projects / oota-llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: fe6f4e4) | patch
AArch64/ARM64: implement BFI optimisation
authorTim Northover <tnorthover@apple.com>
Thu, 24 Apr 2014 12:11:53 +0000 (12:11 +0000)
committerTim Northover <tnorthover@apple.com>
Thu, 24 Apr 2014 12:11:53 +0000 (12:11 +0000)
commitb62ba5eca09b8bf196093279f6c8911d5f90c7cc
tree6705fa6b80f818eb59760327a798d02db50fcce5tree | snapshot
parentfe6f4e4d31aa5fd0840887883ffff45ae0e9295acommit | diff
AArch64/ARM64: implement BFI optimisation

ARM64 was not producing pure BFI instructions for bitfield insertion
operations, unlike AArch64. The approach had to be a little different (in
ISelDAGToDAG rather than ISelLowering), and the outcomes aren't identical but
hopefully this gives it similar power.

This should address PR19424.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@207102 91177308-0d34-0410-b5e6-96231b3b80d8
lib/Target/ARM64/ARM64ISelDAGToDAG.cpp diff | blob | history
test/CodeGen/AArch64/bitfield-insert.ll diff | blob | history
test/CodeGen/ARM64/strict-align.ll diff | blob | history
C/C++ LLVM-based compilers that forbids OOTA behaviors