Merge tag 'amd-drm-next-6.5-2023-06-09' of https://gitlab.freedesktop.org/agd5f/linux into drm-next

amd-drm-next-6.5-2023-06-02:

amdgpu:
- SR-IOV fixes
- Warning fixes
- Misc code cleanups and spelling fixes
- DCN 3.2 updates
- Improved DC FAMS support for better power management
- Improved DC SubVP support for better power management
- DCN 3.1.x fixes
- Max IB size query
- DC GPU reset fixes
- RAS updates
- DCN 3.0.x fixes
- S/G display fixes
- CP shadow buffer support
- Implement connector force callback
- Z8 power improvements
- PSP 13.0.10 vbflash support
- Mode2 reset fixes
- Store MQDs in VRAM to improve queue switch latency
- VCN 3.x fixes
- JPEG 3.x fixes
- Enable DC_FP on LoongArch
- GFXOFF fixes
- GC 9.4.3 partition support
- SDMA 4.4.2 partition support
- VCN/JPEG 4.0.3 partition support
- VCN 4.0.3 updates
- NBIO 7.9 updates
- GC 9.4.3 updates
- Take NUMA into account when allocating memory
- Handle NUMA for partitions
- SMU 13.0.6 updates
- GC 9.4.3 RAS updates
- Stop including unused swiotlb.h
- SMU 13.0.7 fixes
- Fix clock output ordering on some APUs
- Clean up DC FPGA code
- GFX9 preemption fixes
- Misc irq fixes
- S0ix fixes
- Add new DRM_AMDGPU_WERROR config parameter to help with CI
- PCIe fix for RDNA2
- kdoc fixes
- Documentation updates

amdkfd:
- Query TTM mem limit rather than hardcoding it
- GC 9.4.3 partition support
- Handle NUMA for partitions

radeon:
- Fix possible double free
- Stop including unused swiotlb.h
- Fix possible division by zero

ttm:
- Add query for TTM mem limit
- Add NUMA awareness to pools
- Export ttm_pool_fini()

UAPI:
- Add new ctx query flag to better handle GPU resets
  Mesa MR: https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22290
- Add new interface to query and set shadow buffer for RDNA3
  Mesa MR: https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21986
- Add new INFO query for max IB size
  Proposed userspace: https://gitlab.freedesktop.org/bnieuwenhuizen/mesa/-/commits/ib-rejection-v3

amd-drm-next-6.5-2023-06-09:

amdgpu:
- S0ix fixes
- Initial SMU13 Overdrive support
- kdoc fixes
- Misc clode cleanups
- Flexible array fixes
- Display OTG fixes
- SMU 13.0.6 updates
- Revert some broken clock counter updates
- Misc display fixes
- GFX9 preemption fixes
- Add support for newer EEPROM bad page table format
- Add missing radeon secondary id
- Add support for new colorspace KMS API
- CSA fix
- Stable pstate fixes for APUs
- make vbl interface admin only
- Handle PCI accelerator class

amdkfd:
- Add debugger support for gdb

radeon:
- Fix possible UAF

drm:
- Add Colorspace functionality

UAPI:
- Add debugger interface for enabling gdb
  Proposed userspace: https://github.com/ROCm-Developer-Tools/ROCdbgapi/tree/wip-dbgapi
- Add KMS colorspace API
  Discussion: https://lists.freedesktop.org/archives/dri-devel/2023-June/408128.html

From: Alex Deucher <alexander.deucher@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230609174817.7764-1-alexander.deucher@amd.com

1 files changed, 58 insertions, 30 deletions

diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_crat.c b/drivers/gpu/drm/amd/amdkfd/kfd_crat.c
index 475e47027354..3dcd8f8bc98e 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_crat.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_crat.c
@@ -30,6 +30,9 @@
 #include "amdgpu.h"
 #include "amdgpu_amdkfd.h"
 
+/* Fixme: Fake 32GB for 1PNPS1 mode bringup */
+#define DUMMY_VRAM_SIZE 31138512896
+
 /* GPU Processor ID base for dGPUs for which VCRAT needs to be created.
  * GPU processor ID are expressed with Bit[31]=1.
  * The base is set to 0x8000_0000 + 0x1000 to avoid collision with GPU IDs
@@ -1053,6 +1056,8 @@ static int kfd_parse_subtype_mem(struct crat_subtype_memory *mem,
 
 			props->heap_type = heap_type;
 			props->flags = flags;
+			if (size_in_bytes == 0)
+				size_in_bytes = DUMMY_VRAM_SIZE; /* Fixme: TBD */
 			props->size_in_bytes = size_in_bytes;
 			props->width = width;
 
