diff mbox series

[bpf-next,13/15] mm, memcg: Add new helper get_obj_cgroup_from_cgroup

Message ID 20220810151840.16394-14-laoar.shao@gmail.com (mailing list archive)
State New
Headers show
Series bpf: Introduce selectable memcg for bpf map | expand

Commit Message

Yafang Shao Aug. 10, 2022, 3:18 p.m. UTC
Introduce new helper get_obj_cgroup_from_cgroup() to get obj_cgroup from
a specific cgroup.

Signed-off-by: Yafang Shao <laoar.shao@gmail.com>
---
 include/linux/memcontrol.h |  1 +
 mm/memcontrol.c            | 41 +++++++++++++++++++++++++++++++++++++++++
 2 files changed, 42 insertions(+)

Comments

Roman Gushchin Aug. 11, 2022, 4:16 p.m. UTC | #1
On Wed, Aug 10, 2022 at 03:18:38PM +0000, Yafang Shao wrote:
> Introduce new helper get_obj_cgroup_from_cgroup() to get obj_cgroup from
> a specific cgroup.
> 
> Signed-off-by: Yafang Shao <laoar.shao@gmail.com>
> ---
>  include/linux/memcontrol.h |  1 +
>  mm/memcontrol.c            | 41 +++++++++++++++++++++++++++++++++++++++++
>  2 files changed, 42 insertions(+)
> 
> diff --git a/include/linux/memcontrol.h b/include/linux/memcontrol.h
> index 2f0a611..901a921 100644
> --- a/include/linux/memcontrol.h
> +++ b/include/linux/memcontrol.h
> @@ -1713,6 +1713,7 @@ static inline void set_shrinker_bit(struct mem_cgroup *memcg,
>  int __memcg_kmem_charge_page(struct page *page, gfp_t gfp, int order);
>  void __memcg_kmem_uncharge_page(struct page *page, int order);
>  
> +struct obj_cgroup *get_obj_cgroup_from_cgroup(struct cgroup *cgrp);
>  struct obj_cgroup *get_obj_cgroup_from_current(void);
>  struct obj_cgroup *get_obj_cgroup_from_page(struct page *page);
>  
> diff --git a/mm/memcontrol.c b/mm/memcontrol.c
> index 618c366..762cffa 100644
> --- a/mm/memcontrol.c
> +++ b/mm/memcontrol.c
> @@ -2908,6 +2908,47 @@ static struct obj_cgroup *__get_obj_cgroup_from_memcg(struct mem_cgroup *memcg)
>  	return objcg;
>  }
>  
> +static struct obj_cgroup *get_obj_cgroup_from_memcg(struct mem_cgroup *memcg)
> +{
> +	struct obj_cgroup *objcg;
> +
> +	if (memcg_kmem_bypass())
> +		return NULL;
> +
> +	rcu_read_lock();
> +	objcg = __get_obj_cgroup_from_memcg(memcg);
> +	rcu_read_unlock();
> +	return objcg;

