linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [PATCH v3] writeback: fix obtain a reference to a freeing memcg css
@ 2021-04-02  9:11 Muchun Song
  2021-04-04 18:01 ` Tejun Heo
  2021-05-20  3:45 ` Muchun Song
  0 siblings, 2 replies; 5+ messages in thread
From: Muchun Song @ 2021-04-02  9:11 UTC (permalink / raw)
  To: viro, tj, axboe, willy
  Cc: linux-fsdevel, linux-kernel, duanxiongchun, Muchun Song, Michal Hocko

The caller of wb_get_create() should pin the memcg, because
wb_get_create() relies on this guarantee. The rcu read lock
only can guarantee that the memcg css returned by css_from_id()
cannot be released, but the reference of the memcg can be zero.

  rcu_read_lock()
  memcg_css = css_from_id()
  wb_get_create(memcg_css)
      cgwb_create(memcg_css)
          // css_get can change the ref counter from 0 back to 1
          css_get(memcg_css)
  rcu_read_unlock()

Fix it by holding a reference to the css before calling
wb_get_create(). This is not a problem I encountered in the
real world. Just the result of a code review.

Fixes: 682aa8e1a6a1 ("writeback: implement unlocked_inode_to_wb transaction and use it for stat updates")
Signed-off-by: Muchun Song <songmuchun@bytedance.com>
Acked-by: Michal Hocko <mhocko@suse.com>
---
Changelog in v3:
 1. Do not change GFP_ATOMIC.
 2. Update commit log.

 Thanks for Michal's review and suggestions.

Changelog in v2:
 1. Replace GFP_ATOMIC with GFP_NOIO suggested by Matthew.


 fs/fs-writeback.c | 9 +++++++--
 1 file changed, 7 insertions(+), 2 deletions(-)

diff --git a/fs/fs-writeback.c b/fs/fs-writeback.c
index 3ac002561327..dedde99da40d 100644
--- a/fs/fs-writeback.c
+++ b/fs/fs-writeback.c
@@ -506,9 +506,14 @@ static void inode_switch_wbs(struct inode *inode, int new_wb_id)
 	/* find and pin the new wb */
 	rcu_read_lock();
 	memcg_css = css_from_id(new_wb_id, &memory_cgrp_subsys);
-	if (memcg_css)
-		isw->new_wb = wb_get_create(bdi, memcg_css, GFP_ATOMIC);
+	if (memcg_css && !css_tryget(memcg_css))
+		memcg_css = NULL;
 	rcu_read_unlock();
+	if (!memcg_css)
+		goto out_free;
+
+	isw->new_wb = wb_get_create(bdi, memcg_css, GFP_ATOMIC);
+	css_put(memcg_css);
 	if (!isw->new_wb)
 		goto out_free;
 
-- 
2.11.0


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

* Re: [PATCH v3] writeback: fix obtain a reference to a freeing memcg css
  2021-04-02  9:11 [PATCH v3] writeback: fix obtain a reference to a freeing memcg css Muchun Song
@ 2021-04-04 18:01 ` Tejun Heo
  2021-05-20  3:45 ` Muchun Song
  1 sibling, 0 replies; 5+ messages in thread
From: Tejun Heo @ 2021-04-04 18:01 UTC (permalink / raw)
  To: Muchun Song
  Cc: viro, axboe, willy, linux-fsdevel, linux-kernel, duanxiongchun,
	Michal Hocko

On Fri, Apr 02, 2021 at 05:11:45PM +0800, Muchun Song wrote:
> The caller of wb_get_create() should pin the memcg, because
> wb_get_create() relies on this guarantee. The rcu read lock
> only can guarantee that the memcg css returned by css_from_id()
> cannot be released, but the reference of the memcg can be zero.
> 
>   rcu_read_lock()
>   memcg_css = css_from_id()
>   wb_get_create(memcg_css)
>       cgwb_create(memcg_css)
>           // css_get can change the ref counter from 0 back to 1
>           css_get(memcg_css)
>   rcu_read_unlock()
> 
> Fix it by holding a reference to the css before calling
> wb_get_create(). This is not a problem I encountered in the
> real world. Just the result of a code review.
> 
> Fixes: 682aa8e1a6a1 ("writeback: implement unlocked_inode_to_wb transaction and use it for stat updates")
> Signed-off-by: Muchun Song <songmuchun@bytedance.com>
> Acked-by: Michal Hocko <mhocko@suse.com>

Acked-by: Tejun Heo <tj@kernel.org>

Thanks.

