linux_dsm_epyc7002/net/l2tp/l2tp_ppp.c

1919 lines
47 KiB
C
Raw Normal View History

/*****************************************************************************
* Linux PPP over L2TP (PPPoX/PPPoL2TP) Sockets
*
* PPPoX --- Generic PPP encapsulation socket family
* PPPoL2TP --- PPP over L2TP (RFC 2661)
*
* Version: 2.0.0
*
* Authors: James Chapman (jchapman@katalix.com)
*
* Based on original work by Martijn van Oosterhout <kleptog@svana.org>
*
* License:
* This program is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public License
* as published by the Free Software Foundation; either version
* 2 of the License, or (at your option) any later version.
*
*/
/* This driver handles only L2TP data frames; control frames are handled by a
* userspace application.
*
* To send data in an L2TP session, userspace opens a PPPoL2TP socket and
* attaches it to a bound UDP socket with local tunnel_id / session_id and
* peer tunnel_id / session_id set. Data can then be sent or received using
* regular socket sendmsg() / recvmsg() calls. Kernel parameters of the socket
* can be read or modified using ioctl() or [gs]etsockopt() calls.
*
* When a PPPoL2TP socket is connected with local and peer session_id values
* zero, the socket is treated as a special tunnel management socket.
*
* Here's example userspace code to create a socket for sending/receiving data
* over an L2TP session:-
*
* struct sockaddr_pppol2tp sax;
* int fd;
* int session_fd;
*
* fd = socket(AF_PPPOX, SOCK_DGRAM, PX_PROTO_OL2TP);
*
* sax.sa_family = AF_PPPOX;
* sax.sa_protocol = PX_PROTO_OL2TP;
* sax.pppol2tp.fd = tunnel_fd; // bound UDP socket
* sax.pppol2tp.addr.sin_addr.s_addr = addr->sin_addr.s_addr;
* sax.pppol2tp.addr.sin_port = addr->sin_port;
* sax.pppol2tp.addr.sin_family = AF_INET;
* sax.pppol2tp.s_tunnel = tunnel_id;
* sax.pppol2tp.s_session = session_id;
* sax.pppol2tp.d_tunnel = peer_tunnel_id;
* sax.pppol2tp.d_session = peer_session_id;
*
* session_fd = connect(fd, (struct sockaddr *)&sax, sizeof(sax));
*
* A pppd plugin that allows PPP traffic to be carried over L2TP using
* this driver is available from the OpenL2TP project at
* http://openl2tp.sourceforge.net.
*/
#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
#include <linux/module.h>
#include <linux/string.h>
#include <linux/list.h>
#include <linux/uaccess.h>
#include <linux/kernel.h>
#include <linux/spinlock.h>
#include <linux/kthread.h>
#include <linux/sched.h>
#include <linux/slab.h>
#include <linux/errno.h>
#include <linux/jiffies.h>
#include <linux/netdevice.h>
#include <linux/net.h>
#include <linux/inetdevice.h>
#include <linux/skbuff.h>
#include <linux/init.h>
#include <linux/ip.h>
#include <linux/udp.h>
#include <linux/if_pppox.h>
#include <linux/if_pppol2tp.h>
#include <net/sock.h>
#include <linux/ppp_channel.h>
#include <linux/ppp_defs.h>
#include <linux/ppp-ioctl.h>
#include <linux/file.h>
#include <linux/hash.h>
#include <linux/sort.h>
#include <linux/proc_fs.h>
#include <linux/l2tp.h>
#include <linux/nsproxy.h>
#include <net/net_namespace.h>
#include <net/netns/generic.h>
#include <net/dst.h>
#include <net/ip.h>
#include <net/udp.h>
#include <net/xfrm.h>
#include <net/inet_common.h>
#include <asm/byteorder.h>
#include <linux/atomic.h>
#include "l2tp_core.h"
#define PPPOL2TP_DRV_VERSION "V2.0"
/* Space for UDP, L2TP and PPP headers */
#define PPPOL2TP_HEADER_OVERHEAD 40
/* Number of bytes to build transmit L2TP headers.
