drm/amdkfd: Introduce kfd_node struct (v5)

Introduce a new structure, kfd_node, which will now represent
a compute node. kfd_node is carved out of kfd_dev structure.
kfd_dev struct now will become the parent of kfd_node, and will
store common resources such as doorbells, GTT sub-alloctor etc.
kfd_node struct will store all resources specific to a compute
node, such as device queue manager, interrupt handling etc.

This is the first step in adding compute partition support in KFD.

v2: introduce kfd_node struct to gc v11 (Hawking)
v3: make reference to kfd_dev struct through kfd_node (Morris)
v4: use kfd_node instead for kfd isr/mqd functions (Morris)
v5: rebase (Alex)

Signed-off-by: Mukul Joshi <mukul.joshi@amd.com>
Tested-by: Amber Lin <Amber.Lin@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Morris Zhang <Shiwu.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

1 files changed, 94 insertions, 74 deletions

diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_priv.h b/drivers/gpu/drm/amd/amdkfd/kfd_priv.h
index 94a438956868..fdb97e5d0c01 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_priv.h
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_priv.h
@@ -210,11 +210,13 @@ enum cache_policy {
 	((KFD_GC_VERSION(dev) == IP_VERSION(9, 4, 2)) ||	\
 	 (KFD_GC_VERSION(dev) == IP_VERSION(9, 4, 3)))
 
+struct kfd_node;
+
 struct kfd_event_interrupt_class {
-	bool (*interrupt_isr)(struct kfd_dev *dev,
+	bool (*interrupt_isr)(struct kfd_node *dev,
 			const uint32_t *ih_ring_entry, uint32_t *patched_ihre,
 			bool *patched_flag);
-	void (*interrupt_wq)(struct kfd_dev *dev,
+	void (*interrupt_wq)(struct kfd_node *dev,
 			const uint32_t *ih_ring_entry);
 };
 
@@ -236,8 +238,8 @@ struct kfd_device_info {
 	uint64_t reserved_sdma_queues_bitmap;
 };
 
-unsigned int kfd_get_num_sdma_engines(struct kfd_dev *kdev);
-unsigned int kfd_get_num_xgmi_sdma_engines(struct kfd_dev *kdev);
+unsigned int kfd_get_num_sdma_engines(struct kfd_node *kdev);
+unsigned int kfd_get_num_xgmi_sdma_engines(struct kfd_node *kdev);
 
 struct kfd_mem_obj {
 	uint32_t range_start;
@@ -253,13 +255,59 @@ struct kfd_vmid_info {
 	uint32_t vmid_num_kfd;
 };
 
+struct kfd_dev;
+
+struct kfd_node {
+	struct amdgpu_device *adev;     /* Duplicated here along with keeping
+					 * a copy in kfd_dev to save a hop
+					 */
+	const struct kfd2kgd_calls *kfd2kgd; /* Duplicated here along with
+					      * keeping a copy in kfd_dev to
+					      * save a hop
+					      */
+	struct kfd_vmid_info vm_info;
+	unsigned int id;                /* topology stub index */
+	/* Interrupts */
+	struct kfifo ih_fifo;
+	struct workqueue_struct *ih_wq;
+	struct work_struct interrupt_work;
+	spinlock_t interrupt_lock;
+
+	/*
+	 * Interrupts of interest to KFD are copied
+	 * from the HW ring into a SW ring.
+	 */
+	bool interrupts_active;
+
+	/* QCM Device instance */
+	struct device_queue_manager *dqm;
+
+	/* Global GWS resource shared between processes */
+	void *gws;
+	bool gws_debug_workaround;
+
+	/* Clients watching SMI events */
+	struct list_head smi_clients;
+	spinlock_t smi_lock;
+	uint32_t reset_seq_num;
+
+	/* SRAM ECC flag */
+	atomic_t sram_ecc_flag;
+
+	/*spm process id */
+	unsigned int spm_pasid;
+
+	/* Maximum process number mapped to HW scheduler */
+	unsigned int max_proc_per_quantum;
+
+	struct kfd_dev *kfd;
+};
+
 struct kfd_dev {
 	struct amdgpu_device *adev;
 
 	struct kfd_device_info device_info;
 
-	unsigned int id;		/* topology stub index */
-
 	phys_addr_t doorbell_base;	/* Start of actual doorbells used by
 					 * KFD. It is aligned for mapping
 					 * into user mode
@@ -274,7 +322,6 @@ struct kfd_dev {
 					   */
 
 	struct kgd2kfd_shared_resources shared_resources;
-	struct kfd_vmid_info vm_info;
 	struct kfd_local_mem_info local_mem_info;
 
 	const struct kfd2kgd_calls *kfd2kgd;
@@ -290,30 +337,13 @@ struct kfd_dev {
 	unsigned int gtt_sa_chunk_size;
 	unsigned int gtt_sa_num_of_chunks;
 
