[PATCH] Use the 'arcp' fast-math-flag when combining repeated FP divisors

[Sanjay Patel]

Hi hfinkel, echristo, ab, kariddi, milseman,

This is a usage of the IR-level fast-math-flags now that they are propagated to SDNodes (r237046). This was originally part of D8900.

I'm not sure if we need to maintain the global TargetOptions.UnsafeFPMath checks or if we're now free to start deleting those and rely exclusively on the node-level optimization flags. If the latter, I need to update the corresponding fdiv testcases for PowerPC and Aarch64 to use 'arcp'.

http://reviews.llvm.org/D9708

Files:
  lib/CodeGen/SelectionDAG/DAGCombiner.cpp
  test/CodeGen/X86/fdiv-combine.ll

EMAIL PREFERENCES
  http://reviews.llvm.org/settings/panel/emailpreferences/
-------------- next part --------------
A non-text attachment was scrubbed...
Name: D9708.25597.patch
Type: text/x-patch
Size: 4482 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20150512/24251106/attachment.bin>