@@ -1166,7 +1171,7 @@ static int kfd_parse_subtype_iolink(struct crat_subtype_iolink *iolink,
 			if (props->iolink_type == CRAT_IOLINK_TYPE_PCIEXPRESS)
 				props->weight = 20;
 			else if (props->iolink_type == CRAT_IOLINK_TYPE_XGMI)
-				props->weight = 15 * iolink->num_hops_xgmi;
+				props->weight = iolink->weight_xgmi;
 			else
 				props->weight = node_distance(id_from, id_to);
 
@@ -1405,7 +1410,7 @@ static int kfd_fill_gpu_cache_info_from_gfx_config(struct kfd_dev *kdev,
 	return i;
 }
 
-int kfd_get_gpu_cache_info(struct kfd_dev *kdev, struct kfd_gpu_cache_info **pcache_info)
+int kfd_get_gpu_cache_info(struct kfd_node *kdev, struct kfd_gpu_cache_info **pcache_info)
 {
 	int num_of_cache_types = 0;
 
@@ -1524,7 +1529,7 @@ int kfd_get_gpu_cache_info(struct kfd_dev *kdev, struct kfd_gpu_cache_info **pca
 		case IP_VERSION(11, 0, 3):
 		case IP_VERSION(11, 0, 4):
 			num_of_cache_types =
-				kfd_fill_gpu_cache_info_from_gfx_config(kdev, *pcache_info);
+				kfd_fill_gpu_cache_info_from_gfx_config(kdev->kfd, *pcache_info);
 			break;
 		default:
 			*pcache_info = dummy_cache_info;
@@ -1858,7 +1863,7 @@ static int kfd_create_vcrat_image_cpu(void *pcrat_image, size_t *size)
 }
 
 static int kfd_fill_gpu_memory_affinity(int *avail_size,
-		struct kfd_dev *kdev, uint8_t type, uint64_t size,
+		struct kfd_node *kdev, uint8_t type, uint64_t size,
 		struct crat_subtype_memory *sub_type_hdr,
 		uint32_t proximity_domain,
 		const struct kfd_local_mem_info *local_mem_info)
@@ -1887,7 +1892,7 @@ static int kfd_fill_gpu_memory_affinity(int *avail_size,
 }
 
 #ifdef CONFIG_ACPI_NUMA
-static void kfd_find_numa_node_in_srat(struct kfd_dev *kdev)
+static void kfd_find_numa_node_in_srat(struct kfd_node *kdev)
 {
 	struct acpi_table_header *table_header = NULL;
 	struct acpi_subtable_header *sub_header = NULL;
@@ -1972,6 +1977,9 @@ static void kfd_find_numa_node_in_srat(struct kfd_dev *kdev)
 }
 #endif
 
