_SRAM_BAR_BASE,
+	.hwctx_limit    = 16,
 	.sram_offs      = {
 		DEFINE_BAR_OFFSET(MBOX_CHANN_OFF, NPU4_SRAM, MPNPU_SRAM_X2I_MAILBOX_0),
 		DEFINE_BAR_OFFSET(FW_ALIVE_OFF,   NPU4_SRAM, MPNPU_SRAM_X2I_MAILBOX_15),
diff --git a/drivers/accel/amdxdna/npu5_regs.c b/drivers/accel/amdxdna/npu5_regs.c
index dadd72df6263..75ad97f0b937 100644
--- a/drivers/accel/amdxdna/npu5_regs.c
+++ b/drivers/accel/amdxdna/npu5_regs.c
@@ -72,6 +72,7 @@ static const struct amdxdna_dev_priv npu5_dev_priv = {
 	.mbox_dev_addr  = NPU5_MBOX_BAR_BASE,
 	.mbox_size      = 0, /* Use BAR size */
 	.sram_dev_addr  = NPU5_SRAM_BAR_BASE,
+	.hwctx_limit    = 16,
 	.sram_offs      = {
 		DEFINE_BAR_OFFSET(MBOX_CHANN_OFF, NPU5_SRAM, MPNPU_SRAM_X2I_MAILBOX_0),
 		DEFINE_BAR_OFFSET(FW_ALIVE_OFF,   NPU5_SRAM, MPNPU_SRAM_X2I_MAILBOX_15),
diff --git a/drivers/accel/amdxdna/npu6_regs.c b/drivers/accel/amdxdna/npu6_regs.c
index f73f92869b23..758dc013fe13 100644
--- a/drivers/accel/amdxdna/npu6_regs.c
+++ b/drivers/accel/amdxdna/npu6_regs.c
@@ -72,6 +72,7 @@ static const struct amdxdna_dev_priv npu6_dev_priv = {
 	.mbox_dev_addr  = NPU6_MBOX_BAR_BASE,
 	.mbox_size      = 0, /* Use BAR size */
 	.sram_dev_addr  = NPU6_SRAM_BAR_BASE,
+	.hwctx_limit    = 16,
 	.sram_offs      = {
 		DEFINE_BAR_OFFSET(MBOX_CHANN_OFF, NPU6_SRAM, MPNPU_SRAM_X2I_MAILBOX_0),
 		DEFINE_BAR_OFFSET(FW_ALIVE_OFF,   NPU6_SRAM, MPNPU_SRAM_X2I_MAILBOX_15),
-- 
2.34.1

[PATCH 2/3] accel/amdxdna: Add IOCTL parameter for resource data

[Lizhi Hou]

Extend DRM_IOCTL_AMDXDNA_GET_INFO to include additional parameters
that allow collection of resource data.

Signed-off-by: Lizhi Hou <lizhi.hou@amd.com>
---
 drivers/accel/amdxdna/aie2_ctx.c        |  6 ------
 drivers/accel/amdxdna/aie2_message.c    |  2 ++
 drivers/accel/amdxdna/aie2_pci.c        | 27 +++++++++++++++++++++++++
 drivers/accel/amdxdna/amdxdna_pci_drv.c |  3 ++-
 include/uapi/drm/amdxdna_accel.h        | 17 ++++++++++++++++
 5 files changed, 48 insertions(+), 7 deletions(-)

diff --git a/drivers/accel/amdxdna/aie2_ctx.c b/drivers/accel/amdxdna/aie2_ctx.c
index 289a2aaf4cae..b78c47ed0d34 100644
--- a/drivers/accel/amdxdna/aie2_ctx.c
+++ b/drivers/accel/amdxdna/aie2_ctx.c
@@ -556,7 +556,6 @@ int aie2_hwctx_init(struct amdxdna_hwctx *hwctx)
 	struct drm_gpu_scheduler *sched;
 	struct amdxdna_hwctx_priv *priv;
 	struct amdxdna_gem_obj *heap;
-	struct amdxdna_dev_hdl *ndev;
 	int i, ret;
 
 	priv = kzalloc(sizeof(*hwctx->priv), GFP_KERNEL);
@@ -654,8 +653,6 @@ int aie2_hwctx_init(struct amdxdna_hwctx *hwctx)
 	amdxdna_pm_suspend_put(xdna);
 
 	hwctx->status = HWCTX_STAT_INIT;
-	ndev = xdna->dev_handle;
-	ndev->hwctx_num++;
 	init_waitqueue_head(&priv->job_free_wq);
 
 	XDNA_DBG(xdna, "hwctx %s init completed", hwctx->name);
@@ -688,13 +685,10 @@ int aie2_hwctx_init(struct amdxdna_hwctx *hwctx)
 
 void aie2_hwctx_fini(struct amdxdna_hwctx *hwctx)
 {
-	struct amdxdna_dev_hdl *ndev;
 	struct amdxdna_dev *xdna;
 	int idx;
 
 	xdna = hwctx->client->xdna;
-	ndev = xdna->dev_handle;
-	ndev->hwctx_num--;
 
 	XDNA_DBG(xdna, "%s sequence number %lld", hwctx->name, hwctx->priv->seq);
 	drm_sched_entity_destroy(&hwctx->priv->entity);
diff --git a/drivers/accel/amdxdna/aie2_message.c b/drivers/accel/amdxdna/aie2_message.c
index 339dec998247..39214253d804 100644
--- a/drivers/accel/amdxdna/aie2_message.c
+++ b/drivers/accel/amdxdna/aie2_message.c
@@ -235,6 +235,7 @@ int aie2_create_context(struct amdxdna_dev_hdl *ndev, struct amdxdna_hwctx *hwct
 		ret = -EINVAL;
 		goto out_destroy_context;
 	}
+	ndev->hwctx_num++;
 
 	XDNA_DBG(xdna, "%s mailbox channel irq: %d, msix_id: %d",
 		 hwctx->name, ret, resp.msix_id);
@@ -269,6 +270,7 @@ int aie2_destroy_context(struct amdxdna_dev_hdl *ndev, struct amdxdna_hwctx *hwc
 		 hwctx->fw_ctx_id);
 	hwctx->priv->mbox_chann = NULL;
 	hwctx->fw_ctx_id = -1;
+	ndev->hwctx_num--;
 
 	return ret;
 }
diff --git a/drivers/accel/amdxdna/aie2_pci.c b/drivers/accel/amdxdna/aie2_pci.c
index ce57b915004e..396dc6e06007 100644
--- a/drivers/accel/amdxdna/aie2_pci.c
+++ b/drivers/accel/amdxdna/aie2_pci.c
@@ -838,6 +838,30 @@ static int aie2_get_hwctx_status(struct amdxdna_client *client,
 	return 0;
 }
 
