diff options
author | Craig Topper <craig.topper@gmail.com> | 2013-10-07 05:42:48 +0000 |
---|---|---|
committer | Craig Topper <craig.topper@gmail.com> | 2013-10-07 05:42:48 +0000 |
commit | 8fdba75d5b865246455c335adf439def1c3daaeb (patch) | |
tree | 10367cba7f90d1814f80a8513b9bc04724166e27 /lib/Target | |
parent | 36a9b31b981553350f5cc4adad9917656c20e96e (diff) | |
download | llvm-8fdba75d5b865246455c335adf439def1c3daaeb.tar.gz llvm-8fdba75d5b865246455c335adf439def1c3daaeb.tar.bz2 llvm-8fdba75d5b865246455c335adf439def1c3daaeb.tar.xz |
Teach X86 asm parser that VMOVAPSrr and other VEX-encoded register to register moves should be switched from using the MRMSrcReg form to the MRMDestReg form if the source register is a 64-bit extended register and the destination register is not.
This allows the instruction to be encoded using the 2-byte VEX form instead of the 3-byte VEX form. The GNU assembler has similar behavior and instruction selection already does this.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@192088 91177308-0d34-0410-b5e6-96231b3b80d8
Diffstat (limited to 'lib/Target')
-rw-r--r-- | lib/Target/X86/AsmParser/X86AsmParser.cpp | 49 | ||||
-rw-r--r-- | lib/Target/X86/X86InstrSSE.td | 2 |
2 files changed, 50 insertions, 1 deletions
diff --git a/lib/Target/X86/AsmParser/X86AsmParser.cpp b/lib/Target/X86/AsmParser/X86AsmParser.cpp index 93c2169a40..dc9654fa58 100644 --- a/lib/Target/X86/AsmParser/X86AsmParser.cpp +++ b/lib/Target/X86/AsmParser/X86AsmParser.cpp @@ -2271,6 +2271,55 @@ processInstruction(MCInst &Inst, case X86::SBB16i16: return convert16i16to16ri8(Inst, X86::SBB16ri8); case X86::SBB32i32: return convert32i32to32ri8(Inst, X86::SBB32ri8); case X86::SBB64i32: return convert64i32to64ri8(Inst, X86::SBB64ri8); + case X86::VMOVAPDrr: + case X86::VMOVAPDYrr: + case X86::VMOVAPSrr: + case X86::VMOVAPSYrr: + case X86::VMOVDQArr: + case X86::VMOVDQAYrr: + case X86::VMOVDQUrr: + case X86::VMOVDQUYrr: + case X86::VMOVUPDrr: + case X86::VMOVUPDYrr: + case X86::VMOVUPSrr: + case X86::VMOVUPSYrr: { + if (X86II::isX86_64ExtendedReg(Inst.getOperand(0).getReg()) || + !X86II::isX86_64ExtendedReg(Inst.getOperand(1).getReg())) + return false; + + unsigned NewOpc; + switch (Inst.getOpcode()) { + default: llvm_unreachable("Invalid opcode"); + case X86::VMOVAPDrr: NewOpc = X86::VMOVAPDrr_REV; break; + case X86::VMOVAPDYrr: NewOpc = X86::VMOVAPDYrr_REV; break; + case X86::VMOVAPSrr: NewOpc = X86::VMOVAPSrr_REV; break; + case X86::VMOVAPSYrr: NewOpc = X86::VMOVAPSYrr_REV; break; + case X86::VMOVDQArr: NewOpc = X86::VMOVDQArr_REV; break; + case X86::VMOVDQAYrr: NewOpc = X86::VMOVDQAYrr_REV; break; + case X86::VMOVDQUrr: NewOpc = X86::VMOVDQUrr_REV; break; + case X86::VMOVDQUYrr: NewOpc = X86::VMOVDQUYrr_REV; break; + case X86::VMOVUPDrr: NewOpc = X86::VMOVUPDrr_REV; break; + case X86::VMOVUPDYrr: NewOpc = X86::VMOVUPDYrr_REV; break; + case X86::VMOVUPSrr: NewOpc = X86::VMOVUPSrr_REV; break; + case X86::VMOVUPSYrr: NewOpc = X86::VMOVUPSYrr_REV; break; + } + Inst.setOpcode(NewOpc); + return true; + } + case X86::VMOVSDrr: + case X86::VMOVSSrr: { + if (X86II::isX86_64ExtendedReg(Inst.getOperand(0).getReg()) || + !X86II::isX86_64ExtendedReg(Inst.getOperand(2).getReg())) + return false; + unsigned NewOpc; + switch (Inst.getOpcode()) { + default: llvm_unreachable("Invalid opcode"); + case X86::VMOVSDrr: NewOpc = X86::VMOVSDrr_REV; break; + case X86::VMOVSSrr: NewOpc = X86::VMOVSSrr_REV; break; + } + Inst.setOpcode(NewOpc); + return true; + } } } diff --git a/lib/Target/X86/X86InstrSSE.td b/lib/Target/X86/X86InstrSSE.td index 3d8654900c..f6f2266244 100644 --- a/lib/Target/X86/X86InstrSSE.td +++ b/lib/Target/X86/X86InstrSSE.td @@ -1104,7 +1104,7 @@ let Predicates = [UseSSE1] in { // Alias instruction to do FR32 or FR64 reg-to-reg copy using movaps. Upper // bits are disregarded. FIXME: Set encoding to pseudo! -let neverHasSideEffects = 1, SchedRW = [WriteMove] in { +let isCodeGenOnly = 1, neverHasSideEffects = 1, SchedRW = [WriteMove] in { def FsVMOVAPSrr : VPSI<0x28, MRMSrcReg, (outs FR32:$dst), (ins FR32:$src), "movaps\t{$src, $dst|$dst, $src}", [], IIC_SSE_MOVA_P_RR>, VEX; |