ARM64: model pre/post-indexed operations properly.
authorTim Northover <tnorthover@apple.com>
Thu, 22 May 2014 11:56:20 +0000 (11:56 +0000)
committerTim Northover <tnorthover@apple.com>
Thu, 22 May 2014 11:56:20 +0000 (11:56 +0000)
commit369e565f23d34876f1689971b98255dd84131546
treed6a1000ad30999fbdadc47a05605290e7604fdd7
parente072ed71c87ba1be56f1bca1a1a8057760badea0
ARM64: model pre/post-indexed operations properly.

We should be keeping track of the writeback on these instructions,
otherwise we're relying on LLVM's stupidity for correct code.

Fortunately, the MC layer can now handle all required constraints,
which means we can get rid of the CodeGen only PseudoInsts too.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209426 91177308-0d34-0410-b5e6-96231b3b80d8
lib/Target/ARM64/ARM64AsmPrinter.cpp
lib/Target/ARM64/ARM64FrameLowering.cpp
lib/Target/ARM64/ARM64ISelDAGToDAG.cpp
lib/Target/ARM64/ARM64InstrFormats.td
lib/Target/ARM64/ARM64InstrInfo.cpp
lib/Target/ARM64/ARM64InstrInfo.td
lib/Target/ARM64/ARM64LoadStoreOptimizer.cpp
lib/Target/ARM64/AsmParser/ARM64AsmParser.cpp
lib/Target/ARM64/Disassembler/ARM64Disassembler.cpp