apaprocki added inline comments.
================
Comment at: lib/Basic/Targets.cpp:718
@@ +717,3 @@
+ Builder.defineMacro("_LONG_LONG");
+ Builder.defineMacro("_ALL_SOURCE");
+ Builder.defineMacro("_REENTRANT");
----------------
majnemer wrote:
> apaprocki wrote:
> > majnemer wrote:
> > > Are we really supposed to define this macro? Does GCC define this? I
> > > cannot find where it does so in the source.
> > I defined both `_ALL_SOURCE` and `_REENTRANT` because the
> > `SolarisTargetInfo` defines them (`__EXTENSIONS__` is the Solaris
> > equivalent of `_ALL_SOURCE`). If that is an oversight in the Solaris work,
> > I'll remove `_ALL_SOURCE` here. GCC does not define it by default.
> After a closer examination of the gcc sources, it looks like `ALL_SOURCE` is
> defined if gcc is being used for C++ code (via `CPLUSPLUS_CPP_SPEC`).
>
> I don't see where `_REENTRANT` is defined for AIX though...
On AIX, when `gcc` is passed `-pthread`, it defines `_THREAD_SAFE`. I think
this should be changed to
```
if (Opts.POSIXThreads)
Builder.defineMacro("_THREAD_SAFE");
```
On Solaris, `gcc` defines `_REENTRANT` when `-pthread` is passed. I think it is
a separate bug (and I can file a separate revision) to put the Solaris use of
it also on an `Opts.POSIXThreads` check.
https://reviews.llvm.org/D18360
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
