[PATCH] PR5172: Fix for a bug in pragma redefine_extname implementation.

[Andrey Bokhanko]

In http://reviews.llvm.org/D10187#183409, @theraven wrote:

> I'm also not completely sure of the semantics - I just made it work well enough for the Solaris headers that libc++ needed to work.  The changes look fine to me though.


The semantic is described here: http://docs.oracle.com/cd/E19205-01/819-5265/bjacu/index.html

IMHO, clang implements it right; the only thing I did is fixed a corner case.

> Have they been tested with the various things in compiler-rt that use this pragma?  We currently use it as a hack because clang doesn't allow you to implement functions that have the same name as builtins (even when those builtins are ones that need implementing by a function of the same name on some archs).


I did "make check-all" + ran a few internal Intel tests.

What is the best way to test compiler-rt with my changes?

Andrey


http://reviews.llvm.org/D10187

EMAIL PREFERENCES
  http://reviews.llvm.org/settings/panel/emailpreferences/