hsakmt: Enable graphics handle registration with a virtual address

Currently registering graphics memory without specifying a target
node will return a memory handle that's not a virtual address.

As a result, ROCr is forced to register with a target node for
IPC usage.

Mapping memory without specifying a target node afterwards will
result in mapping to the target node that was imported because the
previous import call flags this node targeting action to future mapping.

For ROCr IPC usage, ROCr wants to map to all GPU nodes if the target node
is not specified.

Allow the caller to register graphics handles that returns a virtual
address without having to specify the target node so that the caller
can make a subsequent map call to all GPUs.

Change-Id: I5a935092b885cc3568e4f3a5dd951c7ec6c84fca

libhsakmt/include/hsakmt/hsakmt.h

@@ -512,6 +512,21 @@ hsaKmtRegisterGraphicsHandleToNodes(
     HSAuint32*      NodeArray                      //IN
     );
 
+/**
+  Similar to hsaKmtRegisterGraphicsHandleToNodes but provides registration
+  options via RegisterFlags.
+*/
+
+HSAKMT_STATUS
+HSAKMTAPI
+hsaKmtRegisterGraphicsHandleToNodesExt(
+    HSAuint64       GraphicsResourceHandle,        //IN
+    HsaGraphicsResourceInfo *GraphicsResourceInfo, //OUT
+    HSAuint64       NumberOfNodes,                 //IN
+    HSAuint32*      NodeArray,                     //IN
+    HSA_REGISTER_MEM_FLAGS RegisterFlags           //IN
+    );
+
 /**
  * Export a dmabuf handle and offset for a given memory address
  *

libhsakmt/include/hsakmt/hsakmttypes.h

@@ -1488,6 +1488,15 @@ typedef struct _HsaPcSamplingInfo
 }
 HsaPcSamplingInfo;
 
+typedef union
+{
+    HSAuint32 Value;
+    struct
+    {
+        unsigned int requiresVAddr : 1;  // Requires virtual address
+    } ui32;
+} HSA_REGISTER_MEM_FLAGS;
+
 #pragma pack(pop, hsakmttypes_h)
 
 

libhsakmt/src/fmm.c

@@ -3586,7 +3586,8 @@ HSAKMT_STATUS hsakmt_fmm_register_memory(void *address, uint64_t size_in_bytes,
 HSAKMT_STATUS hsakmt_fmm_register_graphics_handle(HSAuint64 GraphicsResourceHandle,
 					   HsaGraphicsResourceInfo *GraphicsResourceInfo,
 					   uint32_t *gpu_id_array,
-					   uint32_t gpu_id_array_size)
+					   uint32_t gpu_id_array_size,
+					   HSA_REGISTER_MEM_FLAGS RegisterFlags)
 {
 	struct kfd_ioctl_get_dmabuf_info_args infoArgs = {0};
 	struct kfd_ioctl_import_dmabuf_args importArgs = {0};
@@ -3630,7 +3631,7 @@ HSAKMT_STATUS hsakmt_fmm_register_graphics_handle(HSAuint64 GraphicsResourceHand
 		goto error_free_metadata;
 
 	/* import DMA buffer without VA assigned */
-	if (!gpu_id_array && gpu_id_array_size == 0) {
+	if (!gpu_id_array && gpu_id_array_size == 0 && !RegisterFlags.ui32.requiresVAddr) {
 		aperture = &mem_handle_aperture;
 	} else if (hsakmt_topology_is_svm_needed(gpu_mem[gpu_mem_id].EngineId)) {
 		aperture = svm.dgpu_aperture;

libhsakmt/src/fmm.h

@@ -80,7 +80,8 @@ HSAKMT_STATUS hsakmt_fmm_register_memory(void *address, uint64_t size_in_bytes,
 HSAKMT_STATUS hsakmt_fmm_register_graphics_handle(HSAuint64 GraphicsResourceHandle,
 					   HsaGraphicsResourceInfo *GraphicsResourceInfo,
 					   uint32_t *gpu_id_array,
-					   uint32_t gpu_id_array_size);
+					   uint32_t gpu_id_array_size,
+					   HSA_REGISTER_MEM_FLAGS RegisterFlags);
 HSAKMT_STATUS hsakmt_fmm_deregister_memory(void *address);
 HSAKMT_STATUS hsakmt_fmm_export_dma_buf_fd(void *MemoryAddress,
 				    HSAuint64 MemorySizeInBytes,

libhsakmt/src/memory.c

@@ -336,6 +336,23 @@ HSAKMT_STATUS HSAKMTAPI hsaKmtRegisterGraphicsHandleToNodes(HSAuint64 GraphicsRe
 							    HsaGraphicsResourceInfo *GraphicsResourceInfo,
 							    HSAuint64 NumberOfNodes,
 							    HSAuint32 *NodeArray)
+{
+       HSA_REGISTER_MEM_FLAGS regFlags;
+       regFlags.Value = 0;
+        
+       return hsaKmtRegisterGraphicsHandleToNodesExt(GraphicsResourceHandle,
+						     GraphicsResourceInfo,
+						     NumberOfNodes,
+						     NodeArray,
+						     regFlags);
+
+}
+
+HSAKMT_STATUS HSAKMTAPI hsaKmtRegisterGraphicsHandleToNodesExt(HSAuint64 GraphicsResourceHandle,
+							       HsaGraphicsResourceInfo *GraphicsResourceInfo,
+							       HSAuint64 NumberOfNodes,
+							       HSAuint32 *NodeArray,
+							       HSA_REGISTER_MEM_FLAGS RegisterFlags)
 {
 	CHECK_KFD_OPEN();
 	uint32_t *gpu_id_array = NULL;
@@ -351,7 +368,7 @@ HSAKMT_STATUS HSAKMTAPI hsaKmtRegisterGraphicsHandleToNodes(HSAuint64 GraphicsRe
 	if (ret == HSAKMT_STATUS_SUCCESS) {
 		ret = hsakmt_fmm_register_graphics_handle(
 			GraphicsResourceHandle, GraphicsResourceInfo,
-			gpu_id_array, NumberOfNodes * sizeof(uint32_t));
+			gpu_id_array, NumberOfNodes * sizeof(uint32_t), RegisterFlags);
 		if (ret != HSAKMT_STATUS_SUCCESS)
 			free(gpu_id_array);
 	}