Do 'shm_init_ns()' in an early pure_initcall
authorLinus Torvalds <torvalds@linux-foundation.org>
Fri, 5 Aug 2011 05:35:59 +0000 (19:35 -1000)
committerLinus Torvalds <torvalds@linux-foundation.org>
Fri, 5 Aug 2011 05:35:59 +0000 (19:35 -1000)
This isn't really critical any more, since other patches (commit
298507d4d2cf: "shm: optimize exit_shm()") have caused us to not actually
need to touch the rw_mutex unless there are actual shm segments
associated with the namespace, but we really should do tne shm_init_ns()
earlier than we do now.

This, together with commit 288d5abec831 ("Boot up with usermodehelper
disabled") will mean that we really do initialize the initial ipc
namespace data structure before we run any tasks.

Tested-by: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
ipc/shm.c

index b5bae9d945b62103ad8924ecf17dee8ee18a728c..02ecf2c078fce9b62ee2b1ec0ef252c0320acdf5 100644 (file)
--- a/ipc/shm.c
+++ b/ipc/shm.c
@@ -105,9 +105,16 @@ void shm_exit_ns(struct ipc_namespace *ns)
 }
 #endif
 
-void __init shm_init (void)
+static int __init ipc_ns_init(void)
 {
        shm_init_ns(&init_ipc_ns);
+       return 0;
+}
+
+pure_initcall(ipc_ns_init);
+
+void __init shm_init (void)
+{
        ipc_init_proc_interface("sysvipc/shm",
 #if BITS_PER_LONG <= 32
                                "       key      shmid perms       size  cpid  lpid nattch   uid   gid  cuid  cgid      atime      dtime      ctime        rss       swap\n",