+static int aie2_query_resource_info(struct amdxdna_client *client,
+				    struct amdxdna_drm_get_info *args)
+{
+	struct amdxdna_drm_get_resource_info res_info;
+	const struct amdxdna_dev_priv *priv;
+	struct amdxdna_dev_hdl *ndev;
+	struct amdxdna_dev *xdna;
+
+	xdna = client->xdna;
+	ndev = xdna->dev_handle;
+	priv = ndev->priv;
+
+	res_info.npu_clk_max = priv->dpm_clk_tbl[ndev->max_dpm_level].hclk;
+	res_info.npu_tops_max = ndev->max_tops;
+	res_info.npu_task_max = priv->hwctx_limit;
+	res_info.npu_tops_curr = ndev->curr_tops;
+	res_info.npu_task_curr = ndev->hwctx_num;
+
+	if (copy_to_user(u64_to_user_ptr(args->buffer), &res_info, sizeof(res_info)))
+		return -EFAULT;
+
+	return 0;
+}
+
 static int aie2_get_info(struct amdxdna_client *client, struct amdxdna_drm_get_info *args)
 {
 	struct amdxdna_dev *xdna = client->xdna;
@@ -872,6 +896,9 @@ static int aie2_get_info(struct amdxdna_client *client, struct amdxdna_drm_get_i
 	case DRM_AMDXDNA_GET_POWER_MODE:
 		ret = aie2_get_power_mode(client, args);
 		break;
+	case DRM_AMDXDNA_QUERY_RESOURCE_INFO:
+		ret = aie2_query_resource_info(client, args);
+		break;
 	default:
 		XDNA_ERR(xdna, "Not supported request parameter %u", args->param);
 		ret = -EOPNOTSUPP;
diff --git a/drivers/accel/amdxdna/amdxdna_pci_drv.c b/drivers/accel/amdxdna/amdxdna_pci_drv.c
index 3599e713bfcb..af943a603ad1 100644
--- a/drivers/accel/amdxdna/amdxdna_pci_drv.c
+++ b/drivers/accel/amdxdna/amdxdna_pci_drv.c
@@ -29,9 +29,10 @@ MODULE_FIRMWARE("amdnpu/17f0_20/npu.sbin");
  * 0.1: Support getting all hardware contexts by DRM_IOCTL_AMDXDNA_GET_ARRAY
  * 0.2: Support getting last error hardware error
  * 0.3: Support firmware debug buffer
+ * 0.4: Support getting resource information
  */
 #define AMDXDNA_DRIVER_MAJOR		0
-#define AMDXDNA_DRIVER_MINOR		3
+#define AMDXDNA_DRIVER_MINOR		4
 
 /*
  * Bind the driver base on (vendor_id, device_id) pair and later use the
diff --git a/include/uapi/drm/amdxdna_accel.h b/include/uapi/drm/amdxdna_accel.h
index c7eec9ceb2ae..8b679c38d308 100644
--- a/include/uapi/drm/amdxdna_accel.h
+++ b/include/uapi/drm/amdxdna_accel.h
@@ -442,6 +442,23 @@ enum amdxdna_drm_get_param {
 	DRM_AMDXDNA_QUERY_HW_CONTEXTS,
 	DRM_AMDXDNA_QUERY_FIRMWARE_VERSION = 8,
 	DRM_AMDXDNA_GET_POWER_MODE,
+	DRM_AMDXDNA_QUERY_RESOURCE_INFO = 12,
+};
+
+/**
+ * struct amdxdna_drm_get_resource_info - Get resource information
+ */
+struct amdxdna_drm_get_resource_info {
+	/** @npu_clk_max: max H-Clocks */
+	__u64 npu_clk_max;
+	/** @npu_tops_max: max TOPs */
+	__u64 npu_tops_max;
+	/** @npu_task_max: max number of tasks */
+	__u64 npu_task_max;
+	/** @npu_tops_curr: current TOPs */
+	__u64 npu_tops_curr;
+	/** @npu_task_curr: current number of tasks */
+	__u64 npu_task_curr;
 };
 
 /**
-- 
2.34.1

[PATCH 3/3] accel/amdxdna: Add IOCTL parameter for telemetry data

[Lizhi Hou]

Extend DRM_IOCTL_AMDXDNA_GET_INFO to include additional parameters
that allow collection of telemetry data.

Signed-off-by: Lizhi Hou <lizhi.hou@amd.com>
---
 drivers/accel/amdxdna/aie2_message.c          | 56 ++++++++++++--
 drivers/accel/amdxdna/aie2_msg_priv.h         | 25 ++++++-
 drivers/accel/amdxdna/aie2_pci.c              | 73 +++++++++++++++++++
 drivers/accel/amdxdna/aie2_pci.h              |  3 +
 .../accel/amdxdna/amdxdna_mailbox_helper.h    |  6 +-
 drivers/accel/amdxdna/amdxdna_pci_drv.c       |  3 +-
 include/uapi/drm/amdxdna_accel.h              | 17 +++++
 7 files changed, 173 insertions(+), 10 deletions(-)

diff --git a/drivers/accel/amdxdna/aie2_message.c b/drivers/accel/amdxdna/aie2_message.c
index 39214253d804..69cdce9ff208 100644
--- a/drivers/accel/amdxdna/aie2_message.c
+++ b/drivers/accel/amdxdna/aie2_message.c
@@ -47,7 +47,7 @@ static int aie2_send_mgmt_msg_wait(struct amdxdna_dev_hdl *ndev,
 		ndev->mgmt_chann = NULL;
 	}
 
-	if (!ret && *hdl->data != AIE2_STATUS_SUCCESS) {
+	if (!ret && *hdl->status != AIE2_STATUS_SUCCESS) {
 		XDNA_ERR(xdna, "command opcode 0x%x failed, status 0x%x",
 			 msg->opcode, *hdl->data);
 		ret = -EINVAL;
@@ -336,11 +336,6 @@ int aie2_query_status(struct amdxdna_dev_hdl *ndev, char __user *buf,
 		goto fail;
 	}
 
-	if (resp.status != AIE2_STATUS_SUCCESS) {
-		XDNA_ERR(xdna, "Query NPU status failed, status 0x%x", resp.status);
-		ret = -EINVAL;
-		goto fail;
-	}
 	XDNA_DBG(xdna, "Query NPU status completed");
 
 	if (size < resp.size) {
@@ -362,6 +357,55 @@ int aie2_query_status(struct amdxdna_dev_hdl *ndev, char __user *buf,
 	return ret;
 }
 
+int aie2_query_telemetry(struct amdxdna_dev_hdl *ndev,
+			 char __user *buf, u32 size,
+			 struct amdxdna_drm_query_telemetry_header *header)
+{
+	DECLARE_AIE2_MSG(get_telemetry, MSG_OP_GET_TELEMETRY);
+	struct amdxdna_dev *xdna = ndev->xdna;
+	dma_addr_t dma_addr;
+	u8 *addr;
+	int ret;
+
+	if (header->type >= MAX_TELEMETRY_TYPE)
+		return -EINVAL;
+
+	addr = dma_alloc_noncoherent(xdna->ddev.dev, size, &dma_addr,
+				     DMA_FROM_DEVICE, GFP_KERNEL);
+	if (!addr)
+		return -ENOMEM;
+
+	req.buf_addr = dma_addr;
+	req.buf_size = size;
+	req.type = header->type;
+
+	drm_clflush_virt_range(addr, size); /* device can access */
+	ret = aie2_send_mgmt_msg_wait(ndev, &msg);
+	if (ret) {
+		XDNA_ERR(xdna, "Query telemetry failed, status %d", ret);
+		goto free_buf;
+	}
+
+	if (size < resp.size) {
+		ret = -EINVAL;
+		XDNA_ERR(xdna, "Bad buffer size. Available: %u. Needs: %u", size, resp.size);
+		goto free_buf;
+	}
+
+	if (copy_to_user(buf, addr, resp.size)) {
+		ret = -EFAULT;
+		XDNA_ERR(xdna, "Failed to copy telemetry to user space");
+		goto free_buf;
+	}
+
+	header->major = resp.major;
+	header->minor = resp.minor;
+
+free_buf:
+	dma_free_noncoherent(xdna->ddev.dev, size, addr, dma_addr, DMA_FROM_DEVICE);
+	return ret;
+}
+
 int aie2_register_asyn_event_msg(struct amdxdna_dev_hdl *ndev, dma_addr_t addr, u32 size,
 				 void *handle, int (*cb)(void*, void __iomem *, size_t))
 {
diff --git a/drivers/accel/amdxdna/aie2_msg_priv.h b/drivers/accel/amdxdna/aie2_msg_priv.h
index 945140011763..947daa63f064 100644
--- a/drivers/accel/amdxdna/aie2_msg_priv.h
+++ b/drivers/accel/amdxdna/aie2_msg_priv.h
@@ -9,7 +9,8 @@
 enum aie2_msg_opcode {
 	MSG_OP_CREATE_CONTEXT              = 0x2,
 	MSG_OP_DESTROY_CONTEXT             = 0x3,
-	MSG_OP_SYNC_BO			   = 0x7,
+	MSG_OP_GET_TELEMETRY               = 0x4,
+	MSG_OP_SYNC_BO                     = 0x7,
 	MSG_OP_EXECUTE_BUFFER_CF           = 0xC,
 	MSG_OP_QUERY_COL_STATUS            = 0xD,
 	MSG_OP_QUERY_AIE_TILE_INFO         = 0xE,
@@ -137,6 +138,28 @@ struct destroy_ctx_resp {
 	enum aie2_msg_status	status;
 } __packed;
 
