Re: [PATCH v1 3/4] mm: split folio_pte_batch() into folio_pte_batch() and folio_pte_batch_ext()

[Oscar Salvador]

On Wed, Jul 02, 2025 at 11:05:17AM +0200, David Hildenbrand wrote:
> Not necessarily inlined into the body (there might still be a function call,
> depending on what the compiler decides), but inlined into the object file
> and optimized by propagating constants.

I see.

> > With this change, a single function, folio_pte_batch(), identical to folio_pte_batch_ext
> > but without the runtime checks for those arguments will be created (folio_pte_batch()),
> > and so the users of it won't have it inlined in their body ?
> 
> Right. We have a single folio_pte_batch() that is optimized by propagating
> all constants. Instead of having one per object file, we have a single
> shared one.

Alright, clear to me now, thanks for claryfing ;-)!

 

-- 
Oscar Salvador
SUSE Labs