Re: [PATCH v9 2/9] crypto: Add ML-DSA/Dilithium verify support

[James Bottomley]

On Tue, 2025-11-18 at 08:39 +0000, David Howells wrote:
> James Bottomley <James.Bottomley@xxxxxxxxxxxxxxxxxxxxx> wrote:
> 
> > But even if you don't accept that, Google keeps effective joint
> > ownership of the code through their CLA and so could grant a dual
> > licence to the kernel anyway without needing to refer to any
> > contributors.
> 
> Actually, the fact that BoringSSL's ML-DSA implementation uses C++
> with heavy use of integer-parametered templating is more
> insurmountable for borrowing their code.  Yes, it does allow them to
> reduce their LoC to ~3000 and is much more readable, but I can't do
> that in C.

I was only commenting on the legality of copying not the technical
feasibility. 

>   Now, if we want to work on persuading Linus to allow C++ into the
> kernel...

Having worked on it, C++ is a bit of a messy language in that there are
many more idioms for the same construct than in C, so I can see why C
is preferred for cleaner coding.  On the other hand if the safety
profiles work actually produces something useful (unlike safe C++ which
just flamed out) I could see that being a reason to revisit.

Regards,

James