Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1755773AbdDPKvT (ORCPT ); Sun, 16 Apr 2017 06:51:19 -0400 Received: from mail.linuxfoundation.org ([140.211.169.12]:58898 "EHLO mail.linuxfoundation.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754941AbdDPKvQ (ORCPT ); Sun, 16 Apr 2017 06:51:16 -0400 From: Greg Kroah-Hartman To: linux-kernel@vger.kernel.org Cc: Greg Kroah-Hartman , stable@vger.kernel.org, Linus Torvalds , Shmulik Ladkani , "David S. Miller" Subject: [PATCH 3.18 001/145] Revert "af_unix: Fix splice-bind deadlock" Date: Sun, 16 Apr 2017 12:48:14 +0200 Message-Id: <20170416080200.286792038@linuxfoundation.org> X-Mailer: git-send-email 2.12.2 In-Reply-To: <20170416080200.205458595@linuxfoundation.org> References: <20170416080200.205458595@linuxfoundation.org> User-Agent: quilt/0.65 MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 4350 Lines: 161 3.18-stable review patch. If anyone has any objections, please let me know. ------------------ From: Linus Torvalds commit 38f7bd94a97b542de86a2be9229289717e33a7a4 upstream. This reverts commit c845acb324aa85a39650a14e7696982ceea75dc1. It turns out that it just replaces one deadlock with another one: we can still get the wrong lock ordering with the readlock due to overlayfs calling back into the filesystem layer and still taking the vfs locks after the readlock. The proper solution ends up being to just split the readlock into two pieces: the bind lock (taken *outside* the vfs locks) and the IO lock (taken *inside* the filesystem locks). The two locks are independent anyway. Signed-off-by: Linus Torvalds Reviewed-by: Shmulik Ladkani Signed-off-by: David S. Miller Signed-off-by: Greg Kroah-Hartman --- net/unix/af_unix.c | 68 +++++++++++++++++++++-------------------------------- 1 file changed, 27 insertions(+), 41 deletions(-) --- a/net/unix/af_unix.c +++ b/net/unix/af_unix.c @@ -940,20 +940,32 @@ fail: return NULL; } -static int unix_mknod(struct dentry *dentry, struct path *path, umode_t mode, - struct path *res) +static int unix_mknod(const char *sun_path, umode_t mode, struct path *res) { - int err; - - err = security_path_mknod(path, dentry, mode, 0); + struct dentry *dentry; + struct path path; + int err = 0; + /* + * Get the parent directory, calculate the hash for last + * component. + */ + dentry = kern_path_create(AT_FDCWD, sun_path, &path, 0); + err = PTR_ERR(dentry); + if (IS_ERR(dentry)) + return err; + + /* + * All right, let's create it. + */ + err = security_path_mknod(&path, dentry, mode, 0); if (!err) { - err = vfs_mknod(d_inode(path->dentry), dentry, mode, 0); + err = vfs_mknod(d_inode(path.dentry), dentry, mode, 0); if (!err) { - res->mnt = mntget(path->mnt); + res->mnt = mntget(path.mnt); res->dentry = dget(dentry); } } - + done_path_create(&path, dentry); return err; } @@ -964,12 +976,10 @@ static int unix_bind(struct socket *sock struct unix_sock *u = unix_sk(sk); struct sockaddr_un *sunaddr = (struct sockaddr_un *)uaddr; char *sun_path = sunaddr->sun_path; - int err, name_err; + int err; unsigned int hash; struct unix_address *addr; struct hlist_head *list; - struct path path; - struct dentry *dentry; err = -EINVAL; if (sunaddr->sun_family != AF_UNIX) @@ -985,34 +995,14 @@ static int unix_bind(struct socket *sock goto out; addr_len = err; - name_err = 0; - dentry = NULL; - if (sun_path[0]) { - /* Get the parent directory, calculate the hash for last - * component. - */ - dentry = kern_path_create(AT_FDCWD, sun_path, &path, 0); - - if (IS_ERR(dentry)) { - /* delay report until after 'already bound' check */ - name_err = PTR_ERR(dentry); - dentry = NULL; - } - } - err = mutex_lock_interruptible(&u->readlock); if (err) - goto out_path; + goto out; err = -EINVAL; if (u->addr) goto out_up; - if (name_err) { - err = name_err == -EEXIST ? -EADDRINUSE : name_err; - goto out_up; - } - err = -ENOMEM; addr = kmalloc(sizeof(*addr)+addr_len, GFP_KERNEL); if (!addr) @@ -1023,11 +1013,11 @@ static int unix_bind(struct socket *sock addr->hash = hash ^ sk->sk_type; atomic_set(&addr->refcnt, 1); - if (dentry) { - struct path u_path; + if (sun_path[0]) { + struct path path; umode_t mode = S_IFSOCK | (SOCK_INODE(sock)->i_mode & ~current_umask()); - err = unix_mknod(dentry, &path, mode, &u_path); + err = unix_mknod(sun_path, mode, &path); if (err) { if (err == -EEXIST) err = -EADDRINUSE; @@ -1035,9 +1025,9 @@ static int unix_bind(struct socket *sock goto out_up; } addr->hash = UNIX_HASH_SIZE; - hash = d_backing_inode(dentry)->i_ino & (UNIX_HASH_SIZE - 1); + hash = d_backing_inode(path.dentry)->i_ino & (UNIX_HASH_SIZE-1); spin_lock(&unix_table_lock); - u->path = u_path; + u->path = path; list = &unix_socket_table[hash]; } else { spin_lock(&unix_table_lock); @@ -1060,10 +1050,6 @@ out_unlock: spin_unlock(&unix_table_lock); out_up: mutex_unlock(&u->readlock); -out_path: - if (dentry) - done_path_create(&path, dentry); - out: return err; }