-- 
tejun

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

* Re: [PATCH v3] writeback: fix obtain a reference to a freeing memcg css
  2021-04-02  9:11 [PATCH v3] writeback: fix obtain a reference to a freeing memcg css Muchun Song
  2021-04-04 18:01 ` Tejun Heo
@ 2021-05-20  3:45 ` Muchun Song
  2021-06-28 16:32   ` Jan Kara
  1 sibling, 1 reply; 5+ messages in thread
From: Muchun Song @ 2021-05-20  3:45 UTC (permalink / raw)
  To: Alexander Viro, Tejun Heo, axboe, Matthew Wilcox
  Cc: linux-fsdevel, LKML, Xiongchun duan, Michal Hocko

Hi,

It seems like this patch has not been added to the linux-next
tree. Can anyone help with this? Thanks.

On Fri, Apr 2, 2021 at 5:13 PM Muchun Song <songmuchun@bytedance.com> wrote:
>
> The caller of wb_get_create() should pin the memcg, because
> wb_get_create() relies on this guarantee. The rcu read lock
> only can guarantee that the memcg css returned by css_from_id()
> cannot be released, but the reference of the memcg can be zero.
>
>   rcu_read_lock()
>   memcg_css = css_from_id()
>   wb_get_create(memcg_css)
>       cgwb_create(memcg_css)
>           // css_get can change the ref counter from 0 back to 1
>           css_get(memcg_css)
>   rcu_read_unlock()
>
> Fix it by holding a reference to the css before calling
> wb_get_create(). This is not a problem I encountered in the
> real world. Just the result of a code review.
>
> Fixes: 682aa8e1a6a1 ("writeback: implement unlocked_inode_to_wb transaction and use it for stat updates")
> Signed-off-by: Muchun Song <songmuchun@bytedance.com>
> Acked-by: Michal Hocko <mhocko@suse.com>
> ---
> Changelog in v3:
>  1. Do not change GFP_ATOMIC.
>  2. Update commit log.
>
>  Thanks for Michal's review and suggestions.
>
> Changelog in v2:
>  1. Replace GFP_ATOMIC with GFP_NOIO suggested by Matthew.
>
>
>  fs/fs-writeback.c | 9 +++++++--
>  1 file changed, 7 insertions(+), 2 deletions(-)
>
> diff --git a/fs/fs-writeback.c b/fs/fs-writeback.c
> index 3ac002561327..dedde99da40d 100644
> --- a/fs/fs-writeback.c
> +++ b/fs/fs-writeback.c
> @@ -506,9 +506,14 @@ static void inode_switch_wbs(struct inode *inode, int new_wb_id)
>         /* find and pin the new wb */
>         rcu_read_lock();
>         memcg_css = css_from_id(new_wb_id, &memory_cgrp_subsys);
> -       if (memcg_css)
> -               isw->new_wb = wb_get_create(bdi, memcg_css, GFP_ATOMIC);
> +       if (memcg_css && !css_tryget(memcg_css))
> +               memcg_css = NULL;
>         rcu_read_unlock();
> +       if (!memcg_css)
> +               goto out_free;
> +
> +       isw->new_wb = wb_get_create(bdi, memcg_css, GFP_ATOMIC);
> +       css_put(memcg_css);
>         if (!isw->new_wb)
>                 goto out_free;
>
> --
> 2.11.0
>

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

* Re: [PATCH v3] writeback: fix obtain a reference to a freeing memcg css
  2021-05-20  3:45 ` Muchun Song
@ 2021-06-28 16:32   ` Jan Kara
  2021-06-29  3:06     ` [External] " Muchun Song
  0 siblings, 1 reply; 5+ messages in thread
From: Jan Kara @ 2021-06-28 16:32 UTC (permalink / raw)
  To: Muchun Song
  Cc: Alexander Viro, Tejun Heo, axboe, Matthew Wilcox, linux-fsdevel,
	LKML, Xiongchun duan, Michal Hocko

Hi!

