Re: [Libexplain-users] tests t0461a, t0462a, t0463a fail on Fedora 14 x86_64

[Peter M. <pm...@op...>]

> > From: Eric Smith <er...@br...>

> > I'm updating the Fedora package for libexplain to 0.47.D001.  There were 
> > a bunch of tests that failed for me previously that now work.  However, 
> > there are three new failures reported:
> > 
> > PATH=`pwd`/bin:$PATH /bin/sh test/04/t0461a.sh
> > 1,2c1,3
> > < kill(pid = 42, sig = 3200) failed, Invalid argument (EINVAL) because the
> > < sig argument was incorrectly specified
> > ---
> >  > kill(pid = 42 "exe (readlink: Permission denied)", sig = 3200) failed,
> >  > Invalid argument (EINVAL) because the sig argument was incorrectly
> >  > specified
> > FAILED test of kill EINVAL
> > 
> > 
> > PATH=`pwd`/bin:$PATH /bin/sh test/04/t0462a.sh
> > 1,3c1,4
> > < kill(pid = 42, sig = SIGTERM) failed, Operation not permitted (EPERM)
> > < because the process does not have permission to send the signal to any of
> > < the target processes, and the process is not privileged
> > ---
> >  > kill(pid = 42 "exe (readlink: Permission denied)", sig = SIGTERM) failed,
> >  > Operation not permitted (EPERM) because the process does not have
> >  > permission to send the signal to any of the target processes, and the
> >  > process is not privileged
> > FAILED test of kill EPERM
> > 
> > 
> > PATH=`pwd`/bin:$PATH /bin/sh test/04/t0463a.sh
> > 1,2c1,2
> > < kill(pid = 42, sig = SIGKILL) failed, No such process (ESRCH) because the
> > < pid process does not exist
> > ---
> >  > kill(pid = 42 "exe (readlink: Permission denied)", sig = SIGKILL) failed,
> >  > No such process (ESRCH) because the pid process does not exist
> > FAILED test of kill ESRCH

This is caused by process 42 existing.
I'll improve the test scripts to cope with this.

> > What does the extra "exe (readlink: Permission denied)" signify?

It means that when lsof(1) attempted to read the name of the executable,
it couldn't.  The process ID is supplemented with the name of the
executable, when available.

I wish lsof(1) had said nothing, rather than said this.
I'll add code to detected and discard this useless result.

> >   Could 
> > that have something to do with the tests being run in a chroot 
> > environment?  With earlier versions of libexplain and Fedora I had 23 
> > other tests fail seemingly due to the lack of /etc/mtab in the chroot, 
> > but those tests are all passing now.