Re: [PATCH] ppc64: Fix possible race with set_pte on a present PTE

[Benjamin Herrenschmidt <benh@kernel.crashing.org>]

On Mon, 2004-05-24 at 14:36, Linus Torvalds wrote:
> On Mon, 24 May 2004, Benjamin Herrenschmidt wrote:
> > 
> > Typically, you can have a thread faulting on a page. It goes through hash_page,
> > doesn't find the entry, and gets to do_page_fault(). However, just before it
> > takes the mm sem, another thread actually mmap's that page in. Thus we end up
> > in handle_pte_fault() with a present PTE which has a valid mapping already.
> 
> Ok, with you so far. But I don't see how you get to set_pte() that way, 
> since handle_pte_fault() will re-test the pte_present() bit, and never 
> even try to set_pte() if one already exists. Hmm?

No, it doesn't. It tests it, if !present, it goes to do_no_page,
do_file_page or do_swap_page, but the,i if present, it does:

	entry = pte_mkyoung(entry);
	ptep_establish(vma, address, pte, entry);
	update_mmu_cache(vma, address, entry);
	pte_unmap(pte);
	spin_unlock(&mm->page_table_lock);
	return VM_FAULT_MINOR;

Which is, I think, the software PAGE_ACCESSED path used on some archs.

(ptep_establish does set_pte)

Ben.