+enum telemetry_type {
+	TELEMETRY_TYPE_DISABLED,
+	TELEMETRY_TYPE_HEALTH,
+	TELEMETRY_TYPE_ERROR_INFO,
+	TELEMETRY_TYPE_PROFILING,
+	TELEMETRY_TYPE_DEBUG,
+	MAX_TELEMETRY_TYPE
+};
+
+struct get_telemetry_req {
+	enum telemetry_type	type;
+	__u64	buf_addr;
+	__u32	buf_size;
+} __packed;
+
+struct get_telemetry_resp {
+	__u32	major;
+	__u32	minor;
+	__u32	size;
+	enum aie2_msg_status	status;
+} __packed;
+
 struct execute_buffer_req {
 	__u32	cu_idx;
 	__u32	payload[19];
diff --git a/drivers/accel/amdxdna/aie2_pci.c b/drivers/accel/amdxdna/aie2_pci.c
index 396dc6e06007..d7ccbdaf47f5 100644
--- a/drivers/accel/amdxdna/aie2_pci.c
+++ b/drivers/accel/amdxdna/aie2_pci.c
@@ -862,6 +862,76 @@ static int aie2_query_resource_info(struct amdxdna_client *client,
 	return 0;
 }
 
+static int aie2_fill_hwctx_map(struct amdxdna_hwctx *hwctx, void *arg)
+{
+	struct amdxdna_dev *xdna = hwctx->client->xdna;
+	u32 *map = arg;
+
+	if (hwctx->fw_ctx_id >= xdna->dev_handle->priv->hwctx_limit) {
+		XDNA_ERR(xdna, "Invalid fw ctx id %d/%d ", hwctx->fw_ctx_id,
+			 xdna->dev_handle->priv->hwctx_limit);
+		return -EINVAL;
+	}
+
+	map[hwctx->fw_ctx_id] = hwctx->id;
+	return 0;
+}
+
+static int aie2_get_telemetry(struct amdxdna_client *client,
+			      struct amdxdna_drm_get_info *args)
+{
+	struct amdxdna_drm_query_telemetry_header *header __free(kfree) = NULL;
+	u32 telemetry_data_sz, header_sz, elem_num;
+	struct amdxdna_dev *xdna = client->xdna;
+	struct amdxdna_client *tmp_client;
+	int ret;
+
+	elem_num = xdna->dev_handle->priv->hwctx_limit;
+	header_sz = struct_size(header, map, elem_num);
+	if (args->buffer_size <= header_sz) {
+		XDNA_ERR(xdna, "Invalid buffer size");
+		return -EINVAL;
+	}
+
+	telemetry_data_sz = args->buffer_size - header_sz;
+	if (telemetry_data_sz > SZ_4M) {
+		XDNA_ERR(xdna, "Buffer size is too big, %d", telemetry_data_sz);
+		return -EINVAL;
+	}
+
+	header = kzalloc(header_sz, GFP_KERNEL);
+	if (!header)
+		return -ENOMEM;
+
+	if (copy_from_user(header, u64_to_user_ptr(args->buffer), sizeof(*header))) {
+		XDNA_ERR(xdna, "Failed to copy telemetry header from user");
+		return -EFAULT;
+	}
+
+	header->map_num_elements = elem_num;
+	list_for_each_entry(tmp_client, &xdna->client_list, node) {
+		ret = amdxdna_hwctx_walk(tmp_client, &header->map,
+					 aie2_fill_hwctx_map);
+		if (ret)
+			return ret;
+	}
+
+	ret = aie2_query_telemetry(xdna->dev_handle,
+				   u64_to_user_ptr(args->buffer + header_sz),
+				   telemetry_data_sz, header);
+	if (ret) {
+		XDNA_ERR(xdna, "Query telemetry failed ret %d", ret);
+		return ret;
+	}
+
+	if (copy_to_user(u64_to_user_ptr(args->buffer), header, header_sz)) {
+		XDNA_ERR(xdna, "Copy header failed");
+		return -EFAULT;
+	}
+
+	return 0;
+}
+
 static int aie2_get_info(struct amdxdna_client *client, struct amdxdna_drm_get_info *args)
 {
 	struct amdxdna_dev *xdna = client->xdna;
@@ -896,6 +966,9 @@ static int aie2_get_info(struct amdxdna_client *client, struct amdxdna_drm_get_i
 	case DRM_AMDXDNA_GET_POWER_MODE:
 		ret = aie2_get_power_mode(client, args);
 		break;
+	case DRM_AMDXDNA_QUERY_TELEMETRY:
+		ret = aie2_get_telemetry(client, args);
+		break;
 	case DRM_AMDXDNA_QUERY_RESOURCE_INFO:
 		ret = aie2_query_resource_info(client, args);
 		break;
diff --git a/drivers/accel/amdxdna/aie2_pci.h b/drivers/accel/amdxdna/aie2_pci.h
index a79f4f71ff6b..9793cd1e0c55 100644
--- a/drivers/accel/amdxdna/aie2_pci.h
+++ b/drivers/accel/amdxdna/aie2_pci.h
@@ -305,6 +305,9 @@ int aie2_create_context(struct amdxdna_dev_hdl *ndev, struct amdxdna_hwctx *hwct
 int aie2_destroy_context(struct amdxdna_dev_hdl *ndev, struct amdxdna_hwctx *hwctx);
 int aie2_map_host_buf(struct amdxdna_dev_hdl *ndev, u32 context_id, u64 addr, u64 size);
 int aie2_query_status(struct amdxdna_dev_hdl *ndev, char __user *buf, u32 size, u32 *cols_filled);
+int aie2_query_telemetry(struct amdxdna_dev_hdl *ndev,
+			 char __user *buf, u32 size,
+			 struct amdxdna_drm_query_telemetry_header *header);
 int aie2_register_asyn_event_msg(struct amdxdna_dev_hdl *ndev, dma_addr_t addr, u32 size,
 				 void *handle, int (*cb)(void*, void __iomem *, size_t));
 int aie2_config_cu(struct amdxdna_hwctx *hwctx,
diff --git a/drivers/accel/amdxdna/amdxdna_mailbox_helper.h b/drivers/accel/amdxdna/amdxdna_mailbox_helper.h
index 710ff8873d61..556c712cad0a 100644
--- a/drivers/accel/amdxdna/amdxdna_mailbox_helper.h
+++ b/drivers/accel/amdxdna/amdxdna_mailbox_helper.h
@@ -16,16 +16,18 @@ struct xdna_notify {
 	u32			*data;
 	size_t			size;
 	int			error;
+	u32			*status;
 };
 
