This is the mail archive of the libffi-discuss@sourceware.org mailing list for the libffi project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: libffi testsuite failures on OSX Lion, 3.0.10 and HEAD @ 0a1ab12a8d


On Tue, Apr 3, 2012 at 8:04 AM, Anthony Green <green@moxielogic.com> wrote:
> On Tue, Apr 3, 2012 at 12:20 AM, P. Martin <mrsmiley98@lycos.com> wrote:
>> Hello. ?When trying to debug problems with glib for Homebrew,
>> I realized that we needed to test libffi to eliminate any unexpected
>> impact it might be having.
>>
>> I would like to report a dozen or so testsuite failures in both 3.0.10
>> and HEAD @ 0a1ab12a8d when building on OSX Lion 10.7.3 using
>> llvm-gcc-4.2.1 (2336.9) from XCode-4.3.2.
>
> Thanks. ? I can reproduce 8 of these failures on Linux with llvm 2.9
> from this site:
> http://llvm.org/releases/download.html
>
> So at this point I'd say it's a compiler issue, but I'll see if I have
> any time to poke at it today.

I'm going to chalk this up as compiler bugs.

For instance, if I build the promotion.c test with -O0 it runs fine,
but fails as soon as I move to -O1, even when linked with the same
libffi.a.  I double checked the test case and don't see anything
obviously wrong.    So my theory is that llvm is miscompiling
promotion.c when optimized.

I don't plan on investigating llvm 2.9 any more.  The llvm project
doesn't seem to make llvm-gcc binaries available for newer versions of
llvm.  I don't really follow llvm, but perhaps it's all clang now.

If this is a real problem for anybody, please let me know.

AG


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]