[PATCH] PR15546 Avoid NEON SP-FP unless unsafe-math or Darwin

[Renato Golin]

Hi all,

NEON is not IEEE 754 compliant, so we should avoid lowering
single-precision floating point operations with NEON unless unsafe-math is
turned on. The equivalent VFP instructions are IEEE 754 compliant, but in
some cores they're much slower, so some archs/OSs might still request it to
be on by default, such as Swift and Darwin.

A new test was added (neon-spfp.ll) to make sure all combinations are kept
correct, and many other tests were changed (mostly by adding an unsafe
option to A8 but still keeping the old one). Please make sure the
assumptions are correct for Darwin / Swift.

Check-all passes on x86_64, I'm now running on the Chromebook, but I'm not
expecting any problems, just to be safe.

cheers,
--renato
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20130321/72ef5ae7/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: neon-spfp.patch
Type: application/octet-stream
Size: 16441 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20130321/72ef5ae7/attachment.obj>