Re: [RFC][PATCH 1/2] tracking dirty pages in shared mappings -V3

[Peter Zijlstra <a.p.zijlstra@chello.nl>]

On Mon, 2006-05-08 at 22:41 -0700, Christoph Lameter wrote:
> On Mon, 8 May 2006, Peter Zijlstra wrote:
> 
> > @@ -2077,6 +2078,7 @@ static int do_no_page(struct mm_struct *
> >  	unsigned int sequence = 0;
> >  	int ret = VM_FAULT_MINOR;
> >  	int anon = 0;
> > +	int dirty = 0;
> 	dirtied_page = NULL ?

Much nicer indeed!

> > @@ -2150,6 +2152,11 @@ retry:
> >  		entry = mk_pte(new_page, vma->vm_page_prot);
> >  		if (write_access)
> >  			entry = maybe_mkwrite(pte_mkdirty(entry), vma);
> 
> A write fault to a shared mapping does not make the page dirty, just the 
> pte?

We do that here:

> >  			inc_mm_counter(mm, file_rss);
> >  			page_add_file_rmap(new_page);
> > +			if (write_access) {
> > +				get_page(new_page);
> > +				dirty++;



> > +int page_wrprotect(struct page *page)
> 
> The above and related functions look similar to code in 
> rmap.c and migrate.c. Could those be consolidated?

I'll have a look.

Peter

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>