blob: 0182b096a9e813ddf95bbb23ec3ce885941677c1 (
plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
|
From fa27f3e27863e3f4376864b144dc7506de124cf6 Mon Sep 17 00:00:00 2001
From: Jason Jin <Jason.jin@freescale.com>
Date: Thu, 4 Aug 2011 09:59:54 +0800
Subject: [PATCH 43/52] workaround for zero page used on ColdFire platform.
There is some issue for Coldfire platform to use the
zero page for read when there is memory fault. This is
a workaround for this, try to make kernel to use normal
memory instead.
Signed-off-by: Jason Jin <Jason.jin@freescale.com>
---
mm/memory.c | 2 ++
1 files changed, 2 insertions(+), 0 deletions(-)
--- a/mm/memory.c
+++ b/mm/memory.c
@@ -2937,6 +2937,7 @@ static int do_anonymous_page(struct mm_s
if (check_stack_guard_page(vma, address) < 0)
return VM_FAULT_SIGBUS;
+#ifndef CONFIG_COLDFIRE
/* Use the zero-page for reads */
if (!(flags & FAULT_FLAG_WRITE)) {
entry = pte_mkspecial(pfn_pte(my_zero_pfn(address),
@@ -2946,6 +2947,7 @@ static int do_anonymous_page(struct mm_s
goto unlock;
goto setpte;
}
+#endif
/* Allocate our own private page. */
if (unlikely(anon_vma_prepare(vma)))
|
