Re: [PATCH net] selftests: fib_tests: use per-test return value in subtests

[Hangbin Liu <hangbin.liu@linux.dev>]

On Tue, Sep 29, 2026 at 04:45:17PM +0300, Ido Schimmel wrote:
> On Tue, Sep 29, 2026 at 09:57:01AM +0800, Hangbin Liu wrote:
> > From: Hangbin Liu <liuhangbin@kylinos.cn>
> > 
> > fib_tests mixes use of $ret for both per-test return value and global exit
> > code. If an earlier subtest fails and a later function sets ret=0, the
> > script will exit with status 0 even though [FAIL] lines were printed.
> > 
> > Use RET as the per-test return value, as already defined in lib.sh. The
> > exit code ret will only be set in log_test().
> > 
> > Fixes: 607bd2e502f5 ("selftests: fib_tests: Add test cases for IPv4/IPv6 FIB")
> > Reported-by: Sashiko <netdev-bot+sashiko@kernel.org>
> > Closes: https://lore.kernel.org/all/178961184152.22033.9267194009793243294@kernel.org
> > Signed-off-by: Hangbin Liu <liuhangbin@kylinos.cn>
> 
> Please target the patch at net-next and drop the Fixes tag. The patch
> doesn't fix a regression, nothing is failing (or passing when it
> shouldn't) because of it and it cannot be backported cleanly to old
> kernels anyway. I targeted similar patches at net-next in the past. See
> [1], for example.
> 
> [1] https://lore.kernel.org/all/20250908073238.119240-5-idosch@nvidia.com/

OK, got it.

> 
> > ---
> >  tools/testing/selftests/net/fib_tests.sh | 40 ++++++++++++++++----------------
> >  1 file changed, 20 insertions(+), 20 deletions(-)
> > 
> > diff --git a/tools/testing/selftests/net/fib_tests.sh b/tools/testing/selftests/net/fib_tests.sh
> > index b338bfb196a2..9ce5623b049c 100755
> > --- a/tools/testing/selftests/net/fib_tests.sh
> > +++ b/tools/testing/selftests/net/fib_tests.sh
> > @@ -369,7 +369,7 @@ fib_carrier_local_test()
> >  
> >  fib_carrier_unicast_test()
> >  {
> > -	ret=0
> > +	RET=0
> 
> Looks like this line can be removed (similar to fib_carrier_local_test()
> above it) given that RET is never used in this function

Right, I will remove it.

Thanks
Hangbin