Re: [PATCH] kmalloc_percpu

[Andrew Morton]

Rusty Russell <[e-mail]> wrote:
>
> > I think the fixed size pool is perfectly reasonable.
> 
> Yes.  It's a tradeoff.  I think it's worth it at the moment (although
> I'll add a limited printk to __alloc_percpu if it fails).
> 

It's OK as long as nobody uses the feature!  Once it starts to be commonly
used (say, in driver ->open() methods) then we'll get into the same problems
as with vmalloc exhaustion, vmalloc fragmentation, large physically-contig
allocations, etc.

Ho-hum.  Can the magical constant become a __setup thing?