Re: [patch 1/2] x86_64 page fault NMI-safe

[Steven Rostedt <rostedt@goodmis.org>]

On Fri, 2010-07-16 at 12:47 +0200, Frederic Weisbecker wrote:
> > Thus the kernel maps the physical memory with the fewest TLB entries as
> > needed (large pages and large TLB entries). If we can map 64K pages, we
> > do that. Then kmalloc just allocates within this range, it does not need
> > to map any pages. They are already mapped.
> > 
> > Does this make a bit more sense?
> 
> 
> 
> Totally! You've made it very clear to me.
> Moreover I did not know we can have such variable page size. I mean I thought
> we can have variable page size but that would apply to every pages.

In x86_64, if bit 7 in the PDE (Page Directory Entry) is set then it
points to a 2 Meg page. Otherwise it points to a page table which will
have 512 PTE's pointing to 4K pages.

Download:

http://support.amd.com/us/Processor_TechDocs/24593.pdf

It has nice diagrams that explains this. Check out page 207 (fig 5-17)
and 210 (fig 5-22).

The phys_pmd_init() in arch/x86/mm/init_64.c will try to map memory
using 2M pages if it can, otherwise it falls back to 4K pages.

-- Steve