[PATCH v2] proc: use ns_capable instead of capable for timerslack_ns

[PATCH v2] proc: use ns_capable instead of capable for timerslack_ns

[Benjamin Gordon]

Access to timerslack_ns is controlled by a process having CAP_SYS_NICE
in its effective capability set, but the current check looks in the root
namespace instead of the process' user namespace.  Since a process is
allowed to do other activities controlled by CAP_SYS_NICE inside a
namespace, it should also be able to adjust timerslack_ns.

Signed-off-by: Benjamin Gordon <bmgordon@google.com>
Cc: John Stultz <john.stultz@linaro.org>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: Kees Cook <keescook@chromium.org>
Cc: "Serge E. Hallyn" <serge@hallyn.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Arjan van de Ven <arjan@linux.intel.com>
Cc: Oren Laadan <orenl@cellrox.com>
Cc: Ruchi Kandoi <kandoiruchi@google.com>
Cc: Rom Lemarchand <romlem@android.com>
Cc: Todd Kjos <tkjos@google.com>
Cc: Colin Cross <ccross@android.com>
Cc: Nick Kralevich <nnk@google.com>
Cc: Dmitry Shmidt <dimitrysh@google.com>
Cc: Elliott Hughes <enh@google.com>
Cc: Android Kernel Team <kernel-team@android.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
---

Changes from v1:
  - Use the namespace of the target process instead of the file opener.
    Didn't carry over John Stultz' Acked-by since the changes aren't
    cosmetic.

 fs/proc/base.c | 12 +++++++++---
 1 file changed, 9 insertions(+), 3 deletions(-)

diff --git a/fs/proc/base.c b/fs/proc/base.c
index c78d8da09b52c..bdc093ba81dd3 100644
--- a/fs/proc/base.c
+++ b/fs/proc/base.c
@@ -2385,10 +2385,13 @@ static ssize_t timerslack_ns_write(struct file *file, const char __user *buf,
 		return -ESRCH;
 
 	if (p != current) {
-		if (!capable(CAP_SYS_NICE)) {
+		rcu_read_lock();
+		if (!ns_capable(__task_cred(p)->user_ns, CAP_SYS_NICE)) {
+			rcu_read_unlock();
 			count = -EPERM;
 			goto out;
 		}
+		rcu_read_unlock();
 
 		err = security_task_setscheduler(p);
 		if (err) {
@@ -2421,11 +2424,14 @@ static int timerslack_ns_show(struct seq_file *m, void *v)
 		return -ESRCH;
 
 	if (p != current) {
-
-		if (!capable(CAP_SYS_NICE)) {
+		rcu_read_lock();
+		if (!ns_capable(__task_cred(p)->user_ns, CAP_SYS_NICE)) {
+			rcu_read_unlock();
 			err = -EPERM;
 			goto out;
 		}
+		rcu_read_unlock();
+
 		err = security_task_getscheduler(p);
 		if (err)
 			goto out;
-- 
2.19.1.568.g152ad8e336-goog

Re: [PATCH v2] proc: use ns_capable instead of capable for timerslack_ns

[Eric W. Biederman]

Benjamin Gordon <bmgordon@google.com> writes:

> Access to timerslack_ns is controlled by a process having CAP_SYS_NICE
> in its effective capability set, but the current check looks in the root
> namespace instead of the process' user namespace.  Since a process is
> allowed to do other activities controlled by CAP_SYS_NICE inside a
> namespace, it should also be able to adjust timerslack_ns.

Acked-by: "Eric W. Biederman" <ebiederm@xmission.com>

I don't see any fundamental probess with how the processes user
namespace is being accessed.   You can race with setns
and that may result in a descendent user namespace of the current
user namespace being set.  But if you have permissions in the parent
user namespace you will have permissions over a child user namespace.
So the race can't effect the outcome of the ns_capable test.

That and while __task_cred(p) may change it is guaranteed there is a
valid one until __put_task_struct which only happens when a process has
a zero refcount.  Which the success of get_proc_task in before these
checks already ensures is not true.

So from my perspective this looks like a reasonable change.

I don't know how this looks from people who understand the timer bits
and what timerslack does. I suspect it is reasonable as there is no
permission check for changing yourself.

Eric

> Signed-off-by: Benjamin Gordon <bmgordon@google.com>
> Cc: John Stultz <john.stultz@linaro.org>
> Cc: "Eric W. Biederman" <ebiederm@xmission.com>
> Cc: Kees Cook <keescook@chromium.org>
> Cc: "Serge E. Hallyn" <serge@hallyn.com>
> Cc: Thomas Gleixner <tglx@linutronix.de>
> Cc: Arjan van de Ven <arjan@linux.intel.com>
> Cc: Oren Laadan <orenl@cellrox.com>
> Cc: Ruchi Kandoi <kandoiruchi@google.com>
> Cc: Rom Lemarchand <romlem@android.com>
> Cc: Todd Kjos <tkjos@google.com>
> Cc: Colin Cross <ccross@android.com>
> Cc: Nick Kralevich <nnk@google.com>
> Cc: Dmitry Shmidt <dimitrysh@google.com>
> Cc: Elliott Hughes <enh@google.com>
> Cc: Android Kernel Team <kernel-team@android.com>
> Cc: Andrew Morton <akpm@linux-foundation.org>
> ---
>
> Changes from v1:
>   - Use the namespace of the target process instead of the file opener.
>     Didn't carry over John Stultz' Acked-by since the changes aren't
>     cosmetic.
>
>  fs/proc/base.c | 12 +++++++++---
>  1 file changed, 9 insertions(+), 3 deletions(-)
>
> diff --git a/fs/proc/base.c b/fs/proc/base.c
> index c78d8da09b52c..bdc093ba81dd3 100644
> --- a/fs/proc/base.c
> +++ b/fs/proc/base.c
> @@ -2385,10 +2385,13 @@ static ssize_t timerslack_ns_write(struct file *file, const char __user *buf,
>  		return -ESRCH;
>  
>  	if (p != current) {
> -		if (!capable(CAP_SYS_NICE)) {
> +		rcu_read_lock();
> +		if (!ns_capable(__task_cred(p)->user_ns, CAP_SYS_NICE)) {
> +			rcu_read_unlock();
>  			count = -EPERM;
>  			goto out;
>  		}
> +		rcu_read_unlock();
>  
>  		err = security_task_setscheduler(p);
>  		if (err) {
> @@ -2421,11 +2424,14 @@ static int timerslack_ns_show(struct seq_file *m, void *v)
>  		return -ESRCH;
>  
>  	if (p != current) {
> -
> -		if (!capable(CAP_SYS_NICE)) {
> +		rcu_read_lock();
> +		if (!ns_capable(__task_cred(p)->user_ns, CAP_SYS_NICE)) {
> +			rcu_read_unlock();
>  			err = -EPERM;
>  			goto out;
>  		}
> +		rcu_read_unlock();
> +
>  		err = security_task_getscheduler(p);
>  		if (err)
>  			goto out;