diff --git a/lib/Target/PowerPC/PPCInstrInfo.cpp b/lib/Target/PowerPC/PPCInstrInfo.cpp index d2df6645bb0..d9d68446f53 100644 --- a/lib/Target/PowerPC/PPCInstrInfo.cpp +++ b/lib/Target/PowerPC/PPCInstrInfo.cpp @@ -570,12 +570,15 @@ PPCInstrInfo::StoreRegToStackSlot(MachineFunction &MF, // STVX VAL, 0, R0 // // FIXME: We use R0 here, because it isn't available for RA. - NewMIs.push_back(addFrameReference(BuildMI(MF, DL, get(PPC::ADDI), PPC::R0), + bool Is64Bit = TM.getSubtargetImpl()->isPPC64(); + unsigned Instr = Is64Bit ? PPC::ADDI8 : PPC::ADDI; + unsigned GPR0 = Is64Bit ? PPC::X0 : PPC::R0; + NewMIs.push_back(addFrameReference(BuildMI(MF, DL, get(Instr), GPR0), FrameIdx, 0, 0)); NewMIs.push_back(BuildMI(MF, DL, get(PPC::STVX)) .addReg(SrcReg, getKillRegState(isKill)) - .addReg(PPC::R0) - .addReg(PPC::R0)); + .addReg(GPR0) + .addReg(GPR0)); } else { llvm_unreachable("Unknown regclass!"); } @@ -707,10 +710,13 @@ PPCInstrInfo::LoadRegFromStackSlot(MachineFunction &MF, DebugLoc DL, // Dest = LVX 0, R0 // // FIXME: We use R0 here, because it isn't available for RA. - NewMIs.push_back(addFrameReference(BuildMI(MF, DL, get(PPC::ADDI), PPC::R0), + bool Is64Bit = TM.getSubtargetImpl()->isPPC64(); + unsigned Instr = Is64Bit ? PPC::ADDI8 : PPC::ADDI; + unsigned GPR0 = Is64Bit ? PPC::X0 : PPC::R0; + NewMIs.push_back(addFrameReference(BuildMI(MF, DL, get(Instr), GPR0), FrameIdx, 0, 0)); - NewMIs.push_back(BuildMI(MF, DL, get(PPC::LVX),DestReg).addReg(PPC::R0) - .addReg(PPC::R0)); + NewMIs.push_back(BuildMI(MF, DL, get(PPC::LVX),DestReg).addReg(GPR0) + .addReg(GPR0)); } else { llvm_unreachable("Unknown regclass!"); } diff --git a/test/CodeGen/PowerPC/vrspill.ll b/test/CodeGen/PowerPC/vrspill.ll new file mode 100644 index 00000000000..fdd200c1032 --- /dev/null +++ b/test/CodeGen/PowerPC/vrspill.ll @@ -0,0 +1,19 @@ +; RUN: llc -O0 -mtriple=powerpc-unknown-linux-gnu -verify-machineinstrs < %s | FileCheck %s +; RUN: llc -O0 -mtriple=powerpc64-unknown-linux-gnu -verify-machineinstrs < %s | FileCheck %s + +; This verifies that we generate correct spill/reload code for vector regs. + +define void @addrtaken(i32 %i, <4 x float> %w) nounwind { +entry: + %i.addr = alloca i32, align 4 + %w.addr = alloca <4 x float>, align 16 + store i32 %i, i32* %i.addr, align 4 + store <4 x float> %w, <4 x float>* %w.addr, align 16 + call void @foo(i32* %i.addr) + ret void +} + +; CHECK: stvx 2, 0, 0 +; CHECK: lvx 2, 0, 0 + +declare void @foo(i32*)