[llvm-commits] [compiler-rt] r165913 - /compiler-rt/trunk/lib/ubsan/ubsan_value.cc

[Richard Smith]

On Thu, Nov 29, 2012 at 1:53 PM, David Blaikie <dblaikie at gmail.com> wrote:

> On Sun, Oct 14, 2012 at 4:53 PM, Richard Smith
> <richard-llvm at metafoo.co.uk> wrote:
> > Author: rsmith
> > Date: Sun Oct 14 18:53:37 2012
> > New Revision: 165913
> >
> > URL: http://llvm.org/viewvc/llvm-project?rev=165913&view=rev
> > Log:
> > Provide a slightly more helpful diagnostic if -fcatch-undefined-behavior
> finds
> > a problem with __int128 arithmetic but the runtime was not built with
> __int128
> > support.
>
> Would it be possible for the test cases that depend on this would be
> marked as untested/xfail/something on platforms that don't support it?
>

It should be possible, yes. It's a little tricky (it depends on both the
target and on the compiler used to build compiler-rt), but I'll have a look.


> Building with clang is actually failing for me, strangely enough. (I
> assume clang should be supporting __int128? since when?)


Clang has provided __int128 for a while, but it didn't actually work for
32-bit targets. Now it only provides __int128 on targets where it works
(and it defines __SIZEOF_INT128__ on those targets, just like gcc).
Removing __int128 usage for versions of Clang in this window fixed both
32-bit builds with those compilers and builds with older versions of Clang
with no __int128 support (such as the one shipped with the latest Xcode),
at the expense of 64-bit builds with recent-but-not-trunk Clangs losing
__int128 support.

If you try again with a trunk Clang as your $CXX, it should work. Let me
know if not.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20121129/f313ae01/attachment.html>