2022-02-15 20:06:43

by andrey.konovalov

Subject: [PATCH mm] fix for "kasan, fork: reset pointer tags of vmapped stacks"

From: Andrey Konovalov <[email protected]>

That patch didn't update the case when a stack is retrived from
cached_stacks in alloc_thread_stack_node(). As cached_stacks stores
vm_structs and not stack pointers themselves, the pointer tag needs
to be reset there as well.

Signed-off-by: Andrey Konovalov <[email protected]>
---
kernel/fork.c | 10 ++++++----
1 file changed, 6 insertions(+), 4 deletions(-)

diff --git a/kernel/fork.c b/kernel/fork.c
index 57d624f05182..5e3ad2e7a756 100644
--- a/kernel/fork.c
+++ b/kernel/fork.c
@@ -226,15 +226,17 @@ static unsigned long *alloc_thread_stack_node(struct task_struct *tsk, int node)
if (!s)
continue;

- /* Mark stack accessible for KASAN. */
+ /* Reset stack metadata. */
kasan_unpoison_range(s->addr, THREAD_SIZE);

+ stack = kasan_reset_tag(s->addr);
+
/* Clear stale pointers from reused stack. */
- memset(s->addr, 0, THREAD_SIZE);
+ memset(stack, 0, THREAD_SIZE);

tsk->stack_vm_area = s;
- tsk->stack = s->addr;
- return s->addr;
+ tsk->stack = stack;
+ return stack;
}

/*
--
2.25.1

2022-02-16 09:59:27

by Marco Elver

Subject: Re: [PATCH mm] fix for "kasan, fork: reset pointer tags of vmapped stacks"

On Tue, 15 Feb 2022 at 17:52, <[email protected]> wrote:
>
> From: Andrey Konovalov <[email protected]>
>
> That patch didn't update the case when a stack is retrived from
> cached_stacks in alloc_thread_stack_node(). As cached_stacks stores
> vm_structs and not stack pointers themselves, the pointer tag needs
> to be reset there as well.
>
> Signed-off-by: Andrey Konovalov <[email protected]>

Reviewed-by: Marco Elver <[email protected]>

Did the test catch this? If not, can this be tested?

> ---
> kernel/fork.c | 10 ++++++----
> 1 file changed, 6 insertions(+), 4 deletions(-)
>
> diff --git a/kernel/fork.c b/kernel/fork.c
> index 57d624f05182..5e3ad2e7a756 100644
> --- a/kernel/fork.c
> +++ b/kernel/fork.c
> @@ -226,15 +226,17 @@ static unsigned long *alloc_thread_stack_node(struct task_struct *tsk, int node)
> if (!s)
> continue;
>
> - /* Mark stack accessible for KASAN. */
> + /* Reset stack metadata. */
> kasan_unpoison_range(s->addr, THREAD_SIZE);
>
> + stack = kasan_reset_tag(s->addr);
> +
> /* Clear stale pointers from reused stack. */
> - memset(s->addr, 0, THREAD_SIZE);
> + memset(stack, 0, THREAD_SIZE);
>
> tsk->stack_vm_area = s;
> - tsk->stack = s->addr;
> - return s->addr;
> + tsk->stack = stack;
> + return stack;
> }
>
> /*
> --
> 2.25.1
>

2022-02-16 15:43:03

by Andrey Konovalov

Subject: Re: [PATCH mm] fix for "kasan, fork: reset pointer tags of vmapped stacks"

On Wed, Feb 16, 2022 at 10:59 AM Marco Elver <[email protected]> wrote:
>
> On Tue, 15 Feb 2022 at 17:52, <[email protected]> wrote:
> >
> > From: Andrey Konovalov <[email protected]>
> >
> > That patch didn't update the case when a stack is retrived from
> > cached_stacks in alloc_thread_stack_node(). As cached_stacks stores
> > vm_structs and not stack pointers themselves, the pointer tag needs
> > to be reset there as well.
> >
> > Signed-off-by: Andrey Konovalov <[email protected]>
>
> Reviewed-by: Marco Elver <[email protected]>
>
> Did the test catch this? If not, can this be tested?

Kind of, the kernel crashes on boot. I got KASAN_STACK accidentally
disabled in my SW_TAGS config, so I didn't see the crash until now.