Re: [PATCH 1 of 4] mm: add a ptep_modify_prot transaction abstraction

[Jeremy Fitzhardinge]

Benjamin Herrenschmidt wrote:
> > Which architecture are you interested in?  If it isn't x86, you can 
> > probably get anything past Linus ;)
> >
> > I'll do some measurements to see what effect the batchable 
> > ptep_set_wrprotect() has on native.  If it's significant, I'll propose 
> > making it conditional on CONFIG_PARAVIRT.
> >     
>
> Oh, I mostly think about powerpc, I just wondered if I could re-use
> your new stuff in that context. Mostly idle thoughts at this stage, I
> haven't looked seriously.
>   

There are general-purpose hooks in the common code which architectures 
can implement however they like.  In x86-land we hook them up to pvops, 
but you can handle them however you want.

> I have an old patch set to batch forks (and mprotect etc...) TLB
> invalidations (which is what I really want to batch on powerpc, more
> than the actual PTE changes) that involves subtle changes to the
> batching mechanisms etc...
>   

Do you mean setting up batches of per-page tlb shootdowns rather than 
going a global tlb flush at the end?

   J
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/