* Unfortunately the size is different depending on whether sequence numbers
* are enabled.
*/
#define PPPOL2TP_L2TP_HDR_SIZE_SEQ 10
#define PPPOL2TP_L2TP_HDR_SIZE_NOSEQ 6
/* Private data of each session. This data lives at the end of struct
* l2tp_session, referenced via session->priv[].
*/
struct pppol2tp_session {
int owner; /* pid that opened the socket */
l2tp: protect sock pointer of struct pppol2tp_session with RCU pppol2tp_session_create() registers sessions that can't have their corresponding socket initialised. This socket has to be created by userspace, then connected to the session by pppol2tp_connect(). Therefore, we need to protect the pppol2tp socket pointer of L2TP sessions, so that it can safely be updated when userspace is connecting or closing the socket. This will eventually allow pppol2tp_connect() to avoid generating transient states while initialising its parts of the session. To this end, this patch protects the pppol2tp socket pointer using RCU. The pppol2tp socket pointer is still set in pppol2tp_connect(), but only once we know the function isn't going to fail. It's eventually reset by pppol2tp_release(), which now has to wait for a grace period to elapse before it can drop the last reference on the socket. This ensures that pppol2tp_session_get_sock() can safely grab a reference on the socket, even after ps->sk is reset to NULL but before this operation actually gets visible from pppol2tp_session_get_sock(). The rest is standard RCU conversion: pppol2tp_recv(), which already runs in atomic context, is simply enclosed by rcu_read_lock() and rcu_read_unlock(), while other functions are converted to use pppol2tp_session_get_sock() followed by sock_put(). pppol2tp_session_setsockopt() is a special case. It used to retrieve the pppol2tp socket from the L2TP session, which itself was retrieved from the pppol2tp socket. Therefore we can just avoid dereferencing ps->sk and directly use the original socket pointer instead. With all users of ps->sk now handling NULL and concurrent updates, the L2TP ->ref() and ->deref() callbacks aren't needed anymore. Therefore, rather than converting pppol2tp_session_sock_hold() and pppol2tp_session_sock_put(), we can just drop them. Signed-off-by: Guillaume Nault <g.nault@alphalink.fr> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-10-27 21:51:52 +07:00
struct mutex sk_lock; /* Protects .sk */
struct sock __rcu *sk; /* Pointer to the session
* PPPoX socket */
l2tp: protect sock pointer of struct pppol2tp_session with RCU pppol2tp_session_create() registers sessions that can't have their corresponding socket initialised. This socket has to be created by userspace, then connected to the session by pppol2tp_connect(). Therefore, we need to protect the pppol2tp socket pointer of L2TP sessions, so that it can safely be updated when userspace is connecting or closing the socket. This will eventually allow pppol2tp_connect() to avoid generating transient states while initialising its parts of the session. To this end, this patch protects the pppol2tp socket pointer using RCU. The pppol2tp socket pointer is still set in pppol2tp_connect(), but only once we know the function isn't going to fail. It's eventually reset by pppol2tp_release(), which now has to wait for a grace period to elapse before it can drop the last reference on the socket. This ensures that pppol2tp_session_get_sock() can safely grab a reference on the socket, even after ps->sk is reset to NULL but before this operation actually gets visible from pppol2tp_session_get_sock(). The rest is standard RCU conversion: pppol2tp_recv(), which already runs in atomic context, is simply enclosed by rcu_read_lock() and rcu_read_unlock(), while other functions are converted to use pppol2tp_session_get_sock() followed by sock_put(). pppol2tp_session_setsockopt() is a special case. It used to retrieve the pppol2tp socket from the L2TP session, which itself was retrieved from the pppol2tp socket. Therefore we can just avoid dereferencing ps->sk and directly use the original socket pointer instead. With all users of ps->sk now handling NULL and concurrent updates, the L2TP ->ref() and ->deref() callbacks aren't needed anymore. Therefore, rather than converting pppol2tp_session_sock_hold() and pppol2tp_session_sock_put(), we can just drop them. Signed-off-by: Guillaume Nault <g.nault@alphalink.fr> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-10-27 21:51:52 +07:00
struct sock *__sk; /* Copy of .sk, for cleanup */
struct rcu_head rcu; /* For asynchronous release */
int flags; /* accessed by PPPIOCGFLAGS.
