[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#798691: libc6-i686: sin() sometimes broken



On 2015-09-11 20:33, Christian Tacke wrote:
> Package: libc6-i686
> Version: 2.19-18+deb8u1
> Severity: important
> 
> Hi,

Hi,

> I was trying to create a local backport of python-cffi onto
> jessie/i386.
> 
> On a system with libc6-i686 installed, I get a failure in
> the self test of the build:
> 
> >       assert library.sin(12.3) == math.sin(12.3)
> E       assert 0.8444952930965295 == -0.26323179136580094
> 
> Uninstalling libc6-i686 lets the full build and tests run
> through.
> 
> Something very related has already happened before:
> https://www.sourceware.org/bugzilla/show_bug.cgi?id=16625
> 
> This suggests a compiler bug in gcc 4.8. But jessie is
> compiled with 4.9, so it must be something a bit different.

No, the glibc on jessie is actually built with gcc 4.8, as gcc 4.9
introduces some regression on the pthread cleanup functions. This
means it's very likely to be the same bug.

I guess the only way to fix that is to actually fix gcc-4.8 and rebuild
the glibc afterwards.

-- 
Aurelien Jarno                          GPG: 4096R/1DDD8C9B
aurelien@aurel32.net                 http://www.aurel32.net


Reply to: