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 <[email protected]>
Acked-by: Michal Hocko <[email protected]>
---
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
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 <[email protected]>
> Acked-by: Michal Hocko <[email protected]>
Acked-by: Tejun Heo <[email protected]>
Thanks.
--
tejun
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 <[email protected]> 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 <[email protected]>
> Acked-by: Michal Hocko <[email protected]>
> ---
> 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
>
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 <[email protected]> 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 <[email protected]>
> > Acked-by: Michal Hocko <[email protected]>
> > ---
> > 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 <[email protected]>
SUSE Labs, CR
On Tue, Jun 29, 2021 at 12:32 AM Jan Kara <[email protected]> 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 <[email protected]> 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 <[email protected]>
> > > Acked-by: Michal Hocko <[email protected]>
> > > ---
> > > 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 <[email protected]>
> SUSE Labs, CR