libfs: Modify mount_pseudo_xattr to be clear it is not a userspace mount
authorEric W. Biederman <ebiederm@xmission.com>
Wed, 4 Jan 2017 04:37:27 +0000 (17:37 +1300)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Sat, 9 Dec 2017 21:01:51 +0000 (22:01 +0100)
commitc4081f91981b9db6ef27d4cbf9b475ccaf9a56fe
tree737968ed3e00fb37a23c69ecfa0fdae86dae61df
parentd9a4e70f23c3b648a1cb60827f2580a897642963
libfs: Modify mount_pseudo_xattr to be clear it is not a userspace mount

[ Upstream commit 75422726b0f717d67db3283c2eb5bc14fa2619c5 ]

Add MS_KERNMOUNT to the flags that are passed.
Use sget_userns and force &init_user_ns instead of calling sget so that
even if called from a weird context the internal filesystem will be
considered to be in the intial user namespace.

Luis Ressel reported that the the failure to pass MS_KERNMOUNT into
mount_pseudo broke his in development graphics driver that uses the
generic drm infrastructure.  I am not certain the deriver was bug
free in it's usage of that infrastructure but since
mount_pseudo_xattr can never be triggered by userspace it is clearer
and less error prone, and less problematic for the code to be explicit.

Reported-by: Luis Ressel <aranea@aixah.de>
Tested-by: Luis Ressel <aranea@aixah.de>
Acked-by: Al Viro <viro@ZenIV.linux.org.uk>
Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: Sasha Levin <alexander.levin@verizon.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
fs/libfs.c