gpu-compute,dev-hsa: Update CP and HSAPP for full-system

Make the necessary changes to connect Vega pagetable walkers for
full-system mode. Previously the CP and HSA packet processor could only
read AQL packets from system/host memory using proxy port. This allows
for AQL to be read from device memory which is used for non-blit
kernels.

Change-Id: If28eb8be68173da03e15084765e77e92eda178e9
Reviewed-on: https://gem5-review.googlesource.com/c/public/gem5/+/53077
Reviewed-by: Matt Sinclair <mattdsinclair@gmail.com>
Maintainer: Matt Sinclair <mattdsinclair@gmail.com>
Tested-by: kokoro <noreply+kokoro@google.com>

src/gpu-compute/gpu_command_processor.cc

@@ -33,9 +33,11 @@
 
 #include <cassert>
 
+#include "arch/amdgpu/vega/pagetable_walker.hh"
 #include "base/chunk_generator.hh"
 #include "debug/GPUCommandProc.hh"
 #include "debug/GPUKernelInfo.hh"
+#include "dev/amdgpu/amdgpu_device.hh"
 #include "gpu-compute/dispatcher.hh"
 #include "mem/se_translating_port_proxy.hh"
 #include "mem/translating_port_proxy.hh"
@@ -50,7 +52,7 @@ namespace gem5
 
 GPUCommandProcessor::GPUCommandProcessor(const Params &p)
     : DmaVirtDevice(p), dispatcher(*p.dispatcher), _driver(nullptr),
-      hsaPP(p.hsapp)
+      walker(p.walker), hsaPP(p.hsapp)
 {
     assert(hsaPP);
     hsaPP->setDevice(this);
@@ -356,6 +358,13 @@ GPUCommandProcessor::getAddrRanges() const
     return ranges;
 }
 
+void
+GPUCommandProcessor::setGPUDevice(AMDGPUDevice *gpu_device)
+{
+    gpuDevice = gpu_device;
+    walker->setDevRequestor(gpuDevice->vramRequestorId());
+}
+
 void
 GPUCommandProcessor::setShader(Shader *shader)
 {