AMDGPU: add +xnack feature

[oota-llvm.git] / lib / Target / AMDGPU / AMDGPUSubtarget.cpp

diff --git a/lib/Target/AMDGPU/AMDGPUSubtarget.cpp b/lib/Target/AMDGPU/AMDGPUSubtarget.cpp
index c6cb8f9544a55d0b305fb36c378513a66f6e7a34..c6af5b93d257e381aef6048581fda1ef4b990f14 100644 (file)
--- a/lib/Target/AMDGPU/AMDGPUSubtarget.cpp
+++ b/lib/Target/AMDGPU/AMDGPUSubtarget.cpp
@@ -16,6 +16,7 @@
 #include "R600ISelLowering.h"
 #include "R600InstrInfo.h"
 #include "R600MachineScheduler.h"
+#include "SIFrameLowering.h"
 #include "SIISelLowering.h"
 #include "SIInstrInfo.h"
 #include "SIMachineFunctionInfo.h"
@@ -44,6 +45,8 @@ AMDGPUSubtarget::initializeSubtargetDependencies(const Triple &TT,
   // disable it.
 
   SmallString<256> FullFS("+promote-alloca,+fp64-denormals,");
+  if (isAmdHsaOS()) // Turn on FlatForGlobal for HSA.
+    FullFS += "+flat-for-global,";
   FullFS += FS;
 
   if (GPU == "" && TT.getArch() == Triple::amdgcn)
@@ -63,30 +66,41 @@ AMDGPUSubtarget::initializeSubtargetDependencies(const Triple &TT,
 
 AMDGPUSubtarget::AMDGPUSubtarget(const Triple &TT, StringRef GPU, StringRef FS,
                                  TargetMachine &TM)
-    : AMDGPUGenSubtargetInfo(TargetTuple(TT), GPU, FS), DevName(GPU),
-      Is64bit(false), DumpCode(false), R600ALUInst(false),
-      HasVertexCache(false), TexVTXClauseSize(0), Gen(AMDGPUSubtarget::R600),
-      FP64(false), FP64Denormals(false), FP32Denormals(false),
-      FastFMAF32(false), CaymanISA(false), FlatAddressSpace(false),
+    : AMDGPUGenSubtargetInfo(TT, GPU, FS), DevName(GPU), Is64bit(false),
+      DumpCode(false), R600ALUInst(false), HasVertexCache(false),
+      TexVTXClauseSize(0), Gen(AMDGPUSubtarget::R600), FP64(false),
+      FP64Denormals(false), FP32Denormals(false), FastFMAF32(false),
+      CaymanISA(false), FlatAddressSpace(false), FlatForGlobal(false),
       EnableIRStructurizer(true), EnablePromoteAlloca(false), EnableIfCvt(true),
       EnableLoadStoreOpt(false), EnableUnsafeDSOffsetFolding(false),
+      EnableXNACK(false),
       WavefrontSize(0), CFALUBug(false), LocalMemorySize(0),
       EnableVGPRSpilling(false), SGPRInitBug(false), IsGCN(false),
       GCN1Encoding(false), GCN3Encoding(false), CIInsts(false), LDSBankCount(0),
       IsaVersion(ISAVersion0_0_0), EnableHugeScratchBuffer(false),
-      FrameLowering(TargetFrameLowering::StackGrowsUp,
-                    64 * 16, // Maximum stack alignment (long16)
-                    0),
+      FrameLowering(nullptr),
       InstrItins(getInstrItineraryForCPU(GPU)), TargetTriple(TT) {
 
   initializeSubtargetDependencies(TT, GPU, FS);
 
+  const unsigned MaxStackAlign = 64 * 16; // Maximum stack alignment (long16)
+
   if (getGeneration() <= AMDGPUSubtarget::NORTHERN_ISLANDS) {
     InstrInfo.reset(new R600InstrInfo(*this));
     TLInfo.reset(new R600TargetLowering(TM, *this));
+
+    // FIXME: Should have R600 specific FrameLowering
+    FrameLowering.reset(new AMDGPUFrameLowering(
+                          TargetFrameLowering::StackGrowsUp,
+                          MaxStackAlign,
+                          0));
   } else {
     InstrInfo.reset(new SIInstrInfo(*this));
     TLInfo.reset(new SITargetLowering(TM, *this));
+    FrameLowering.reset(new SIFrameLowering(
+                          TargetFrameLowering::StackGrowsUp,
+                          MaxStackAlign,
+                          0));
   }
 }