Skip to contentSkip to conversation

drm/amdgpu: Create hqd info structure

PATCH Superseded Amber Lin 1 participant 2 messages Apr 1

Subsystems (3)
lore
Share & tools
lore

Amber Lin

author
#0 lore raw

PATCH v4 4/10 drm/amdgpu: Create hqd info structure

Create hung_queue_hqd_info structure and fill in hung queses information passed by MES, including queue type, pipe id, and queue id.

---
 drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c | 20 ++++++++------------
 drivers/gpu/drm/amd/amdgpu/amdgpu_mes.h | 13 +++++++++++++
 2 files changed, 21 insertions(+), 12 deletions(-)
drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c +8 -12
index f1f8bbfc31e0..47c989980824 100644
@@ -447,7 +447,7 @@ int amdgpu_mes_detect_and_reset_hung_queues(struct amdgpu_device *adev,
 {
 	struct mes_detect_and_reset_queue_input input;
 	u32 *db_array = adev->mes.hung_queue_db_array_cpu_addr[xcc_id];
-	int r, i;
+	int hqd_info_offset = adev->mes.hung_queue_hqd_info_offset, r, i;
 
 	if (!hung_db_num || !hung_db_array)
 		return -EINVAL;
@@ -466,8 +466,9 @@ int amdgpu_mes_detect_and_reset_hung_queues(struct amdgpu_device *adev,
 	r = adev->mes.funcs->detect_and_reset_hung_queues(&adev->mes,
 							  &input);
 
-	if (r && detect_only) {
-		dev_err(adev->dev, "Failed to detect hung queues\n");
+	if (r) {
+		dev_warn(adev->dev, "Failed to %s hung queues\n",
+			detect_only? "detect" : "reset");
 		return r;
 	}
 
@@ -481,18 +482,13 @@ int amdgpu_mes_detect_and_reset_hung_queues(struct amdgpu_device *adev,
 		}
 	}
 
-	if (r && !hung_db_num) {
-		dev_err(adev->dev, "Failed to detect and reset hung queues\n");
+	if (!hung_db_num) {
+		dev_warn(adev->dev, "No hung queues info from MES\n");
 		return r;
 	}
 
-	/*
-	 * TODO: return HQD info for MES scheduled user compute queue reset cases
-	 * stored in hung_db_array hqd info offset to full array size
-	 */
-
-	if (r)
-		dev_err(adev->dev, "failed to reset\n");
+	for (i = hqd_info_offset; i < hqd_info_offset + *hung_db_num; i++)
+		hung_db_array[i] = db_array[i];
 
 	return r;
 }
drivers/gpu/drm/amd/amdgpu/amdgpu_mes.h +13 -0
index f80e3aca9c78..2e6ae9f84db0 100644
@@ -170,6 +170,19 @@ struct amdgpu_mes {
 	uint64_t            shared_cmd_buf_gpu_addr[AMDGPU_MAX_MES_INST_PIPES];
 };
 
+struct amdgpu_mes_hung_queue_hqd_info {
+	union {
+		struct {
+			uint32_t queue_type: 3; // queue type
+			uint32_t pipe_index: 4; // pipe index
+			uint32_t queue_index: 8; // queue index
+			uint32_t reserved: 17;
+		};
+
+		uint32_t bit0_31;
+	};
+};
+
 struct amdgpu_mes_gang {
 	int 				gang_id;
 	int 				priority;
Signature · 1 line
2.43.0
author
#1 lore raw

PATCH v4 9/10 drm/amdkfd: Reset queue/pipe in MES

When removing queues fails, KFD calls amdgpu_mes to detect and reset hung queues, then cleans up those hung queues in KFD.

---
 drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c       |   6 +
 drivers/gpu/drm/amd/amdgpu/amdgpu_mes.h       |   1 +
 .../drm/amd/amdkfd/kfd_device_queue_manager.c | 147 +++++++++++++++++-
 .../drm/amd/amdkfd/kfd_device_queue_manager.h |   4 +-
 drivers/gpu/drm/amd/amdkfd/kfd_priv.h         |   1 +
 5 files changed, 156 insertions(+), 3 deletions(-)
drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c +6 -0
index c7e4b882809d..ab62203743c0 100644
@@ -788,6 +788,12 @@ bool amdgpu_mes_suspend_resume_all_supported(struct amdgpu_device *adev)
 		amdgpu_ip_version(adev, GC_HWIP, 0) >= IP_VERSION(12, 0, 0));
 }
 
+bool amdgpu_mes_queue_reset_by_mes_supported(struct amdgpu_device *adev)
+{
+	return (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(12, 1, 0) &&
+		(adev->mes.sched_version & AMDGPU_MES_VERSION_MASK) >= 0x73);
+}
+
 /* Fix me -- node_id is used to identify the correct MES instances in the future */
 static int amdgpu_mes_set_enforce_isolation(struct amdgpu_device *adev,
 					    uint32_t node_id, bool enable)
drivers/gpu/drm/amd/amdgpu/amdgpu_mes.h +1 -0
index 643b4f8d757a..44fa4d73bce8 100644
@@ -548,6 +548,7 @@ static inline void amdgpu_mes_unlock(struct amdgpu_mes *mes)
 }
 
 bool amdgpu_mes_suspend_resume_all_supported(struct amdgpu_device *adev);
+bool amdgpu_mes_queue_reset_by_mes_supported(struct amdgpu_device *adev);
 
 int amdgpu_mes_update_enforce_isolation(struct amdgpu_device *adev);
 
drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c +145 -2

225 lines not on this page · raw message

drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.h +3 -1
index 3272328da11f..e6eca38cae4e 100644
@@ -32,7 +32,6 @@
 #include "kfd_priv.h"
 #include "kfd_mqd_manager.h"
 
-
 #define VMID_NUM 16
 
 #define KFD_MES_PROCESS_QUANTUM		100000
@@ -285,6 +284,9 @@ struct device_queue_manager {
 	struct dqm_detect_hang_info *detect_hang_info;
 	size_t detect_hang_info_size;
 	int detect_hang_count;
+	/* for per-queue reset with mes */
+	uint32_t *hung_db_array;
+	struct amdgpu_mes_hung_queue_hqd_info *hqd_info;
 };
 
 void device_queue_manager_init_cik(
drivers/gpu/drm/amd/amdkfd/kfd_priv.h +1 -0
index e7a8f3e17872..7e0d4b83c2cf 100644
@@ -523,6 +523,7 @@ struct queue_properties {
 	uint32_t pm4_target_xcc;
 	bool is_dbg_wa;
 	bool is_user_cu_masked;
+	bool is_reset;
 	/* Not relevant for user mode queues in cp scheduling */
 	unsigned int vmid;
 	/* Relevant only for sdma queues*/
Signature · 1 line
2.43.0