Re: [RFC] virtual memmap for sparsemem [1/2] arch independent part

From: Christoph Lameter <clameter_at_sgi.com>
Date: 2006-10-20 11:42:24
On Fri, 20 Oct 2006, KAMEZAWA Hiroyuki wrote:

> Yes. but it seems to need per-arch implementation (in page fault handler).
> like this (from ia64)

If we have a statically assigned virtual memory area then this is not a 
big problem. With sharing the VMALLOC address space this may be a problem. 
I think a static address space is no problem for 64 bit platforms where 
we have lots of virtual address space. 32 bit platforms may have a dense 
address space where vmemmap is not needed. 
Maybe switch to a static address range range ? You saw my ia64 patch 
that did this right?


> Maybe extra optimization patch can be discussed after this generic code is settled.

Ok.

> > > +#ifdef CONFIG_VMEMMAP_SPARSEMEM
> > > +extern struct page *virt_memmap_start;
> > 
> > extern struct page[] would be better performance wise. Use the definitions 
> > for FLATMEM?
> Okay. will make it as array. or some constant value.

See my IA64 patchset for vmemmap static. We could define the mem_map 
address statically in the linker.

> > The virtual memmap has the potential of becoming the default for x86_64 
> > and many other platforms that already map memory. There is no performance 
> > difference between FLATMEM and this virtual memmap approach if there are 
> > already mappings in play.
> > 
> Hmm, adding CONFIG_HAVE_ARCH_LARGE_KERNEL_PAGE_MAPPING will be good ?
> We can add per-arch patches afterwards.

Great!
-
To unsubscribe from this list: send the line "unsubscribe linux-ia64" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Received on Fri Oct 20 11:42:41 2006

This archive was generated by hypermail 2.1.8 : 2006-10-20 11:42:52 EST