Re: [PULL] cpumask: finally make them variable size w/ CPUMASK_OFFSTACK.

[Rusty Russell <rusty@rustcorp.com.au>]

On Thu, 10 May 2012 09:42:15 +0200, Ingo Molnar <mingo@kernel.org> wrote:
> 
> * Rusty Russell <rusty@rustcorp.com.au> wrote:
> 
> > Mainly because I didn't want to disturb the archs which don't 
> > care at all about large cpumasks.  After all, putting a struct 
> > cpumask on the stack is pretty convenient.
> 
> Yes.
> 
> > But we could add a new arch config which removes it, and set 
> > it from x86.
> 
> Could we just use a single cpumask type, cpumask_t or so, which 
> would be the *only* generic method to use cpumasks?
> 
> (Current cpumask_t would move to cpumask_full_t.)
> 
> This would be the 'final' destiation for the cpumask code: the 
> natural type to use in new code is cpumask_t, while in special 
> cases we could use cpumask_full_t - but the name signals that 
> it's a potentially large structure.
> 
> On architectures that don't worry about large cpumasks (yet ...) 
> cpumask_t and cpumask_full_t maps to the same thing, so there's 
> no difference.
> 
> This would make things more natural IMO.
> 
> There would be no 'struct cpumask'. (and 'cpumask_var_t' would 
> disappear too due to the rename.)
> 
> Thoughts?

I don't understand, sorry.  I think I'd need some code to understand.

Unfortunately I was wrong about being able to remove struct cpumask's
definition when CONFIG_CPUMASK_OFFSTACK=n: we need it for cpumask_var_t
in that case :(

Cheers,
Rusty.