This code doesn't make sense to me. What does rcu read lock protect here?
Yafang Shao Aug. 12, 2022, 12:35 a.m. UTC | #2
On Fri, Aug 12, 2022 at 12:16 AM Roman Gushchin
<roman.gushchin@linux.dev> wrote:
>
> On Wed, Aug 10, 2022 at 03:18:38PM +0000, Yafang Shao wrote:
> > Introduce new helper get_obj_cgroup_from_cgroup() to get obj_cgroup from
> > a specific cgroup.
> >
> > Signed-off-by: Yafang Shao <laoar.shao@gmail.com>
> > ---
> >  include/linux/memcontrol.h |  1 +
> >  mm/memcontrol.c            | 41 +++++++++++++++++++++++++++++++++++++++++
> >  2 files changed, 42 insertions(+)
> >
> > diff --git a/include/linux/memcontrol.h b/include/linux/memcontrol.h
> > index 2f0a611..901a921 100644
> > --- a/include/linux/memcontrol.h
> > +++ b/include/linux/memcontrol.h
> > @@ -1713,6 +1713,7 @@ static inline void set_shrinker_bit(struct mem_cgroup *memcg,
> >  int __memcg_kmem_charge_page(struct page *page, gfp_t gfp, int order);
> >  void __memcg_kmem_uncharge_page(struct page *page, int order);
> >
> > +struct obj_cgroup *get_obj_cgroup_from_cgroup(struct cgroup *cgrp);
> >  struct obj_cgroup *get_obj_cgroup_from_current(void);
> >  struct obj_cgroup *get_obj_cgroup_from_page(struct page *page);
> >
> > diff --git a/mm/memcontrol.c b/mm/memcontrol.c
> > index 618c366..762cffa 100644
> > --- a/mm/memcontrol.c
> > +++ b/mm/memcontrol.c
> > @@ -2908,6 +2908,47 @@ static struct obj_cgroup *__get_obj_cgroup_from_memcg(struct mem_cgroup *memcg)
> >       return objcg;
> >  }
> >
> > +static struct obj_cgroup *get_obj_cgroup_from_memcg(struct mem_cgroup *memcg)
> > +{
> > +     struct obj_cgroup *objcg;
> > +
> > +     if (memcg_kmem_bypass())
> > +             return NULL;
> > +
> > +     rcu_read_lock();
> > +     objcg = __get_obj_cgroup_from_memcg(memcg);
> > +     rcu_read_unlock();
> > +     return objcg;
>
> This code doesn't make sense to me. What does rcu read lock protect here?

To protect rcu_dereference(memcg->objcg);.
Doesn't it need the read rcu lock ?
Shakeel Butt Aug. 12, 2022, 4:57 p.m. UTC | #3
On Wed, Aug 10, 2022 at 03:18:38PM +0000, Yafang Shao wrote:
> Introduce new helper get_obj_cgroup_from_cgroup() to get obj_cgroup from
> a specific cgroup.

Can you please add couple of lines on why you need objcg?

> 
> Signed-off-by: Yafang Shao <laoar.shao@gmail.com>
> ---
>  include/linux/memcontrol.h |  1 +
>  mm/memcontrol.c            | 41 +++++++++++++++++++++++++++++++++++++++++
>  2 files changed, 42 insertions(+)
> 
> diff --git a/include/linux/memcontrol.h b/include/linux/memcontrol.h
> index 2f0a611..901a921 100644
> --- a/include/linux/memcontrol.h
> +++ b/include/linux/memcontrol.h
> @@ -1713,6 +1713,7 @@ static inline void set_shrinker_bit(struct mem_cgroup *memcg,
>  int __memcg_kmem_charge_page(struct page *page, gfp_t gfp, int order);
>  void __memcg_kmem_uncharge_page(struct page *page, int order);
>  
> +struct obj_cgroup *get_obj_cgroup_from_cgroup(struct cgroup *cgrp);
>  struct obj_cgroup *get_obj_cgroup_from_current(void);
>  struct obj_cgroup *get_obj_cgroup_from_page(struct page *page);
>  
> diff --git a/mm/memcontrol.c b/mm/memcontrol.c
> index 618c366..762cffa 100644
> --- a/mm/memcontrol.c
> +++ b/mm/memcontrol.c
> @@ -2908,6 +2908,47 @@ static struct obj_cgroup *__get_obj_cgroup_from_memcg(struct mem_cgroup *memcg)
>  	return objcg;
>  }
>  
> +static struct obj_cgroup *get_obj_cgroup_from_memcg(struct mem_cgroup *memcg)
> +{
> +	struct obj_cgroup *objcg;
> +
> +	if (memcg_kmem_bypass())
> +		return NULL;
> +
> +	rcu_read_lock();
> +	objcg = __get_obj_cgroup_from_memcg(memcg);
> +	rcu_read_unlock();
> +	return objcg;
> +}
> +
> +struct obj_cgroup *get_obj_cgroup_from_cgroup(struct cgroup *cgrp)
> +{
> +	struct cgroup_subsys_state *css;
> +	struct mem_cgroup *memcg;
> +	struct obj_cgroup *objcg;
> +
> +	rcu_read_lock();
> +	css = rcu_dereference(cgrp->subsys[memory_cgrp_id]);
> +	if (!css || !css_tryget_online(css)) {
> +		rcu_read_unlock();
> +		cgroup_put(cgrp);
> +		return ERR_PTR(-EINVAL);
> +	}
> +	rcu_read_unlock();
> +	cgroup_put(cgrp);