On Thu 20-05-21 11:45:23, Muchun Song wrote:
> It seems like this patch has not been added to the linux-next
> tree. Can anyone help with this? Thanks.

Muchun, did someone pickup this patch? I don't see it merged so unless
somebody yells, I'll pick it up to my tree and send it to Linus for rc2.

								Honza

> 
> On Fri, Apr 2, 2021 at 5:13 PM Muchun Song <songmuchun@bytedance.com> wrote:
> >
> > The caller of wb_get_create() should pin the memcg, because
> > wb_get_create() relies on this guarantee. The rcu read lock
> > only can guarantee that the memcg css returned by css_from_id()
> > cannot be released, but the reference of the memcg can be zero.
> >
> >   rcu_read_lock()
> >   memcg_css = css_from_id()
> >   wb_get_create(memcg_css)
> >       cgwb_create(memcg_css)
> >           // css_get can change the ref counter from 0 back to 1
> >           css_get(memcg_css)
> >   rcu_read_unlock()
> >
> > Fix it by holding a reference to the css before calling
> > wb_get_create(). This is not a problem I encountered in the
> > real world. Just the result of a code review.
> >
> > Fixes: 682aa8e1a6a1 ("writeback: implement unlocked_inode_to_wb transaction and use it for stat updates")
> > Signed-off-by: Muchun Song <songmuchun@bytedance.com>
> > Acked-by: Michal Hocko <mhocko@suse.com>
> > ---
> > Changelog in v3:
> >  1. Do not change GFP_ATOMIC.
> >  2. Update commit log.
> >
> >  Thanks for Michal's review and suggestions.
> >
> > Changelog in v2:
> >  1. Replace GFP_ATOMIC with GFP_NOIO suggested by Matthew.
> >
> >
> >  fs/fs-writeback.c | 9 +++++++--
> >  1 file changed, 7 insertions(+), 2 deletions(-)
> >
> > diff --git a/fs/fs-writeback.c b/fs/fs-writeback.c
> > index 3ac002561327..dedde99da40d 100644
> > --- a/fs/fs-writeback.c
> > +++ b/fs/fs-writeback.c
> > @@ -506,9 +506,14 @@ static void inode_switch_wbs(struct inode *inode, int new_wb_id)
> >         /* find and pin the new wb */
> >         rcu_read_lock();
> >         memcg_css = css_from_id(new_wb_id, &memory_cgrp_subsys);
> > -       if (memcg_css)
> > -               isw->new_wb = wb_get_create(bdi, memcg_css, GFP_ATOMIC);
> > +       if (memcg_css && !css_tryget(memcg_css))
> > +               memcg_css = NULL;
> >         rcu_read_unlock();
> > +       if (!memcg_css)
> > +               goto out_free;
> > +
> > +       isw->new_wb = wb_get_create(bdi, memcg_css, GFP_ATOMIC);
> > +       css_put(memcg_css);
> >         if (!isw->new_wb)
> >                 goto out_free;
> >
> > --
> > 2.11.0
> >
-- 
Jan Kara <jack@suse.com>
SUSE Labs, CR

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

* Re: [External] Re: [PATCH v3] writeback: fix obtain a reference to a freeing memcg css
  2021-06-28 16:32   ` Jan Kara
@ 2021-06-29  3:06     ` Muchun Song
  0 siblings, 0 replies; 5+ messages in thread
From: Muchun Song @ 2021-06-29  3:06 UTC (permalink / raw)
  To: Jan Kara
  Cc: Alexander Viro, Tejun Heo, axboe, Matthew Wilcox, linux-fsdevel,
	LKML, Xiongchun duan, Michal Hocko

On Tue, Jun 29, 2021 at 12:32 AM Jan Kara <jack@suse.cz> wrote:
>
> Hi!
>
> On Thu 20-05-21 11:45:23, Muchun Song wrote:
> > It seems like this patch has not been added to the linux-next
> > tree. Can anyone help with this? Thanks.
>
Hi,

> Muchun, did someone pickup this patch? I don't see it merged so unless

No, not yet.