-	/* Interrupts */
-	struct kfifo ih_fifo;
-	struct workqueue_struct *ih_wq;
-	struct work_struct interrupt_work;
-	spinlock_t interrupt_lock;
-
-	/* QCM Device instance */
-	struct device_queue_manager *dqm;
-
 	bool init_complete;
-	/*
-	 * Interrupts of interest to KFD are copied
-	 * from the HW ring into a SW ring.
-	 */
-	bool interrupts_active;
 
 	/* Firmware versions */
 	uint16_t mec_fw_version;
 	uint16_t mec2_fw_version;
 	uint16_t sdma_fw_version;
 
-	/* Maximum process number mapped to HW scheduler */
-	unsigned int max_proc_per_quantum;
-
 	/* CWSR */
 	bool cwsr_enabled;
 	const void *cwsr_isa;
@@ -327,21 +357,9 @@ struct kfd_dev {
 	/* Use IOMMU v2 flag */
 	bool use_iommu_v2;
 
-	/* SRAM ECC flag */
-	atomic_t sram_ecc_flag;
-
 	/* Compute Profile ref. count */
 	atomic_t compute_profile;
 
-	/* Global GWS resource shared between processes */
-	void *gws;
-
-	/* Clients watching SMI events */
-	struct list_head smi_clients;
-	spinlock_t smi_lock;
-
-	uint32_t reset_seq_num;
-
 	struct ida doorbell_ida;
 	unsigned int max_doorbell_slices;
 
@@ -349,6 +367,8 @@ struct kfd_dev {
 
 	/* HMM page migration MEMORY_DEVICE_PRIVATE mapping */
 	struct dev_pagemap pgmap;
+
+	struct kfd_node *node;
 };
 
 enum kfd_mempool {
@@ -563,7 +583,7 @@ struct queue {
 	unsigned int doorbell_id;
 
 	struct kfd_process	*process;
-	struct kfd_dev		*device;
+	struct kfd_node		*device;
 	void *gws;
 
 	/* procfs */
@@ -697,7 +717,7 @@ enum kfd_pdd_bound {
 /* Data that is per-process-per device. */
 struct kfd_process_device {
 	/* The device that owns this data. */
-	struct kfd_dev *dev;
+	struct kfd_node *dev;
 
 	/* The process that owns this kfd_process_device. */
 	struct kfd_process *process;
@@ -925,7 +945,7 @@ struct amdkfd_ioctl_desc {
 	unsigned int cmd_drv;
 	const char *name;
 };
-bool kfd_dev_is_large_bar(struct kfd_dev *dev);
+bool kfd_dev_is_large_bar(struct kfd_node *dev);
 
 int kfd_process_create_wq(void);
 void kfd_process_destroy_wq(void);
@@ -961,16 +981,16 @@ int kfd_process_get_user_gpu_id(struct kfd_process *p, uint32_t actual_gpu_id);
 
 int kfd_process_device_init_vm(struct kfd_process_device *pdd,
 			       struct file *drm_file);
-struct kfd_process_device *kfd_bind_process_to_device(struct kfd_dev *dev,
+struct kfd_process_device *kfd_bind_process_to_device(struct kfd_node *dev,
 						struct kfd_process *p);
-struct kfd_process_device *kfd_get_process_device_data(struct kfd_dev *dev,
+struct kfd_process_device *kfd_get_process_device_data(struct kfd_node *dev,
 							struct kfd_process *p);
-struct kfd_process_device *kfd_create_process_device_data(struct kfd_dev *dev,
+struct kfd_process_device *kfd_create_process_device_data(struct kfd_node *dev,
 							struct kfd_process *p);
 
 bool kfd_process_xnack_mode(struct kfd_process *p, bool supported);
 
-int kfd_reserved_mem_mmap(struct kfd_dev *dev, struct kfd_process *process,
+int kfd_reserved_mem_mmap(struct kfd_node *dev, struct kfd_process *process,
 			  struct vm_area_struct *vma);
 
