[PATCH] Fix end_of_stack() fn and location of stack canary for archs using STACK_GROWSUP

[PATCH] Fix end_of_stack() fn and location of stack canary for archs using STACK_GROWSUP

[Chuck Ebbert]

Aaron Tomlin recently posted patches [1] to enable checking the stack canary on
every task switch. Looking at the canary code, I realized that every arch
(except ia64, which adds some space for register spill above the stack) shares a
definition of end_of_stack() that makes it the first long after the threadinfo.

For stacks that grow down, this low address is correct because the stack starts
at the end of the thread area and grows toward lower addresses. However, for
stacks that grow up, toward higher addresses, this is wrong. (The stack actually
grows away from the canary.) On these archs end_of_stack() should return the 
address of the last long, at the highest possible address for the stack.

[1] http://lkml.org/lkml/2014/9/12/293

Signed-off-by: Chuck Ebbert <cebbert.lkml@gmail.com>

---

Compile tested only, with Aaron's patches applied and the new option
CONFIG_SCHED_STACK_END_CHECK they add enabled. I have no way to test
this any further.

diff a/include/linux/sched.h b/include/linux/sched.h
--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@ -2610,7 +2610,11 @@ static inline void setup_thread_stack(struct task_struct *p, struct task_struct
 
 static inline unsigned long *end_of_stack(struct task_struct *p)
 {
+#ifdef CONFIG_STACK_GROWSUP
+	return (unsigned long *)((unsigned long)task_thread_info(p) + THREAD_SIZE) - 1;
+#else
 	return (unsigned long *)(task_thread_info(p) + 1);
+#endif
 }
 
 #endif

Re: [PATCH] Fix end_of_stack() fn and location of stack canary for archs using STACK_GROWSUP

[James Hogan]

Hi Chuck,

On 16/09/14 08:37, Chuck Ebbert wrote:
>  static inline unsigned long *end_of_stack(struct task_struct *p)
>  {
> +#ifdef CONFIG_STACK_GROWSUP
> +	return (unsigned long *)((unsigned long)task_thread_info(p) + THREAD_SIZE) - 1;

Nit: this line should probably be wrapped to 80 columns.

Other than that, I've tested this on metag and can confirm that it fixes
the following BUG which you would otherwise get during boot with Aaron's
patches:

BUG: failure at kernel/sched/core.c:2664/schedule_debug()!
Kernel panic - not syncing: BUG!

Tested-by: James Hogan <james.hogan@imgtec.com> [metag]
Acked-by: James Hogan <james.hogan@imgtec.com>

Aaron: please can you try to get this patch applied before your patch
series.

Thanks
James

Re: [PATCH] Fix end_of_stack() fn and location of stack canary for archs using STACK_GROWSUP

[Aaron Tomlin]

On Fri, Sep 19, 2014 at 12:21:37PM +0100, James Hogan wrote:
> Hi Chuck,
> 
> On 16/09/14 08:37, Chuck Ebbert wrote:
> >  static inline unsigned long *end_of_stack(struct task_struct *p)
> >  {
> > +#ifdef CONFIG_STACK_GROWSUP
> > +	return (unsigned long *)((unsigned long)task_thread_info(p) + THREAD_SIZE) - 1;
> 
> Nit: this line should probably be wrapped to 80 columns.
> 
> Other than that, I've tested this on metag and can confirm that it fixes
> the following BUG which you would otherwise get during boot with Aaron's
> patches:
> 
> BUG: failure at kernel/sched/core.c:2664/schedule_debug()!
> Kernel panic - not syncing: BUG!
> 
> Tested-by: James Hogan <james.hogan@imgtec.com> [metag]
> Acked-by: James Hogan <james.hogan@imgtec.com>

OK.

Acked-by: Aaron Tomlin <atomlin@redhat.com>

> Aaron: please can you try to get this patch applied before your patch
> series.

Ingo,


I hope it's not too late to get this patch in (once the nit has been
addressed) for CONFIG_STACK_GROWSUP?

Regards,

-- 
Aaron Tomlin