2015-06-24 10:24:06

by Kirill A. Shutemov

Subject: [PATCH] mm: fix status code move_pages() returns for zero page

Man page for move_pages(2) specifies that status code for zero page is
supposed to be -EFAULT. Currently kernel return -ENOENT in this case.

follow_page() can do it for us, if we would ask for FOLL_DUMP.

Signed-off-by: Kirill A. Shutemov <[email protected]>
Cc: Christoph Lameter <[email protected]>
Cc: Hugh Dickins <[email protected]>
---
mm/migrate.c | 18 ++++++------------
1 file changed, 6 insertions(+), 12 deletions(-)

diff --git a/mm/migrate.c b/mm/migrate.c
index 236ee25e79d9..d3529d620a5b 100644
--- a/mm/migrate.c
+++ b/mm/migrate.c
@@ -1222,7 +1222,9 @@ static int do_move_page_to_node_array(struct mm_struct *mm,
if (!vma || pp->addr < vma->vm_start || !vma_migratable(vma))
goto set_status;

- page = follow_page(vma, pp->addr, FOLL_GET|FOLL_SPLIT);
+ /* FOLL_DUMP to ignore special (like zero) pages */
+ page = follow_page(vma, pp->addr,
+ FOLL_GET | FOLL_SPLIT | FOLL_DUMP);

err = PTR_ERR(page);
if (IS_ERR(page))
@@ -1232,10 +1234,6 @@ static int do_move_page_to_node_array(struct mm_struct *mm,
if (!page)
goto set_status;

- /* Use PageReserved to check for zero page */
- if (PageReserved(page))
- goto put_and_set;
-
pp->page = page;
err = page_to_nid(page);

@@ -1392,18 +1390,14 @@ static void do_pages_stat_array(struct mm_struct *mm, unsigned long nr_pages,
if (!vma || addr < vma->vm_start)
goto set_status;

- page = follow_page(vma, addr, 0);
+ /* FOLL_DUMP to ignore special (like zero) pages */
+ page = follow_page(vma, addr, FOLL_DUMP);

err = PTR_ERR(page);
if (IS_ERR(page))
goto set_status;

- err = -ENOENT;
- /* Use PageReserved to check for zero page */
- if (!page || PageReserved(page))
- goto set_status;
-
- err = page_to_nid(page);
+ err = page ? page_to_nid(page) : -ENOENT;
set_status:
*status = err;

--
2.1.4

2015-06-29 14:59:07

by Christoph Lameter (Ampere)

Subject: Re: [PATCH] mm: fix status code move_pages() returns for zero page

On Wed, 24 Jun 2015, Kirill A. Shutemov wrote:

> Man page for move_pages(2) specifies that status code for zero page is
> supposed to be -EFAULT. Currently kernel return -ENOENT in this case.
>
> follow_page() can do it for us, if we would ask for FOLL_DUMP.

FOLL_DUMP also has the consequence that the upper layer page tables pages
are no longer allocated.

Otherwise this looks ok.

Reviewed-by: Christoph Lameter <[email protected]>