drm/amdgpu: add bitmask to iterate vmhubs

As the layout of VMHUB definition has been changed to cover multiple
XCD/AID case, the original num_vmhubs is not appropriate to do vmhub
iteration any more.

Drop num_vmhubs and introduce vmhubs_mask instead.

v2: switch to the new VMHUB layout
v3: use DECLARE_BITMAP to define vmhubs_mask
Signed-off-by: Le Ma <le.ma@amd.com>
Reviewed-by: Christian König <christian.koenig@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.h

@@ -829,7 +829,7 @@ struct amdgpu_device {
 	dma_addr_t			dummy_page_addr;
 	struct amdgpu_vm_manager	vm_manager;
 	struct amdgpu_vmhub             vmhub[AMDGPU_MAX_VMHUBS];
-	unsigned			num_vmhubs;
+	DECLARE_BITMAP(vmhubs_mask, AMDGPU_MAX_VMHUBS);
 
 	/* memory management */
 	struct amdgpu_mman		mman;

drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c

@@ -733,7 +733,7 @@ int amdgpu_amdkfd_flush_gpu_tlb_vmid(struct amdgpu_device *adev,
 	if (adev->family == AMDGPU_FAMILY_AI) {
 		int i;
 
-		for (i = 0; i < adev->num_vmhubs; i++)
+		for_each_set_bit(i, adev->vmhubs_mask, AMDGPU_MAX_VMHUBS)
 			amdgpu_gmc_flush_gpu_tlb(adev, vmid, i, 0);
 	} else {
 		amdgpu_gmc_flush_gpu_tlb(adev, vmid, AMDGPU_GFXHUB(0), 0);

drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c

@@ -182,7 +182,7 @@ void amdgpu_gart_unbind(struct amdgpu_device *adev, uint64_t offset,
 	}
 	mb();
 	amdgpu_device_flush_hdp(adev, NULL);
-	for (i = 0; i < adev->num_vmhubs; i++)
+	for_each_set_bit(i, adev->vmhubs_mask, AMDGPU_MAX_VMHUBS)
 		amdgpu_gmc_flush_gpu_tlb(adev, 0, i, 0);
 
 	drm_dev_exit(idx);
@@ -264,7 +264,7 @@ void amdgpu_gart_invalidate_tlb(struct amdgpu_device *adev)
 
 	mb();
 	amdgpu_device_flush_hdp(adev, NULL);
-	for (i = 0; i < adev->num_vmhubs; i++)
+	for_each_set_bit(i, adev->vmhubs_mask, AMDGPU_MAX_VMHUBS)
 		amdgpu_gmc_flush_gpu_tlb(adev, 0, i, 0);
 }
 

drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c

@@ -460,7 +460,7 @@ static int gmc_v10_0_flush_gpu_tlb_pasid(struct amdgpu_device *adev,
 				&queried_pasid);
 		if (ret	&& queried_pasid == pasid) {
 			if (all_hub) {
-				for (i = 0; i < adev->num_vmhubs; i++)
+				for_each_set_bit(i, adev->vmhubs_mask, AMDGPU_MAX_VMHUBS)
 					gmc_v10_0_flush_gpu_tlb(adev, vmid,
 							i, flush_type);
 			} else {
@@ -928,7 +928,8 @@ static int gmc_v10_0_sw_init(void *handle)
 	case IP_VERSION(10, 3, 6):
 	case IP_VERSION(10, 3, 3):
 	case IP_VERSION(10, 3, 7):
-		adev->num_vmhubs = 2;
+		set_bit(AMDGPU_GFXHUB(0), adev->vmhubs_mask);
+		set_bit(AMDGPU_MMHUB0(0), adev->vmhubs_mask);
 		/*
 		 * To fulfill 4-level page support,
 		 * vm size is 256TB (48bit), maximum size of Navi10/Navi14/Navi12,

drivers/gpu/drm/amd/amdgpu/gmc_v11_0.c

@@ -364,7 +364,7 @@ static int gmc_v11_0_flush_gpu_tlb_pasid(struct amdgpu_device *adev,
 				&queried_pasid);
 		if (ret	&& queried_pasid == pasid) {
 			if (all_hub) {
-				for (i = 0; i < adev->num_vmhubs; i++)
+				for_each_set_bit(i, adev->vmhubs_mask, AMDGPU_MAX_VMHUBS)
 					gmc_v11_0_flush_gpu_tlb(adev, vmid,
 							i, flush_type);
 			} else {
@@ -779,7 +779,8 @@ static int gmc_v11_0_sw_init(void *handle)
 	case IP_VERSION(11, 0, 2):
 	case IP_VERSION(11, 0, 3):
 	case IP_VERSION(11, 0, 4):
-		adev->num_vmhubs = 2;
+		set_bit(AMDGPU_GFXHUB(0), adev->vmhubs_mask);
+		set_bit(AMDGPU_MMHUB0(0), adev->vmhubs_mask);
 		/*
 		 * To fulfill 4-level page support,
 		 * vm size is 256TB (48bit), maximum size,

drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c

@@ -808,7 +808,7 @@ static int gmc_v6_0_sw_init(void *handle)
 	int r;
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
-	adev->num_vmhubs = 1;
+	set_bit(AMDGPU_GFXHUB(0), adev->vmhubs_mask);
 
 	if (adev->flags & AMD_IS_APU) {
 		adev->gmc.vram_type = AMDGPU_VRAM_TYPE_UNKNOWN;

drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c

@@ -977,7 +977,7 @@ static int gmc_v7_0_sw_init(void *handle)
 	int r;
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
-	adev->num_vmhubs = 1;
+	set_bit(AMDGPU_GFXHUB(0), adev->vmhubs_mask);
 
 	if (adev->flags & AMD_IS_APU) {
 		adev->gmc.vram_type = AMDGPU_VRAM_TYPE_UNKNOWN;

drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c

@@ -1093,7 +1093,7 @@ static int gmc_v8_0_sw_init(void *handle)
 	int r;
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
-	adev->num_vmhubs = 1;
+	set_bit(AMDGPU_GFXHUB(0), adev->vmhubs_mask);
 
 	if (adev->flags & AMD_IS_APU) {
 		adev->gmc.vram_type = AMDGPU_VRAM_TYPE_UNKNOWN;

drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c

@@ -481,7 +481,7 @@ static int gmc_v9_0_vm_fault_interrupt_state(struct amdgpu_device *adev,
 
 	switch (state) {
 	case AMDGPU_IRQ_STATE_DISABLE:
-		for (j = 0; j < adev->num_vmhubs; j++) {
+		for_each_set_bit(j, adev->vmhubs_mask, AMDGPU_MAX_VMHUBS) {
 			hub = &adev->vmhub[j];
 			for (i = 0; i < 16; i++) {
 				reg = hub->vm_context0_cntl + i;
@@ -509,7 +509,7 @@ static int gmc_v9_0_vm_fault_interrupt_state(struct amdgpu_device *adev,
 		}
 		break;
 	case AMDGPU_IRQ_STATE_ENABLE:
-		for (j = 0; j < adev->num_vmhubs; j++) {
+		for_each_set_bit(j, adev->vmhubs_mask, AMDGPU_MAX_VMHUBS) {
 			hub = &adev->vmhub[j];
 			for (i = 0; i < 16; i++) {
 				reg = hub->vm_context0_cntl + i;
@@ -803,7 +803,7 @@ static void gmc_v9_0_flush_gpu_tlb(struct amdgpu_device *adev, uint32_t vmid,
 	u32 j, inv_req, inv_req2, tmp;
 	struct amdgpu_vmhub *hub;
 
-	BUG_ON(vmhub >= adev->num_vmhubs);
+	BUG_ON(vmhub >= AMDGPU_MAX_VMHUBS);
 
 	hub = &adev->vmhub[vmhub];
 	if (adev->gmc.xgmi.num_physical_nodes &&
@@ -987,7 +987,7 @@ static int gmc_v9_0_flush_gpu_tlb_pasid(struct amdgpu_device *adev,
 				&queried_pasid);
 		if (ret && queried_pasid == pasid) {
 			if (all_hub) {
-				for (i = 0; i < adev->num_vmhubs; i++)
+				for_each_set_bit(i, adev->vmhubs_mask, AMDGPU_MAX_VMHUBS)
 					gmc_v9_0_flush_gpu_tlb(adev, vmid,
 							i, flush_type);
 			} else {
@@ -1684,7 +1684,8 @@ static int gmc_v9_0_sw_init(void *handle)
 	switch (adev->ip_versions[GC_HWIP][0]) {
 	case IP_VERSION(9, 1, 0):
 	case IP_VERSION(9, 2, 2):
-		adev->num_vmhubs = 2;
+		set_bit(AMDGPU_GFXHUB(0), adev->vmhubs_mask);
+		set_bit(AMDGPU_MMHUB0(0), adev->vmhubs_mask);
 
 		if (adev->rev_id == 0x0 || adev->rev_id == 0x1) {
 			amdgpu_vm_adjust_size(adev, 256 * 1024, 9, 3, 48);
@@ -1701,8 +1702,8 @@ static int gmc_v9_0_sw_init(void *handle)
 	case IP_VERSION(9, 3, 0):
 	case IP_VERSION(9, 4, 2):
 	case IP_VERSION(9, 4, 3):
-		adev->num_vmhubs = 2;
-
+		set_bit(AMDGPU_GFXHUB(0), adev->vmhubs_mask);
+		set_bit(AMDGPU_MMHUB0(0), adev->vmhubs_mask);
 
 		/*
 		 * To fulfill 4-level page support,
@@ -1718,7 +1719,9 @@ static int gmc_v9_0_sw_init(void *handle)
 			adev->gmc.translate_further = adev->vm_manager.num_level > 1;
 		break;
 	case IP_VERSION(9, 4, 1):
-		adev->num_vmhubs = 3;
+		set_bit(AMDGPU_GFXHUB(0), adev->vmhubs_mask);
+		set_bit(AMDGPU_MMHUB0(0), adev->vmhubs_mask);
+		set_bit(AMDGPU_MMHUB1(0), adev->vmhubs_mask);
 
 		/* Keep the vm size same with Vega20 */
 		amdgpu_vm_adjust_size(adev, 256 * 1024, 9, 3, 48);
@@ -1944,7 +1947,7 @@ static int gmc_v9_0_hw_init(void *handle)
 			adev->gfxhub.funcs->set_fault_enable_default(adev, value);
 		adev->mmhub.funcs->set_fault_enable_default(adev, value);
 	}
-	for (i = 0; i < adev->num_vmhubs; ++i) {
+	for_each_set_bit(i, adev->vmhubs_mask, AMDGPU_MAX_VMHUBS) {
 		if (adev->in_s0ix && (i == AMDGPU_GFXHUB(0)))
 			continue;
 		gmc_v9_0_flush_gpu_tlb(adev, 0, i, 0);