Re: [PATCH] netdevsim: fix deadlock in nsim_bus_dev_max_vfs_write()

[Slawomir Stepien]

On sie 07, 2026 14:47, Jakub Kicinski wrote:
> On Fri, 7 Aug 2026 09:31:20 +0200 Slawomir Stepien wrote:
> > That's true, but is netdevsim used *just* by the selftests (or was
> > designed with only selftests in mind)? What if someone is using it
> > without selftests?
> 
> Quoting documentation:
> 
>   netdevsim
>   ~~~~~~~~~
>   
>   ``netdevsim`` is a test driver which can be used to exercise driver
>   configuration APIs without requiring capable hardware.
>   Mock-ups and tests based on ``netdevsim`` are encouraged when
>   adding new APIs with complex logic in the stack. The tests should
>   be written so that they can run both against ``netdevsim`` and a real
>   device (see ``tools/testing/selftests/drivers/net/README.rst``).
>   ``netdevsim``-only tests should focus on testing corner cases
>   and failure paths in the core which are hard to exercise with a real driver.
>   
>   ``netdevsim`` in itself is **not** considered
>   a use case/user. You must also implement the new APIs in a real driver.
>   
>   We give no guarantees that ``netdevsim`` won't change in the future
>   in a way which would break what would normally be considered uAPI.
>   
>   ``netdevsim`` is reserved for use by upstream tests only, so any
>   new ``netdevsim`` features must be accompanied by selftests under
>   ``tools/testing/selftests/``.
>   
> See: https://www.kernel.org/doc/html/next/process/maintainer-netdev.html#netdevsim
> 
> > On the other hand: what sashiko found:
> > https://netdev-ai.bots.linux.dev/sashiko/#/patchset/b7bf56ea-7522-4163-acd5-aaa69ad03b3a%40mail.kernel.org
> > is true: the same issue will be with e.g. break_health. So it seems
> > to me that a better approach would be to change when the debugfs
> > files are removed.
> 
> I seem to recall being annoyed at the fact that the health API takes
> devlink lock. It should be callable from IRQ even. Forcing drivers
> to worry about calling context is annoying for real drivers too.
> 
> > It seems to me that change in nsim_drv_remove() might be easy, but
> > what about nsim_dev_reload_down()...it seems it will have the same
> > deadlock. Or am I missing something for this case?
> 
> netdevsim is just a test mock. Fixing it for the sake of fixing
> netdevsim is a waste of everyone's time. The first question you should
> be asking yourself is "do I understand what this code was *designed
> for*"..

Thanks Jakub, this is now clear for me. I will create a patch that removes this not used part then.

-- 
Slawomir Stepien