[PATCH] mm/memcg: return value of the function mem_cgroup_from_css() is not checked

[PATCH] mm/memcg: return value of the function mem_cgroup_from_css() is not checked

[Yizhuo]

Inside function mem_cgroup_wb_domain(), the pointer memcg
could be NULL via mem_cgroup_from_css(). However, this pointer is
not checked and directly dereferenced in the if statement,
which is potentially unsafe.

Signed-off-by: Yizhuo <yzhai003@ucr.edu>
---
 mm/memcontrol.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index 661f046ad318..bd84bdaed3b0 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -3665,7 +3665,7 @@ struct wb_domain *mem_cgroup_wb_domain(struct bdi_writeback *wb)
 {
 	struct mem_cgroup *memcg = mem_cgroup_from_css(wb->memcg_css);
 
-	if (!memcg->css.parent)
+	if (!memcg || !memcg->css.parent)
 		return NULL;
 
 	return &memcg->cgwb_domain;
-- 
2.17.1

[PATCH] mm/memcg: return value of the function mem_cgroup_from_css() is not checked

[Yizhuo]

Inside function mem_cgroup_wb_domain(), the pointer memcg
could be NULL via mem_cgroup_from_css(). However, this pointer is
not checked and directly dereferenced in the if statement,
which is potentially unsafe.

Signed-off-by: Yizhuo <yzhai003@ucr.edu>
---
 mm/memcontrol.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index 661f046ad318..bd84bdaed3b0 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -3665,7 +3665,7 @@ struct wb_domain *mem_cgroup_wb_domain(struct bdi_writeback *wb)
 {
 	struct mem_cgroup *memcg = mem_cgroup_from_css(wb->memcg_css);
 
-	if (!memcg->css.parent)
+	if (!memcg || !memcg->css.parent)
 		return NULL;
 
 	return &memcg->cgwb_domain;
-- 
2.17.1

Re: [PATCH] mm/memcg: return value of the function mem_cgroup_from_css() is not checked

[Michal Hocko]

On Wed 21-08-19 23:22:09, Yizhuo wrote:
> Inside function mem_cgroup_wb_domain(), the pointer memcg
> could be NULL via mem_cgroup_from_css(). However, this pointer is
> not checked and directly dereferenced in the if statement,
> which is potentially unsafe.

Could you describe circumstances when this would happen? The code is
this way for 5 years without any issues. Are we just lucky or something
has changed recently to make this happen?
 
> Signed-off-by: Yizhuo <yzhai003@ucr.edu>
> ---
>  mm/memcontrol.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/mm/memcontrol.c b/mm/memcontrol.c
> index 661f046ad318..bd84bdaed3b0 100644
> --- a/mm/memcontrol.c
> +++ b/mm/memcontrol.c
> @@ -3665,7 +3665,7 @@ struct wb_domain *mem_cgroup_wb_domain(struct bdi_writeback *wb)
>  {
>  	struct mem_cgroup *memcg = mem_cgroup_from_css(wb->memcg_css);
>  
> -	if (!memcg->css.parent)
> +	if (!memcg || !memcg->css.parent)
>  		return NULL;
>  
>  	return &memcg->cgwb_domain;
> -- 
> 2.17.1
> 

-- 
Michal Hocko
SUSE Labs

Re: [PATCH] mm/memcg: return value of the function mem_cgroup_from_css() is not checked

[Yizhuo Zhai]

[-- Attachment #1: Type: text/plain, Size: 1522 bytes --]

This will happen if variable "wb->memcg_css" is NULL. This case is reported
by our analysis tool.
Since the function mem_cgroup_wb_domain() is visible to the global, we
cannot control caller's behavior.

