From: Brad Peters <bpeters@redhat.com>
Date: Tue, 5 Aug 2008 11:22:48 -0400
Subject: [x86] hugetlb: inconsistent get_user_pages (x86 piece)
Message-id: 48987048.1080703@redhat.com
O-Subject: Re: [PATCH RHEL5.3] inconsistent get_user_pages() and memory mapped - missing piece
Bugzilla: 456449
RH-Acked-by: Anton Arapov <aarapov@redhat.com>
RH-Acked-by: Rik van Riel <riel@redhat.com>
RH-Acked-by: David Howells <dhowells@redhat.com>

RHBZ#:
======
https://bugzilla.redhat.com/show_bug.cgi?id=456449

Description:
===========
Bug Fix / All archs

This is a backport piece that was missed from RHBZ 408781, which made
RHEL5.2.  While very unlikely to be hit under normal
usage, it is possible, and could result in a data integrity issue.  Hitting
this fault would involve a situation where either
a device driver is working with user space huge pages, or (theoretically),
with huge page direct IO.

RHEL Version Found:
================
RHEL 5.2

kABI Status:
============
No symbols were harmed.

Brew:
=====
Built on all platforms.
http://brewweb.devel.redhat.com/brew/taskinfo?taskID=1407248

Upstream Status:
================
Already upstream at:
http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=72fad7139b6829f71d7f41f39eb30da5760d90a8

Test Status:
============
One line simple fix - will test on beta unless required before

===============================================================

Brad Peters 1-978-392-1000 x 23183
IBM on-site partner.

Proposed Patch:
===============
This patch is based on 2.6.18-99.el5
-----------------

diff --git a/mm/hugetlb.c b/mm/hugetlb.c
index ebbffa0..e80a916 100644
--- a/mm/hugetlb.c
+++ b/mm/hugetlb.c
@@ -628,7 +628,8 @@ int follow_hugetlb_page(struct mm_struct *mm, struct vm_area_struct *vma,
 		 */
 		pte = huge_pte_offset(mm, vaddr & HPAGE_MASK);
 
-		if (!pte || huge_pte_none(huge_ptep_get(pte))) {
+		if (!pte || huge_pte_none(huge_ptep_get(pte)) ||
+		    (write && !pte_write(huge_ptep_get(pte)))) {
 			int ret;
 
 			spin_unlock(&mm->page_table_lock);