linux-kernel - Re: BUG_ON(nd->inode != parent->d

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Date:	Sat, 09 Mar 2013 00:28:27 -0800
From:	ebiederm@...ssion.com (Eric W. Biederman)
To:	Dave Jones <davej@...hat.com>
Cc:	Linus Torvalds <torvalds@...ux-foundation.org>,
	Al Viro <viro@...iv.linux.org.uk>,
	Linux Kernel <linux-kernel@...r.kernel.org>
Subject: Re: BUG_ON(nd->inode != parent->d_inode);

Dave Jones <davej@...hat.com> writes:

> On Fri, Mar 08, 2013 at 07:38:33PM -0800, Eric W. Biederman wrote:
>  > Dave Jones <davej@...hat.com> writes:
>  > 
>  > > On Fri, Mar 08, 2013 at 07:16:09PM -0800, Linus Torvalds wrote:
>  > >  > Goodie. Your bug reports gave me heartburn. But it sounds like we have an
>  > >  > angle on all of the ones you've seen now?
>  > >  > 
>  > >  > Or have I forgotten about some case?
>  > >  
>  > > To be honest I've lost track of the whole collection.
>  > > Let me repull your latest tree, and see what falls out.
>  > > (I'll turn off CONFIG_USER_NS for now too until that gets fixed)
>  > 
>  > It was CONFIG_UTS_NS that tripped you.  
>  > 
>  > Since I can trigger this with /proc/self/ns/mnt/a you are going to be
>  > able to compile this one out.
>
> fwiw, the other namespace procfs files look like they have the same bug
>
> I just triggered it again on /proc/571/task/571/ns/net

Yep same code.

I will do a proper submission of this after I have slept.  But
here is the fix.

From: "Eric W. Biederman" <ebiederm@...ssion.com>
Date: Sat, 9 Mar 2013 00:14:45 -0800
Subject: [PATCH] proc: Use nd_jump_link in proc_ns_follow_link

Update proc_ns_follow_link to use nd_jump_link instead of just
manually updating nd.path.dentry.

This fixes the BUG_ON(nd->inode != parent->d_inode) reported by Dave
Jones and reproduced trivially with mkdir /proc/self/ns/uts/a.

Sigh it looks like the VFS change to require use of nd_jump_link
happend while proc_ns_follow_link was baking and since the common case
of proc_ns_follow_link continued to work without problems the need for
making this change was overlooked.

Cc: stable@...r.kernel.org
Signed-off-by: "Eric W. Biederman" <ebiederm@...ssion.com>
---
 fs/proc/namespaces.c |   12 ++++++------
 1 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/fs/proc/namespaces.c b/fs/proc/namespaces.c
index b7a4719..66b51c0 100644
--- a/fs/proc/namespaces.c
+++ b/fs/proc/namespaces.c
@@ -118,7 +118,7 @@ static void *proc_ns_follow_link(struct dentry *dentry, struct nameidata *nd)
 	struct super_block *sb = inode->i_sb;
 	struct proc_inode *ei = PROC_I(inode);
 	struct task_struct *task;
-	struct dentry *ns_dentry;
+	struct path ns_path;
 	void *error = ERR_PTR(-EACCES);
 
 	task = get_proc_task(inode);
@@ -128,14 +128,14 @@ static void *proc_ns_follow_link(struct dentry *dentry, struct nameidata *nd)
 	if (!ptrace_may_access(task, PTRACE_MODE_READ))
 		goto out_put_task;
 
-	ns_dentry = proc_ns_get_dentry(sb, task, ei->ns_ops);
-	if (IS_ERR(ns_dentry)) {
-		error = ERR_CAST(ns_dentry);
+	ns_path.dentry = proc_ns_get_dentry(sb, task, ei->ns_ops);
+	if (IS_ERR(ns_path.dentry)) {
+		error = ERR_CAST(ns_path.dentry);
 		goto out_put_task;
 	}
 
-	dput(nd->path.dentry);
-	nd->path.dentry = ns_dentry;
+	ns_path.mnt = mntget(nd->path.mnt);
+	nd_jump_link(nd, &ns_path);
 	error = NULL;
 
 out_put_task:
-- 
1.7.5.4

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/