-#define DECLARE_XDNA_MSG_COMMON(name, op, status)			\
+#define DECLARE_XDNA_MSG_COMMON(name, op, s)				\
 	struct name##_req	req = { 0 };				\
-	struct name##_resp	resp = { status	};			\
+	struct name##_resp	resp = { .status = s };			\
 	struct xdna_notify	hdl = {					\
 		.error = 0,						\
 		.data = (u32 *)&resp,					\
 		.size = sizeof(resp),					\
 		.comp = COMPLETION_INITIALIZER_ONSTACK(hdl.comp),	\
+		.status = (u32 *)&resp.status,				\
 	};								\
 	struct xdna_mailbox_msg msg = {					\
 		.send_data = (u8 *)&req,				\
diff --git a/drivers/accel/amdxdna/amdxdna_pci_drv.c b/drivers/accel/amdxdna/amdxdna_pci_drv.c
index af943a603ad1..7590265d4485 100644
--- a/drivers/accel/amdxdna/amdxdna_pci_drv.c
+++ b/drivers/accel/amdxdna/amdxdna_pci_drv.c
@@ -30,9 +30,10 @@ MODULE_FIRMWARE("amdnpu/17f0_20/npu.sbin");
  * 0.2: Support getting last error hardware error
  * 0.3: Support firmware debug buffer
  * 0.4: Support getting resource information
+ * 0.5: Support getting telemetry data
  */
 #define AMDXDNA_DRIVER_MAJOR		0
-#define AMDXDNA_DRIVER_MINOR		4
+#define AMDXDNA_DRIVER_MINOR		5
 
 /*
  * Bind the driver base on (vendor_id, device_id) pair and later use the
diff --git a/include/uapi/drm/amdxdna_accel.h b/include/uapi/drm/amdxdna_accel.h
index 8b679c38d308..8ad254bc35a5 100644
--- a/include/uapi/drm/amdxdna_accel.h
+++ b/include/uapi/drm/amdxdna_accel.h
@@ -442,6 +442,7 @@ enum amdxdna_drm_get_param {
 	DRM_AMDXDNA_QUERY_HW_CONTEXTS,
 	DRM_AMDXDNA_QUERY_FIRMWARE_VERSION = 8,
 	DRM_AMDXDNA_GET_POWER_MODE,
+	DRM_AMDXDNA_QUERY_TELEMETRY,
 	DRM_AMDXDNA_QUERY_RESOURCE_INFO = 12,
 };
 
@@ -461,6 +462,22 @@ struct amdxdna_drm_get_resource_info {
 	__u64 npu_task_curr;
 };
 
+/**
+ * struct amdxdna_drm_query_telemetry_header - Telemetry data header
+ */
+struct amdxdna_drm_query_telemetry_header {
+	/** @major: Firmware telemetry interface major version number */
+	__u32 major;
+	/** @minor: Firmware telemetry interface minor version number */
+	__u32 minor;
+	/** @type: Telemetry query type */
+	__u32 type;
+	/** @map_num_elements: Total number of elements in the map table */
+	__u32 map_num_elements;
+	/** @map: Element map */
+	__u32 map[];
+};
+
 /**
  * struct amdxdna_drm_get_info - Get some information from the AIE hardware.
  * @param: Value in enum amdxdna_drm_get_param. Specifies the structure passed in the buffer.
-- 
2.34.1

Re: [PATCH 1/3] accel/amdxdna: Add hardware specific attributes

[Mario Limonciello]