* Unused. */
};
static int pppol2tp_xmit(struct ppp_channel *chan, struct sk_buff *skb);
static const struct ppp_channel_ops pppol2tp_chan_ops = {
.start_xmit = pppol2tp_xmit,
};
static const struct proto_ops pppol2tp_ops;
l2tp: protect sock pointer of struct pppol2tp_session with RCU pppol2tp_session_create() registers sessions that can't have their corresponding socket initialised. This socket has to be created by userspace, then connected to the session by pppol2tp_connect(). Therefore, we need to protect the pppol2tp socket pointer of L2TP sessions, so that it can safely be updated when userspace is connecting or closing the socket. This will eventually allow pppol2tp_connect() to avoid generating transient states while initialising its parts of the session. To this end, this patch protects the pppol2tp socket pointer using RCU. The pppol2tp socket pointer is still set in pppol2tp_connect(), but only once we know the function isn't going to fail. It's eventually reset by pppol2tp_release(), which now has to wait for a grace period to elapse before it can drop the last reference on the socket. This ensures that pppol2tp_session_get_sock() can safely grab a reference on the socket, even after ps->sk is reset to NULL but before this operation actually gets visible from pppol2tp_session_get_sock(). The rest is standard RCU conversion: pppol2tp_recv(), which already runs in atomic context, is simply enclosed by rcu_read_lock() and rcu_read_unlock(), while other functions are converted to use pppol2tp_session_get_sock() followed by sock_put(). pppol2tp_session_setsockopt() is a special case. It used to retrieve the pppol2tp socket from the L2TP session, which itself was retrieved from the pppol2tp socket. Therefore we can just avoid dereferencing ps->sk and directly use the original socket pointer instead. With all users of ps->sk now handling NULL and concurrent updates, the L2TP ->ref() and ->deref() callbacks aren't needed anymore. Therefore, rather than converting pppol2tp_session_sock_hold() and pppol2tp_session_sock_put(), we can just drop them. Signed-off-by: Guillaume Nault <g.nault@alphalink.fr> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-10-27 21:51:52 +07:00
/* Retrieves the pppol2tp socket associated to a session.
* A reference is held on the returned socket, so this function must be paired
* with sock_put().
*/
static struct sock *pppol2tp_session_get_sock(struct l2tp_session *session)
{
struct pppol2tp_session *ps = l2tp_session_priv(session);
struct sock *sk;
rcu_read_lock();
sk = rcu_dereference(ps->sk);
if (sk)
sock_hold(sk);
rcu_read_unlock();
return sk;
}
/* Helpers to obtain tunnel/session contexts from sockets.
*/
static inline struct l2tp_session *pppol2tp_sock_to_session(struct sock *sk)
{
struct l2tp_session *session;
if (sk == NULL)
return NULL;
sock_hold(sk);
session = (struct l2tp_session *)(sk->sk_user_data);
if (session == NULL) {
sock_put(sk);
goto out;
}
BUG_ON(session->magic != L2TP_SESSION_MAGIC);
out:
return session;
}
/*****************************************************************************
* Receive data handling
*****************************************************************************/
/* Receive message. This is the recvmsg for the PPPoL2TP socket.
*/
static int pppol2tp_recvmsg(struct socket *sock, struct msghdr *msg,
size_t len, int flags)
{
int err;
struct sk_buff *skb;
struct sock *sk = sock->sk;
err = -EIO;
if (sk->sk_state & PPPOX_BOUND)
goto end;
err = 0;
skb = skb_recv_datagram(sk, flags & ~MSG_DONTWAIT,
flags & MSG_DONTWAIT, &err);
if (!skb)
goto end;
if (len > skb->len)
len = skb->len;
else if (len < skb->len)
msg->msg_flags |= MSG_TRUNC;
err = skb_copy_datagram_msg(skb, 0, msg, len);
if (likely(err == 0))
err = len;
kfree_skb(skb);
end:
return err;
}
static void pppol2tp_recv(struct l2tp_session *session, struct sk_buff *skb, int data_len)
{
struct pppol2tp_session *ps = l2tp_session_priv(session);
struct sock *sk = NULL;
/* If the socket is bound, send it in to PPP's input queue. Otherwise
* queue it on the session socket.
