diff options
| author | Craig Topper <craig.topper@gmail.com> | 2013-03-14 07:09:57 +0000 |
|---|---|---|
| committer | Craig Topper <craig.topper@gmail.com> | 2013-03-14 07:09:57 +0000 |
| commit | 599521f1671d720a2c786058810537287920f44a (patch) | |
| tree | 29c5c4bb78de3e832137e37a0bb637ae1bd224de /lib/Target | |
| parent | 6f8c6852a08df9d41f4e0242624ec744cd7e0d2f (diff) | |
| download | llvm-599521f1671d720a2c786058810537287920f44a.tar.gz llvm-599521f1671d720a2c786058810537287920f44a.tar.bz2 llvm-599521f1671d720a2c786058810537287920f44a.tar.xz | |
Teach X86 MC instruction lowering 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.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@177011 91177308-0d34-0410-b5e6-96231b3b80d8
Diffstat (limited to 'lib/Target')
| -rw-r--r-- | lib/Target/X86/X86MCInstLower.cpp | 42 |
1 files changed, 42 insertions, 0 deletions
diff --git a/lib/Target/X86/X86MCInstLower.cpp b/lib/Target/X86/X86MCInstLower.cpp index 3af1b3e06b..1e8347dfe4 100644 --- a/lib/Target/X86/X86MCInstLower.cpp +++ b/lib/Target/X86/X86MCInstLower.cpp @@ -407,6 +407,48 @@ ReSimplify: LowerUnaryToTwoAddr(OutMI, X86::XOR32rr); // MOV32r0 -> XOR32rr break; + // Commute operands to get a smaller encoding by using VEX.R instead of VEX.B + // if one of the registers is extended, but other isn't. + 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::VMOVSDrr: + case X86::VMOVSSrr: + case X86::VMOVUPDrr: + case X86::VMOVUPDYrr: + case X86::VMOVUPSrr: + case X86::VMOVUPSYrr: { + if (X86II::isX86_64ExtendedReg(OutMI.getOperand(0).getReg()) && + !X86II::isX86_64ExtendedReg(OutMI.getOperand(1).getReg())) + break; + + unsigned NewOpc; + switch (OutMI.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::VMOVSDrr: NewOpc = X86::VMOVSDrr_REV; break; + case X86::VMOVSSrr: NewOpc = X86::VMOVSSrr_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; + } + OutMI.setOpcode(NewOpc); + break; + } + // TAILJMPr64, CALL64r, CALL64pcrel32 - These instructions have register // inputs modeled as normal uses instead of implicit uses. As such, truncate // off all but the first operand (the callee). FIXME: Change isel. |