The above put seems out of place and buggy.

> +
> +	memcg = mem_cgroup_from_css(css);
> +	if (!memcg) {
> +		css_put(css);
> +		return ERR_PTR(-EINVAL);
> +	}
> +
> +	objcg = get_obj_cgroup_from_memcg(memcg);
> +	css_put(css);
> +
> +	return objcg;
> +}
> +
>  __always_inline struct obj_cgroup *get_obj_cgroup_from_current(void)
>  {
>  	struct obj_cgroup *objcg = NULL;
> -- 
> 1.8.3.1
>
Roman Gushchin Aug. 12, 2022, 5:40 p.m. UTC | #4
On Fri, Aug 12, 2022 at 08:35:19AM +0800, Yafang Shao wrote:
> On Fri, Aug 12, 2022 at 12:16 AM Roman Gushchin
> <roman.gushchin@linux.dev> wrote:
> >
> > On Wed, Aug 10, 2022 at 03:18:38PM +0000, Yafang Shao wrote:
> > > Introduce new helper get_obj_cgroup_from_cgroup() to get obj_cgroup from
> > > a specific cgroup.
> > >
> > > Signed-off-by: Yafang Shao <laoar.shao@gmail.com>
> > > ---
> > >  include/linux/memcontrol.h |  1 +
> > >  mm/memcontrol.c            | 41 +++++++++++++++++++++++++++++++++++++++++
> > >  2 files changed, 42 insertions(+)
> > >
> > > diff --git a/include/linux/memcontrol.h b/include/linux/memcontrol.h
> > > index 2f0a611..901a921 100644
> > > --- a/include/linux/memcontrol.h
> > > +++ b/include/linux/memcontrol.h
> > > @@ -1713,6 +1713,7 @@ static inline void set_shrinker_bit(struct mem_cgroup *memcg,
> > >  int __memcg_kmem_charge_page(struct page *page, gfp_t gfp, int order);
> > >  void __memcg_kmem_uncharge_page(struct page *page, int order);
> > >
> > > +struct obj_cgroup *get_obj_cgroup_from_cgroup(struct cgroup *cgrp);
> > >  struct obj_cgroup *get_obj_cgroup_from_current(void);
> > >  struct obj_cgroup *get_obj_cgroup_from_page(struct page *page);
> > >
> > > diff --git a/mm/memcontrol.c b/mm/memcontrol.c
> > > index 618c366..762cffa 100644
> > > --- a/mm/memcontrol.c
> > > +++ b/mm/memcontrol.c
> > > @@ -2908,6 +2908,47 @@ static struct obj_cgroup *__get_obj_cgroup_from_memcg(struct mem_cgroup *memcg)
> > >       return objcg;
> > >  }
> > >
> > > +static struct obj_cgroup *get_obj_cgroup_from_memcg(struct mem_cgroup *memcg)
> > > +{
> > > +     struct obj_cgroup *objcg;
> > > +
> > > +     if (memcg_kmem_bypass())
> > > +             return NULL;
> > > +
> > > +     rcu_read_lock();
> > > +     objcg = __get_obj_cgroup_from_memcg(memcg);
> > > +     rcu_read_unlock();
> > > +     return objcg;
> >
> > This code doesn't make sense to me. What does rcu read lock protect here?
> 
> To protect rcu_dereference(memcg->objcg);.
> Doesn't it need the read rcu lock ?

