drm/amdgpu: Use ttm_pages_limit to override vram reporting

On GFXIP9.4.3 APU, allow the memory reporting as per the ttm pages
limit in NPS1 mode.
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Rajneesh Bhardwaj <rajneesh.bhardwaj@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

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

@@ -28,6 +28,7 @@
 #include "amdgpu.h"
 #include "amdgpu_gfx.h"
 #include "amdgpu_dma_buf.h"
+#include <drm/ttm/ttm_tt.h>
 #include <linux/module.h>
 #include <linux/dma-buf.h>
 #include "amdgpu_xgmi.h"
@@ -783,11 +784,23 @@ void amdgpu_amdkfd_unlock_kfd(struct amdgpu_device *adev)
 
 u64 amdgpu_amdkfd_xcp_memory_size(struct amdgpu_device *adev, int xcp_id)
 {
-	u64 tmp;
 	s8 mem_id = KFD_XCP_MEM_ID(adev, xcp_id);
+	u64 tmp;
 
 	if (adev->gmc.num_mem_partitions && xcp_id >= 0 && mem_id >= 0) {
+		if (adev->gmc.is_app_apu && adev->gmc.num_mem_partitions == 1) {
+			/* In NPS1 mode, we should restrict the vram reporting
+			 * tied to the ttm_pages_limit which is 1/2 of the system
+			 * memory. For other partition modes, the HBM is uniformly
+			 * divided already per numa node reported. If user wants to
+			 * go beyond the default ttm limit and maximize the ROCm
+			 * allocations, they can go up to max ttm and sysmem limits.
+			 */
+
+			tmp = (ttm_tt_pages_limit() << PAGE_SHIFT) / num_online_nodes();
+		} else {
 			tmp = adev->gmc.mem_partitions[mem_id].size;
+		}
 		do_div(tmp, adev->xcp_mgr->num_xcp_per_mem_partition);
 		return ALIGN_DOWN(tmp, PAGE_SIZE);
 	} else {

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

@@ -1816,15 +1816,14 @@ static void
 gmc_v9_0_init_acpi_mem_ranges(struct amdgpu_device *adev,
 			      struct amdgpu_mem_partition_info *mem_ranges)
 {
-	int num_ranges = 0, ret, mem_groups;
 	struct amdgpu_numa_info numa_info;
 	int node_ids[MAX_MEM_RANGES];
+	int num_ranges = 0, ret;
 	int num_xcc, xcc_id;
 	uint32_t xcc_mask;
 
 	num_xcc = NUM_XCC(adev->gfx.xcc_mask);
 	xcc_mask = (1U << num_xcc) - 1;
-	mem_groups = hweight32(adev->aid_mask);
 
 	for_each_inst(xcc_id, xcc_mask)	{
 		ret = amdgpu_acpi_get_mem_info(adev, xcc_id, &numa_info);
@@ -1849,12 +1848,6 @@ gmc_v9_0_init_acpi_mem_ranges(struct amdgpu_device *adev,
 	}
 
 	adev->gmc.num_mem_partitions = num_ranges;
-
-	/* If there is only partition, don't use entire size */
-	if (adev->gmc.num_mem_partitions == 1) {
-		mem_ranges[0].size = mem_ranges[0].size * (mem_groups - 1);
-		do_div(mem_ranges[0].size, mem_groups);
-	}
 }
 
 static void