Re: Another possible use for LKMM, or a subset (strengthening) thereof

[Paul E. McKenney]

On Tue, Oct 19, 2021 at 07:56:35AM +0800, Boqun Feng wrote:
> Hi Paul,
> 
> On Mon, Oct 18, 2021 at 03:53:13PM -0700, Paul E. McKenney wrote:
> > On Thu, Oct 07, 2021 at 01:56:21PM -0700, Paul E. McKenney wrote:
> > > Hello!
> > > 
> > > On the perhaps unlikely chance that this is new news of interest...
> > > 
> > > I have finally prototyped the full "So You Want to Rust the Linux
> > > Kernel?" series (as in marked "under construction").
> > > 
> > > https://paulmck.livejournal.com/62436.html
> > 
> > And this blog series is now proclaimed to be feature complete.
> > 
> > Recommendations (both short- and long-term) may be found in the last post,
> > "TL;DR: Memory-Model Recommendations for Rusting the Linux Kernel",
> > at https://paulmck.livejournal.com/65341.html.
> 
> Thanks for putting this together! For the short-term recommendations, I
> think one practical goal would be having the equivalent (or stronger)
> litmus tests in Rust for the ones in tools/memory-model/litmus-tests.
> The translation of litmus tests may be trivial, but it at least ensure
> us that Rust can support the existing patterns widely used in Linux
> kernel. Of course, the Rust litmus tests don't have to be able to run
> with herd, we just need some code snippest to check our understanding of
> Rust memory model. ;-)

It would be very helpful for klitmus to be able to check Rust-code memory
ordering, now that you mention it!  This would be useful (for example)
to test the Rust wrappers on weakly ordered systems, such as ARM's.

> Besides, it's interesting to how things react with each if one function
> in the litmus test is in Rust and the other is in C ;-) Maybe this is a
> long-term goal.
> 
> Thoughts?

These issues are quite important.  How do you feel that they should be
tracked?

							Thanx, Paul