*/
l2tp: protect sock pointer of struct pppol2tp_session with RCU pppol2tp_session_create() registers sessions that can't have their corresponding socket initialised. This socket has to be created by userspace, then connected to the session by pppol2tp_connect(). Therefore, we need to protect the pppol2tp socket pointer of L2TP sessions, so that it can safely be updated when userspace is connecting or closing the socket. This will eventually allow pppol2tp_connect() to avoid generating transient states while initialising its parts of the session. To this end, this patch protects the pppol2tp socket pointer using RCU. The pppol2tp socket pointer is still set in pppol2tp_connect(), but only once we know the function isn't going to fail. It's eventually reset by pppol2tp_release(), which now has to wait for a grace period to elapse before it can drop the last reference on the socket. This ensures that pppol2tp_session_get_sock() can safely grab a reference on the socket, even after ps->sk is reset to NULL but before this operation actually gets visible from pppol2tp_session_get_sock(). The rest is standard RCU conversion: pppol2tp_recv(), which already runs in atomic context, is simply enclosed by rcu_read_lock() and rcu_read_unlock(), while other functions are converted to use pppol2tp_session_get_sock() followed by sock_put(). pppol2tp_session_setsockopt() is a special case. It used to retrieve the pppol2tp socket from the L2TP session, which itself was retrieved from the pppol2tp socket. Therefore we can just avoid dereferencing ps->sk and directly use the original socket pointer instead. With all users of ps->sk now handling NULL and concurrent updates, the L2TP ->ref() and ->deref() callbacks aren't needed anymore. Therefore, rather than converting pppol2tp_session_sock_hold() and pppol2tp_session_sock_put(), we can just drop them. Signed-off-by: Guillaume Nault <g.nault@alphalink.fr> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-10-27 21:51:52 +07:00
rcu_read_lock();
sk = rcu_dereference(ps->sk);
if (sk == NULL)
goto no_sock;
/* If the first two bytes are 0xFF03, consider that it is the PPP's
* Address and Control fields and skip them. The L2TP module has always
* worked this way, although, in theory, the use of these fields should
* be negociated and handled at the PPP layer. These fields are
* constant: 0xFF is the All-Stations Address and 0x03 the Unnumbered
* Information command with Poll/Final bit set to zero (RFC 1662).
*/
if (pskb_may_pull(skb, 2) && skb->data[0] == PPP_ALLSTATIONS &&
skb->data[1] == PPP_UI)
skb_pull(skb, 2);
if (sk->sk_state & PPPOX_BOUND) {
struct pppox_sock *po;
l2tp_dbg(session, L2TP_MSG_DATA,
"%s: recv %d byte data frame, passing to ppp\n",
session->name, data_len);
po = pppox_sk(sk);
ppp_input(&po->chan, skb);
} else {
l2tp_dbg(session, L2TP_MSG_DATA,
"%s: recv %d byte data frame, passing to L2TP socket\n",
session->name, data_len);
if (sock_queue_rcv_skb(sk, skb) < 0) {
atomic_long_inc(&session->stats.rx_errors);
kfree_skb(skb);
}
}
l2tp: protect sock pointer of struct pppol2tp_session with RCU pppol2tp_session_create() registers sessions that can't have their corresponding socket initialised. This socket has to be created by userspace, then connected to the session by pppol2tp_connect(). Therefore, we need to protect the pppol2tp socket pointer of L2TP sessions, so that it can safely be updated when userspace is connecting or closing the socket. This will eventually allow pppol2tp_connect() to avoid generating transient states while initialising its parts of the session. To this end, this patch protects the pppol2tp socket pointer using RCU. The pppol2tp socket pointer is still set in pppol2tp_connect(), but only once we know the function isn't going to