[PATCH] kthread: fix use-after-free if kthread fork fails

[PATCH] kthread: fix use-after-free if kthread fork fails

[Vegard Nossum]

If a kthread forks (e.g. usermodehelper since commit 1da5c46fa965) but
fails in copy_process() between calling dup_task_struct() and setting
p->set_child_tid, then the value of p->set_child_tid will be inherited
from the parent and get prematurely freed by free_kthread_struct().

    kthread()
     - worker_thread()
        - process_one_work()
        |  - call_usermodehelper_exec_work()
        |     - kernel_thread()
        |        - _do_fork()
        |           - copy_process()
        |              - dup_task_struct()
        |                 - arch_dup_task_struct()
        |                    - tsk->set_child_tid = current->set_child_tid // implied
        |              - ...
        |              - goto bad_fork_*
        |              - ...
        |              - free_task(tsk)
        |                 - free_kthread_struct(tsk)
        |                    - kfree(tsk->set_child_tid)
        - ...
        - schedule()
           - __schedule()
              - wq_worker_sleeping()
                 - kthread_data(task)->flags // UAF

The problem started showing up with commit 1da5c46fa965 since it reused
->set_child_tid for the kthread worker data.

A better long-term solution might be to get rid of the ->set_child_tid
abuse. The comment in set_kthread_struct() also looks slightly wrong.

Fixes: 1da5c46fa965ff90f5ffc080b6ab3fae5e227bc3 ("kthread: Make struct kthread kmalloc'ed")
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Andy Lutomirski <luto@kernel.org>
Debugged-by: Jamie Iles <jamie.iles@oracle.com>
Signed-off-by: Vegard Nossum <vegard.nossum@oracle.com>
---
 kernel/fork.c | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/kernel/fork.c b/kernel/fork.c
index dd5a371c392a..fbdc29365b83 100644
--- a/kernel/fork.c
+++ b/kernel/fork.c
@@ -518,6 +518,13 @@ static struct task_struct *dup_task_struct(struct task_struct *orig, int node)
 	atomic_set(&tsk->stack_refcount, 1);
 #endif
 
+	/*
+	 * Forking kthreads (e.g. usermodehelper) should not inherit this
+	 * field since it's a pointer to a 'struct kthread' which is not
+	 * reference counted.
+	 */
+	tsk->set_child_tid = NULL;
+
 	if (err)
 		goto free_stack;
 
-- 
2.12.0.rc0

Re: [PATCH] kthread: fix use-after-free if kthread fork fails

[Oleg Nesterov]

On 05/05, Vegard Nossum wrote:
>
> If a kthread forks (e.g. usermodehelper since commit 1da5c46fa965) but
> fails in copy_process() between calling dup_task_struct() and setting
> p->set_child_tid, then the value of p->set_child_tid will be inherited
> from the parent and get prematurely freed by free_kthread_struct().

Aaah... thanks!

> --- a/kernel/fork.c
> +++ b/kernel/fork.c
> @@ -518,6 +518,13 @@ static struct task_struct *dup_task_struct(struct task_struct *orig, int node)
>  	atomic_set(&tsk->stack_refcount, 1);
>  #endif
>  
> +	/*
> +	 * Forking kthreads (e.g. usermodehelper) should not inherit this
> +	 * field since it's a pointer to a 'struct kthread' which is not
> +	 * reference counted.
> +	 */
> +	tsk->set_child_tid = NULL;
> +

Can't we just move both

	p->set_child_tid = (clone_flags & CLONE_CHILD_SETTID) ? child_tidptr : NULL;
	/*
	 * Clear TID on mm_release()?
	 */
	p->clear_child_tid = (clone_flags & CLONE_CHILD_CLEARTID) ? child_tidptr : NULL;

lines here?

Oleg.