 /* KFD process API for creating and translating handles */
@@ -994,7 +1014,7 @@ void kfd_pasid_free(u32 pasid);
 size_t kfd_doorbell_process_slice(struct kfd_dev *kfd);
 int kfd_doorbell_init(struct kfd_dev *kfd);
 void kfd_doorbell_fini(struct kfd_dev *kfd);
-int kfd_doorbell_mmap(struct kfd_dev *dev, struct kfd_process *process,
+int kfd_doorbell_mmap(struct kfd_node *dev, struct kfd_process *process,
 		      struct vm_area_struct *vma);
 void __iomem *kfd_get_kernel_doorbell(struct kfd_dev *kfd,
 					unsigned int *doorbell_off);
@@ -1012,10 +1032,10 @@ void kfd_free_process_doorbells(struct kfd_dev *kfd,
 				unsigned int doorbell_index);
 /* GTT Sub-Allocator */
 
-int kfd_gtt_sa_allocate(struct kfd_dev *kfd, unsigned int size,
+int kfd_gtt_sa_allocate(struct kfd_node *node, unsigned int size,
 			struct kfd_mem_obj **mem_obj);
 
-int kfd_gtt_sa_free(struct kfd_dev *kfd, struct kfd_mem_obj *mem_obj);
+int kfd_gtt_sa_free(struct kfd_node *node, struct kfd_mem_obj *mem_obj);
 
 extern struct device *kfd_device;
 
@@ -1028,25 +1048,25 @@ void kfd_procfs_del_queue(struct queue *q);
 /* Topology */
 int kfd_topology_init(void);
 void kfd_topology_shutdown(void);
-int kfd_topology_add_device(struct kfd_dev *gpu);
-int kfd_topology_remove_device(struct kfd_dev *gpu);
+int kfd_topology_add_device(struct kfd_node *gpu);
+int kfd_topology_remove_device(struct kfd_node *gpu);
 struct kfd_topology_device *kfd_topology_device_by_proximity_domain(
 						uint32_t proximity_domain);
 struct kfd_topology_device *kfd_topology_device_by_proximity_domain_no_lock(
 						uint32_t proximity_domain);
 struct kfd_topology_device *kfd_topology_device_by_id(uint32_t gpu_id);
-struct kfd_dev *kfd_device_by_id(uint32_t gpu_id);
-struct kfd_dev *kfd_device_by_pci_dev(const struct pci_dev *pdev);
-struct kfd_dev *kfd_device_by_adev(const struct amdgpu_device *adev);
-int kfd_topology_enum_kfd_devices(uint8_t idx, struct kfd_dev **kdev);
+struct kfd_node *kfd_device_by_id(uint32_t gpu_id);
+struct kfd_node *kfd_device_by_pci_dev(const struct pci_dev *pdev);
+struct kfd_node *kfd_device_by_adev(const struct amdgpu_device *adev);
+int kfd_topology_enum_kfd_devices(uint8_t idx, struct kfd_node **kdev);
 int kfd_numa_node_to_apic_id(int numa_node_id);
 void kfd_double_confirm_iommu_support(struct kfd_dev *gpu);
 
 /* Interrupts */
-int kfd_interrupt_init(struct kfd_dev *dev);
-void kfd_interrupt_exit(struct kfd_dev *dev);
-bool enqueue_ih_ring_entry(struct kfd_dev *kfd,	const void *ih_ring_entry);
-bool interrupt_is_wanted(struct kfd_dev *dev,
+int kfd_interrupt_init(struct kfd_node *dev);
+void kfd_interrupt_exit(struct kfd_node *dev);
+bool enqueue_ih_ring_entry(struct kfd_node *kfd, const void *ih_ring_entry);
+bool interrupt_is_wanted(struct kfd_node *dev,
 				const uint32_t *ih_ring_entry,
 				uint32_t *patched_ihre, bool *flag);
 
@@ -1174,22 +1194,22 @@ void print_queue_properties(struct queue_properties *q);
 void print_queue(struct queue *q);
 
