[PATCH] D19370: [InstCombine] Preserve fast math flags when combining PHIs

[silviu.baranga@arm.com via llvm-commits]

sbaranga created this revision.
sbaranga added a reviewer: majnemer.
sbaranga added a subscriber: llvm-commits.

When optimizing PHIs which have inputs floating point binary
operators, we preserve all IR flags except the fast math
flags.

This change removes the logic which tracked some of the IR flags
(no wrap, exact) and replaces it by doing an and on the IR flags of
all inputs to the PHI - which will also handle the fast math
flags.

http://reviews.llvm.org/D19370

Files:
  lib/Transforms/InstCombine/InstCombinePHI.cpp
  test/Transforms/InstCombine/phi-preserve-ir-flags.ll

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D19370.54518.patch
Type: text/x-patch
Size: 6294 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20160421/44c05dc3/attachment.bin>