> somebody yells, I'll pick it up to my tree and send it to Linus for rc2.

I'll be happy if you do that. Thanks.

>
>                                                                 Honza
>
> >
> > On Fri, Apr 2, 2021 at 5:13 PM Muchun Song <songmuchun@bytedance.com> wrote:
> > >
> > > The caller of wb_get_create() should pin the memcg, because
> > > wb_get_create() relies on this guarantee. The rcu read lock
> > > only can guarantee that the memcg css returned by css_from_id()
> > > cannot be released, but the reference of the memcg can be zero.
> > >
> > >   rcu_read_lock()
> > >   memcg_css = css_from_id()
> > >   wb_get_create(memcg_css)
> > >       cgwb_create(memcg_css)
> > >           // css_get can change the ref counter from 0 back to 1
> > >           css_get(memcg_css)
> > >   rcu_read_unlock()
> > >
> > > Fix it by holding a reference to the css before calling
> > > wb_get_create(). This is not a problem I encountered in the
> > > real world. Just the result of a code review.
> > >
> > > Fixes: 682aa8e1a6a1 ("writeback: implement unlocked_inode_to_wb transaction and use it for stat updates")
> > > Signed-off-by: Muchun Song <songmuchun@bytedance.com>
> > > Acked-by: Michal Hocko <mhocko@suse.com>
> > > ---
> > > Changelog in v3:
> > >  1. Do not change GFP_ATOMIC.
> > >  2. Update commit log.
> > >
> > >  Thanks for Michal's review and suggestions.
> > >
> > > Changelog in v2:
> > >  1. Replace GFP_ATOMIC with GFP_NOIO suggested by Matthew.
> > >
> > >
> > >  fs/fs-writeback.c | 9 +++++++--
> > >  1 file changed, 7 insertions(+), 2 deletions(-)
> > >
> > > diff --git a/fs/fs-writeback.c b/fs/fs-writeback.c
> > > index 3ac002561327..dedde99da40d 100644
> > > --- a/fs/fs-writeback.c
> > > +++ b/fs/fs-writeback.c
> > > @@ -506,9 +506,14 @@ static void inode_switch_wbs(struct inode *inode, int new_wb_id)
> > >         /* find and pin the new wb */
> > >         rcu_read_lock();
> > >         memcg_css = css_from_id(new_wb_id, &memory_cgrp_subsys);
> > > -       if (memcg_css)
> > > -               isw->new_wb = wb_get_create(bdi, memcg_css, GFP_ATOMIC);
> > > +       if (memcg_css && !css_tryget(memcg_css))
> > > +               memcg_css = NULL;
> > >         rcu_read_unlock();
> > > +       if (!memcg_css)
> > > +               goto out_free;
> > > +
> > > +       isw->new_wb = wb_get_create(bdi, memcg_css, GFP_ATOMIC);
> > > +       css_put(memcg_css);
> > >         if (!isw->new_wb)
> > >                 goto out_free;
> > >
> > > --
> > > 2.11.0
> > >
> --
> Jan Kara <jack@suse.com>
> SUSE Labs, CR

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

end of thread, other threads:[~2021-06-29  3:06 UTC | newest]

Thread overview: 5+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2021-04-02  9:11 [PATCH v3] writeback: fix obtain a reference to a freeing memcg css Muchun Song
2021-04-04 18:01 ` Tejun Heo
2021-05-20  3:45 ` Muchun Song
2021-06-28 16:32   ` Jan Kara
2021-06-29  3:06     ` [External] " Muchun Song

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).