Re: Status of llvm/clang 3.4?

From: Don Lewis <truckman_at_FreeBSD.org>
Date: Tue, 4 Mar 2014 11:22:17 -0800 (PST)
On  4 Mar, Ed Maste wrote:
> On 4 March 2014 12:22, Steve Kargl <sgk_at_troutmask.apl.washington.edu> wrote:
>>
>> On Tue, Mar 04, 2014 at 12:17:18PM +0000, Thomas Mueller wrote:
>> > What is the current status of clang, regarding known bugs, on FreeBSD-current?
>> >
>> > There were reports of www/firefox failing to build because of bug in llvm.
>>
>> Still broken of i386.  Given the google results for "llvm ud2"
>> it is likely that clang will never be fixed.
> 
> Except in this case it's just a plain old Clang bug, not a general
> undefined behaviour issue.  The Clang PR is here:
> http://llvm.org/bugs/show_bug.cgi?id=19007
> 
> Dimitry has a patch to work around the issue in FreeBSD PR 187103:
> http://www.freebsd.org/cgi/query-pr.cgi?pr=187103

In addtion to this firefox (and libxul - same fix applies) problem, I
also ran into problems with:
	multimedia/libxine		PR 187017	already fixed
	multimedia/gstreamer-ffmpeg	PR 187018	already fixed
	math/atlas			PR 186997	i386 only / patch in PR
	math/fftw*			PR 186998	i386 only / patch in PR
These are not clang bugs, but differences from gcc.

Also, not a clang bug, but caused by the lack of gcc in base:
	graphics/jogl			PR 187186	patch in PR

I also ran into a problem bulding math/py-numeric for i386 on a machine
with an Athlon 64 CPU, which causes -march=athlon64 to be set.  In this
case clang aborts.  I filed this bug bug report upstream:
<http://llvm.org/bugs/show_bug.cgi?id=19029>.

I ran into a few more broken ports, but the problems are not clang
related.
Received on Tue Mar 04 2014 - 18:22:33 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:40:47 UTC