 struct mqd_manager *mqd_manager_init_cik(enum KFD_MQD_TYPE type,
-		struct kfd_dev *dev);
+		struct kfd_node *dev);
 struct mqd_manager *mqd_manager_init_cik_hawaii(enum KFD_MQD_TYPE type,
-		struct kfd_dev *dev);
+		struct kfd_node *dev);
 struct mqd_manager *mqd_manager_init_vi(enum KFD_MQD_TYPE type,
-		struct kfd_dev *dev);
+		struct kfd_node *dev);
 struct mqd_manager *mqd_manager_init_vi_tonga(enum KFD_MQD_TYPE type,
-		struct kfd_dev *dev);
+		struct kfd_node *dev);
 struct mqd_manager *mqd_manager_init_v9(enum KFD_MQD_TYPE type,
-		struct kfd_dev *dev);
+		struct kfd_node *dev);
 struct mqd_manager *mqd_manager_init_v10(enum KFD_MQD_TYPE type,
-		struct kfd_dev *dev);
+		struct kfd_node *dev);
 struct mqd_manager *mqd_manager_init_v11(enum KFD_MQD_TYPE type,
-		struct kfd_dev *dev);
-struct device_queue_manager *device_queue_manager_init(struct kfd_dev *dev);
+		struct kfd_node *dev);
+struct device_queue_manager *device_queue_manager_init(struct kfd_node *dev);
 void device_queue_manager_uninit(struct device_queue_manager *dqm);
-struct kernel_queue *kernel_queue_init(struct kfd_dev *dev,
+struct kernel_queue *kernel_queue_init(struct kfd_node *dev,
 					enum kfd_queue_type type);
 void kernel_queue_uninit(struct kernel_queue *kq, bool hanging);
 int kfd_dqm_evict_pasid(struct device_queue_manager *dqm, u32 pasid);
@@ -1206,7 +1226,7 @@ void kfd_process_dequeue_from_all_devices(struct kfd_process *p);
 int pqm_init(struct process_queue_manager *pqm, struct kfd_process *p);
 void pqm_uninit(struct process_queue_manager *pqm);
 int pqm_create_queue(struct process_queue_manager *pqm,
-			    struct kfd_dev *dev,
+			    struct kfd_node *dev,
 			    struct file *f,
 			    struct queue_properties *properties,
 			    unsigned int *qid,
@@ -1323,7 +1343,7 @@ int kfd_wait_on_events(struct kfd_process *p,
 		       uint32_t *wait_result);
 void kfd_signal_event_interrupt(u32 pasid, uint32_t partial_id,
 				uint32_t valid_id_bits);
-void kfd_signal_iommu_event(struct kfd_dev *dev,
+void kfd_signal_iommu_event(struct kfd_node *dev,
 			    u32 pasid, unsigned long address,
 			    bool is_write_requested, bool is_execute_requested);
 void kfd_signal_hw_exception_event(u32 pasid);
@@ -1339,12 +1359,12 @@ int kfd_event_create(struct file *devkfd, struct kfd_process *p,
 int kfd_get_num_events(struct kfd_process *p);
 int kfd_event_destroy(struct kfd_process *p, uint32_t event_id);
 
-void kfd_signal_vm_fault_event(struct kfd_dev *dev, u32 pasid,
+void kfd_signal_vm_fault_event(struct kfd_node *dev, u32 pasid,
 				struct kfd_vm_fault_info *info);
 
-void kfd_signal_reset_event(struct kfd_dev *dev);
+void kfd_signal_reset_event(struct kfd_node *dev);
 
-void kfd_signal_poison_consumed_event(struct kfd_dev *dev, u32 pasid);
+void kfd_signal_poison_consumed_event(struct kfd_node *dev, u32 pasid);
 
 void kfd_flush_tlb(struct kfd_process_device *pdd, enum TLB_FLUSH_TYPE type);
 
@@ -1359,12 +1379,12 @@ static inline bool kfd_flush_tlb_after_unmap(struct kfd_dev *dev)
 bool kfd_is_locked(void);
 
 /* Compute profile */
-void kfd_inc_compute_active(struct kfd_dev *dev);
-void kfd_dec_compute_active(struct kfd_dev *dev);
+void kfd_inc_compute_active(struct kfd_node *dev);
+void kfd_dec_compute_active(struct kfd_node *dev);
 
 /* Cgroup Support */
 /* Check with device cgroup if @kfd device is accessible */
-static inline int kfd_devcgroup_check_permission(struct kfd_dev *kfd)
+static inline int kfd_devcgroup_check_permission(struct kfd_node *kfd)
 {
 #if defined(CONFIG_CGROUP_DEVICE) || defined(CONFIG_CGROUP_BPF)
 	struct drm_device *ddev = adev_to_drm(kfd->adev);
@@ -1389,7 +1409,7 @@ int dqm_debugfs_hqds(struct seq_file *m, void *data);
 int kfd_debugfs_rls_by_device(struct seq_file *m, void *data);
 int pm_debugfs_runlist(struct seq_file *m, void *data);
 
-int kfd_debugfs_hang_hws(struct kfd_dev *dev);
+int kfd_debugfs_hang_hws(struct kfd_node *dev);
 int pm_debugfs_hang_hws(struct packet_manager *pm);
 int dqm_debugfs_hang_hws(struct device_queue_manager *dqm);