cgroups.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [PATCH v2] cgroup: avoid false positive gcc-6 warning
@ 2016-03-14 23:21 Arnd Bergmann
       [not found] ` <1457997681-2011105-1-git-send-email-arnd-r2nGTMty4D4@public.gmane.org>
  0 siblings, 1 reply; 2+ messages in thread
From: Arnd Bergmann @ 2016-03-14 23:21 UTC (permalink / raw)
  To: Tejun Heo, Li Zefan, Johannes Weiner
  Cc: Arnd Bergmann, Aditya Kali, Aleksa Sarai,
	cgroups-u79uwXL29TY76Z2rM5mHXA,
	linux-kernel-u79uwXL29TY76Z2rM5mHXA

When all subsystems are disabled, gcc notices that cgroup_subsys_enabled_key
is a zero-length array and that any access to it must be out of bounds:

In file included from ../include/linux/cgroup.h:19:0,
                 from ../kernel/cgroup.c:31:
../kernel/cgroup.c: In function 'cgroup_add_cftypes':
../kernel/cgroup.c:261:53: error: array subscript is above array bounds [-Werror=array-bounds]
  return static_key_enabled(cgroup_subsys_enabled_key[ssid]);
                            ~~~~~~~~~~~~~~~~~~~~~~~~~^~~~~~
../include/linux/jump_label.h:271:40: note: in definition of macro 'static_key_enabled'
  static_key_count((struct static_key *)x) > 0;    \
                                        ^

We should never call the function in this particular case, so this is
not a bug. In order to silence the warning, this adds an explicit check
for the CGROUP_SUBSYS_COUNT==0 case.

Signed-off-by: Arnd Bergmann <arnd-r2nGTMty4D4@public.gmane.org>
---
v2: use 'false' as the return code instead of '0', as suggested by
the friendly kbuild test robot.

 kernel/cgroup.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/kernel/cgroup.c b/kernel/cgroup.c
index 155c88470543..5f65c1002b4b 100644
--- a/kernel/cgroup.c
+++ b/kernel/cgroup.c
@@ -258,6 +258,9 @@ static int cgroup_addrm_files(struct cgroup_subsys_state *css,
  */
 static bool cgroup_ssid_enabled(int ssid)
 {
+	if (CGROUP_SUBSYS_COUNT == 0)
+		return false;
+
 	return static_key_enabled(cgroup_subsys_enabled_key[ssid]);
 }
 
-- 
2.7.0

^ permalink raw reply related	[flat|nested] 2+ messages in thread

* Re: [PATCH v2] cgroup: avoid false positive gcc-6 warning
       [not found] ` <1457997681-2011105-1-git-send-email-arnd-r2nGTMty4D4@public.gmane.org>
@ 2016-03-16 20:33   ` Tejun Heo
  0 siblings, 0 replies; 2+ messages in thread
From: Tejun Heo @ 2016-03-16 20:33 UTC (permalink / raw)
  To: Arnd Bergmann
  Cc: Li Zefan, Johannes Weiner, Aditya Kali, Aleksa Sarai,
	cgroups-u79uwXL29TY76Z2rM5mHXA,
	linux-kernel-u79uwXL29TY76Z2rM5mHXA

On Tue, Mar 15, 2016 at 12:21:06AM +0100, Arnd Bergmann wrote:
> When all subsystems are disabled, gcc notices that cgroup_subsys_enabled_key
> is a zero-length array and that any access to it must be out of bounds:
> 
> In file included from ../include/linux/cgroup.h:19:0,
>                  from ../kernel/cgroup.c:31:
> ../kernel/cgroup.c: In function 'cgroup_add_cftypes':
> ../kernel/cgroup.c:261:53: error: array subscript is above array bounds [-Werror=array-bounds]
>   return static_key_enabled(cgroup_subsys_enabled_key[ssid]);
>                             ~~~~~~~~~~~~~~~~~~~~~~~~~^~~~~~
> ../include/linux/jump_label.h:271:40: note: in definition of macro 'static_key_enabled'
>   static_key_count((struct static_key *)x) > 0;    \
>                                         ^
> 
> We should never call the function in this particular case, so this is
> not a bug. In order to silence the warning, this adds an explicit check
> for the CGROUP_SUBSYS_COUNT==0 case.
> 
> Signed-off-by: Arnd Bergmann <arnd-r2nGTMty4D4@public.gmane.org>

Applied to cgroup/for-4.6.

Thanks.

-- 
tejun

^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2016-03-16 20:33 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2016-03-14 23:21 [PATCH v2] cgroup: avoid false positive gcc-6 warning Arnd Bergmann
     [not found] ` <1457997681-2011105-1-git-send-email-arnd-r2nGTMty4D4@public.gmane.org>
2016-03-16 20:33   ` Tejun Heo

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).