Re: [PATCH 3/3] test: Re-implement pasta NDP tests using tunbridge & exeter

[Stefano Brivio <sbrivio@redhat.com>]

On Wed, 8 Oct 2025 13:32:27 +1100
David Gibson <david@gibson.dropbear.id.au> wrote:

> On Tue, Oct 07, 2025 at 10:01:10PM +0200, Stefano Brivio wrote:
> > On Thu,  2 Oct 2025 17:57:08 +1000
> > David Gibson <david@gibson.dropbear.id.au> wrote:
> >   
> > > Convert the pasta NDP tests from shell and our own DSL to Python using
> > > the exeter test protocol and tunbridge network simulation library.
> > > 
> > > Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
> > > ---
> > >  test/Makefile          |  2 +-
> > >  test/pasta/dhcp        |  5 ++++
> > >  test/pasta/ndp.py      | 59 ++++++++++++++++++++++++++++++++++++++++++
> > >  test/run               |  6 +++--
> > >  test/tasst/__init__.py |  4 +++
> > >  test/tasst/pasta.py    | 40 ++++++++++++++++++++++++++++
> > >  6 files changed, 113 insertions(+), 3 deletions(-)
> > >  create mode 100755 test/pasta/ndp.py
> > >  create mode 100644 test/tasst/pasta.py
> > > 
> > > diff --git a/test/Makefile b/test/Makefile
> > > index f66c7e7e..95e3d75e 100644
> > > --- a/test/Makefile
> > > +++ b/test/Makefile
> > > @@ -67,7 +67,7 @@ ASSETS = $(DOWNLOAD_ASSETS) $(LOCAL_ASSETS)
> > >  
> > >  EXETER_SH = smoke/smoke.sh build/static_checkers.sh
> > >  EXETER_PYPATH = exeter/py3:tunbridge/:.
> > > -EXETER_PYTHON = smoke/smoke.py build/build.py
> > > +EXETER_PYTHON = smoke/smoke.py build/build.py pasta/ndp.py
> > >  EXETER_BATS = $(EXETER_SH:%=%.bats) $(EXETER_PYTHON:%=%.bats)
> > >  BATS_FILES = $(EXETER_BATS) \
> > >  	podman/test/system/505-networking-pasta.bats
> > > diff --git a/test/pasta/dhcp b/test/pasta/dhcp
> > > index e1c66be6..61279fbf 100644
> > > --- a/test/pasta/dhcp
> > > +++ b/test/pasta/dhcp
> > > @@ -18,6 +18,11 @@ test	Interface name
> > >  nsout	IFNAME ip -j link show | jq -rM '.[] | select(.link_type == "ether").ifname'
> > >  check	[ -n "__IFNAME__" ]
> > >  
> > > +# Bring up the interface
> > > +ns	ip link set dev __IFNAME__ up
> > > +# Wait for SLAAC & DAD to complete
> > > +ns	while ! ip -j -6 addr show dev __IFNAME__ | jq -e '.[].addr_info.[] | select(.protocol == "kernel_ra")'; do sleep 0.1; done
> > > +
> > >  test	DHCP: address
> > >  ns	/sbin/dhclient -4 --no-pid __IFNAME__
> > >  nsout	ADDR ip -j -4 addr show|jq -rM '.[] | select(.ifname == "__IFNAME__").addr_info[0].local'
> > > diff --git a/test/pasta/ndp.py b/test/pasta/ndp.py
> > > new file mode 100755
> > > index 00000000..8c7ce31e
> > > --- /dev/null
> > > +++ b/test/pasta/ndp.py
> > > @@ -0,0 +1,59 @@
> > > +#! /usr/bin/env python3
> > > +#
> > > +# SPDX-License-Identifier: GPL-2.0-or-later
> > > +#
> > > +# test/pasta/ndp.py - pasta NDP functionality
> > > +#
> > > +# Copyright Red Hat
> > > +# Author: David Gibson <david@gibson.dropbear.id.au>
> > > +
> > > +import contextlib
> > > +import dataclasses
> > > +from typing import Iterator
> > > +
> > > +import exeter
> > > +import tunbridge
> > > +import tasst
> > > +
> > > +
> > > +@dataclasses.dataclass
> > > +class UnconfiguredScenario(exeter.Scenario):
> > > +    """Tests for a pasta instance without --config-net"""
> > > +
> > > +    host: tunbridge.Site
> > > +    guest: tunbridge.Site
> > > +    ifname: str
> > > +    addr6: tunbridge.ip.AddrMask6
> > > +    gw6: tunbridge.ip.Addr6  
> > 
> > Until this point, it looks like stuff I can happily copy and paste,
> > and grasp, even. But then:
> >   
> > > +    @exeter.scenariotest
> > > +    def test_ifname(self) -> None:
> > > +        ifs = tunbridge.ip.ifs(self.guest)
> > > +        exeter.assert_eq(set(ifs), {'lo', self.ifname})  
> > 
> > ...why does a "Scenario" have a .ifname?  
> 
> Yeah, the readability of the Scenario mechanism was something I was
> particularly concerned about.  I think the concept is valuable, but
> I'm very open to different ways of naming or organising it, if we can
> up with something better.

