arch-gcn3,gpu-compute: Move TLB to common folder in amdgpu

This TLB is more of an "APU" TLB than anything GCN3 specific. It can be
used with either GCN3 or Vega. With this change, VEGA_X86 builds and one
can run binaries with Vega ISA code using the same steps as GCN3 but
building the Vega ISA instead.

Change-Id: I0c92bcd0379a18628dc05cb5af070bdc7e692c7c
Reviewed-on: https://gem5-review.googlesource.com/c/public/gem5/+/53803
Maintainer: Bobby Bruce <bbruce@ucdavis.edu>
Reviewed-by: Matt Sinclair <mattdsinclair@gmail.com>
Maintainer: Matt Sinclair <mattdsinclair@gmail.com>
Tested-by: kokoro <noreply+kokoro@google.com>

src/gpu-compute/compute_unit.cc

@@ -35,6 +35,7 @@
 
 #include <limits>
 
+#include "arch/amdgpu/common/tlb.hh"
 #include "base/output.hh"
 #include "debug/GPUDisp.hh"
 #include "debug/GPUExec.hh"

src/gpu-compute/fetch_unit.cc

@@ -33,6 +33,7 @@
 
 #include "gpu-compute/fetch_unit.hh"
 
+#include "arch/amdgpu/common/tlb.hh"
 #include "base/bitfield.hh"
 #include "debug/GPUFetch.hh"
 #include "debug/GPUPort.hh"

src/gpu-compute/shader.cc

@@ -35,6 +35,7 @@
 
 #include <limits>
 
+#include "arch/amdgpu/common/tlb.hh"
 #include "base/chunk_generator.hh"
 #include "debug/GPUAgentDisp.hh"
 #include "debug/GPUDisp.hh"