projects / oota-llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 214fd3d) | patch
Drop the limitation to IEEE floating point types from the fdiv of pow2 -> fmul transform.
authorBenjamin Kramer <benny.kra@googlemail.com>
Thu, 8 Nov 2012 13:58:10 +0000 (13:58 +0000)
committerBenjamin Kramer <benny.kra@googlemail.com>
Thu, 8 Nov 2012 13:58:10 +0000 (13:58 +0000)
commit6b370e615ca5b3ef17ed4fe4fa891fab1e2bdd9a
treec8cf9df9e818cd256d670ce14bd4a1e0ddf2f8a5tree | snapshot
parent214fd3d2449738bbe0215dce24406dc29d9e49f7commit | diff
Drop the limitation to IEEE floating point types from the fdiv of pow2 -> fmul transform.

This is safe for x87 long doubles and ppc double doubles too.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167582 91177308-0d34-0410-b5e6-96231b3b80d8
lib/Support/APFloat.cpp diff | blob | history
unittests/ADT/APFloatTest.cpp diff | blob | history
C/C++ LLVM-based compilers that forbids OOTA behaviors