From the description you give below, the name seems to fit.

> A "Scenario" (specifically a subclass of exeter.Scenario) is a group
> of tests with a common set of parameters.  In this case
> UnconfiguredScenario is a bunch of tests about the behaviour of pasta
> without --config-net.  Each of those tests has access to the host and
> guest sites, the expected interface name, address and gateway in the
> guest - that is, the contents of an UncofiguredScenario instance.

I'm not sure if I understand this correctly, but if each guest has a
single interface, that sounds a bit limiting.

Actually, I think any abstraction that doesn't offer arbitrary sets of
(and relationships between) the objects shown via netlink (or, at
least, namespaces, links, routes, addresses, neighbours) might be
limiting and not generic enough.

> That instance describes a real (simulated) environment in which we can
> run those tests.
> 
> You use this by supplying a function which sets things up, then yields
> an UnconfiguredScenario instance describing what it set up.  exeter
> will run all of the UnconfiguredScenario tests on the environment the
> setup function created, each one as a separate test case.

This part is now clear to me, and I think it's not complicated to grasp
the concept vaguely but enough to copy, paste, and modify code doing
this.

It would be even better to hide this entirely, because "yielding a
scenario" is a Python thing. In general, there's an imperative part in
all this (bordering functional programming, but still, not descriptive)
which I struggle to see as beneficial.

Here the tasks at hand are, roughly:

1. represent two network namespaces, with two interfaces each (loopback
   and non-loopback), with pasta connecting one of the interfaces of the
   inner one

2. bring up one of the interfaces

3. compare addresses

...and doing 1. like that is simply not... intuitive, I think.

> Usually, there are multiple ways to set up a suitable enviroment:
> running pasta with an existing guest ns vs. pasta creating the ns is a
> simple example.  You can create different setup functions for each of
> those, and re-use all the tests in the Scenario against each of those
> setups.
> 
> > > +
> > > +    @tunbridge.ndp.NdpScenario.subscenario
> > > +    def test_ndp(self) -> tunbridge.ndp.NdpScenario:
> > > +        tunbridge.ip.ifup(self.guest, self.ifname)  
> > 
> > This raises the question of how much of tunbridge one needs to know to
> > be able to write a basic test. Why is ifup() in 'ip'? I thought it
> > would be more of a "link" thing.  
> 
> Finding misleading names is a big reason for seeking early feedback.
> There's kind of a reason for ifup to be in ip: it optionally takes IP
> addresses to configure on the interface.  But... there's no inherent
> reason it couldn't take other sorts of network address too, so I'll
> look into moving that into a "link" module or something like it.

I think sticking to netlink objects would make this a bit more
familiar, if possible.

> > I admit I haven't had time to browse tunbridge recently, I'm just
> > looking at this series right now.  
> 
> That's fine.  At some point it would be good to have you look at
> tunbridge too, but reading this series _without_ reading tunbridge is
> a very useful perspective at this stage.
> 
> >   
> > > +        return tunbridge.ndp.NdpScenario(client=self.guest,
> > > +                                         ifname=self.ifname,
> > > +                                         network=self.addr6.network,
> > > +                                         gateway=self.gw6)  
> > 
> > This makes sense to me.  
> 
> Ok, good.  The Scenario stuff might not be as impenetrable as I
> feared.

Here I was simply commenting on the fact that I intuitively understand
those arguments and how they belong to the scenario, not on the
Scenario abstraction itself, but in any case, yes, given a bit of time
and sufficient motivation, I don't think it's impenetrable either.

So, while at it, let me share my most substantial worry about all this
at the moment. While not impenetrable implies it's usable, I'm not sure
how much further that goes.

