Re: [patch 2.6.13-rc4] fix get_user_pages bug

[Nick Piggin <nickpiggin@yahoo.com.au>]

[-- Attachment #1: Type: text/plain, Size: 683 bytes --]

Hugh Dickins wrote:
> On Tue, 2 Aug 2005, Martin Schwidefsky wrote:

>>With the additional !pte_write(pte) check (and if I haven't overlooked
>>something which is not unlikely) s390 should work fine even without the
>>software-dirty bit hack.
> 
> 
> I agree the pte_write check ought to go back in next to the pte_dirty
> check, and that will leave s390 handling most uses of get_user_pages
> correctly, but still failing to handle the peculiar case of strace
> modifying a page to which the user does not currently have write access
> (e.g. setting a breakpoint in readonly text).
> 

Oh, here is the patch I sent Linus and forgot to CC
everyone else.

-- 
SUSE Labs, Novell Inc.


[-- Attachment #2: mm-opt-gup.patch --]
[-- Type: text/plain, Size: 1025 bytes --]

Allow __follow_page to succeed when encountering a clean, writeable
pte. Requires reintroduction of the direct page dirtying. Means
get_user_pages doesn't have to drop the page_table_lock and enter
the page fault handler for every clean, writeable pte it encounters
(when being called for write).

Signed-off-by: Nick Piggin <npiggin@suse.de>

Index: linux-2.6/mm/memory.c
===================================================================
--- linux-2.6.orig/mm/memory.c
+++ linux-2.6/mm/memory.c
@@ -811,15 +811,18 @@ static struct page *__follow_page(struct
 	pte = *ptep;
 	pte_unmap(ptep);
 	if (pte_present(pte)) {
-		if (write && !pte_dirty(pte))
+		if (write && !pte_write(pte) && !pte_dirty(pte))
 			goto out;
 		if (read && !pte_read(pte))
 			goto out;
 		pfn = pte_pfn(pte);
 		if (pfn_valid(pfn)) {
 			page = pfn_to_page(pfn);
-			if (accessed)
+			if (accessed) {
+				if (write && !pte_dirty(pte)&& !PageDirty(page))
+					set_page_dirty(page);
 				mark_page_accessed(page);
+			}
 			return page;
 		}
 	}