On 11/4/25 12:25 AM, Lizhi Hou wrote:
> Add three hardware specific attributes to describe device capabilities:
>    hwctx_limit: The maximum number of hardware context supported.
>    max_tops: The maximum TOPS supported.
>    curr_tops: The TOPS achievable with the current power and frequency
>               configuration.
> 
> Signed-off-by: Lizhi Hou <lizhi.hou@amd.com>
Reviewed-by: Mario Limonciello (AMD) <superm1@kernel.org>
> ---
>   drivers/accel/amdxdna/aie2_pci.h  |  3 +++
>   drivers/accel/amdxdna/aie2_smu.c  | 11 +++++++++++
>   drivers/accel/amdxdna/npu1_regs.c |  1 +
>   drivers/accel/amdxdna/npu2_regs.c |  1 +
>   drivers/accel/amdxdna/npu4_regs.c |  1 +
>   drivers/accel/amdxdna/npu5_regs.c |  1 +
>   drivers/accel/amdxdna/npu6_regs.c |  1 +
>   7 files changed, 19 insertions(+)
> 
> diff --git a/drivers/accel/amdxdna/aie2_pci.h b/drivers/accel/amdxdna/aie2_pci.h
> index 6cc24641d3db..a79f4f71ff6b 100644
> --- a/drivers/accel/amdxdna/aie2_pci.h
> +++ b/drivers/accel/amdxdna/aie2_pci.h
> @@ -195,6 +195,8 @@ struct amdxdna_dev_hdl {
>   	u32				clk_gating;
>   	u32				npuclk_freq;
>   	u32				hclk_freq;
> +	u32				max_tops;
> +	u32				curr_tops;
>   
>   	/* Mailbox and the management channel */
>   	struct mailbox			*mbox;
> @@ -246,6 +248,7 @@ struct amdxdna_dev_priv {
>   	u32				mbox_dev_addr;
>   	/* If mbox_size is 0, use BAR size. See MBOX_SIZE macro */
>   	u32				mbox_size;
> +	u32				hwctx_limit;
>   	u32				sram_dev_addr;
>   	struct aie2_bar_off_pair	sram_offs[SRAM_MAX_INDEX];
>   	struct aie2_bar_off_pair	psp_regs_off[PSP_MAX_REGS];
> diff --git a/drivers/accel/amdxdna/aie2_smu.c b/drivers/accel/amdxdna/aie2_smu.c
> index 7f292a615ed8..11c0e9e7b03a 100644
> --- a/drivers/accel/amdxdna/aie2_smu.c
> +++ b/drivers/accel/amdxdna/aie2_smu.c
> @@ -23,6 +23,13 @@
>   #define AIE2_SMU_SET_SOFT_DPMLEVEL	0x7
>   #define AIE2_SMU_SET_HARD_DPMLEVEL	0x8
>   
> +#define NPU4_DPM_TOPS(ndev, dpm_level) \
> +({ \
> +	typeof(ndev) _ndev = ndev; \
> +	(4096 * (_ndev)->total_col * \
> +	 (_ndev)->priv->dpm_clk_tbl[dpm_level].hclk / 1000000); \
> +})
> +
>   static int aie2_smu_exec(struct amdxdna_dev_hdl *ndev, u32 reg_cmd,
>   			 u32 reg_arg, u32 *out)
>   {
> @@ -84,6 +91,8 @@ int npu1_set_dpm(struct amdxdna_dev_hdl *ndev, u32 dpm_level)
>   	amdxdna_pm_suspend_put(ndev->xdna);
>   	ndev->hclk_freq = freq;
>   	ndev->dpm_level = dpm_level;
> +	ndev->max_tops = 2 * ndev->total_col;
> +	ndev->curr_tops = ndev->max_tops * freq / 1028;
>   
>   	XDNA_DBG(ndev->xdna, "MP-NPU clock %d, H clock %d\n",
>   		 ndev->npuclk_freq, ndev->hclk_freq);
> @@ -121,6 +130,8 @@ int npu4_set_dpm(struct amdxdna_dev_hdl *ndev, u32 dpm_level)
>   	ndev->npuclk_freq = ndev->priv->dpm_clk_tbl[dpm_level].npuclk;
>   	ndev->hclk_freq = ndev->priv->dpm_clk_tbl[dpm_level].hclk;
>   	ndev->dpm_level = dpm_level;
> +	ndev->max_tops = NPU4_DPM_TOPS(ndev, ndev->max_dpm_level);
> +	ndev->curr_tops = NPU4_DPM_TOPS(ndev, dpm_level);
>   
>   	XDNA_DBG(ndev->xdna, "MP-NPU clock %d, H clock %d\n",
>   		 ndev->npuclk_freq, ndev->hclk_freq);
> diff --git a/drivers/accel/amdxdna/npu1_regs.c b/drivers/accel/amdxdna/npu1_regs.c
> index 4a43d02404d1..ec407f3b48fc 100644
> --- a/drivers/accel/amdxdna/npu1_regs.c
> +++ b/drivers/accel/amdxdna/npu1_regs.c
> @@ -79,6 +79,7 @@ static const struct amdxdna_dev_priv npu1_dev_priv = {
>   	.mbox_dev_addr  = NPU1_MBOX_BAR_BASE,
>   	.mbox_size      = 0, /* Use BAR size */
>   	.sram_dev_addr  = NPU1_SRAM_BAR_BASE,
> +	.hwctx_limit    = 6,
>   	.sram_offs      = {
>   		DEFINE_BAR_OFFSET(MBOX_CHANN_OFF, NPU1_SRAM, MPNPU_SRAM_X2I_MAILBOX_0),
>   		DEFINE_BAR_OFFSET(FW_ALIVE_OFF,   NPU1_SRAM, MPNPU_SRAM_I2X_MAILBOX_15),
> diff --git a/drivers/accel/amdxdna/npu2_regs.c b/drivers/accel/amdxdna/npu2_regs.c
> index 97df2f09356a..86f87d0d1354 100644
> --- a/drivers/accel/amdxdna/npu2_regs.c
> +++ b/drivers/accel/amdxdna/npu2_regs.c
> @@ -72,6 +72,7 @@ static const struct amdxdna_dev_priv npu2_dev_priv = {
>   	.mbox_dev_addr  = NPU2_MBOX_BAR_BASE,
>   	.mbox_size      = 0, /* Use BAR size */
>   	.sram_dev_addr  = NPU2_SRAM_BAR_BASE,
> +	.hwctx_limit    = 16,
>   	.sram_offs      = {
>   		DEFINE_BAR_OFFSET(MBOX_CHANN_OFF, NPU2_SRAM, MPNPU_SRAM_X2I_MAILBOX_0),
>   		DEFINE_BAR_OFFSET(FW_ALIVE_OFF,   NPU2_SRAM, MPNPU_SRAM_X2I_MAILBOX_15),
> diff --git a/drivers/accel/amdxdna/npu4_regs.c b/drivers/accel/amdxdna/npu4_regs.c
> index 5a4ed0c363f8..d90777275a9f 100644
> --- a/drivers/accel/amdxdna/npu4_regs.c
> +++ b/drivers/accel/amdxdna/npu4_regs.c
> @@ -99,6 +99,7 @@ static const struct amdxdna_dev_priv npu4_dev_priv = {
>   	.mbox_dev_addr  = NPU4_MBOX_BAR_BASE,
>   	.mbox_size      = 0, /* Use BAR size */
>   	.sram_dev_addr  = NPU4_SRAM_BAR_BASE,
> +	.hwctx_limit    = 16,
>   	.sram_offs      = {
>   		DEFINE_BAR_OFFSET(MBOX_CHANN_OFF, NPU4_SRAM, MPNPU_SRAM_X2I_MAILBOX_0),
>   		DEFINE_BAR_OFFSET(FW_ALIVE_OFF,   NPU4_SRAM, MPNPU_SRAM_X2I_MAILBOX_15),
> diff --git a/drivers/accel/amdxdna/npu5_regs.c b/drivers/accel/amdxdna/npu5_regs.c
> index dadd72df6263..75ad97f0b937 100644
> --- a/drivers/accel/amdxdna/npu5_regs.c
> +++ b/drivers/accel/amdxdna/npu5_regs.c
> @@ -72,6 +72,7 @@ static const struct amdxdna_dev_priv npu5_dev_priv = {
>   	.mbox_dev_addr  = NPU5_MBOX_BAR_BASE,
>   	.mbox_size      = 0, /* Use BAR size */
>   	.sram_dev_addr  = NPU5_SRAM_BAR_BASE,
> +	.hwctx_limit    = 16,
>   	.sram_offs      = {
>   		DEFINE_BAR_OFFSET(MBOX_CHANN_OFF, NPU5_SRAM, MPNPU_SRAM_X2I_MAILBOX_0),
>   		DEFINE_BAR_OFFSET(FW_ALIVE_OFF,   NPU5_SRAM, MPNPU_SRAM_X2I_MAILBOX_15),
> diff --git a/drivers/accel/amdxdna/npu6_regs.c b/drivers/accel/amdxdna/npu6_regs.c
> index f73f92869b23..758dc013fe13 100644
> --- a/drivers/accel/amdxdna/npu6_regs.c
> +++ b/drivers/accel/amdxdna/npu6_regs.c
> @@ -72,6 +72,7 @@ static const struct amdxdna_dev_priv npu6_dev_priv = {
>   	.mbox_dev_addr  = NPU6_MBOX_BAR_BASE,
>   	.mbox_size      = 0, /* Use BAR size */
>   	.sram_dev_addr  = NPU6_SRAM_BAR_BASE,
> +	.hwctx_limit    = 16,
>   	.sram_offs      = {
>   		DEFINE_BAR_OFFSET(MBOX_CHANN_OFF, NPU6_SRAM, MPNPU_SRAM_X2I_MAILBOX_0),
>   		DEFINE_BAR_OFFSET(FW_ALIVE_OFF,   NPU6_SRAM, MPNPU_SRAM_X2I_MAILBOX_15),

Re: [PATCH 2/3] accel/amdxdna: Add IOCTL parameter for resource data

[Mario Limonciello]