That's mostly fine if the only goal is to develop and run tests for
passt (and I say "mostly" because to run these tests as part of
automatic distribution testing you need to package them, and have
packages for many distributions, which is a bit difficult to justify if
you have a single usage, but let's set this aside for a moment).

Still, that single-goal perspective doesn't look sustainable to me.
That's the case for the current test suite, but it was never meant to
be a real "framework" or simulator or anything anybody would like to
use for anything else.

If I'm looking for a tool that lets me quickly set up a VXLAN tunnel
between two nodes and try to flip offloads on and off I think it's
unreasonable to expect I'll go for some Scenario abstraction on the
basis of being, after all... not impenetrable.

And this kind of stuff is a very recurrent need in Linux networking
development, in my experience, as well as an unsatisfied need in
testing of many related projects.

Of course, one pressing goal right now is to have a more structured way
to define tests for passt, and anything that lets us achieve that goal
with a reasonable amount of time and effort is welcome.

But not having an interface that lets people build a test tunnel
between two nodes in a couple of minutes of reading examples carries a
serious risk that this gets stuck "forever" to passt and its tests.

> > > +
> > > +
> > > +@UnconfiguredScenario.test
> > > +@contextlib.contextmanager
> > > +def simh_pasta_setup() -> Iterator[UnconfiguredScenario]:
> > > +    with (tunbridge.sample.simple_host('host') as simh,
> > > +          tunbridge.sample.isolated('guest', simh.site) as guest):
> > > +        assert simh.ip6 is not None
> > > +        assert simh.gw6_ll is not None
> > > +        with tasst.pasta.pasta(simh.site, guest):
> > > +            yield UnconfiguredScenario(host=simh.site,
> > > +                                       guest=guest,
> > > +                                       ifname=simh.ifname,
> > > +                                       addr6=simh.ip6,
> > > +                                       gw6=simh.gw6_ll)  
> > 
> > ...and this too.
> > 
> > But there's one thing I'm missing: if it's a network simulator, why do
> > you need to call a simple_host() method to *describe* the fact that you
> > have a host / site? That looks rather unexpected.
> > 
> > I mean, I would have expected a syntax, in pseudocode, expressing:
> > 
> > 1. x := node (properties such as a list of interfaces a, b, c)
> > 
> > 2. pasta implements/connects a
> > 
> > ...I think this is mostly embedded in the sample.simple_host() thing,
> > but I'm not sure how. Maybe it will become clearer once I actually look
> > into tunbridge, though.  
> 
> Right.  "simple_host" isn't just an arbitrary node, but a (small)
> predefined network topology: a node configured with a single default
> gateway (also simulated, albeit minimally) - that is, the "classic"
> pasta host.  The idea is that the tunbridge.sample module will have a
> bunch of such example networks - so far there's:
>  - isolated() (node with loopback only)
>  - back_to_back() (two nodes connected by a veth)
>  - simple_host()
> 
> Suggestions for better names welcome, as always.

I'm a bit worried by the mere fact that those example networks (and
they're all methods instead of some kind of grammar!) are needed.

Anyway, I don't find back_to_back() particularly descriptive (what
makes it not front-to-front?). Perhaps a more mundane "two_nodes()"
makes it more obvious (they won't be isolated, of course).

> > Of course, I'm trying to push away my bias coming from the fact I was,
> > several years ago, for kselftests, aiming at something like this
> > instead:
> > 
> > 	A veth B
> > 	x=$(addr A veth)
> > 	B ping -c1 $x
> > 	A $x vxlan B $(addr B veth)
> > 	...
> > 
> > (where 'veth', 'vxlan' were both reserved keywords). Maybe once
> > non-trivial links are implemented in tunbridge it will all become more
> > obvious.  
> 
> I think tunbridge is not dissimilar to this, though with functions
> rather than reserved words.

That's pretty much the whole difference I was trying to convey, though.
Syntax is not entirely irrelevant. Of course, it doesn't need to be
reserved words in arbitrary positions, but probably there are other
ways to consider.

