drm/amdkfd: Add XCC instance to kgd2kfd interface (v3)

Gfx 9 starts to have multiple XCC instances in one device. Add instance
parameter to kgd2kfd functions where XCC instance was hard coded as 0.
Also, update code to pass the correct instance number when running
on a multi-XCC setup.

v2: introduce the XCC instance to gfx v11 (Morris)
v3: rebase (Alex)

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

1 files changed, 12 insertions, 10 deletions

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v10.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v10.c
index f599e1e74fcc..7b60268d93c0 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v10.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v10.c
@@ -79,7 +79,7 @@ static void kgd_program_sh_mem_settings(struct amdgpu_device *adev, uint32_t vmi
 					uint32_t sh_mem_config,
 					uint32_t sh_mem_ape1_base,
 					uint32_t sh_mem_ape1_limit,
-					uint32_t sh_mem_bases)
+					uint32_t sh_mem_bases, uint32_t inst)
 {
 	lock_srbm(adev, 0, 0, 0, vmid);
 
@@ -91,7 +91,7 @@ static void kgd_program_sh_mem_settings(struct amdgpu_device *adev, uint32_t vmi
 }
 
 static int kgd_set_pasid_vmid_mapping(struct amdgpu_device *adev, u32 pasid,
-					unsigned int vmid)
+					unsigned int vmid, uint32_t inst)
 {
 	/*
 	 * We have to assume that there is no outstanding mapping.
@@ -135,7 +135,8 @@ static int kgd_set_pasid_vmid_mapping(struct amdgpu_device *adev, u32 pasid,
  * but still works
  */
 
-static int kgd_init_interrupts(struct amdgpu_device *adev, uint32_t pipe_id)
+static int kgd_init_interrupts(struct amdgpu_device *adev, uint32_t pipe_id,
+				uint32_t inst)
 {
 	uint32_t mec;
 	uint32_t pipe;
@@ -205,7 +206,7 @@ static inline struct v10_sdma_mqd *get_sdma_mqd(void *mqd)
 static int kgd_hqd_load(struct amdgpu_device *adev, void *mqd,
 			uint32_t pipe_id, uint32_t queue_id,
 			uint32_t __user *wptr, uint32_t wptr_shift,
-			uint32_t wptr_mask, struct mm_struct *mm)
+			uint32_t wptr_mask, struct mm_struct *mm, uint32_t inst)
 {
 	struct v10_compute_mqd *m;
 	uint32_t *mqd_hqd;
@@ -286,7 +287,7 @@ static int kgd_hqd_load(struct amdgpu_device *adev, void *mqd,
 
 static int kgd_hiq_mqd_load(struct amdgpu_device *adev, void *mqd,
 			    uint32_t pipe_id, uint32_t queue_id,
-			    uint32_t doorbell_off)
+			    uint32_t doorbell_off, uint32_t inst)
 {
 	struct amdgpu_ring *kiq_ring = &adev->gfx.kiq[0].ring;
 	struct v10_compute_mqd *m;
@@ -338,7 +339,7 @@ out_unlock:
 
 static int kgd_hqd_dump(struct amdgpu_device *adev,
 			uint32_t pipe_id, uint32_t queue_id,
-			uint32_t (**dump)[2], uint32_t *n_regs)
+			uint32_t (**dump)[2], uint32_t *n_regs, uint32_t inst)
 {
 	uint32_t i = 0, reg;
 #define HQD_N_REGS 56
@@ -469,7 +470,7 @@ static int kgd_hqd_sdma_dump(struct amdgpu_device *adev,
 
 static bool kgd_hqd_is_occupied(struct amdgpu_device *adev,
 				uint64_t queue_address, uint32_t pipe_id,
-				uint32_t queue_id)
+				uint32_t queue_id, uint32_t inst)
 {
 	uint32_t act;
 	bool retval = false;
@@ -510,7 +511,7 @@ static bool kgd_hqd_sdma_is_occupied(struct amdgpu_device *adev, void *mqd)
 static int kgd_hqd_destroy(struct amdgpu_device *adev, void *mqd,
 				enum kfd_preempt_type reset_type,
 				unsigned int utimeout, uint32_t pipe_id,
-				uint32_t queue_id)
+				uint32_t queue_id, uint32_t inst)
 {
 	enum hqd_dequeue_request_type type;
 	unsigned long end_jiffies;
@@ -673,7 +674,7 @@ static bool get_atc_vmid_pasid_mapping_info(struct amdgpu_device *adev,
 
 static int kgd_wave_control_execute(struct amdgpu_device *adev,
 					uint32_t gfx_index_val,
-					uint32_t sq_cmd)
+					uint32_t sq_cmd, uint32_t inst)
 {
 	uint32_t data = 0;
 
@@ -709,7 +710,8 @@ static void set_vm_context_page_table_base(struct amdgpu_device *adev,
 }
 
 static void program_trap_handler_settings(struct amdgpu_device *adev,
-		uint32_t vmid, uint64_t tba_addr, uint64_t tma_addr)
+		uint32_t vmid, uint64_t tba_addr, uint64_t tma_addr,
+		uint32_t inst)
 {
 	lock_srbm(adev, 0, 0, 0, vmid);