[v2] security: device_cgroup: fix RCU lockdep splat
diff mbox

Message ID 1441199548-29633-1-git-send-email-balbi@ti.com
State New
Headers show

Commit Message

Felipe Balbi Sept. 2, 2015, 1:12 p.m. UTC
while booting AM437x device, the following splat
triggered:

[   12.005238] ===============================
[   12.009749] [ INFO: suspicious RCU usage. ]
[   12.014116] 4.2.0-next-20150831 #1154 Not tainted
[   12.019050] -------------------------------
[   12.023408] security/device_cgroup.c:405 device_cgroup:verify_new_ex called without proper synchronization!
[   12.033576] other info that might help us debug this:

[   12.041942] rcu_scheduler_active = 1, debug_locks = 0
[   12.048796] 4 locks held by systemd/1:
[   12.052700]  #0:  (sb_writers#7){.+.+.+}, at: [<c017af84>] __sb_start_write+0x8c/0xb0
[   12.060954]  #1:  (&of->mutex){+.+.+.}, at: [<c01f1600>] kernfs_fop_write+0x50/0x1b8
[   12.069085]  #2:  (s_active#30){++++.+}, at: [<c01f1608>] kernfs_fop_write+0x58/0x1b8
[   12.077310]  #3:  (devcgroup_mutex){+.+...}, at: [<c0317bfc>] devcgroup_access_write+0x20/0x658
[   12.086575] stack backtrace:
[   12.091124] CPU: 0 PID: 1 Comm: systemd Not tainted 4.2.0-next-20150831 #1154
[   12.098609] Hardware name: Generic AM43 (Flattened Device Tree)
[   12.104807] [<c001770c>] (unwind_backtrace) from [<c0013a58>] (show_stack+0x10/0x14)
[   12.112924] [<c0013a58>] (show_stack) from [<c034f014>] (dump_stack+0x84/0x9c)
[   12.120491] [<c034f014>] (dump_stack) from [<c0317a04>] (verify_new_ex+0xc4/0xdc)
[   12.128326] [<c0317a04>] (verify_new_ex) from [<c0317f50>] (devcgroup_access_write+0x374/0x658)
[   12.137426] [<c0317f50>] (devcgroup_access_write) from [<c00d2800>] (cgroup_file_write+0x28/0x1bc)
[   12.146796] [<c00d2800>] (cgroup_file_write) from [<c01f1670>] (kernfs_fop_write+0xc0/0x1b8)
[   12.155620] [<c01f1670>] (kernfs_fop_write) from [<c0177c94>] (__vfs_write+0x1c/0xd8)
[   12.163783] [<c0177c94>] (__vfs_write) from [<c0178594>] (vfs_write+0x90/0x16c)
[   12.171426] [<c0178594>] (vfs_write) from [<c0178db4>] (SyS_write+0x44/0x9c)
[   12.178806] [<c0178db4>] (SyS_write) from [<c000f680>] (ret_fast_syscall+0x0/0x1c)

Fix it by making sure rcu_read_lock() is held
around calls to parent_has_perm().

Signed-off-by: Felipe Balbi <balbi@ti.com>
---

Changes since v1:
	- move rcu_read_lock/unlock to wrap parent_has_perm()

 security/device_cgroup.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

Comments

Josh Boyer Sept. 2, 2015, 3:27 p.m. UTC | #1
On Wed, Sep 2, 2015 at 9:12 AM, Felipe Balbi <balbi@ti.com> wrote:
> while booting AM437x device, the following splat
> triggered:
>
> [   12.005238] ===============================
> [   12.009749] [ INFO: suspicious RCU usage. ]
> [   12.014116] 4.2.0-next-20150831 #1154 Not tainted
> [   12.019050] -------------------------------
> [   12.023408] security/device_cgroup.c:405 device_cgroup:verify_new_ex called without proper synchronization!
> [   12.033576] other info that might help us debug this:
>
> [   12.041942] rcu_scheduler_active = 1, debug_locks = 0
> [   12.048796] 4 locks held by systemd/1:
> [   12.052700]  #0:  (sb_writers#7){.+.+.+}, at: [<c017af84>] __sb_start_write+0x8c/0xb0
> [   12.060954]  #1:  (&of->mutex){+.+.+.}, at: [<c01f1600>] kernfs_fop_write+0x50/0x1b8
> [   12.069085]  #2:  (s_active#30){++++.+}, at: [<c01f1608>] kernfs_fop_write+0x58/0x1b8
> [   12.077310]  #3:  (devcgroup_mutex){+.+...}, at: [<c0317bfc>] devcgroup_access_write+0x20/0x658
> [   12.086575] stack backtrace:
> [   12.091124] CPU: 0 PID: 1 Comm: systemd Not tainted 4.2.0-next-20150831 #1154
> [   12.098609] Hardware name: Generic AM43 (Flattened Device Tree)
> [   12.104807] [<c001770c>] (unwind_backtrace) from [<c0013a58>] (show_stack+0x10/0x14)
> [   12.112924] [<c0013a58>] (show_stack) from [<c034f014>] (dump_stack+0x84/0x9c)
> [   12.120491] [<c034f014>] (dump_stack) from [<c0317a04>] (verify_new_ex+0xc4/0xdc)
> [   12.128326] [<c0317a04>] (verify_new_ex) from [<c0317f50>] (devcgroup_access_write+0x374/0x658)
> [   12.137426] [<c0317f50>] (devcgroup_access_write) from [<c00d2800>] (cgroup_file_write+0x28/0x1bc)
> [   12.146796] [<c00d2800>] (cgroup_file_write) from [<c01f1670>] (kernfs_fop_write+0xc0/0x1b8)
> [   12.155620] [<c01f1670>] (kernfs_fop_write) from [<c0177c94>] (__vfs_write+0x1c/0xd8)
> [   12.163783] [<c0177c94>] (__vfs_write) from [<c0178594>] (vfs_write+0x90/0x16c)
> [   12.171426] [<c0178594>] (vfs_write) from [<c0178db4>] (SyS_write+0x44/0x9c)
> [   12.178806] [<c0178db4>] (SyS_write) from [<c000f680>] (ret_fast_syscall+0x0/0x1c)
>
> Fix it by making sure rcu_read_lock() is held
> around calls to parent_has_perm().
>
> Signed-off-by: Felipe Balbi <balbi@ti.com>