+#define KFD_CRAT_INTRA_SOCKET_WEIGHT	13
+#define KFD_CRAT_XGMI_WEIGHT		15
+
 /* kfd_fill_gpu_direct_io_link - Fill in direct io link from GPU
  * to its NUMA node
  *	@avail_size: Available size in the memory
@@ -1982,7 +1990,7 @@ static void kfd_find_numa_node_in_srat(struct kfd_dev *kdev)
  *	Return 0 if successful else return -ve value
  */
 static int kfd_fill_gpu_direct_io_link_to_cpu(int *avail_size,
-			struct kfd_dev *kdev,
+			struct kfd_node *kdev,
 			struct crat_subtype_iolink *sub_type_hdr,
 			uint32_t proximity_domain)
 {
@@ -2002,7 +2010,16 @@ static int kfd_fill_gpu_direct_io_link_to_cpu(int *avail_size,
 	/* Fill in IOLINK subtype.
 	 * TODO: Fill-in other fields of iolink subtype
 	 */
-	if (kdev->adev->gmc.xgmi.connected_to_cpu) {
+	if (kdev->adev->gmc.xgmi.connected_to_cpu ||
+	    (KFD_GC_VERSION(kdev) == IP_VERSION(9, 4, 3) &&
+	     kdev->adev->smuio.funcs->get_pkg_type(kdev->adev) ==
+	     AMDGPU_PKG_TYPE_APU)) {
+		bool ext_cpu = KFD_GC_VERSION(kdev) != IP_VERSION(9, 4, 3);
+		int mem_bw = 819200, weight = ext_cpu ? KFD_CRAT_XGMI_WEIGHT :
+							KFD_CRAT_INTRA_SOCKET_WEIGHT;
+		uint32_t bandwidth = ext_cpu ? amdgpu_amdkfd_get_xgmi_bandwidth_mbytes(
+							kdev->adev, NULL, true) : mem_bw;
+
 		/*
 		 * with host gpu xgmi link, host can access gpu memory whether
 		 * or not pcie bar type is large, so always create bidirectional
@@ -2010,14 +2027,9 @@ static int kfd_fill_gpu_direct_io_link_to_cpu(int *avail_size,
 		 */
 		sub_type_hdr->flags |= CRAT_IOLINK_FLAGS_BI_DIRECTIONAL;
 		sub_type_hdr->io_interface_type = CRAT_IOLINK_TYPE_XGMI;
-		sub_type_hdr->num_hops_xgmi = 1;
-		if (KFD_GC_VERSION(kdev) == IP_VERSION(9, 4, 2)) {
-			sub_type_hdr->minimum_bandwidth_mbs =
-					amdgpu_amdkfd_get_xgmi_bandwidth_mbytes(
-							kdev->adev, NULL, true);
-			sub_type_hdr->maximum_bandwidth_mbs =
-					sub_type_hdr->minimum_bandwidth_mbs;
-		}
+		sub_type_hdr->weight_xgmi = weight;
+		sub_type_hdr->minimum_bandwidth_mbs = bandwidth;
+		sub_type_hdr->maximum_bandwidth_mbs = bandwidth;
 	} else {
 		sub_type_hdr->io_interface_type = CRAT_IOLINK_TYPE_PCIEXPRESS;
 		sub_type_hdr->minimum_bandwidth_mbs =
@@ -2029,7 +2041,8 @@ static int kfd_fill_gpu_direct_io_link_to_cpu(int *avail_size,
 	sub_type_hdr->proximity_domain_from = proximity_domain;
 
 #ifdef CONFIG_ACPI_NUMA
-	if (kdev->adev->pdev->dev.numa_node == NUMA_NO_NODE)
+	if (kdev->adev->pdev->dev.numa_node == NUMA_NO_NODE &&
+	    num_possible_nodes() > 1)
 		kfd_find_numa_node_in_srat(kdev);
 #endif
 #ifdef CONFIG_NUMA
@@ -2044,12 +2057,14 @@ static int kfd_fill_gpu_direct_io_link_to_cpu(int *avail_size,
 }
 
