Merge tag 'amd-drm-next-5.13-2021-03-23' of https://gitlab.freedesktop.org/agd5f/linux into drm-next

amd-drm-next-5.13-2021-03-23:

amdgpu:
- Debugfs cleanup
- Various cleanups and spelling fixes
- Flexible array cleanups
- Initial AMD Freesync HDMI
- Display fixes
- 10bpc dithering improvements
- Display ASSR support
- Clean up and unify powerplay and swsmu interfaces
- Vangogh fixes
- Add SMU gfx busy queues for RV/PCO
- PCIE DPM fixes
- S0ix fixes
- GPU metrics data fixes
- DCN secure display support
- Backlight type override
- Add initial support for Aldebaran
- RAS fixes
- Prime fixes for A+A systems
- Reset fixes
- Initial resource cursor support
- Drop legacy IO BAR requirements
- Various power fixes

amdkfd:
- MMU notifier fixes
- APU fixes

radeon:
- Debugfs cleanups
- Flexible array cleanups

UAPI:
- amdgpu: Add a new INFO ioctl interface to query video capabilities
  rather than hardcoding them in userspace.  This allows us to provide
  fine grained asic capabilities (e.g., if a particular part is
  bandwidth limited, we can limit the capabilities).  Proposed userspace:
  https://gitlab.freedesktop.org/leoliu/drm/-/commits/info_video_caps
  https://gitlab.freedesktop.org/leoliu/mesa/-/commits/info_video_caps
- amdkfd: bump the driver version.  There was a problem with reporting
  some RAS features on older versions of the driver. Proposed userspace:
  7cdd63475c

Danvet: A bunch of conflicts all over, but it seems to compile ... I
did put the call to dc_allow_idle_optimizations() on a single line
since it looked a bit too jarring to be left alone.

Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
From: Alex Deucher <alexander.deucher@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210324040147.1990338-1-alexander.deucher@amd.com

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

@@ -31,6 +31,7 @@
 #include "amdgpu_amdkfd.h"
 #include "amdgpu_dma_buf.h"
 #include <uapi/linux/kfd_ioctl.h>
+#include "amdgpu_xgmi.h"
 
 /* BO flag to indicate a KFD userptr BO */
 #define AMDGPU_AMDKFD_USERPTR_BO (1ULL << 63)
@@ -96,7 +97,7 @@ void amdgpu_amdkfd_gpuvm_init_mem_limits(void)
 	uint64_t mem;
 
 	si_meminfo(&si);
-	mem = si.totalram - si.totalhigh;
+	mem = si.freeram - si.freehigh;
 	mem *= si.mem_unit;
 
 	spin_lock_init(&kfd_mem_limit.mem_limit_lock);
@@ -412,7 +413,10 @@ static uint64_t get_pte_flags(struct amdgpu_device *adev, struct kgd_mem *mem)
 {
 	struct amdgpu_device *bo_adev = amdgpu_ttm_adev(mem->bo->tbo.bdev);
 	bool coherent = mem->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_COHERENT;
+	bool uncached = mem->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_UNCACHED;
 	uint32_t mapping_flags;
+	uint64_t pte_flags;
+	bool snoop = false;
 
 	mapping_flags = AMDGPU_VM_PAGE_READABLE;
 	if (mem->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_WRITABLE)
@@ -433,12 +437,41 @@ static uint64_t get_pte_flags(struct amdgpu_device *adev, struct kgd_mem *mem)
 				AMDGPU_VM_MTYPE_UC : AMDGPU_VM_MTYPE_NC;
 		}
 		break;
+	case CHIP_ALDEBARAN:
+		if (coherent && uncached) {
+			if (adev->gmc.xgmi.connected_to_cpu ||
+				!(mem->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_VRAM))
+				snoop = true;
+			mapping_flags |= AMDGPU_VM_MTYPE_UC;
+		} else if (mem->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_VRAM) {
+			if (bo_adev == adev) {
+				mapping_flags |= AMDGPU_VM_MTYPE_RW;
+				if (adev->gmc.xgmi.connected_to_cpu)
+					snoop = true;
+			} else {
+				mapping_flags |= AMDGPU_VM_MTYPE_NC;
+				if (amdgpu_xgmi_same_hive(adev, bo_adev))
+					snoop = true;
+			}
+		} else {
+			snoop = true;
+			if (adev->gmc.xgmi.connected_to_cpu)
+				/* system memory uses NC on A+A */
+				mapping_flags |= AMDGPU_VM_MTYPE_NC;
+			else
+				mapping_flags |= coherent ?
+					AMDGPU_VM_MTYPE_UC : AMDGPU_VM_MTYPE_NC;
+		}
+		break;
 	default:
 		mapping_flags |= coherent ?
 			AMDGPU_VM_MTYPE_UC : AMDGPU_VM_MTYPE_NC;
 	}
 
-	return amdgpu_gem_va_map_flags(adev, mapping_flags);
+	pte_flags = amdgpu_gem_va_map_flags(adev, mapping_flags);
+	pte_flags |= snoop ? AMDGPU_PTE_SNOOPED : 0;
+
+	return pte_flags;
 }
 
 /* add_bo_to_vm - Add a BO to a VM