[cfe-dev] recent change broke -fcatch-undefined-behavior

[Jean-Daniel Dupas]

Le 14 nov. 2012 à 01:17, Richard Smith <richard at metafoo.co.uk> a écrit :

> On Tue, Nov 13, 2012 at 3:08 PM, Sean McBride <sean at rogue-research.com> wrote:
>> On Tue, 16 Oct 2012 23:35:37 -0700, Richard Smith said:
>> 
>>>>> Right now, we could just emit calls to @llvm.trap on unsupported
>>>> platforms,
>>>>> but some of the remaining checks actually need runtime library support. I
>>>>> would prefer to get the runtime working on more platforms, rather than
>>>>> working around its absence.
>>>> 
>>>> Thanks to all for the explanations.
>>>> 
>>>> I guess I'm playing with fire by using svn trunk, but for me, this is a
>>>> big regression: -fcatch-undefined-behavior has gone from a imperfect but
>>>> helpful tool to be totally broken.  I do hope it'll work again before 3.2?
>>> 
>>> 
>>> Yes, I've been promised access to a Mac, and I'll get it working there, if
>>> no-one else gets there first :)
>> 
>> Hi,
>> 
>> With 3.2 now in RC phase, I'd thought I'd pester about -fcatch-undefined-behavior on OS X...  It worked in 3.1, will it be restored in 3.2?  I don't see anything connected to it here:
>> 
>> <http://llvm.org/bugs/show_bug.cgi?id=13893>
>> 
>> Do you need a bug report?
> 
> Thanks for the prod, I've checked in some pending patches to add OS X
> support, in r167888, r167889, and r167890. If you could build
> clang_rt.ubsan_osx and let me know whether everything is working as
> intended, that'd be great; we can then ask to get those patches pulled
> onto the 3.2 branch.


If nothing is done in the mean time, I will have a look at the actual state of ubsan to see what remain to be done to make it work on OS X tonight (CET), and will post the required patches (for compiler-rt and the clang driver).

-- Jean-Daniel