[PATCH] userns: fix return value in map_write

[PATCH] userns: fix return value in map_write

[Gao feng]

If the map is already set,we should return -EINVAL.

Signed-off-by: Gao feng <gaofeng-BthXqXjhjHXQFUHtdCDX3A@public.gmane.org>
---
 kernel/user_namespace.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/kernel/user_namespace.c b/kernel/user_namespace.c
index 054471d..0231f87 100644
--- a/kernel/user_namespace.c
+++ b/kernel/user_namespace.c
@@ -560,11 +560,11 @@ static ssize_t map_write(struct file *file, const char __user *buf,
 	 */
 	mutex_lock(&id_map_mutex);
 
-	ret = -EPERM;
 	/* Only allow one successful write to the map */
 	if (map->nr_extents != 0)
 		goto out;
 
+	ret = -EPERM;
 	/* Require the appropriate privilege CAP_SETUID or CAP_SETGID
 	 * over the user namespace in order to set the id mapping.
 	 */
-- 
1.7.7.6

Re: [PATCH] userns: fix return value in map_write

[Eric W. Biederman]

Gao feng <gaofeng-BthXqXjhjHXQFUHtdCDX3A@public.gmane.org> writes:

> If the map is already set,we should return -EINVAL.

What makes -EINVAL a better return code?

Eric

Re: [PATCH] userns: fix return value in map_write

[Gao feng]

于 2012年10月10日 12:06, Eric W. Biederman 写道:
> Gao feng <gaofeng@cn.fujitsu.com> writes:
> 
>> If the map is already set,we should return -EINVAL.
> 
> What makes -EINVAL a better return code?
> 

It shouldn't be -EPERM,this will cause user confused.
The first time to write to the uid_map file is success,
and the second time,kernel will return -EPERM.actually
the user has right to write to the uid_map.

And if you think -EPERM is better,it's not necessary
to initialize the value of ret.because we set it to
-EPERM after we get mutex lock id_map_mutex.
_______________________________________________
Containers mailing list
Containers@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/containers