[PATCH] debugfs: fix security_locked_down() call for SELinux

Make sure that security_locked_down() is checked last so that a bogus
denial is not reported by SELinux when (ia->ia_valid & (ATTR_MODE |
ATTR_UID | ATTR_GID)) is zero.

Note: this was introduced by commit 5496197f9b08 ("debugfs: Restrict
debugfs when the kernel is locked down"), but it didn't matter at that
time, as the SELinux support came in later.

Fixes: 59438b46471a ("security,lockdown,selinux: implement SELinux lockdown")
Signed-off-by: Ondrej Mosnacek <omosnace@redhat.com>
---
 fs/debugfs/inode.c | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)

diff --git a/fs/debugfs/inode.c b/fs/debugfs/inode.c
index 22e86ae4dd5a..bbfc7898c1aa 100644
--- a/fs/debugfs/inode.c
+++ b/fs/debugfs/inode.c

@@ -45,10 +45,13 @@ static unsigned int debugfs_allow = DEFAULT_DEBUGFS_ALLOW_BITS;
 static int debugfs_setattr(struct user_namespace *mnt_userns,
 			   struct dentry *dentry, struct iattr *ia)
 {
-	int ret = security_locked_down(LOCKDOWN_DEBUGFS);
+	int ret;
 
-	if (ret && (ia->ia_valid & (ATTR_MODE | ATTR_UID | ATTR_GID)))
-		return ret;
+	if (ia->ia_valid & (ATTR_MODE | ATTR_UID | ATTR_GID)) {
+		ret = security_locked_down(LOCKDOWN_DEBUGFS);
+		if (ret)
+			return ret;
+	}
 	return simple_setattr(&init_user_ns, dentry, ia);
 }
 

-- 
2.31.1

On Fri, May 07, 2021 at 01:41:50PM +0200, Ondrej Mosnacek wrote:

Make sure that security_locked_down() is checked last so that a bogus
denial is not reported by SELinux when (ia->ia_valid & (ATTR_MODE |
ATTR_UID | ATTR_GID)) is zero.

Why would this be "bogus"?

Note: this was introduced by commit 5496197f9b08 ("debugfs: Restrict
debugfs when the kernel is locked down"), but it didn't matter at that
time, as the SELinux support came in later.

Fixes: 59438b46471a ("security,lockdown,selinux: implement SELinux lockdown")

What does this "fix"?

What is happening in selinux that it can not handle this sequence now?
That commit showed up a long time ago, this feels "odd"...

thanks,

greg k-h

On Fri, May 07, 2021 at 02:03:04PM +0200, Greg Kroah-Hartman wrote:

On Fri, May 07, 2021 at 01:41:50PM +0200, Ondrej Mosnacek wrote:

quoted

Make sure that security_locked_down() is checked last so that a bogus
denial is not reported by SELinux when (ia->ia_valid & (ATTR_MODE |
ATTR_UID | ATTR_GID)) is zero.

Why would this be "bogus"?

I presume selinux is logging a denial ... but we don't then actually
deny the operation.

quoted

Note: this was introduced by commit 5496197f9b08 ("debugfs: Restrict
debugfs when the kernel is locked down"), but it didn't matter at that
time, as the SELinux support came in later.

Fixes: 59438b46471a ("security,lockdown,selinux: implement SELinux lockdown")

What does this "fix"?

What is happening in selinux that it can not handle this sequence now?
That commit showed up a long time ago, this feels "odd"...

thanks,

greg k-h

On Fri, May 07, 2021 at 01:12:18PM +0100, Matthew Wilcox wrote:

On Fri, May 07, 2021 at 02:03:04PM +0200, Greg Kroah-Hartman wrote:

quoted

On Fri, May 07, 2021 at 01:41:50PM +0200, Ondrej Mosnacek wrote:

quoted

Make sure that security_locked_down() is checked last so that a bogus
denial is not reported by SELinux when (ia->ia_valid & (ATTR_MODE |
ATTR_UID | ATTR_GID)) is zero.

Why would this be "bogus"?

I presume selinux is logging a denial ... but we don't then actually
deny the operation.

That would be nice to note here...

On Fri, May 7, 2021 at 2:16 PM Greg Kroah-Hartman
[off-list ref] wrote:

On Fri, May 07, 2021 at 01:12:18PM +0100, Matthew Wilcox wrote:

quoted

On Fri, May 07, 2021 at 02:03:04PM +0200, Greg Kroah-Hartman wrote:

quoted

On Fri, May 07, 2021 at 01:41:50PM +0200, Ondrej Mosnacek wrote:

quoted

Make sure that security_locked_down() is checked last so that a bogus
denial is not reported by SELinux when (ia->ia_valid & (ATTR_MODE |
ATTR_UID | ATTR_GID)) is zero.

Why would this be "bogus"?

I presume selinux is logging a denial ... but we don't then actually
deny the operation.

That would be nice to note here...

Granted, I didn't do a good job of describing the issue in the patch
description... I'll send a v2 with hopefully a better description.

-- 
Ondrej Mosnacek
Software Engineer, Linux Security - SELinux kernel
Red Hat, Inc.