On 11/4/25 12:25 AM, Lizhi Hou wrote:
> Extend DRM_IOCTL_AMDXDNA_GET_INFO to include additional parameters
> that allow collection of resource data.
> 
> Signed-off-by: Lizhi Hou <lizhi.hou@amd.com>
Reviewed-by: Mario Limonciello (AMD) <superm1@kernel.org>> ---
>   drivers/accel/amdxdna/aie2_ctx.c        |  6 ------
>   drivers/accel/amdxdna/aie2_message.c    |  2 ++
>   drivers/accel/amdxdna/aie2_pci.c        | 27 +++++++++++++++++++++++++
>   drivers/accel/amdxdna/amdxdna_pci_drv.c |  3 ++-
>   include/uapi/drm/amdxdna_accel.h        | 17 ++++++++++++++++
>   5 files changed, 48 insertions(+), 7 deletions(-)
> 
> diff --git a/drivers/accel/amdxdna/aie2_ctx.c b/drivers/accel/amdxdna/aie2_ctx.c
> index 289a2aaf4cae..b78c47ed0d34 100644
> --- a/drivers/accel/amdxdna/aie2_ctx.c
> +++ b/drivers/accel/amdxdna/aie2_ctx.c
> @@ -556,7 +556,6 @@ int aie2_hwctx_init(struct amdxdna_hwctx *hwctx)
>   	struct drm_gpu_scheduler *sched;
>   	struct amdxdna_hwctx_priv *priv;
>   	struct amdxdna_gem_obj *heap;
> -	struct amdxdna_dev_hdl *ndev;
>   	int i, ret;
>   
>   	priv = kzalloc(sizeof(*hwctx->priv), GFP_KERNEL);
> @@ -654,8 +653,6 @@ int aie2_hwctx_init(struct amdxdna_hwctx *hwctx)
>   	amdxdna_pm_suspend_put(xdna);
>   
>   	hwctx->status = HWCTX_STAT_INIT;
> -	ndev = xdna->dev_handle;
> -	ndev->hwctx_num++;
>   	init_waitqueue_head(&priv->job_free_wq);
>   
>   	XDNA_DBG(xdna, "hwctx %s init completed", hwctx->name);
> @@ -688,13 +685,10 @@ int aie2_hwctx_init(struct amdxdna_hwctx *hwctx)
>   
>   void aie2_hwctx_fini(struct amdxdna_hwctx *hwctx)
>   {
> -	struct amdxdna_dev_hdl *ndev;
>   	struct amdxdna_dev *xdna;
>   	int idx;
>   
>   	xdna = hwctx->client->xdna;
> -	ndev = xdna->dev_handle;
> -	ndev->hwctx_num--;
>   
>   	XDNA_DBG(xdna, "%s sequence number %lld", hwctx->name, hwctx->priv->seq);
>   	drm_sched_entity_destroy(&hwctx->priv->entity);
> diff --git a/drivers/accel/amdxdna/aie2_message.c b/drivers/accel/amdxdna/aie2_message.c
> index 339dec998247..39214253d804 100644
> --- a/drivers/accel/amdxdna/aie2_message.c
> +++ b/drivers/accel/amdxdna/aie2_message.c
> @@ -235,6 +235,7 @@ int aie2_create_context(struct amdxdna_dev_hdl *ndev, struct amdxdna_hwctx *hwct
>   		ret = -EINVAL;
>   		goto out_destroy_context;
>   	}
> +	ndev->hwctx_num++;
>   
>   	XDNA_DBG(xdna, "%s mailbox channel irq: %d, msix_id: %d",
>   		 hwctx->name, ret, resp.msix_id);
> @@ -269,6 +270,7 @@ int aie2_destroy_context(struct amdxdna_dev_hdl *ndev, struct amdxdna_hwctx *hwc
>   		 hwctx->fw_ctx_id);
>   	hwctx->priv->mbox_chann = NULL;
>   	hwctx->fw_ctx_id = -1;
> +	ndev->hwctx_num--;
>   
>   	return ret;
>   }
> diff --git a/drivers/accel/amdxdna/aie2_pci.c b/drivers/accel/amdxdna/aie2_pci.c
> index ce57b915004e..396dc6e06007 100644
> --- a/drivers/accel/amdxdna/aie2_pci.c
> +++ b/drivers/accel/amdxdna/aie2_pci.c
> @@ -838,6 +838,30 @@ static int aie2_get_hwctx_status(struct amdxdna_client *client,
>   	return 0;
>   }
>   
> +static int aie2_query_resource_info(struct amdxdna_client *client,
> +				    struct amdxdna_drm_get_info *args)
> +{
> +	struct amdxdna_drm_get_resource_info res_info;
> +	const struct amdxdna_dev_priv *priv;
> +	struct amdxdna_dev_hdl *ndev;
> +	struct amdxdna_dev *xdna;
> +
> +	xdna = client->xdna;
> +	ndev = xdna->dev_handle;
> +	priv = ndev->priv;
> +
> +	res_info.npu_clk_max = priv->dpm_clk_tbl[ndev->max_dpm_level].hclk;
> +	res_info.npu_tops_max = ndev->max_tops;
> +	res_info.npu_task_max = priv->hwctx_limit;
> +	res_info.npu_tops_curr = ndev->curr_tops;
> +	res_info.npu_task_curr = ndev->hwctx_num;
> +
> +	if (copy_to_user(u64_to_user_ptr(args->buffer), &res_info, sizeof(res_info)))
> +		return -EFAULT;
> +
> +	return 0;
> +}
> +
>   static int aie2_get_info(struct amdxdna_client *client, struct amdxdna_drm_get_info *args)
>   {
>   	struct amdxdna_dev *xdna = client->xdna;
> @@ -872,6 +896,9 @@ static int aie2_get_info(struct amdxdna_client *client, struct amdxdna_drm_get_i
>   	case DRM_AMDXDNA_GET_POWER_MODE:
>   		ret = aie2_get_power_mode(client, args);
>   		break;
> +	case DRM_AMDXDNA_QUERY_RESOURCE_INFO:
> +		ret = aie2_query_resource_info(client, args);
> +		break;
>   	default:
>   		XDNA_ERR(xdna, "Not supported request parameter %u", args->param);
>   		ret = -EOPNOTSUPP;
> diff --git a/drivers/accel/amdxdna/amdxdna_pci_drv.c b/drivers/accel/amdxdna/amdxdna_pci_drv.c
> index 3599e713bfcb..af943a603ad1 100644
> --- a/drivers/accel/amdxdna/amdxdna_pci_drv.c
> +++ b/drivers/accel/amdxdna/amdxdna_pci_drv.c
> @@ -29,9 +29,10 @@ MODULE_FIRMWARE("amdnpu/17f0_20/npu.sbin");
>    * 0.1: Support getting all hardware contexts by DRM_IOCTL_AMDXDNA_GET_ARRAY
>    * 0.2: Support getting last error hardware error
>    * 0.3: Support firmware debug buffer
> + * 0.4: Support getting resource information
>    */
>   #define AMDXDNA_DRIVER_MAJOR		0
> -#define AMDXDNA_DRIVER_MINOR		3
> +#define AMDXDNA_DRIVER_MINOR		4
>   
>   /*
>    * Bind the driver base on (vendor_id, device_id) pair and later use the
> diff --git a/include/uapi/drm/amdxdna_accel.h b/include/uapi/drm/amdxdna_accel.h
> index c7eec9ceb2ae..8b679c38d308 100644
> --- a/include/uapi/drm/amdxdna_accel.h
> +++ b/include/uapi/drm/amdxdna_accel.h
> @@ -442,6 +442,23 @@ enum amdxdna_drm_get_param {
>   	DRM_AMDXDNA_QUERY_HW_CONTEXTS,
>   	DRM_AMDXDNA_QUERY_FIRMWARE_VERSION = 8,
>   	DRM_AMDXDNA_GET_POWER_MODE,
> +	DRM_AMDXDNA_QUERY_RESOURCE_INFO = 12,
> +};
> +
> +/**
> + * struct amdxdna_drm_get_resource_info - Get resource information
> + */
> +struct amdxdna_drm_get_resource_info {
> +	/** @npu_clk_max: max H-Clocks */
> +	__u64 npu_clk_max;
> +	/** @npu_tops_max: max TOPs */
> +	__u64 npu_tops_max;
> +	/** @npu_task_max: max number of tasks */
> +	__u64 npu_task_max;
> +	/** @npu_tops_curr: current TOPs */
> +	__u64 npu_tops_curr;
> +	/** @npu_task_curr: current number of tasks */
> +	__u64 npu_task_curr;
>   };
>   
>   /**

Re: [PATCH 3/3] accel/amdxdna: Add IOCTL parameter for telemetry data

[Mario Limonciello]

