netdev - Re: [PATCH net-next] net: netdevsim: Support setting dev->perm

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <20250204085624.39b0dc69@kernel.org>
Date: Tue, 4 Feb 2025 08:56:24 -0800
From: Jakub Kicinski <kuba@...nel.org>
To: Toke Høiland-Jørgensen <toke@...hat.com>
Cc: Andrew Lunn <andrew+netdev@...n.ch>, "David S. Miller"
 <davem@...emloft.net>, Eric Dumazet <edumazet@...gle.com>, Paolo Abeni
 <pabeni@...hat.com>, netdev@...r.kernel.org
Subject: Re: [PATCH net-next] net: netdevsim: Support setting dev->perm_addr

On Tue, 04 Feb 2025 12:20:56 +0100 Toke Høiland-Jørgensen wrote:
> > netdevsim is not for user space testing. We have gone down the path
> > of supporting random features in it already, and then wasted time trying
> > to maintain them thru various devlink related perturbations, just to
> > find out that the features weren't actually used any more.
> >
> > NetworkManager can do the HW testing using virtme-ng.  
> 
> Sorry if I'm being dense, but how would that work? What device type
> would one create inside a virtme-ng environment that would have a
> perm_addr set?

virtme-ng is just a qemu wrapper. Qemu supports a bunch of emulated HW.

> > If you want to go down the netdevsim path you must provide a meaningful 
> > in-tree test, but let's be clear that we will 100% delete both the test
> > and the netdevsim functionality if it causes any issues.  
> 
> Can certainly add a test case, sure! Any preference for where to put it?
> Somewhere in selftests/net, I guess, but where? rtnetlink.sh and
> bpf_offload.py seem to be the only files currently doing anything with
> netdevsim. I could add a case to the former?

No preference, just an emphasis on _meaningful_.

Kernel supports loading OOT modules, too. I really don't want us
to be in the business of carrying test harnesses for random pieces
of user space code.