All of lore.kernel.org
 help / color / mirror / Atom feed
* [PATCH] mm: do not free shared swap slots
@ 2019-10-22 11:21 Vinayak Menon
  2019-10-30  0:24 ` Minchan Kim
  0 siblings, 1 reply; 5+ messages in thread
From: Vinayak Menon @ 2019-10-22 11:21 UTC (permalink / raw)
  To: minchan, linux-mm; +Cc: kernel-team, Vinayak Menon, Michal Hocko

The following race is observed due to which a processes faulting
on a swap entry, finds the page neither in swapcache nor swap. This
causes zram to give a zero filled page that gets mapped to the
process, resulting in a user space crash later.

Consider parent and child processes Pa and Pb sharing the same swap
slot with swap_count 2. Swap is on zram with SWP_SYNCHRONOUS_IO set.
Virtual address 'VA' of Pa and Pb points to the shared swap entry.

Pa                                       Pb

fault on VA                              fault on VA
do_swap_page                             do_swap_page
lookup_swap_cache fails                  lookup_swap_cache fails
                                         Pb scheduled out
swapin_readahead (deletes zram entry)
swap_free (makes swap_count 1)
                                         Pb scheduled in
                                         swap_readpage (swap_count == 1)
                                         Takes SWP_SYNCHRONOUS_IO path
                                         zram enrty absent
                                         zram gives a zero filled page

Fix this by making sure that swap slot is freed only when swap count
drops down to one.

Signed-off-by: Vinayak Menon <vinmenon@codeaurora.org>
Suggested-by: Minchan Kim <minchan@google.com>
Cc: Michal Hocko <mhocko@suse.com>
---
 mm/page_io.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/mm/page_io.c b/mm/page_io.c
index 83db25a..3a198de 100644
--- a/mm/page_io.c
+++ b/mm/page_io.c
@@ -74,6 +74,7 @@ static void swap_slot_free_notify(struct page *page)
 {
 	struct swap_info_struct *sis;
 	struct gendisk *disk;
+	swp_entry_t entry;
 
 	/*
 	 * There is no guarantee that the page is in swap cache - the software
@@ -105,11 +106,10 @@ static void swap_slot_free_notify(struct page *page)
 	 * we again wish to reclaim it.
 	 */
 	disk = sis->bdev->bd_disk;
-	if (disk->fops->swap_slot_free_notify) {
-		swp_entry_t entry;
+	entry.val = page_private(page);
+	if (disk->fops->swap_slot_free_notify && __swap_count(entry) == 1) {
 		unsigned long offset;
 
-		entry.val = page_private(page);
 		offset = swp_offset(entry);
 
 		SetPageDirty(page);
-- 
QUALCOMM INDIA, on behalf of Qualcomm Innovation Center, Inc. is a
member of the Code Aurora Forum, hosted by The Linux Foundation



^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: [PATCH] mm: do not free shared swap slots
  2019-10-22 11:21 [PATCH] mm: do not free shared swap slots Vinayak Menon
@ 2019-10-30  0:24 ` Minchan Kim
  2019-10-30 10:24   ` Michal Hocko
  0 siblings, 1 reply; 5+ messages in thread
From: Minchan Kim @ 2019-10-30  0:24 UTC (permalink / raw)
  To: Vinayak Menon; +Cc: linux-mm, kernel-team, Michal Hocko

On Tue, Oct 22, 2019 at 04:51:34PM +0530, Vinayak Menon wrote:
> The following race is observed due to which a processes faulting
> on a swap entry, finds the page neither in swapcache nor swap. This
> causes zram to give a zero filled page that gets mapped to the
> process, resulting in a user space crash later.
> 
> Consider parent and child processes Pa and Pb sharing the same swap
> slot with swap_count 2. Swap is on zram with SWP_SYNCHRONOUS_IO set.
> Virtual address 'VA' of Pa and Pb points to the shared swap entry.
> 
> Pa                                       Pb
> 
> fault on VA                              fault on VA
> do_swap_page                             do_swap_page
> lookup_swap_cache fails                  lookup_swap_cache fails
>                                          Pb scheduled out
> swapin_readahead (deletes zram entry)
> swap_free (makes swap_count 1)
>                                          Pb scheduled in
>                                          swap_readpage (swap_count == 1)
>                                          Takes SWP_SYNCHRONOUS_IO path
>                                          zram enrty absent
>                                          zram gives a zero filled page
> 
> Fix this by making sure that swap slot is freed only when swap count
> drops down to one.
> 
> Signed-off-by: Vinayak Menon <vinmenon@codeaurora.org>
> Suggested-by: Minchan Kim <minchan@google.com>
> Cc: Michal Hocko <mhocko@suse.com>
Acked-by: Minchan Kim <minchan@kernel.org>

-stable material from v4.15.

