Re: Lock the page

[Daniel Phillips]

On Friday 27 July 2001 04:55, [e-mail] wrote:
> After getting a page from alloc_pages(mask, order), is it necessary
> to lock the page to prevent it from being swapped out? Can I assume
> that I already have exclusive access to the page? Can I assume that
> the page is free and no other kernel control paths would use it?

You alloced it, it's yours.  You haven't entered it into any managed
structure like the page cache, it's not on a lru list so it won't be
scanned by reclaim, and no page table entry points at it (except for
the kernel's own linear mapping if it's not a highuser page) so
virtual scanning won't find it either.

--
Daniel