fail. It's eventually reset by pppol2tp_release(), which now has to wait for a grace period to elapse before it can drop the last reference on the socket. This ensures that pppol2tp_session_get_sock() can safely grab a reference on the socket, even after ps->sk is reset to NULL but before this operation actually gets visible from pppol2tp_session_get_sock(). The rest is standard RCU conversion: pppol2tp_recv(), which already runs in atomic context, is simply enclosed by rcu_read_lock() and rcu_read_unlock(), while other functions are converted to use pppol2tp_session_get_sock() followed by sock_put(). pppol2tp_session_setsockopt() is a special case. It used to retrieve the pppol2tp socket from the L2TP session, which itself was retrieved from the pppol2tp socket. Therefore we can just avoid dereferencing ps->sk and directly use the original socket pointer instead. With all users of ps->sk now handling NULL and concurrent updates, the L2TP ->ref() and ->deref() callbacks aren't needed anymore. Therefore, rather than converting pppol2tp_session_sock_hold() and pppol2tp_session_sock_put(), we can just drop them. Signed-off-by: Guillaume Nault <g.nault@alphalink.fr> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-10-27 21:51:52 +07:00
rcu_read_unlock();
return;
no_sock:
l2tp: protect sock pointer of struct pppol2tp_session with RCU pppol2tp_session_create() registers sessions that can't have their corresponding socket initialised. This socket has to be created by userspace, then connected to the session by pppol2tp_connect(). Therefore, we need to protect the pppol2tp socket pointer of L2TP sessions, so that it can safely be updated when userspace is connecting or closing the socket. This will eventually allow pppol2tp_connect() to avoid generating transient states while initialising its parts of the session. To this end, this patch protects the pppol2tp socket pointer using RCU. The pppol2tp socket pointer is still set in pppol2tp_connect(), but only once we know the function isn't going to fail. It's eventually reset by pppol2tp_release(), which now has to wait for a grace period to elapse before it can drop the last reference on the socket. This ensures that pppol2tp_session_get_sock() can safely grab a reference on the socket, even after ps->sk is reset to NULL but before this operation actually gets visible from pppol2tp_session_get_sock(). The rest is standard RCU conversion: pppol2tp_recv(), which already runs in atomic context, is simply enclosed by rcu_read_lock() and rcu_read_unlock(), while other functions are converted to use pppol2tp_session_get_sock() followed by sock_put(). pppol2tp_session_setsockopt() is a special case. It used to retrieve the pppol2tp socket from the L2TP session, which itself was retrieved from the pppol2tp socket. Therefore we can just avoid dereferencing ps->sk and directly use the original socket pointer instead. With all users of ps->sk now handling NULL and concurrent updates, the L2TP ->ref() and ->deref() callbacks aren't needed anymore. Therefore, rather than converting pppol2tp_session_sock_hold() and pppol2tp_session_sock_put(), we can just drop them. Signed-off-by: Guillaume Nault <g.nault@alphalink.fr> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-10-27 21:51:52 +07:00
rcu_read_unlock();
l2tp_info(session, L2TP_MSG_DATA, "%s: no socket\n", session->name);
kfree_skb(skb);
}
/************************************************************************
* Transmit handling
***********************************************************************/
/* This is the sendmsg for the PPPoL2TP pppol2tp_session socket. We come here
* when a user application does a sendmsg() on the session socket. L2TP and
* PPP headers must be inserted into the user's data.