No, it's not how rcu works. Please, take a look at the docs here:
https://docs.kernel.org/RCU/whatisRCU.html#whatisrcu .
In particular, it describes this specific case very well.

In 2 words, you don't protect the rcu_dereference() call, you protect the pointer
you get, cause it's valid only inside the rcu read section. After rcu_read_unlock()
it might point at a random data, because the protected object can be already freed.

Thanks!
Yafang Shao Aug. 12, 2022, 11:56 p.m. UTC | #5
On Sat, Aug 13, 2022 at 1:40 AM Roman Gushchin <roman.gushchin@linux.dev> wrote:
>
> On Fri, Aug 12, 2022 at 08:35:19AM +0800, Yafang Shao wrote:
> > On Fri, Aug 12, 2022 at 12:16 AM Roman Gushchin
> > <roman.gushchin@linux.dev> wrote:
> > >
> > > On Wed, Aug 10, 2022 at 03:18:38PM +0000, Yafang Shao wrote:
> > > > Introduce new helper get_obj_cgroup_from_cgroup() to get obj_cgroup from
> > > > a specific cgroup.
> > > >
> > > > Signed-off-by: Yafang Shao <laoar.shao@gmail.com>
> > > > ---
> > > >  include/linux/memcontrol.h |  1 +
> > > >  mm/memcontrol.c            | 41 +++++++++++++++++++++++++++++++++++++++++
> > > >  2 files changed, 42 insertions(+)
> > > >
> > > > diff --git a/include/linux/memcontrol.h b/include/linux/memcontrol.h
> > > > index 2f0a611..901a921 100644
> > > > --- a/include/linux/memcontrol.h
> > > > +++ b/include/linux/memcontrol.h
> > > > @@ -1713,6 +1713,7 @@ static inline void set_shrinker_bit(struct mem_cgroup *memcg,
> > > >  int __memcg_kmem_charge_page(struct page *page, gfp_t gfp, int order);
> > > >  void __memcg_kmem_uncharge_page(struct page *page, int order);
> > > >
> > > > +struct obj_cgroup *get_obj_cgroup_from_cgroup(struct cgroup *cgrp);
> > > >  struct obj_cgroup *get_obj_cgroup_from_current(void);
> > > >  struct obj_cgroup *get_obj_cgroup_from_page(struct page *page);
> > > >
> > > > diff --git a/mm/memcontrol.c b/mm/memcontrol.c
> > > > index 618c366..762cffa 100644
> > > > --- a/mm/memcontrol.c
> > > > +++ b/mm/memcontrol.c
> > > > @@ -2908,6 +2908,47 @@ static struct obj_cgroup *__get_obj_cgroup_from_memcg(struct mem_cgroup *memcg)
> > > >       return objcg;
> > > >  }
> > > >
> > > > +static struct obj_cgroup *get_obj_cgroup_from_memcg(struct mem_cgroup *memcg)
> > > > +{
> > > > +     struct obj_cgroup *objcg;
> > > > +
> > > > +     if (memcg_kmem_bypass())
> > > > +             return NULL;
> > > > +
> > > > +     rcu_read_lock();
> > > > +     objcg = __get_obj_cgroup_from_memcg(memcg);
> > > > +     rcu_read_unlock();
> > > > +     return objcg;
> > >
> > > This code doesn't make sense to me. What does rcu read lock protect here?
> >
> > To protect rcu_dereference(memcg->objcg);.
> > Doesn't it need the read rcu lock ?
>
> No, it's not how rcu works. Please, take a look at the docs here:
> https://docs.kernel.org/RCU/whatisRCU.html#whatisrcu .
> In particular, it describes this specific case very well.
>
> In 2 words, you don't protect the rcu_dereference() call, you protect the pointer

I just copied and pasted rcu_dereference(memcg->objcg) there to make it clear.
Actually it protects memcg->objcg, doesn't it ?