On Thu, Aug 22, 2019 at 12:06 AM Michal Hocko <mhocko@kernel.org> wrote:

> On Wed 21-08-19 23:22:09, Yizhuo wrote:
> > Inside function mem_cgroup_wb_domain(), the pointer memcg
> > could be NULL via mem_cgroup_from_css(). However, this pointer is
> > not checked and directly dereferenced in the if statement,
> > which is potentially unsafe.
>
> Could you describe circumstances when this would happen? The code is
> this way for 5 years without any issues. Are we just lucky or something
> has changed recently to make this happen?
>
> > Signed-off-by: Yizhuo <yzhai003@ucr.edu>
> > ---
> >  mm/memcontrol.c | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/mm/memcontrol.c b/mm/memcontrol.c
> > index 661f046ad318..bd84bdaed3b0 100644
> > --- a/mm/memcontrol.c
> > +++ b/mm/memcontrol.c
> > @@ -3665,7 +3665,7 @@ struct wb_domain *mem_cgroup_wb_domain(struct
> bdi_writeback *wb)
> >  {
> >       struct mem_cgroup *memcg = mem_cgroup_from_css(wb->memcg_css);
> >
> > -     if (!memcg->css.parent)
> > +     if (!memcg || !memcg->css.parent)
> >               return NULL;
> >
> >       return &memcg->cgwb_domain;
> > --
> > 2.17.1
> >
>
> --
> Michal Hocko
> SUSE Labs
>


-- 
Kind Regards,

*Yizhuo Zhai*

*Computer Science, Graduate Student*
*University of California, Riverside *

[-- Attachment #2: Type: text/html, Size: 2588 bytes --]

Re: [PATCH] mm/memcg: return value of the function mem_cgroup_from_css() is not checked

[Michal Hocko]

On Thu 22-08-19 13:07:17, Yizhuo Zhai wrote:
> This will happen if variable "wb->memcg_css" is NULL. This case is reported
> by our analysis tool.

Does your tool report the particular call path and conditions when that
happen? Or is it just a "it mignt happen" kinda thing?

> Since the function mem_cgroup_wb_domain() is visible to the global, we
> cannot control caller's behavior.

I am sorry but I do not understand what is this supposed to mean.
-- 
Michal Hocko
SUSE Labs

Re: [PATCH] mm/memcg: return value of the function mem_cgroup_from_css() is not checked

[Yizhuo Zhai]

Our tool did not trace back the whole path, so, now we could say it
might happen.

On Thu, Aug 22, 2019 at 1:12 PM Michal Hocko <mhocko@kernel.org> wrote:
>
> On Thu 22-08-19 13:07:17, Yizhuo Zhai wrote:
> > This will happen if variable "wb->memcg_css" is NULL. This case is reported
> > by our analysis tool.
>
> Does your tool report the particular call path and conditions when that
> happen? Or is it just a "it mignt happen" kinda thing?
>
> > Since the function mem_cgroup_wb_domain() is visible to the global, we
> > cannot control caller's behavior.
>
> I am sorry but I do not understand what is this supposed to mean.
> --
> Michal Hocko
> SUSE Labs



-- 
Kind Regards,

Yizhuo Zhai

Computer Science, Graduate Student
University of California, Riverside

Re: [PATCH] mm/memcg: return value of the function mem_cgroup_from_css() is not checked

[Yizhuo Zhai]

Our tool did not trace back the whole path, so, now we could say it
might happen.

On Thu, Aug 22, 2019 at 1:12 PM Michal Hocko <mhocko@kernel.org> wrote:
>
> On Thu 22-08-19 13:07:17, Yizhuo Zhai wrote:
> > This will happen if variable "wb->memcg_css" is NULL. This case is reported
> > by our analysis tool.
>
> Does your tool report the particular call path and conditions when that
> happen? Or is it just a "it mignt happen" kinda thing?
>
> > Since the function mem_cgroup_wb_domain() is visible to the global, we
> > cannot control caller's behavior.
>
> I am sorry but I do not understand what is this supposed to mean.
> --
> Michal Hocko
> SUSE Labs



-- 
Kind Regards,

Yizhuo Zhai

Computer Science, Graduate Student
University of California, Riverside