Re: Possible pthread_cond_timedwait() regression on kfreebsd-i386

[Stefan Fritsch <sf@sfritsch.de>]

On Saturday 02 June 2012, Steven Chamberlain wrote:
> On 02/06/12 20:42, Stefan Fritsch wrote:
> > I will just wait for an eglibc update and then see if it fixes
> > the problem.
> 
> I tested this on my kfreebsd-i386 system where I already rebuilt
> eglibc
> 
> with the patch from SVN.  Looks good to me:
> > steven@kfreebsd-i386:~/apr-1.4.6/build-kfreebsd-i386/test$ m=0 ;
> > for n in $(seq 1 100) ; do ./testall -v testsock && m=$((m+1)) ;
> > done ; echo "Passed $m/$n runs" testsock            : SUCCESS
> > All tests passed.
> > testsock            : SUCCESS
> > All tests passed.
> > testsock            : SUCCESS
> > All tests passed.
> 
> ...
> 
> > Passed 100/100 runs
> 
> So it seems we just need to give back apr and perl after the fixed
> eglibc is ready on the buildds.

It seems you got confused by the apr package's logic to ignore some 
test results on some systems. You have to run testlock, not testsock.

Cheers,
Stefan