On 11/4/25 12:25 AM, Lizhi Hou wrote:
> Extend DRM_IOCTL_AMDXDNA_GET_INFO to include additional parameters
> that allow collection of telemetry data.
> 
> Signed-off-by: Lizhi Hou <lizhi.hou@amd.com>
Reviewed-by: Mario Limonciello (AMD) <superm1@kernel.org>> ---
>   drivers/accel/amdxdna/aie2_message.c          | 56 ++++++++++++--
>   drivers/accel/amdxdna/aie2_msg_priv.h         | 25 ++++++-
>   drivers/accel/amdxdna/aie2_pci.c              | 73 +++++++++++++++++++
>   drivers/accel/amdxdna/aie2_pci.h              |  3 +
>   .../accel/amdxdna/amdxdna_mailbox_helper.h    |  6 +-
>   drivers/accel/amdxdna/amdxdna_pci_drv.c       |  3 +-
>   include/uapi/drm/amdxdna_accel.h              | 17 +++++
>   7 files changed, 173 insertions(+), 10 deletions(-)
> 
> diff --git a/drivers/accel/amdxdna/aie2_message.c b/drivers/accel/amdxdna/aie2_message.c
> index 39214253d804..69cdce9ff208 100644
> --- a/drivers/accel/amdxdna/aie2_message.c
> +++ b/drivers/accel/amdxdna/aie2_message.c
> @@ -47,7 +47,7 @@ static int aie2_send_mgmt_msg_wait(struct amdxdna_dev_hdl *ndev,
>   		ndev->mgmt_chann = NULL;
>   	}
>   
> -	if (!ret && *hdl->data != AIE2_STATUS_SUCCESS) {
> +	if (!ret && *hdl->status != AIE2_STATUS_SUCCESS) {
>   		XDNA_ERR(xdna, "command opcode 0x%x failed, status 0x%x",
>   			 msg->opcode, *hdl->data);
>   		ret = -EINVAL;
> @@ -336,11 +336,6 @@ int aie2_query_status(struct amdxdna_dev_hdl *ndev, char __user *buf,
>   		goto fail;
>   	}
>   
> -	if (resp.status != AIE2_STATUS_SUCCESS) {
> -		XDNA_ERR(xdna, "Query NPU status failed, status 0x%x", resp.status);
> -		ret = -EINVAL;
> -		goto fail;
> -	}
>   	XDNA_DBG(xdna, "Query NPU status completed");
>   
>   	if (size < resp.size) {
> @@ -362,6 +357,55 @@ int aie2_query_status(struct amdxdna_dev_hdl *ndev, char __user *buf,
>   	return ret;
>   }
>   
> +int aie2_query_telemetry(struct amdxdna_dev_hdl *ndev,
> +			 char __user *buf, u32 size,
> +			 struct amdxdna_drm_query_telemetry_header *header)
> +{
> +	DECLARE_AIE2_MSG(get_telemetry, MSG_OP_GET_TELEMETRY);
> +	struct amdxdna_dev *xdna = ndev->xdna;
> +	dma_addr_t dma_addr;
> +	u8 *addr;
> +	int ret;
> +
> +	if (header->type >= MAX_TELEMETRY_TYPE)
> +		return -EINVAL;
> +
> +	addr = dma_alloc_noncoherent(xdna->ddev.dev, size, &dma_addr,
> +				     DMA_FROM_DEVICE, GFP_KERNEL);
> +	if (!addr)
> +		return -ENOMEM;
> +
> +	req.buf_addr = dma_addr;
> +	req.buf_size = size;
> +	req.type = header->type;
> +
> +	drm_clflush_virt_range(addr, size); /* device can access */
> +	ret = aie2_send_mgmt_msg_wait(ndev, &msg);
> +	if (ret) {
> +		XDNA_ERR(xdna, "Query telemetry failed, status %d", ret);
> +		goto free_buf;
> +	}
> +
> +	if (size < resp.size) {
> +		ret = -EINVAL;
> +		XDNA_ERR(xdna, "Bad buffer size. Available: %u. Needs: %u", size, resp.size);
> +		goto free_buf;
> +	}
> +
> +	if (copy_to_user(buf, addr, resp.size)) {
> +		ret = -EFAULT;
> +		XDNA_ERR(xdna, "Failed to copy telemetry to user space");
> +		goto free_buf;
> +	}
> +
> +	header->major = resp.major;
> +	header->minor = resp.minor;
> +
> +free_buf:
> +	dma_free_noncoherent(xdna->ddev.dev, size, addr, dma_addr, DMA_FROM_DEVICE);
> +	return ret;
> +}
> +
>   int aie2_register_asyn_event_msg(struct amdxdna_dev_hdl *ndev, dma_addr_t addr, u32 size,
>   				 void *handle, int (*cb)(void*, void __iomem *, size_t))
>   {
> diff --git a/drivers/accel/amdxdna/aie2_msg_priv.h b/drivers/accel/amdxdna/aie2_msg_priv.h
> index 945140011763..947daa63f064 100644
> --- a/drivers/accel/amdxdna/aie2_msg_priv.h
> +++ b/drivers/accel/amdxdna/aie2_msg_priv.h
> @@ -9,7 +9,8 @@
>   enum aie2_msg_opcode {
>   	MSG_OP_CREATE_CONTEXT              = 0x2,
>   	MSG_OP_DESTROY_CONTEXT             = 0x3,
> -	MSG_OP_SYNC_BO			   = 0x7,
> +	MSG_OP_GET_TELEMETRY               = 0x4,
> +	MSG_OP_SYNC_BO                     = 0x7,
>   	MSG_OP_EXECUTE_BUFFER_CF           = 0xC,
>   	MSG_OP_QUERY_COL_STATUS            = 0xD,
>   	MSG_OP_QUERY_AIE_TILE_INFO         = 0xE,
> @@ -137,6 +138,28 @@ struct destroy_ctx_resp {
>   	enum aie2_msg_status	status;
>   } __packed;
>   
> +enum telemetry_type {
> +	TELEMETRY_TYPE_DISABLED,
> +	TELEMETRY_TYPE_HEALTH,
> +	TELEMETRY_TYPE_ERROR_INFO,
> +	TELEMETRY_TYPE_PROFILING,
> +	TELEMETRY_TYPE_DEBUG,
> +	MAX_TELEMETRY_TYPE
> +};
> +
> +struct get_telemetry_req {
> +	enum telemetry_type	type;
> +	__u64	buf_addr;
> +	__u32	buf_size;
> +} __packed;
> +
> +struct get_telemetry_resp {
> +	__u32	major;
> +	__u32	minor;
> +	__u32	size;
> +	enum aie2_msg_status	status;
> +} __packed;
> +
>   struct execute_buffer_req {
>   	__u32	cu_idx;
>   	__u32	payload[19];
> diff --git a/drivers/accel/amdxdna/aie2_pci.c b/drivers/accel/amdxdna/aie2_pci.c
> index 396dc6e06007..d7ccbdaf47f5 100644
> --- a/drivers/accel/amdxdna/aie2_pci.c
> +++ b/drivers/accel/amdxdna/aie2_pci.c
> @@ -862,6 +862,76 @@ static int aie2_query_resource_info(struct amdxdna_client *client,
>   	return 0;
>   }
>   
> +static int aie2_fill_hwctx_map(struct amdxdna_hwctx *hwctx, void *arg)
> +{
> +	struct amdxdna_dev *xdna = hwctx->client->xdna;
> +	u32 *map = arg;
> +
> +	if (hwctx->fw_ctx_id >= xdna->dev_handle->priv->hwctx_limit) {
> +		XDNA_ERR(xdna, "Invalid fw ctx id %d/%d ", hwctx->fw_ctx_id,
> +			 xdna->dev_handle->priv->hwctx_limit);
> +		return -EINVAL;
> +	}
> +
> +	map[hwctx->fw_ctx_id] = hwctx->id;
> +	return 0;
> +}
> +
> +static int aie2_get_telemetry(struct amdxdna_client *client,
> +			      struct amdxdna_drm_get_info *args)
> +{
> +	struct amdxdna_drm_query_telemetry_header *header __free(kfree) = NULL;
> +	u32 telemetry_data_sz, header_sz, elem_num;
> +	struct amdxdna_dev *xdna = client->xdna;
> +	struct amdxdna_client *tmp_client;
> +	int ret;
> +
> +	elem_num = xdna->dev_handle->priv->hwctx_limit;
> +	header_sz = struct_size(header, map, elem_num);
> +	if (args->buffer_size <= header_sz) {
> +		XDNA_ERR(xdna, "Invalid buffer size");
> +		return -EINVAL;
> +	}
> +
> +	telemetry_data_sz = args->buffer_size - header_sz;
> +	if (telemetry_data_sz > SZ_4M) {
> +		XDNA_ERR(xdna, "Buffer size is too big, %d", telemetry_data_sz);
> +		return -EINVAL;
> +	}
> +
> +	header = kzalloc(header_sz, GFP_KERNEL);
> +	if (!header)
> +		return -ENOMEM;
> +
> +	if (copy_from_user(header, u64_to_user_ptr(args->buffer), sizeof(*header))) {
> +		XDNA_ERR(xdna, "Failed to copy telemetry header from user");
> +		return -EFAULT;
> +	}
> +
> +	header->map_num_elements = elem_num;
> +	list_for_each_entry(tmp_client, &xdna->client_list, node) {
> +		ret = amdxdna_hwctx_walk(tmp_client, &header->map,
> +					 aie2_fill_hwctx_map);
> +		if (ret)
> +			return ret;
> +	}
> +
> +	ret = aie2_query_telemetry(xdna->dev_handle,
> +				   u64_to_user_ptr(args->buffer + header_sz),
> +				   telemetry_data_sz, header);
> +	if (ret) {
> +		XDNA_ERR(xdna, "Query telemetry failed ret %d", ret);
> +		return ret;
> +	}
> +
> +	if (copy_to_user(u64_to_user_ptr(args->buffer), header, header_sz)) {
> +		XDNA_ERR(xdna, "Copy header failed");
> +		return -EFAULT;
> +	}
> +
> +	return 0;
> +}
> +
>   static int aie2_get_info(struct amdxdna_client *client, struct amdxdna_drm_get_info *args)
>   {
>   	struct amdxdna_dev *xdna = client->xdna;
> @@ -896,6 +966,9 @@ static int aie2_get_info(struct amdxdna_client *client, struct amdxdna_drm_get_i
>   	case DRM_AMDXDNA_GET_POWER_MODE:
>   		ret = aie2_get_power_mode(client, args);
>   		break;
> +	case DRM_AMDXDNA_QUERY_TELEMETRY:
> +		ret = aie2_get_telemetry(client, args);
> +		break;
>   	case DRM_AMDXDNA_QUERY_RESOURCE_INFO:
>   		ret = aie2_query_resource_info(client, args);
>   		break;
> diff --git a/drivers/accel/amdxdna/aie2_pci.h b/drivers/accel/amdxdna/aie2_pci.h
> index a79f4f71ff6b..9793cd1e0c55 100644
> --- a/drivers/accel/amdxdna/aie2_pci.h
> +++ b/drivers/accel/amdxdna/aie2_pci.h
> @@ -305,6 +305,9 @@ int aie2_create_context(struct amdxdna_dev_hdl *ndev, struct amdxdna_hwctx *hwct
>   int aie2_destroy_context(struct amdxdna_dev_hdl *ndev, struct amdxdna_hwctx *hwctx);
>   int aie2_map_host_buf(struct amdxdna_dev_hdl *ndev, u32 context_id, u64 addr, u64 size);
>   int aie2_query_status(struct amdxdna_dev_hdl *ndev, char __user *buf, u32 size, u32 *cols_filled);
> +int aie2_query_telemetry(struct amdxdna_dev_hdl *ndev,
> +			 char __user *buf, u32 size,
> +			 struct amdxdna_drm_query_telemetry_header *header);
>   int aie2_register_asyn_event_msg(struct amdxdna_dev_hdl *ndev, dma_addr_t addr, u32 size,
>   				 void *handle, int (*cb)(void*, void __iomem *, size_t));
>   int aie2_config_cu(struct amdxdna_hwctx *hwctx,
> diff --git a/drivers/accel/amdxdna/amdxdna_mailbox_helper.h b/drivers/accel/amdxdna/amdxdna_mailbox_helper.h
> index 710ff8873d61..556c712cad0a 100644
> --- a/drivers/accel/amdxdna/amdxdna_mailbox_helper.h
> +++ b/drivers/accel/amdxdna/amdxdna_mailbox_helper.h
> @@ -16,16 +16,18 @@ struct xdna_notify {
>   	u32			*data;
>   	size_t			size;
>   	int			error;
> +	u32			*status;
>   };
>   
> -#define DECLARE_XDNA_MSG_COMMON(name, op, status)			\
> +#define DECLARE_XDNA_MSG_COMMON(name, op, s)				\
>   	struct name##_req	req = { 0 };				\
> -	struct name##_resp	resp = { status	};			\
> +	struct name##_resp	resp = { .status = s };			\
>   	struct xdna_notify	hdl = {					\
>   		.error = 0,						\
>   		.data = (u32 *)&resp,					\
>   		.size = sizeof(resp),					\
>   		.comp = COMPLETION_INITIALIZER_ONSTACK(hdl.comp),	\
> +		.status = (u32 *)&resp.status,				\
>   	};								\
>   	struct xdna_mailbox_msg msg = {					\
>   		.send_data = (u8 *)&req,				\
> diff --git a/drivers/accel/amdxdna/amdxdna_pci_drv.c b/drivers/accel/amdxdna/amdxdna_pci_drv.c
> index af943a603ad1..7590265d4485 100644
> --- a/drivers/accel/amdxdna/amdxdna_pci_drv.c
> +++ b/drivers/accel/amdxdna/amdxdna_pci_drv.c
> @@ -30,9 +30,10 @@ MODULE_FIRMWARE("amdnpu/17f0_20/npu.sbin");
>    * 0.2: Support getting last error hardware error
>    * 0.3: Support firmware debug buffer
>    * 0.4: Support getting resource information
> + * 0.5: Support getting telemetry data
>    */
>   #define AMDXDNA_DRIVER_MAJOR		0
> -#define AMDXDNA_DRIVER_MINOR		4
> +#define AMDXDNA_DRIVER_MINOR		5
>   
>   /*
>    * Bind the driver base on (vendor_id, device_id) pair and later use the
> diff --git a/include/uapi/drm/amdxdna_accel.h b/include/uapi/drm/amdxdna_accel.h
> index 8b679c38d308..8ad254bc35a5 100644
> --- a/include/uapi/drm/amdxdna_accel.h
> +++ b/include/uapi/drm/amdxdna_accel.h
> @@ -442,6 +442,7 @@ enum amdxdna_drm_get_param {
>   	DRM_AMDXDNA_QUERY_HW_CONTEXTS,
>   	DRM_AMDXDNA_QUERY_FIRMWARE_VERSION = 8,
>   	DRM_AMDXDNA_GET_POWER_MODE,
> +	DRM_AMDXDNA_QUERY_TELEMETRY,
>   	DRM_AMDXDNA_QUERY_RESOURCE_INFO = 12,
>   };
>   
> @@ -461,6 +462,22 @@ struct amdxdna_drm_get_resource_info {
>   	__u64 npu_task_curr;
>   };
>   
> +/**
> + * struct amdxdna_drm_query_telemetry_header - Telemetry data header
> + */
> +struct amdxdna_drm_query_telemetry_header {
> +	/** @major: Firmware telemetry interface major version number */
> +	__u32 major;
> +	/** @minor: Firmware telemetry interface minor version number */
> +	__u32 minor;
> +	/** @type: Telemetry query type */
> +	__u32 type;
> +	/** @map_num_elements: Total number of elements in the map table */
> +	__u32 map_num_elements;
> +	/** @map: Element map */
> +	__u32 map[];
> +};
> +
>   /**
>    * struct amdxdna_drm_get_info - Get some information from the AIE hardware.
>    * @param: Value in enum amdxdna_drm_get_param. Specifies the structure passed in the buffer.