 static int kfd_fill_gpu_xgmi_link_to_gpu(int *avail_size,
-			struct kfd_dev *kdev,
-			struct kfd_dev *peer_kdev,
+			struct kfd_node *kdev,
+			struct kfd_node *peer_kdev,
 			struct crat_subtype_iolink *sub_type_hdr,
 			uint32_t proximity_domain_from,
 			uint32_t proximity_domain_to)
 {
+	bool use_ta_info = kdev->kfd->num_nodes == 1;
+
 	*avail_size -= sizeof(struct crat_subtype_iolink);
 	if (*avail_size < 0)
 		return -ENOMEM;
@@ -2064,12 +2079,25 @@ static int kfd_fill_gpu_xgmi_link_to_gpu(int *avail_size,
 	sub_type_hdr->io_interface_type = CRAT_IOLINK_TYPE_XGMI;
 	sub_type_hdr->proximity_domain_from = proximity_domain_from;
 	sub_type_hdr->proximity_domain_to = proximity_domain_to;
-	sub_type_hdr->num_hops_xgmi =
-		amdgpu_amdkfd_get_xgmi_hops_count(kdev->adev, peer_kdev->adev);
-	sub_type_hdr->maximum_bandwidth_mbs =
-		amdgpu_amdkfd_get_xgmi_bandwidth_mbytes(kdev->adev, peer_kdev->adev, false);
-	sub_type_hdr->minimum_bandwidth_mbs = sub_type_hdr->maximum_bandwidth_mbs ?
-		amdgpu_amdkfd_get_xgmi_bandwidth_mbytes(kdev->adev, NULL, true) : 0;
+
+	if (use_ta_info) {
+		sub_type_hdr->weight_xgmi = KFD_CRAT_XGMI_WEIGHT *
+			amdgpu_amdkfd_get_xgmi_hops_count(kdev->adev, peer_kdev->adev);
+		sub_type_hdr->maximum_bandwidth_mbs =
+			amdgpu_amdkfd_get_xgmi_bandwidth_mbytes(kdev->adev,
+							peer_kdev->adev, false);
+		sub_type_hdr->minimum_bandwidth_mbs = sub_type_hdr->maximum_bandwidth_mbs ?
+			amdgpu_amdkfd_get_xgmi_bandwidth_mbytes(kdev->adev, NULL, true) : 0;
+	} else {
+		bool is_single_hop = kdev->kfd == peer_kdev->kfd;
+		int weight = is_single_hop ? KFD_CRAT_INTRA_SOCKET_WEIGHT :
+			(2 * KFD_CRAT_INTRA_SOCKET_WEIGHT) + KFD_CRAT_XGMI_WEIGHT;
+		int mem_bw = 819200;
+
+		sub_type_hdr->weight_xgmi = weight;
+		sub_type_hdr->maximum_bandwidth_mbs = is_single_hop ? mem_bw : 0;
+		sub_type_hdr->minimum_bandwidth_mbs = is_single_hop ? mem_bw : 0;
+	}
 
 	return 0;
 }
@@ -2081,7 +2109,7 @@ static int kfd_fill_gpu_xgmi_link_to_gpu(int *avail_size,
  *		[OUT] actual size of data filled in crat_image
  */
 static int kfd_create_vcrat_image_gpu(void *pcrat_image,
-				      size_t *size, struct kfd_dev *kdev,
+				      size_t *size, struct kfd_node *kdev,
 				      uint32_t proximity_domain)
 {
 	struct crat_header *crat_table = (struct crat_header *)pcrat_image;
@@ -2153,7 +2181,7 @@ static int kfd_create_vcrat_image_gpu(void *pcrat_image,
 	/* Check if this node supports IOMMU. During parsing this flag will
 	 * translate to HSA_CAP_ATS_PRESENT
 	 */
-	if (!kfd_iommu_check_device(kdev))
+	if (!kfd_iommu_check_device(kdev->kfd))
 		cu->hsa_capability |= CRAT_CU_FLAGS_IOMMU_PRESENT;
 
 	crat_table->length += sub_type_hdr->length;
@@ -2216,12 +2244,12 @@ static int kfd_create_vcrat_image_gpu(void *pcrat_image,
 	 * (from other GPU to this GPU) will be added
 	 * in kfd_parse_subtype_iolink.
 	 */
-	if (kdev->hive_id) {
+	if (kdev->kfd->hive_id) {
 		for (nid = 0; nid < proximity_domain; ++nid) {
 			peer_dev = kfd_topology_device_by_proximity_domain_no_lock(nid);
 			if (!peer_dev->gpu)
 				continue;
-			if (peer_dev->gpu->hive_id != kdev->hive_id)
+			if (peer_dev->gpu->kfd->hive_id != kdev->kfd->hive_id)
 				continue;
 			sub_type_hdr = (typeof(sub_type_hdr))(
 				(char *)sub_type_hdr +
@@ -2255,12 +2283,12 @@ static int kfd_create_vcrat_image_gpu(void *pcrat_image,
  *		(COMPUTE_UNIT_CPU | COMPUTE_UNIT_GPU) - Create VCRAT for APU
  *			-- this option is not currently implemented.
  *			The assumption is that all AMD APUs will have CRAT
- *	@kdev: Valid kfd_device required if flags contain COMPUTE_UNIT_GPU
+ *	@kdev: Valid kfd_node required if flags contain COMPUTE_UNIT_GPU
  *
  *	Return 0 if successful else return -ve value
  */
 int kfd_create_crat_image_virtual(void **crat_image, size_t *size,
-				  int flags, struct kfd_dev *kdev,
+				  int flags, struct kfd_node *kdev,
 				  uint32_t proximity_domain)
 {
 	void *pcrat_image = NULL;