Re: [PATCH] kthread: fix use-after-free if kthread fork fails

[Vegard Nossum]

[-- Attachment #1: Type: text/plain, Size: 1505 bytes --]

On 05/05/17 18:44, Oleg Nesterov wrote:
> On 05/05, Vegard Nossum wrote:
>>
>> If a kthread forks (e.g. usermodehelper since commit 1da5c46fa965) but
>> fails in copy_process() between calling dup_task_struct() and setting
>> p->set_child_tid, then the value of p->set_child_tid will be inherited
>> from the parent and get prematurely freed by free_kthread_struct().
>
> Aaah... thanks!
>
>> --- a/kernel/fork.c
>> +++ b/kernel/fork.c
>> @@ -518,6 +518,13 @@ static struct task_struct *dup_task_struct(struct task_struct *orig, int node)
>>  	atomic_set(&tsk->stack_refcount, 1);
>>  #endif
>>
>> +	/*
>> +	 * Forking kthreads (e.g. usermodehelper) should not inherit this
>> +	 * field since it's a pointer to a 'struct kthread' which is not
>> +	 * reference counted.
>> +	 */
>> +	tsk->set_child_tid = NULL;
>> +
>
> Can't we just move both
>
> 	p->set_child_tid = (clone_flags & CLONE_CHILD_SETTID) ? child_tidptr : NULL;
> 	/*
> 	 * Clear TID on mm_release()?
> 	 */
> 	p->clear_child_tid = (clone_flags & CLONE_CHILD_CLEARTID) ? child_tidptr : NULL;
>
> lines here?

clone_flags is not available in dup_task_struct(), but we could move
those lines higher in copy_process(). The reason we didn't do it was
that we thought it was a little fragile/unobvious that this has to
happen before free_task() is called and that it was safer to clear it in
dup_task_struct() (which also contains zeroing of other fields).

The newly attached patch has been tested and seems to work, if you
prefer it.


Vegard

[-- Attachment #2: set_child_tid.patch --]
[-- Type: text/x-patch, Size: 1093 bytes --]

diff --git a/kernel/fork.c b/kernel/fork.c
index fbdc29365b83..c52e22fdf7ca 100644
--- a/kernel/fork.c
+++ b/kernel/fork.c
@@ -1561,6 +1561,18 @@ static __latent_entropy struct task_struct *copy_process(
 	if (!p)
 		goto fork_out;
 
+	/*
+	 * This _must_ happen before we call free_task(), i.e. before we jump
+	 * to any of the bad_fork_* labels. This is to avoid freeing
+	 * p->set_child_tid which is (ab)used as a kthread's data pointer for
+	 * kernel threads (PF_KTHREAD).
+	 */
+	p->set_child_tid = (clone_flags & CLONE_CHILD_SETTID) ? child_tidptr : NULL;
+	/*
+	 * Clear TID on mm_release()?
+	 */
+	p->clear_child_tid = (clone_flags & CLONE_CHILD_CLEARTID) ? child_tidptr : NULL;
+
 	ftrace_graph_init_task(p);
 
 	rt_mutex_init_task(p);
@@ -1727,11 +1739,6 @@ static __latent_entropy struct task_struct *copy_process(
 		}
 	}
 
-	p->set_child_tid = (clone_flags & CLONE_CHILD_SETTID) ? child_tidptr : NULL;
-	/*
-	 * Clear TID on mm_release()?
-	 */
-	p->clear_child_tid = (clone_flags & CLONE_CHILD_CLEARTID) ? child_tidptr : NULL;
 #ifdef CONFIG_BLOCK
 	p->plug = NULL;
 #endif

Re: [PATCH] kthread: fix use-after-free if kthread fork fails

[Oleg Nesterov]

On 05/05, Vegard Nossum wrote:
>
> On 05/05/17 18:44, Oleg Nesterov wrote:
> >
> >Can't we just move both
> >
> >	p->set_child_tid = (clone_flags & CLONE_CHILD_SETTID) ? child_tidptr : NULL;
> >	/*
> >	 * Clear TID on mm_release()?
> >	 */
> >	p->clear_child_tid = (clone_flags & CLONE_CHILD_CLEARTID) ? child_tidptr : NULL;
> >
> >lines here?
>
> clone_flags is not available in dup_task_struct(), but we could move
> those lines higher in copy_process().

Yes, yes, this is what I meant.

> The newly attached patch has been tested and seems to work, if you
> prefer it.

Yes, please, this loos a bit better simply because we do not need to set it twice.

And I agree this needs cleanups. Even if we forget about this particular problem
and the usage of set_child_tid, we should add copy_misc() which should absorb a
lot of chaotic initializations from copy_process() imo.

Oleg.