mgorny added inline comments.
================
Comment at: lib/Driver/Tools.cpp:3267
+ if (llvm::sys::fs::is_directory(CandidateLibPath))
+ CmdArgs.push_back(Args.MakeArgString("-L" + CandidateLibPath));
+
----------------
pirama wrote:
> mgorny wrote:
> > Don't you also need rpath for it? Or is this purely for static runtime?
> I am doing this for a cross-compiling toolchain (Android NDK) where the
> actual rpath is not valid at runtime. The runtime is packaged with the
> application and made available to the loader behind the scenes.
>
> That said, I don't think providing an rpath doesn't hurt. I'll wait for
> input from @cbergstrom and OpenMP folks to see if this'd be useful.
@cbergstrom says: 'let them know that your comments are equivalent (or better)
than mine'. But yeah, PathScale's adding rpaths for the libraries on our end,
so this is consistent with what we do.
https://reviews.llvm.org/D30015
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
