Re: [PATCH] kernel/exit.c: pointer sighand could be uninitialized

Re: [PATCH] kernel/exit.c: pointer sighand could be uninitialized

[Kees Cook]

On Wed, May 2, 2018 at 6:48 PM, Yizhuo Zhai <yzhai003@ucr.edu> wrote:
> Variable 'sighand' could be uninitialized if probe_kernel_address fails
> (-EFAULT). The later use in the if statement may lead to undefined behavior.
>
> Signed-off-by: yzhai003@ucr.edu <yzhai003@ucr.edu>
> ---
>  kernel/exit.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/kernel/exit.c b/kernel/exit.c
> index f6cad39..a353bd1 100644
> --- a/kernel/exit.c
> +++ b/kernel/exit.c
> @@ -232,7 +232,7 @@ void release_task(struct task_struct *p)
>   */
>  struct task_struct *task_rcu_dereference(struct task_struct **ptask)
>  {
> - struct sighand_struct *sighand;
> + struct sighand_struct *sighand = NULL;
>   struct task_struct *task;
>
>   /*

Better would probably be to check the return of probe_kernel_address()
and take appropriate action...

-Kees

-- 
Kees Cook
Pixel Security

Re: [PATCH] kernel/exit.c: pointer sighand could be uninitialized

[Al Viro]

On Wed, May 02, 2018 at 06:48:57PM -0700, Yizhuo Zhai wrote:
> Variable 'sighand' could be uninitialized if probe_kernel_address fails
> (-EFAULT). The later use in the if statement may lead to undefined behavior.

Excuse me, but that's nonsense.  The value *copied* into it (in case
probe_kernel_address() has not failed) may be just as uninitialized.
If mere "compare uninitialized pointer value to NULL" can cause nasal demons to fly,
	* we are screwed anyway
	* the piece of crap compiler should be printed on sandpaper and used to
polish its authors.

Read the comments in there, please.  Especially the one regarding the second case.

Re: [PATCH] kernel/exit.c: pointer sighand could be uninitialized

[Chengyu Song]

Right. Should have checked the commit history ... this has been brought up before:

> Peter Zijlstra - May 18, 2016, 5:02 p.m.
> 
> > +struct task_struct *task_rcu_dereference(struct task_struct **ptask)
> > +{
> > +	struct task_struct *task;
> > +	struct sighand_struct *sighand;
> 
> I think that needs: ' = NULL',

and led to the comments added there.

Sorry for wasting your time.

-chengyu

> On May 2, 2018, at 10:06 PM, Al Viro <viro@zeniv.linux.org.uk> wrote:
> 
> On Wed, May 02, 2018 at 06:48:57PM -0700, Yizhuo Zhai wrote:
>> Variable 'sighand' could be uninitialized if probe_kernel_address fails
>> (-EFAULT). The later use in the if statement may lead to undefined behavior.
> 
> Excuse me, but that's nonsense.  The value *copied* into it (in case
> probe_kernel_address() has not failed) may be just as uninitialized.
> If mere "compare uninitialized pointer value to NULL" can cause nasal demons to fly,
> 	* we are screwed anyway
> 	* the piece of crap compiler should be printed on sandpaper and used to
> polish its authors.
> 
> Read the comments in there, please.  Especially the one regarding the second case.