Thanks.


^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: [PATCH] mm: do not free shared swap slots
  2019-10-30  0:24 ` Minchan Kim
@ 2019-10-30 10:24   ` Michal Hocko
  2019-10-31 14:46     ` Minchan Kim
  0 siblings, 1 reply; 5+ messages in thread
From: Michal Hocko @ 2019-10-30 10:24 UTC (permalink / raw)
  To: Minchan Kim; +Cc: Vinayak Menon, linux-mm, kernel-team

On Tue 29-10-19 17:24:47, Minchan Kim wrote:
> On Tue, Oct 22, 2019 at 04:51:34PM +0530, Vinayak Menon wrote:
> > The following race is observed due to which a processes faulting
> > on a swap entry, finds the page neither in swapcache nor swap. This
> > causes zram to give a zero filled page that gets mapped to the
> > process, resulting in a user space crash later.
> > 
> > Consider parent and child processes Pa and Pb sharing the same swap
> > slot with swap_count 2. Swap is on zram with SWP_SYNCHRONOUS_IO set.
> > Virtual address 'VA' of Pa and Pb points to the shared swap entry.
> > 
> > Pa                                       Pb
> > 
> > fault on VA                              fault on VA
> > do_swap_page                             do_swap_page
> > lookup_swap_cache fails                  lookup_swap_cache fails
> >                                          Pb scheduled out
> > swapin_readahead (deletes zram entry)
> > swap_free (makes swap_count 1)
> >                                          Pb scheduled in
> >                                          swap_readpage (swap_count == 1)
> >                                          Takes SWP_SYNCHRONOUS_IO path
> >                                          zram enrty absent
> >                                          zram gives a zero filled page
> > 
> > Fix this by making sure that swap slot is freed only when swap count
> > drops down to one.
> > 
> > Signed-off-by: Vinayak Menon <vinmenon@codeaurora.org>
> > Suggested-by: Minchan Kim <minchan@google.com>
> > Cc: Michal Hocko <mhocko@suse.com>
> Acked-by: Minchan Kim <minchan@kernel.org>
> 
> -stable material from v4.15.

Which commit does this fix? Fixes: tag would be highly appreciated.

-- 
Michal Hocko
SUSE Labs


^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: [PATCH] mm: do not free shared swap slots
  2019-10-30 10:24   ` Michal Hocko
@ 2019-10-31 14:46     ` Minchan Kim
  2019-10-31 17:15       ` Michal Hocko
  0 siblings, 1 reply; 5+ messages in thread
From: Minchan Kim @ 2019-10-31 14:46 UTC (permalink / raw)
  To: Michal Hocko; +Cc: Vinayak Menon, linux-mm, kernel-team

On Wed, Oct 30, 2019 at 11:24:09AM +0100, Michal Hocko wrote:
> On Tue 29-10-19 17:24:47, Minchan Kim wrote:
> > On Tue, Oct 22, 2019 at 04:51:34PM +0530, Vinayak Menon wrote:
> > > The following race is observed due to which a processes faulting
> > > on a swap entry, finds the page neither in swapcache nor swap. This
> > > causes zram to give a zero filled page that gets mapped to the
> > > process, resulting in a user space crash later.
> > > 
> > > Consider parent and child processes Pa and Pb sharing the same swap
> > > slot with swap_count 2. Swap is on zram with SWP_SYNCHRONOUS_IO set.
> > > Virtual address 'VA' of Pa and Pb points to the shared swap entry.
> > > 
> > > Pa                                       Pb
> > > 
> > > fault on VA                              fault on VA
> > > do_swap_page                             do_swap_page
> > > lookup_swap_cache fails                  lookup_swap_cache fails
> > >                                          Pb scheduled out
> > > swapin_readahead (deletes zram entry)
> > > swap_free (makes swap_count 1)
> > >                                          Pb scheduled in
> > >                                          swap_readpage (swap_count == 1)
> > >                                          Takes SWP_SYNCHRONOUS_IO path
> > >                                          zram enrty absent
> > >                                          zram gives a zero filled page
> > > 
> > > Fix this by making sure that swap slot is freed only when swap count
> > > drops down to one.
> > > 
> > > Signed-off-by: Vinayak Menon <vinmenon@codeaurora.org>
> > > Suggested-by: Minchan Kim <minchan@google.com>
> > > Cc: Michal Hocko <mhocko@suse.com>
> > Acked-by: Minchan Kim <minchan@kernel.org>
> > 
> > -stable material from v4.15.
> 
> Which commit does this fix? Fixes: tag would be highly appreciated.