> you get, cause it's valid only inside the rcu read section. After rcu_read_unlock()
> it might point at a random data, because the protected object can be already freed.
>

Are you sure?
Can't the obj_cgroup_tryget(objcg) prevent it from being freed ?
Yafang Shao Aug. 13, 2022, 12:07 a.m. UTC | #6
On Sat, Aug 13, 2022 at 12:57 AM Shakeel Butt <shakeelb@google.com> wrote:
>
> On Wed, Aug 10, 2022 at 03:18:38PM +0000, Yafang Shao wrote:
> > Introduce new helper get_obj_cgroup_from_cgroup() to get obj_cgroup from
> > a specific cgroup.
>
> Can you please add couple of lines on why you need objcg?
>

Sure. will update in the next version.

> >
> > Signed-off-by: Yafang Shao <laoar.shao@gmail.com>
> > ---
> >  include/linux/memcontrol.h |  1 +
> >  mm/memcontrol.c            | 41 +++++++++++++++++++++++++++++++++++++++++
> >  2 files changed, 42 insertions(+)
> >
> > diff --git a/include/linux/memcontrol.h b/include/linux/memcontrol.h
> > index 2f0a611..901a921 100644
> > --- a/include/linux/memcontrol.h
> > +++ b/include/linux/memcontrol.h
> > @@ -1713,6 +1713,7 @@ static inline void set_shrinker_bit(struct mem_cgroup *memcg,
> >  int __memcg_kmem_charge_page(struct page *page, gfp_t gfp, int order);
> >  void __memcg_kmem_uncharge_page(struct page *page, int order);
> >
> > +struct obj_cgroup *get_obj_cgroup_from_cgroup(struct cgroup *cgrp);
> >  struct obj_cgroup *get_obj_cgroup_from_current(void);
> >  struct obj_cgroup *get_obj_cgroup_from_page(struct page *page);
> >
> > diff --git a/mm/memcontrol.c b/mm/memcontrol.c
> > index 618c366..762cffa 100644
> > --- a/mm/memcontrol.c
> > +++ b/mm/memcontrol.c
> > @@ -2908,6 +2908,47 @@ static struct obj_cgroup *__get_obj_cgroup_from_memcg(struct mem_cgroup *memcg)
> >       return objcg;
> >  }
> >
> > +static struct obj_cgroup *get_obj_cgroup_from_memcg(struct mem_cgroup *memcg)
> > +{
> > +     struct obj_cgroup *objcg;
> > +
> > +     if (memcg_kmem_bypass())
> > +             return NULL;
> > +
> > +     rcu_read_lock();
> > +     objcg = __get_obj_cgroup_from_memcg(memcg);
> > +     rcu_read_unlock();
> > +     return objcg;
> > +}
> > +
> > +struct obj_cgroup *get_obj_cgroup_from_cgroup(struct cgroup *cgrp)
> > +{
> > +     struct cgroup_subsys_state *css;
> > +     struct mem_cgroup *memcg;
> > +     struct obj_cgroup *objcg;
> > +
> > +     rcu_read_lock();
> > +     css = rcu_dereference(cgrp->subsys[memory_cgrp_id]);
> > +     if (!css || !css_tryget_online(css)) {
> > +             rcu_read_unlock();
> > +             cgroup_put(cgrp);
> > +             return ERR_PTR(-EINVAL);
> > +     }
> > +     rcu_read_unlock();
> > +     cgroup_put(cgrp);
>
> The above put seems out of place and buggy.
>

