[PATCH] DRAFT: Add new test which covers force user bug

[Robin Hack]

Hello.

I know, I know.. I rewrited it again from the ground. If you will have
some time left, you can always spend it with recent patch.

Polite ping :).


Have nice day
Robin Hack

On Fri, Jan 8, 2016 at 11:56 AM, Robin Hack <rhack at redhat.com> wrote:
> HI all.
>
> Thank you for review.
>
> I discovered other bugs so I decided to do some changes:
> * test were moved to fileserver test env
> * new samba/unix accound named force_user were created
>    - change solves previous issue with make test TESTS=samba3.local.nss
>    - additionally it solves "bug" in test itself. While setup, 2 groups with
> same GID were created which is not nice and taint test env
> * test script were renamed to:
> samba3.blackbox.smbclient.forceuser_validusers
>   - user name has been hard coded but set of arguments for script remains
> (it's for future use :))
>
> New patch is attached.
>
> Have nice day
> Robin Hack
>
>
>
> On Thu, Jan 7, 2016 at 11:31 AM, Michael Adam <obnox at samba.org> wrote:
>>
>> On 2016-01-07 at 23:07 +1300, Andrew Bartlett wrote:
>> > On Thu, 2016-01-07 at 09:43 +0100, Michael Adam wrote:
>> > > Hi Robin,
>> > >
>> > > Thanks a lot for working on the testsuite!!
>> > >
>> > > Some comments:
>> > >
>> > > - Have you run a full 'make test' with your patch?
>> > >   (You should, in this case, since it modifies the test
>> > >   environment...)
>> > >
>> > > - I don't quite understand the patch yet:
>> > >   See comments inline below.
>> > >
>> > >
>> > > On 2016-01-04 at 14:35 +0100, Robin Hack wrote:
>> > > > Patch is here :).
>> > > >
>> > > > On Mon, Jan 4, 2016 at 2:28 PM, Robin Hack <rhack at redhat.com>
>> > > > wrote:
>> > > >
>> > > > > Hi Jeremy.
>> > > > >
>> > > > > Sorry for such a big delay. I was on vacation. And thank you very
>> > > > > much for
>> > > > > heads up.
>> > > > >
>> > > > > I looked at issue which you described and I hope that I fixed it.
>> > > > >
>> > > > > In previous case I used $unix_gids[0] for $unix_name user.
>> > > > > I use newly generated gid now.
>> > > > >
>> > > > > Could you please check and review my changes again and test them?
>> > > > > I ran
>> > > > > $ make test TESTS=samba3.local.nss
>> > > > > and issue seems gone. I'm not sure if other issues will not
>> > > > > raise.
>> > > > >
>> > > > > Sorry for issues with my patch. I didn't realized how fragile it
>> > > > > can be :(.
>> > > > >
>> > > > >
>> > > > > Have nice day
>> > > > > Robin Hack
>> > >
>> > >
>> > > > From cae478de1f577b181eab68f0bef7d6e49ba1c1a7 Mon Sep 17 00:00:00
>> > > > 2001
>> > > > From: Robin Hack <rhack at redhat.com>
>> > > > Date: Fri, 4 Dec 2015 14:50:56 +0100
>> > > > Subject: [PATCH] samba3.blackbox.smbclient.forceuser: Add new test
>> > > > for force
>> > > >  user option.
>> > > >
>> > > > Test covers commit
>> > > > a85d6eb355e8277fe69db6e5837a7f65f57d9c1d
>> > > >
>> > > > BUG: https://bugzilla.samba.org/show_bug.cgi?id=9878
>> > > > RH BUG: https://bugzilla.redhat.com/show_bug.cgi?id=1077651
>> > > >
>> > > > Signed-off-by: Robin Hack <rhack at redhat.com>
>> > > > ---
>> > > >  selftest/target/Samba3.pm              | 31 ++++++++++++++++
>> > > >  source3/script/tests/test_forceuser.sh | 66
>> > > > ++++++++++++++++++++++++++++++++++
>> > > >  source3/selftest/tests.py              |  6 ++--
>> > > >  3 files changed, 101 insertions(+), 2 deletions(-)
>> > > >  create mode 100755 source3/script/tests/test_forceuser.sh
>> > > >
>> > > > diff --git a/selftest/target/Samba3.pm b/selftest/target/Samba3.pm
>> > > > index 94612b2..ccb2751 100755
>> > > > --- a/selftest/target/Samba3.pm
>> > > > +++ b/selftest/target/Samba3.pm
>> > > > @@ -790,11 +790,36 @@ $ret->{USERNAME} = KTEST\\Administrator
>> > > >  sub setup_maptoguest($$)
>> > > >  {
>> > > >   my ($self, $path) = @_;
>> > > > + my $prefix_abs = abs_path($path);
>> > > >
>> > > >   print "PROVISIONING maptoguest...";
>> > > >
>> > > > + my @dirs = ();
>> > > > +
>> > > > + mkdir($prefix_abs, 0777);
>> > > > + my $share_dir="$prefix_abs/share";
>> > > > + push(@dirs, $share_dir);
>> > > > +
>> > > > + my $force_user_not_work_dir =
>> > > > "$share_dir/force_user_not_work";
>> > > > + push(@dirs, $force_user_not_work_dir);
>> > > > +
>> > > > + my $unix_name = ($ENV{USER} or $ENV{LOGNAME} or
>> > > > `PATH=/usr/ucb:$ENV{PATH} whoami`);
>> > > > + chomp $unix_name;
>> > > > +
>> > > > +
>> > > >   my $options = "
>> > > >  map to guest = bad user
>> > > > +
>> > > > +# BUG: https://bugzilla.samba.org/show_bug.cgi?id=9878
>> > > > +# BUG: https://bugzilla.redhat.com/show_bug.cgi?id=1077651
>> > >
>> > > These bugs are not at all related to map to guest.
>> > > But your test and comments seem to suggest that.
>> > >
>> > > I think you should either remove the references to map-to-guest
>> > > and do it (maybe additionally?) with a non-map-to-guest share,
>> > > or else you should remove the above BZ references.
>> >
>> > The tests are of the combination of force user and map to guest = bad
>> > user.
>>
>> Correct. My point is that the referenced bugs have no
>> relationship whatsoever to map-to-guest, as far as I can tell.
>>
>> Michael
>
>