AMDGPU/SI: Emit amd_kernel_code_t in EmitFunctionBodyStart()

[• tstellarAMD]

Summary

This way the function symbol points to the start of amd_kernel_code_t
rather than the start of the function.

[arsenm]

LGTM

lib/Target/AMDGPU/AMDGPUAsmPrinter.cpp

119

Is this alignment not required anymore?

[• tstellarAMD]

lib/Target/AMDGPU/AMDGPUAsmPrinter.cpp

119

Alignment for the function is emitted in EmitFunctionHeader(). We don't need to re-emit it after the AMDKernelCodeT object, because the object is 256 bytes (which is the same as the alignment), so the start of the function, is already aligned properly.

lib/Target/AMDGPU/AMDGPUAsmPrinter.h

public:
 
   const char *getPassName() const override {
     return "AMDGPU Assembly Printer";
   }
 
   /// Implemented in AMDGPUMCInstLower.cpp
   void EmitInstruction(const MachineInstr *MI) override;
 
+  void EmitFunctionBodyStart() override;
+
   void EmitEndOfAsmFile(Module &M) override;
 
   bool PrintAsmOperand(const MachineInstr *MI, unsigned OpNo,
                        unsigned AsmVariant, const char *ExtraCode,
                        raw_ostream &O) override;
 
 protected:
   std::vector<std::string> DisasmLines, HexLines;
   size_t DisasmLineMaxLen;
 };
 
 } // End anonymous llvm
 
 #endif

lib/Target/AMDGPU/AMDGPUAsmPrinter.cpp

extern "C" void LLVMInitializeAMDGPUAsmPrinter() {
   TargetRegistry::RegisterAsmPrinter(TheAMDGPUTarget, createAMDGPUAsmPrinterPass);
   TargetRegistry::RegisterAsmPrinter(TheGCNTarget, createAMDGPUAsmPrinterPass);
 }
 
 AMDGPUAsmPrinter::AMDGPUAsmPrinter(TargetMachine &TM,
                                    std::unique_ptr<MCStreamer> Streamer)
     : AsmPrinter(TM, std::move(Streamer)) {}
 
+void AMDGPUAsmPrinter::EmitFunctionBodyStart() {
+  const AMDGPUSubtarget &STM = MF->getSubtarget<AMDGPUSubtarget>();
+  SIProgramInfo KernelInfo;
+  if (STM.isAmdHsaOS()) {
+    getSIProgramInfo(KernelInfo, *MF);
+    EmitAmdKernelCodeT(*MF, KernelInfo);
+  }
+}
+
 void AMDGPUAsmPrinter::EmitEndOfAsmFile(Module &M) {
 
   // This label is used to mark the end of the .text section.
   const TargetLoweringObjectFile &TLOF = getObjFileLowering();
   OutStreamer->SwitchSection(TLOF.getTextSection());
   MCSymbol *EndOfTextLabel =
       OutContext.getOrCreateSymbol(StringRef(END_OF_TEXT_LABEL_NAME));
   OutStreamer->EmitLabel(EndOfTextLabel);
 }
 
 bool AMDGPUAsmPrinter::runOnMachineFunction(MachineFunction &MF) {
 
   // The starting address of all shader programs must be 256 bytes aligned.
   MF.setAlignment(8);
 
   SetupMachineFunction(MF);
 
   MCContext &Context = getObjFileLowering().getContext();
   MCSectionELF *ConfigSection =
       Context.getELFSection(".AMDGPU.config", ELF::SHT_PROGBITS, 0);
   OutStreamer->SwitchSection(ConfigSection);
 
   const AMDGPUSubtarget &STM = MF.getSubtarget<AMDGPUSubtarget>();
   SIProgramInfo KernelInfo;
-  if (STM.isAmdHsaOS()) {
-    getSIProgramInfo(KernelInfo, MF);
-    EmitAmdKernelCodeT(MF, KernelInfo);
-    OutStreamer->EmitCodeAlignment(2 << (MF.getAlignment() - 1));
-  } else if (STM.getGeneration() >= AMDGPUSubtarget::SOUTHERN_ISLANDS) {
+  if (STM.getGeneration() >= AMDGPUSubtarget::SOUTHERN_ISLANDS) {
+    if (!STM.isAmdHsaOS()) {
       getSIProgramInfo(KernelInfo, MF);
       EmitProgramInfoSI(MF, KernelInfo);
+    }
   } else {
     EmitProgramInfoR600(MF);
   }
 
   DisasmLines.clear();
   HexLines.clear();
   DisasmLineMaxLen = 0;
 

test/CodeGen/AMDGPU/hsa.ll

 ; RUN: llc < %s -mtriple=amdgcn--amdhsa -mcpu=kaveri | FileCheck --check-prefix=HSA %s
 
+; HSA: {{^}}simple:
 ; HSA: .section        .hsa.version
 ; HSA-NEXT: .ascii  "HSA Code Unit:0.0:AMD:0.1:GFX8.1:0"
-; HSA: {{^}}simple:
+; Test that the amd_kernel_code_t object is emitted
+; HSA: .asciz
 ; Make sure we are setting the ATC bit:
 ; HSA: s_mov_b32 s[[HI:[0-9]]], 0x100f000
 ; HSA: buffer_store_dword v{{[0-9]+}}, s[0:[[HI]]], 0
 
 define void @simple(i32 addrspace(1)* %out) {
 entry:
   store i32 0, i32 addrspace(1)* %out
   ret void
 }