> It's a bit hidden here, because we're
> using these pre-built chunks - I expect that would be the case for
> your system as well, once you get to complex enough setups that you
> want to re-use non-trivial pieces.
> 
> For example the guts of back_to_back() is:
> 
>     with isolated(f'{name}0', sb) as s0, \
>          isolated(f'{name}1', sb) as s1:
>         if0, if1 = f'veth{name}0', f'veth{name}1'
>         with veth.veth(s0, if0, s1, if1):
> 	    ...
> 
> There's more, but that's mostly about IP allocation (it optionally
> does that).
> 
> > > +
> > > +
> > > +if __name__ == '__main__':
> > > +    exeter.main()
> > > diff --git a/test/run b/test/run
> > > index 3872a56e..4f09d767 100755
> > > --- a/test/run
> > > +++ b/test/run
> > > @@ -43,8 +43,10 @@ KERNEL=${KERNEL:-"/boot/vmlinuz-$(uname -r)"}
> > >  
> > >  COMMIT="$(git log --oneline --no-decorate -1)"
> > >  
> > > -# Let exeter tests written in Python find their modules
> > > +# Let exeter tests written in Python find their modules and binaries to run
> > >  export PYTHONPATH=${BASEPATH}/exeter/py3:${BASEPATH}/tunbridge:${BASEPATH}
> > > +export PASTA=${PASTA:-${BASEPATH}/../pasta}
> > > +
> > >  
> > >  . lib/util
> > >  . lib/context
> > > @@ -75,8 +77,8 @@ run() {
> > >  	exeter build/build.py
> > >  	exeter build/static_checkers.sh
> > >  
> > > +	exeter pasta/ndp.py
> > >  	setup pasta
> > > -	test pasta/ndp
> > >  	test pasta/dhcp
> > >  	test pasta/tcp
> > >  	test pasta/udp
> > > diff --git a/test/tasst/__init__.py b/test/tasst/__init__.py
> > > index fd4fe9a8..f5386b3a 100644
> > > --- a/test/tasst/__init__.py
> > > +++ b/test/tasst/__init__.py
> > > @@ -8,3 +8,7 @@
> > >  #
> > >  # Copyright Red Hat
> > >  # Author: David Gibson <david@gibson.dropbear.id.au>
> > > +
> > > +from . import pasta
> > > +
> > > +__all__ = ['pasta']
> > > diff --git a/test/tasst/pasta.py b/test/tasst/pasta.py
> > > new file mode 100644
> > > index 00000000..91f59036
> > > --- /dev/null
> > > +++ b/test/tasst/pasta.py
> > > @@ -0,0 +1,40 @@
> > > +#! /usr/bin/env python3
> > > +#
> > > +# SPDX-License-Identifier: GPL-2.0-or-later
> > > +#
> > > +# TASST - Test A Simple Socket Transport
> > > +#
> > > +# test/tasst/pasta.py - Helpers for seeting up pasta instances
> > > +#
> > > +# Copyright Red Hat
> > > +# Author: David Gibson <david@gibson.dropbear.id.au>
> > > +
> > > +import contextlib
> > > +import os
> > > +from typing import Iterator
> > > +
> > > +import tunbridge
> > > +
> > > +
> > > +@contextlib.contextmanager
> > > +def pasta(host: tunbridge.Site, guest: tunbridge.Site, *opts: str) \
> > > +        -> Iterator[tunbridge.site.SiteProcess]:
> > > +    if tunbridge.unshare.parent(guest) is not host:
> > > +        raise ValueError("pasta guest must be a namespace under host site")
> > > +
> > > +    # This implies guest is a namespace site
> > > +    assert isinstance(guest, tunbridge.unshare.NsenterSite)
> > > +
> > > +    exe = os.environ['PASTA']
> > > +
> > > +    with host.tempdir() as piddir:
> > > +        pidfile = os.path.join(piddir, 'pasta.pid')
> > > +        cmd = [exe, '-f', '-P', pidfile] + list(opts) + [f'{guest.pid}']
> > > +        with host.bg(*cmd, stop=True) as pasta:
> > > +            # Wait for the PID file to be written
> > > +            pidstr = None
> > > +            while not pidstr:
> > > +                pidstr = host.readfile(pidfile, check=False)
> > > +            pid = int(pidstr)
> > > +            print(f'pasta started, host: {host}, guest: {guest}, pid: {pid}')
> > > +            yield pasta  
> > 
> > ...perhaps we could also print version and path.  
> 
> Path I can easily add.  Version would require an extra invocation of
> pasta, which I don't really want to do.

Ah, right, never mind. The path will be good enough for that.

> > This part also looks
> > quite readable and intuitive to me without having looked into tunbridge
> > recently.  
> 
> Ok, that's promising.

I mean, I think it's all usable for the moment, and perhaps a starting
point for some other kind of... front-end? I'm not sure. As I mentioned
I'm a bit worried about the potential for universal intuitiveness and
usability.

-- 
Stefano