Oops, I missed to mention it.
Fixes: aa8d22a11da9 (mm: swap: SWP_SYNCHRONOUS_IO: skip swapcache only if swapped page has no other reference)

> 
> -- 
> Michal Hocko
> SUSE Labs


^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: [PATCH] mm: do not free shared swap slots
  2019-10-31 14:46     ` Minchan Kim
@ 2019-10-31 17:15       ` Michal Hocko
  0 siblings, 0 replies; 5+ messages in thread
From: Michal Hocko @ 2019-10-31 17:15 UTC (permalink / raw)
  To: Minchan Kim; +Cc: Vinayak Menon, linux-mm, kernel-team

On Thu 31-10-19 07:46:08, Minchan Kim wrote:
> On Wed, Oct 30, 2019 at 11:24:09AM +0100, Michal Hocko wrote:
> > On Tue 29-10-19 17:24:47, Minchan Kim wrote:
> > > On Tue, Oct 22, 2019 at 04:51:34PM +0530, Vinayak Menon wrote:
> > > > The following race is observed due to which a processes faulting
> > > > on a swap entry, finds the page neither in swapcache nor swap. This
> > > > causes zram to give a zero filled page that gets mapped to the
> > > > process, resulting in a user space crash later.
> > > > 
> > > > Consider parent and child processes Pa and Pb sharing the same swap
> > > > slot with swap_count 2. Swap is on zram with SWP_SYNCHRONOUS_IO set.
> > > > Virtual address 'VA' of Pa and Pb points to the shared swap entry.
> > > > 
> > > > Pa                                       Pb
> > > > 
> > > > fault on VA                              fault on VA
> > > > do_swap_page                             do_swap_page
> > > > lookup_swap_cache fails                  lookup_swap_cache fails
> > > >                                          Pb scheduled out
> > > > swapin_readahead (deletes zram entry)
> > > > swap_free (makes swap_count 1)
> > > >                                          Pb scheduled in
> > > >                                          swap_readpage (swap_count == 1)
> > > >                                          Takes SWP_SYNCHRONOUS_IO path
> > > >                                          zram enrty absent
> > > >                                          zram gives a zero filled page
> > > > 
> > > > Fix this by making sure that swap slot is freed only when swap count
> > > > drops down to one.
> > > > 
> > > > Signed-off-by: Vinayak Menon <vinmenon@codeaurora.org>
> > > > Suggested-by: Minchan Kim <minchan@google.com>
> > > > Cc: Michal Hocko <mhocko@suse.com>
> > > Acked-by: Minchan Kim <minchan@kernel.org>
> > > 
> > > -stable material from v4.15.
> > 
> > Which commit does this fix? Fixes: tag would be highly appreciated.
> 
> Oops, I missed to mention it.
> Fixes: aa8d22a11da9 (mm: swap: SWP_SYNCHRONOUS_IO: skip swapcache only if swapped page has no other reference)

Thanks!

-- 
Michal Hocko
SUSE Labs


^ permalink raw reply	[flat|nested] 5+ messages in thread

end of thread, other threads:[~2019-10-31 17:15 UTC | newest]

Thread overview: 5+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2019-10-22 11:21 [PATCH] mm: do not free shared swap slots Vinayak Menon
2019-10-30  0:24 ` Minchan Kim
2019-10-30 10:24   ` Michal Hocko
2019-10-31 14:46     ` Minchan Kim
2019-10-31 17:15       ` Michal Hocko

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.