This cleared up the splat on all my machines and I don't see any other
side effects (even with lockdep enabled).  Thanks!

Tested-by: Josh Boyer <jwboyer@fedoraproject.org>

josh

> ---
>
> Changes since v1:
>         - move rcu_read_lock/unlock to wrap parent_has_perm()
>
>  security/device_cgroup.c | 7 ++++++-
>  1 file changed, 6 insertions(+), 1 deletion(-)
>
> diff --git a/security/device_cgroup.c b/security/device_cgroup.c
> index 73455089feef..dd77ed206fa4 100644
> --- a/security/device_cgroup.c
> +++ b/security/device_cgroup.c
> @@ -608,6 +608,7 @@ static int devcgroup_update_access(struct dev_cgroup *devcgroup,
>         int count, rc = 0;
>         struct dev_exception_item ex;
>         struct dev_cgroup *parent = css_to_devcgroup(devcgroup->css.parent);
> +       int ret;
>
>         if (!capable(CAP_SYS_ADMIN))
>                 return -EPERM;
> @@ -734,7 +735,11 @@ static int devcgroup_update_access(struct dev_cgroup *devcgroup,
>                         break;
>                 }
>
> -               if (!parent_has_perm(devcgroup, &ex))
> +               rcu_read_lock();
> +               ret = parent_has_perm(devcgroup, &ex);
> +               rcu_read_unlock();
> +
> +               if (!ret)
>                         return -EPERM;
>                 rc = dev_exception_add(devcgroup, &ex);
>                 break;
> --
> 2.5.0
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
--
To unsubscribe from this list: send the line "unsubscribe linux-omap" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Tejun Heo Sept. 2, 2015, 4:24 p.m. UTC | #2
cc'ing Paul.

On Wed, Sep 02, 2015 at 08:12:28AM -0500, Felipe Balbi wrote:
> while booting AM437x device, the following splat
> triggered:
> 
> [   12.005238] ===============================
> [   12.009749] [ INFO: suspicious RCU usage. ]
> [   12.014116] 4.2.0-next-20150831 #1154 Not tainted
> [   12.019050] -------------------------------
> [   12.023408] security/device_cgroup.c:405 device_cgroup:verify_new_ex called without proper synchronization!
...
> [   12.128326] [<c0317a04>] (verify_new_ex) from [<c0317f50>] (devcgroup_access_write+0x374/0x658)
> [   12.137426] [<c0317f50>] (devcgroup_access_write) from [<c00d2800>] (cgroup_file_write+0x28/0x1bc)
> [   12.146796] [<c00d2800>] (cgroup_file_write) from [<c01f1670>] (kernfs_fop_write+0xc0/0x1b8)
> [   12.155620] [<c01f1670>] (kernfs_fop_write) from [<c0177c94>] (__vfs_write+0x1c/0xd8)
> [   12.163783] [<c0177c94>] (__vfs_write) from [<c0178594>] (vfs_write+0x90/0x16c)
> [   12.171426] [<c0178594>] (vfs_write) from [<c0178db4>] (SyS_write+0x44/0x9c)
> [   12.178806] [<c0178db4>] (SyS_write) from [<c000f680>] (ret_fast_syscall+0x0/0x1c)

This shouldn't be happening because devcgroup_access_write() always
grabs devcgroup_mutex.  Looking at the log, the culprit seems to be
f78f5b90c4ff ("rcu: Rename rcu_lockdep_assert() to
RCU_LOCKDEP_WARN()").  It missed the bang for the second test while
inverting it, so adding rcu_read_lock() isn't the right fix here.

Paul, can you please fix it?

Thanks.

Patch
diff mbox

diff --git a/security/device_cgroup.c b/security/device_cgroup.c
index 73455089feef..dd77ed206fa4 100644
--- a/security/device_cgroup.c
+++ b/security/device_cgroup.c
@@ -608,6 +608,7 @@  static int devcgroup_update_access(struct dev_cgroup *devcgroup,
 	int count, rc = 0;
 	struct dev_exception_item ex;
 	struct dev_cgroup *parent = css_to_devcgroup(devcgroup->css.parent);
+	int ret;
 
 	if (!capable(CAP_SYS_ADMIN))
 		return -EPERM;
@@ -734,7 +735,11 @@  static int devcgroup_update_access(struct dev_cgroup *devcgroup,
 			break;
 		}
 
-		if (!parent_has_perm(devcgroup, &ex))
+		rcu_read_lock();
+		ret = parent_has_perm(devcgroup, &ex);
+		rcu_read_unlock();
+
+		if (!ret)
 			return -EPERM;
 		rc = dev_exception_add(devcgroup, &ex);
 		break;