Thanks for pointing it out.
The cgroup_put should be used in bpf_map_save_memcg().
I will update it.
Roman Gushchin Aug. 13, 2022, 6:30 p.m. UTC | #7
On Sat, Aug 13, 2022 at 07:56:54AM +0800, Yafang Shao wrote:
> On Sat, Aug 13, 2022 at 1:40 AM Roman Gushchin <roman.gushchin@linux.dev> wrote:
> >
> > On Fri, Aug 12, 2022 at 08:35:19AM +0800, Yafang Shao wrote:
> > > On Fri, Aug 12, 2022 at 12:16 AM Roman Gushchin
> > > <roman.gushchin@linux.dev> wrote:
> > > >
> > > > On Wed, Aug 10, 2022 at 03:18:38PM +0000, Yafang Shao wrote:
> > > > > Introduce new helper get_obj_cgroup_from_cgroup() to get obj_cgroup from
> > > > > a specific cgroup.
> > > > >
> > > > > Signed-off-by: Yafang Shao <laoar.shao@gmail.com>
> > > > > ---
> > > > >  include/linux/memcontrol.h |  1 +
> > > > >  mm/memcontrol.c            | 41 +++++++++++++++++++++++++++++++++++++++++
> > > > >  2 files changed, 42 insertions(+)
> > > > >
> > > > > diff --git a/include/linux/memcontrol.h b/include/linux/memcontrol.h
> > > > > index 2f0a611..901a921 100644
> > > > > --- a/include/linux/memcontrol.h
> > > > > +++ b/include/linux/memcontrol.h
> > > > > @@ -1713,6 +1713,7 @@ static inline void set_shrinker_bit(struct mem_cgroup *memcg,
> > > > >  int __memcg_kmem_charge_page(struct page *page, gfp_t gfp, int order);
> > > > >  void __memcg_kmem_uncharge_page(struct page *page, int order);
> > > > >
> > > > > +struct obj_cgroup *get_obj_cgroup_from_cgroup(struct cgroup *cgrp);
> > > > >  struct obj_cgroup *get_obj_cgroup_from_current(void);
> > > > >  struct obj_cgroup *get_obj_cgroup_from_page(struct page *page);
> > > > >
> > > > > diff --git a/mm/memcontrol.c b/mm/memcontrol.c
> > > > > index 618c366..762cffa 100644
> > > > > --- a/mm/memcontrol.c
> > > > > +++ b/mm/memcontrol.c
> > > > > @@ -2908,6 +2908,47 @@ static struct obj_cgroup *__get_obj_cgroup_from_memcg(struct mem_cgroup *memcg)
> > > > >       return objcg;
> > > > >  }
> > > > >
> > > > > +static struct obj_cgroup *get_obj_cgroup_from_memcg(struct mem_cgroup *memcg)
> > > > > +{
> > > > > +     struct obj_cgroup *objcg;
> > > > > +
> > > > > +     if (memcg_kmem_bypass())
> > > > > +             return NULL;
> > > > > +
> > > > > +     rcu_read_lock();
> > > > > +     objcg = __get_obj_cgroup_from_memcg(memcg);
> > > > > +     rcu_read_unlock();
> > > > > +     return objcg;
> > > >
> > > > This code doesn't make sense to me. What does rcu read lock protect here?
> > >
> > > To protect rcu_dereference(memcg->objcg);.
> > > Doesn't it need the read rcu lock ?
> >
> > No, it's not how rcu works. Please, take a look at the docs here:
> > https://docs.kernel.org/RCU/whatisRCU.html#whatisrcu .
> > In particular, it describes this specific case very well.
> >
> > In 2 words, you don't protect the rcu_dereference() call, you protect the pointer
> 
> I just copied and pasted rcu_dereference(memcg->objcg) there to make it clear.
> Actually it protects memcg->objcg, doesn't it ?
> 
> > you get, cause it's valid only inside the rcu read section. After rcu_read_unlock()
> > it might point at a random data, because the protected object can be already freed.
> >
> 
> Are you sure?
> Can't the obj_cgroup_tryget(objcg) prevent it from being freed ?

