Re: [PATCH] proc: Fix timerslack_ns CAP_SYS_NICE check when adjusting self

[Kees Cook]

On Wed, Aug 10, 2016 at 2:12 PM, John Stultz <john.stultz@xxxxxxxxxx> wrote:
> On Wed, Aug 10, 2016 at 2:02 PM, Kees Cook <keescook@xxxxxxxxxxxx> wrote:
>> On Wed, Aug 10, 2016 at 11:36 AM, Kees Cook <keescook@xxxxxxxxxxxx> wrote:
>>> On Tue, Aug 9, 2016 at 4:54 PM, John Stultz <john.stultz@xxxxxxxxxx> wrote:
>>>> In changing from checking ptrace_may_access(p, PTRACE_MODE_ATTACH_FSCREDS)
>>>> to capable(CAP_SYS_NICE), I missed that ptrace_my_access succeeds
>>>> when p == current, but the CAP_SYS_NICE doesn't.
>>>>
>>>> Thus while the previous commit was intended to loosen the needed
>>>> privledges to modify a processes timerslack, it needlessly restricted
>>>> a task modifying its own timerslack via the proc/<tid>/timerslack_ns
>>>> (which is permitted also via the PR_SET_TIMERSLACK method).
>>>>
>>>> This patch corrects this by checking if p == current before checking
>>>> the CAP_SYS_NICE value.
>>>>
>>>> This patch applies on top of my two previous patches currently in -mm
>>>>
>>>> Cc: Kees Cook <keescook@xxxxxxxxxxxx>
>>>> Cc: "Serge E. Hallyn" <serge@xxxxxxxxxx>
>>>> Cc: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
>>>> Cc: Thomas Gleixner <tglx@xxxxxxxxxxxxx>
>>>> CC: Arjan van de Ven <arjan@xxxxxxxxxxxxxxx>
>>>> Cc: Oren Laadan <orenl@xxxxxxxxxxx>
>>>> Cc: Ruchi Kandoi <kandoiruchi@xxxxxxxxxx>
>>>> Cc: Rom Lemarchand <romlem@xxxxxxxxxxx>
>>>> Cc: Todd Kjos <tkjos@xxxxxxxxxx>
>>>> Cc: Colin Cross <ccross@xxxxxxxxxxx>
>>>> Cc: Nick Kralevich <nnk@xxxxxxxxxx>
>>>> Cc: Dmitry Shmidt <dimitrysh@xxxxxxxxxx>
>>>> Cc: Elliott Hughes <enh@xxxxxxxxxx>
>>>> Cc: Android Kernel Team <kernel-team@xxxxxxxxxxx>
>>>> Signed-off-by: John Stultz <john.stultz@xxxxxxxxxx>
>>>> ---
>>>>  fs/proc/base.c | 34 +++++++++++++++++++---------------
>>>>  1 file changed, 19 insertions(+), 15 deletions(-)
>>>>
>>>> diff --git a/fs/proc/base.c b/fs/proc/base.c
>>>> index 02f8389..01c3c2d 100644
>>>> --- a/fs/proc/base.c
>>>> +++ b/fs/proc/base.c
>>>> @@ -2281,15 +2281,17 @@ static ssize_t timerslack_ns_write(struct file *file, const char __user *buf,
>>>>         if (!p)
>>>>                 return -ESRCH;
>>>>
>>>> -       if (!capable(CAP_SYS_NICE)) {
>>>> -               count = -EPERM;
>>>> -               goto out;
>>>> -       }
>>>> +       if (p != current) {
>>>> +               if (!capable(CAP_SYS_NICE)) {
>>>> +                       count = -EPERM;
>>>> +                       goto out;
>>>> +               }
>>>>
>>>> -       err = security_task_setscheduler(p);
>>>> -       if (err) {
>>>> -               count = err;
>>>> -               goto out;
>>>> +               err = security_task_setscheduler(p);
>>>> +               if (err) {
>>>> +                       count = err;
>>>> +                       goto out;
>>>> +               }
>>>>         }
>>>
>>> This entirely bypasses LSM when p == current. Is that intended?
>>
>> I take back my concern. :) I think this is correct (as you mention in
>> the thread: the prctl LSM hook already fired), so until there is a
>
> But did it? The prctrl hook is just for the prctrl interface. The
> proc/<tid>/timerslack_ns is separate.

Oh, hrm, well, I think I'm still fine with it as-is: if we end up
needing to tighten this to block current, we can change it. Nick,
would this be something you'd want to be able to do with the hook?

-Kees

> This is part of my confusion here, mostly in that I'm not really sure
> I have a good sense of philosophy for LSM hooks.
> Are these just interface guards/hooks, or are we trying to map the
> hook to the underlying action being taken?
>
> As with the prctrl interface, it seems like its just an interface
> guard, but the /proc/<tid>/timerslack_ns interface checking
> security_task_setscheduler() seems to be more connected to the
> underlying action being done by changing the timerslack value.



-- 
Kees Cook
Nexus Security