SWDEV-468424 - hipgraph capture memset node

Capture AQL packets during GraphInstantiation and enqueue AQL packets during graph launch.

Added support to capture single graph memset node.
Capture support for memset node is currently disabled.
Memset capture will be enabled when capture for multiple packets are supported..

Change-Id: I14dfbc41731025cc3a548a730558915def3fa384

rocclr/device/rocm/rocvirtual.cpp

@@ -1497,6 +1497,9 @@ void VirtualGPU::profilingBegin(amd::Command& command, bool sdmaProfiling) {
       }
     }
   }
+  if (command.getCapturingState()) {
+    currCmd_ = &command;
+  }
 }
 
 // ================================================================================================
@@ -1514,6 +1517,7 @@ void VirtualGPU::profilingEnd(amd::Command& command) {
   if (AMD_DIRECT_DISPATCH) {
     assert(retainExternalSignals_ || Barriers().IsExternalSignalListEmpty());
   }
+  currCmd_ = nullptr;
 }
 
 // ================================================================================================
@@ -3018,7 +3022,7 @@ bool VirtualGPU::submitKernelInternal(const amd::NDRangeContainer& sizes,
 
   amd::Memory* const* memories =
       reinterpret_cast<amd::Memory* const*>(parameters + kernelParams.memoryObjOffset());
-  bool isGraphCapture = vcmd != nullptr && vcmd->getCapturingState();
+  bool isGraphCapture = currCmd_ != nullptr && currCmd_->getCapturingState();
   for (int j = 0; j < iteration; j++) {
     // Reset global size for dimension dim if split is needed
     if (dim != -1) {
@@ -3238,7 +3242,8 @@ bool VirtualGPU::submitKernelInternal(const amd::NDRangeContainer& sizes,
     if (!kernel.parameters().deviceKernelArgs() || gpuKernel.isInternalKernel()) {
       // Allocate buffer to hold kernel arguments
       if (isGraphCapture) {
-        argBuffer = vcmd->getKernArgOffset();
+        argBuffer = currCmd_->getKernArgOffset();
+        currCmd_->SetKernelName(gpuKernel.name());
       } else {
 
         argBuffer = reinterpret_cast<address>(
@@ -3324,18 +3329,18 @@ bool VirtualGPU::submitKernelInternal(const amd::NDRangeContainer& sizes,
       aql_packet->setup = sizes.dimensions() << HSA_KERNEL_DISPATCH_PACKET_SETUP_DIMENSIONS;
     }
 
-    if (vcmd == nullptr) {
+    if (isGraphCapture) {
       // Dispatch the packet
       if (!dispatchAqlPacket(&dispatchPacket, aqlHeaderWithOrder,
-                             (sizes.dimensions() << HSA_KERNEL_DISPATCH_PACKET_SETUP_DIMENSIONS),
-                             GPU_FLUSH_ON_EXECUTION)) {
+                        (sizes.dimensions() << HSA_KERNEL_DISPATCH_PACKET_SETUP_DIMENSIONS),
+                        GPU_FLUSH_ON_EXECUTION, currCmd_->getCapturingState(),
+                        currCmd_->getAqlPacket())) {
         return false;
       }
     } else {
       if (!dispatchAqlPacket(&dispatchPacket, aqlHeaderWithOrder,
                              (sizes.dimensions() << HSA_KERNEL_DISPATCH_PACKET_SETUP_DIMENSIONS),
-                             GPU_FLUSH_ON_EXECUTION, vcmd->getCapturingState(),
-                             vcmd->getAqlPacket())) {
+                             GPU_FLUSH_ON_EXECUTION)) {
         return false;
       }
     }