Ok, now I see where it comes from. You copy-pasted it from get_obj_cgroup_from_current()?
There rcu read lock section protects memcg, not objcg.
In your case you don't need it, because memcg is passed as a parameter to the function,
so it's the duty of the caller to ensure the lifetime of memcg.

Thanks!
Yafang Shao Aug. 14, 2022, 2:35 a.m. UTC | #8
On Sun, Aug 14, 2022 at 2:30 AM Roman Gushchin <roman.gushchin@linux.dev> wrote:
>
> On Sat, Aug 13, 2022 at 07:56:54AM +0800, Yafang Shao wrote:
> > On Sat, Aug 13, 2022 at 1:40 AM Roman Gushchin <roman.gushchin@linux.dev> wrote:
> > >
> > > On Fri, Aug 12, 2022 at 08:35:19AM +0800, Yafang Shao wrote:
> > > > On Fri, Aug 12, 2022 at 12:16 AM Roman Gushchin
> > > > <roman.gushchin@linux.dev> wrote:
> > > > >
> > > > > On Wed, Aug 10, 2022 at 03:18:38PM +0000, Yafang Shao wrote:
> > > > > > Introduce new helper get_obj_cgroup_from_cgroup() to get obj_cgroup from
> > > > > > a specific cgroup.
> > > > > >
> > > > > > Signed-off-by: Yafang Shao <laoar.shao@gmail.com>
> > > > > > ---
> > > > > >  include/linux/memcontrol.h |  1 +
> > > > > >  mm/memcontrol.c            | 41 +++++++++++++++++++++++++++++++++++++++++
> > > > > >  2 files changed, 42 insertions(+)
> > > > > >
> > > > > > diff --git a/include/linux/memcontrol.h b/include/linux/memcontrol.h
> > > > > > index 2f0a611..901a921 100644
> > > > > > --- a/include/linux/memcontrol.h
> > > > > > +++ b/include/linux/memcontrol.h
> > > > > > @@ -1713,6 +1713,7 @@ static inline void set_shrinker_bit(struct mem_cgroup *memcg,
> > > > > >  int __memcg_kmem_charge_page(struct page *page, gfp_t gfp, int order);
> > > > > >  void __memcg_kmem_uncharge_page(struct page *page, int order);
> > > > > >
> > > > > > +struct obj_cgroup *get_obj_cgroup_from_cgroup(struct cgroup *cgrp);
> > > > > >  struct obj_cgroup *get_obj_cgroup_from_current(void);
> > > > > >  struct obj_cgroup *get_obj_cgroup_from_page(struct page *page);
> > > > > >
> > > > > > diff --git a/mm/memcontrol.c b/mm/memcontrol.c
> > > > > > index 618c366..762cffa 100644
> > > > > > --- a/mm/memcontrol.c
> > > > > > +++ b/mm/memcontrol.c
> > > > > > @@ -2908,6 +2908,47 @@ static struct obj_cgroup *__get_obj_cgroup_from_memcg(struct mem_cgroup *memcg)
> > > > > >       return objcg;
> > > > > >  }
> > > > > >
> > > > > > +static struct obj_cgroup *get_obj_cgroup_from_memcg(struct mem_cgroup *memcg)
> > > > > > +{
> > > > > > +     struct obj_cgroup *objcg;
> > > > > > +
> > > > > > +     if (memcg_kmem_bypass())
> > > > > > +             return NULL;
> > > > > > +
> > > > > > +     rcu_read_lock();
> > > > > > +     objcg = __get_obj_cgroup_from_memcg(memcg);
> > > > > > +     rcu_read_unlock();
> > > > > > +     return objcg;
> > > > >
> > > > > This code doesn't make sense to me. What does rcu read lock protect here?
> > > >
> > > > To protect rcu_dereference(memcg->objcg);.
> > > > Doesn't it need the read rcu lock ?
> > >
> > > No, it's not how rcu works. Please, take a look at the docs here:
> > > https://docs.kernel.org/RCU/whatisRCU.html#whatisrcu .
> > > In particular, it describes this specific case very well.
> > >
> > > In 2 words, you don't protect the rcu_dereference() call, you protect the pointer
> >
> > I just copied and pasted rcu_dereference(memcg->objcg) there to make it clear.
> > Actually it protects memcg->objcg, doesn't it ?
> >
> > > you get, cause it's valid only inside the rcu read section. After rcu_read_unlock()
> > > it might point at a random data, because the protected object can be already freed.
> > >
> >
> > Are you sure?
> > Can't the obj_cgroup_tryget(objcg) prevent it from being freed ?
>
> Ok, now I see where it comes from. You copy-pasted it from get_obj_cgroup_from_current()?
> There rcu read lock section protects memcg, not objcg.

Could you pls explain in detail why we should protect memcg instead of objcg ?
Why does the memcg need the read rcu lock ?

> In your case you don't need it, because memcg is passed as a parameter to the function,
> so it's the duty of the caller to ensure the lifetime of memcg.
>

I'm still a bit confused. See below,

objcg = rcu_dereference(memcg->objcg);
percpu_ref_tryget(&objcg->refcnt);    <<<< what if the objcg is freed
before this operation ??
diff mbox series

Patch

diff --git a/include/linux/memcontrol.h b/include/linux/memcontrol.h
index 2f0a611..901a921 100644
--- a/include/linux/memcontrol.h
+++ b/include/linux/memcontrol.h
@@ -1713,6 +1713,7 @@  static inline void set_shrinker_bit(struct mem_cgroup *memcg,
 int __memcg_kmem_charge_page(struct page *page, gfp_t gfp, int order);
 void __memcg_kmem_uncharge_page(struct page *page, int order);
 
+struct obj_cgroup *get_obj_cgroup_from_cgroup(struct cgroup *cgrp);
 struct obj_cgroup *get_obj_cgroup_from_current(void);
 struct obj_cgroup *get_obj_cgroup_from_page(struct page *page);
 
diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index 618c366..762cffa 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -2908,6 +2908,47 @@  static struct obj_cgroup *__get_obj_cgroup_from_memcg(struct mem_cgroup *memcg)
 	return objcg;
 }
 
+static struct obj_cgroup *get_obj_cgroup_from_memcg(struct mem_cgroup *memcg)
+{
+	struct obj_cgroup *objcg;
+
+	if (memcg_kmem_bypass())
+		return NULL;
+
+	rcu_read_lock();
+	objcg = __get_obj_cgroup_from_memcg(memcg);
+	rcu_read_unlock();
+	return objcg;
+}
+
+struct obj_cgroup *get_obj_cgroup_from_cgroup(struct cgroup *cgrp)
+{
+	struct cgroup_subsys_state *css;
+	struct mem_cgroup *memcg;
+	struct obj_cgroup *objcg;
+
+	rcu_read_lock();
+	css = rcu_dereference(cgrp->subsys[memory_cgrp_id]);
+	if (!css || !css_tryget_online(css)) {
+		rcu_read_unlock();
+		cgroup_put(cgrp);
+		return ERR_PTR(-EINVAL);
+	}
+	rcu_read_unlock();
+	cgroup_put(cgrp);
+
+	memcg = mem_cgroup_from_css(css);
+	if (!memcg) {
+		css_put(css);
+		return ERR_PTR(-EINVAL);
+	}
+
+	objcg = get_obj_cgroup_from_memcg(memcg);
+	css_put(css);
+
+	return objcg;
+}
+
 __always_inline struct obj_cgroup *get_obj_cgroup_from_current(void)
 {
 	struct obj_cgroup *objcg = NULL;