*/
static int pppol2tp_sendmsg(struct socket *sock, struct msghdr *m,
size_t total_len)
{
struct sock *sk = sock->sk;
struct sk_buff *skb;
int error;
struct l2tp_session *session;
struct l2tp_tunnel *tunnel;
int uhlen;
error = -ENOTCONN;
if (sock_flag(sk, SOCK_DEAD) || !(sk->sk_state & PPPOX_CONNECTED))
goto error;
/* Get session and tunnel contexts */
error = -EBADF;
session = pppol2tp_sock_to_session(sk);
if (session == NULL)
goto error;
tunnel = session->tunnel;
uhlen = (tunnel->encap == L2TP_ENCAPTYPE_UDP) ? sizeof(struct udphdr) : 0;
/* Allocate a socket buffer */
error = -ENOMEM;
skb = sock_wmalloc(sk, NET_SKB_PAD + sizeof(struct iphdr) +
uhlen + session->hdr_len +
2 + total_len, /* 2 bytes for PPP_ALLSTATIONS & PPP_UI */
0, GFP_KERNEL);
if (!skb)
goto error_put_sess;
/* Reserve space for headers. */
skb_reserve(skb, NET_SKB_PAD);
skb_reset_network_header(skb);
skb_reserve(skb, sizeof(struct iphdr));
skb_reset_transport_header(skb);
skb_reserve(skb, uhlen);
/* Add PPP header */
skb->data[0] = PPP_ALLSTATIONS;
skb->data[1] = PPP_UI;
skb_put(skb, 2);
/* Copy user data into skb */
error = memcpy_from_msg(skb_put(skb, total_len), m, total_len);
if (error < 0) {
kfree_skb(skb);
goto error_put_sess;
}
l2tp: must disable bh before calling l2tp_xmit_skb() François Cachereul made a very nice bug report and suspected the bh_lock_sock() / bh_unlok_sock() pair used in l2tp_xmit_skb() from process context was not good. This problem was added by commit 6af88da14ee284aaad6e4326da09a89191ab6165 ("l2tp: Fix locking in l2tp_core.c"). l2tp_eth_dev_xmit() runs from BH context, so we must disable BH from other l2tp_xmit_skb() users. [ 452.060011] BUG: soft lockup - CPU#1 stuck for 23s! [accel-pppd:6662] [ 452.061757] Modules linked in: l2tp_ppp l2tp_netlink l2tp_core pppoe pppox ppp_generic slhc ipv6 ext3 mbcache jbd virtio_balloon xfs exportfs dm_mod virtio_blk ata_generic virtio_net floppy ata_piix libata virtio_pci virtio_ring virtio [last unloaded: scsi_wait_scan] [ 452.064012] CPU 1 [ 452.080015] BUG: soft lockup - CPU#2 stuck for 23s! [accel-pppd:6643] [ 452.080015] CPU 2 [ 452.080015] [ 452.080015] Pid: 6643, comm: accel-pppd Not tainted 3.2.46.mini #1 Bochs Bochs [ 452.080015] RIP: 0010:[<ffffffff81059f6c>] [<ffffffff81059f6c>] do_raw_spin_lock+0x17/0x1f [ 452.080015] RSP: 0018:ffff88007125fc18 EFLAGS: 00000293 [ 452.080015] RAX: 000000000000aba9 RBX: ffffffff811d0703 RCX: 0000000000000000 [ 452.080015] RDX: 00000000000000ab RSI: ffff8800711f6896 RDI: ffff8800745c8110 [ 452.080015] RBP: ffff88007125fc18 R08: 0000000000000020 R09: 0000000000000000 [ 452.080015] R10: 0000000000000000 R11: 0000000000000280 R12: 0000000000000286 [ 452.080015] R13: 0000000000000020 R14: 0000000000000240 R15: 0000000000000000 [ 452.080015] FS: 00007fdc0cc24700(0000) GS:ffff8800b6f00000(0000) knlGS:0000000000000000 [ 452.080015] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 452.080015] CR2: 00007fdb054899b8 CR3: 0000000074404000 CR4: 00000000000006a0 [ 452.080015] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 [ 452.080015] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400 [ 452.080015] Process accel-pppd (pid: 6643, threadinfo ffff88007125e000, task ffff8800b27e6dd0) [ 452.080015] Stack: [ 452.080015] ffff88007125fc28 ffffffff81256559 ffff88007125fc98 ffffffffa01b2bd1 [ 452.080015] ffff88007125fc58 000000000000000c 00000000029490d0 0000009c71dbe25e [ 452.080015] 000000000000005c 000000080000000e 0000000000000000 ffff880071170600 [ 452.080015] Call Trace: [ 452.080015] [<ffffffff81256559>] _raw_spin_lock+0xe/0x10 [ 452.080015] [<ffffffffa01b2bd1>] l2tp_xmit_skb+0x189/0x4ac [l2tp_core] [ 452.080015] [<ffffffffa01c2d36>] pppol2tp_sendmsg+0x15e/0x19c [l2tp_ppp] [ 452.080015] [<ffffffff811c7872>] __sock_sendmsg_nosec+0x22/0x24 [ 452.080015] [<ffffffff811c83bd>] sock_sendmsg+0xa1/0xb6 [ 452.080015] [<ffffffff81254e88>] ? __schedule+0x5c1/0x616 [ 452.080015] [<ffffffff8103c7c6>] ? __dequeue_signal+0xb7/0x10c [ 452.080015] [<ffffffff810bbd21>] ? fget_light+0x75/0x89 [ 452.080015] [<ffffffff811c8444>] ? sockfd_lookup_light+0x20/0x56 [ 452.080015] [<ffffffff811c9b34>] sys_sendto+0x10c/0x13b [ 452.080015] [<ffffffff8125cac2>] system_call_fastpath+0x16/0x1b [ 452.080015] Code: 81 48 89 e5 72 0c 31 c0 48 81 ff 45 66 25 81 0f 92 c0 5d c3 55 b8 00 01 00 00 48 89 e5 f0 66 0f c1 07 0f b6 d4 38 d0 74 06 f3 90 <8a> 07 eb f6 5d c3 90 90 55 48 89 e5 9c 58 0f 1f 44 00 00 5d c3 [ 452.080015] Call Trace: [ 452.080015] [<ffffffff81256559>] _raw_spin_lock+0xe/0x10 [ 452.080015] [<ffffffffa01b2bd1>] l2tp_xmit_skb+0x189/0x4ac [l2tp_core] [ 452.080015] [<ffffffffa01c2d36>] pppol2tp_sendmsg+0x15e/0x19c [l2tp_ppp] [ 452.080015] [<ffffffff811c7872>] __sock_sendmsg_nosec+0x22/0x24 [ 452.080015] [<ffffffff811c83bd>] sock_sendmsg+0xa1/0xb6 [ 452.080015] [<ffffffff81254e88>] ? __schedule+0x5c1/0x616 [ 452.080015] [<ffffffff8103c7c6>] ? __dequeue_signal+0xb7/0x10c [ 452.080015] [<ffffffff810bbd21>] ? fget_light+0x75/0x89 [ 452.080015] [<ffffffff811c8444>] ? sockfd_lookup_light+0x20/0x56 [ 452.080015] [<ffffffff811c9b34>] sys_sendto+0x10c/0x13b [ 452.080015] [<ffffffff8125cac2>] system_call_fastpath+0x16/0x1b [ 452.064012] [ 452.064012] Pid: 6662, comm: accel-pppd Not tainted 3.2.46.mini #1 Bochs Bochs [ 452.064012] RIP: 0010:[<ffffffff81059f6e>] [<ffffffff81059f6e>] do_raw_spin_lock+0x19/0x1f [ 452.064012] RSP: 0018:ffff8800b6e83ba0 EFLAGS: 00000297 [ 452.064012] RAX: 000000000000aaa9 RBX: ffff8800b6e83b40 RCX: 0000000000000002 [ 452.064012] RDX: 00000000000000aa RSI: 000000000000000a RDI: ffff8800745c8110 [ 452.064012] RBP: ffff8800b6e83ba0 R08: 000000000000c802 R09: 000000000000001c [ 452.064012] R10: ffff880071096c4e R11: 0000000000000006 R12: ffff8800b6e83b18 [ 452.064012] R13: ffffffff8125d51e R14: ffff8800b6e83ba0 R15: ffff880072a589c0 [ 452.064012] FS: 00007fdc0b81e700(0000) GS:ffff8800b6e80000(0000) knlGS:0000000000000000 [ 452.064012] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 452.064012] CR2: 0000000000625208 CR3: 0000000074404000 CR4: 00000000000006a0 [ 452.064012] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 [ 452.064012] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400 [ 452.064012] Process accel-pppd (pid: 6662, threadinfo ffff88007129a000, task ffff8800744f7410) [ 452.064012] Stack: [ 452.064012] ffff8800b6e83bb0 ffffffff81256559 ffff8800b6e83bc0 ffffffff8121c64a [ 452.064012] ffff8800b6e83bf0 ffffffff8121ec7a ffff880072a589c0 ffff880071096c62 [ 452.064012] 0000000000000011 ffffffff81430024 ffff8800b6e83c80 ffffffff8121f276 [ 452.064012] Call Trace: [ 452.064012] <IRQ> [ 452.064012] [<ffffffff81256559>] _raw_spin_lock+0xe/0x10 [ 452.064012] [<ffffffff8121c64a>] spin_lock+0x9/0xb [ 452.064012] [<ffffffff8121ec7a>] udp_queue_rcv_skb+0x186/0x269 [ 452.064012] [<ffffffff8121f276>] __udp4_lib_rcv+0x297/0x4ae [ 452.064012] [<ffffffff8121c178>] ? raw_rcv+0xe9/0xf0 [ 452.064012] [<ffffffff8121f4a7>] udp_rcv+0x1a/0x1c [ 452.064012] [<ffffffff811fe385>] ip_local_deliver_finish+0x12b/0x1a5 [ 452.064012] [<ffffffff811fe54e>] ip_local_deliver+0x53/0x84 [ 452.064012] [<ffffffff811fe1d0>] ip_rcv_finish+0x2bc/0x2f3 [ 452.064012] [<ffffffff811fe78f>] ip_rcv+0x210/0x269 [ 452.064012] [<ffffffff8101911e>] ? kvm_clock_get_cycles+0x9/0xb [ 452.064012] [<ffffffff811d88cd>] __netif_receive_skb+0x3a5/0x3f7 [ 452.064012] [<ffffffff811d8eba>] netif_receive_skb+0x57/0x5e [ 452.064012] [<ffffffff811cf30f>] ? __netdev_alloc_skb+0x1f/0x3b [ 452.064012] [<ffffffffa0049126>] virtnet_poll+0x4ba/0x5a4 [virtio_net] [ 452.064012] [<ffffffff811d9417>] net_rx_action+0x73/0x184 [ 452.064012] [<ffffffffa01b2cc2>] ? l2tp_xmit_skb+0x27a/0x4ac [l2tp_core] [ 452.064012] [<ffffffff810343b9>] __do_softirq+0xc3/0x1a8 [ 452.064012] [<ffffffff81013b56>] ? ack_APIC_irq+0x10/0x12 [ 452.064012] [<ffffffff81256559>] ? _raw_spin_lock+0xe/0x10 [ 452.064012] [<ffffffff8125e0ac>] call_softirq+0x1c/0x26 [ 452.064012] [<ffffffff81003587>] do_softirq+0x45/0x82 [ 452.064012] [<ffffffff81034667>] irq_exit+0x42/0x9c [ 452.064012] [<ffffffff8125e146>] do_IRQ+0x8e/0xa5 [ 452.064012] [<ffffffff8125676e>] common_interrupt+0x6e/0x6e [ 452.064012] <EOI> [ 452.064012] [<ffffffff810b82a1>] ? kfree+0x8a/0xa3 [ 452.064012] [<ffffffffa01b2cc2>] ? l2tp_xmit_skb+0x27a/0x4ac [l2tp_core] [ 452.064012] [<ffffffffa01b2c25>] ? l2tp_xmit_skb+0x1dd/0x4ac [l2tp_core] [ 452.064012] [<ffffffffa01c2d36>] pppol2tp_sendmsg+0x15e/0x19c [l2tp_ppp] [ 452.064012] [<ffffffff811c7872>] __sock_sendmsg_nosec+0x22/0x24 [ 452.064012] [<ffffffff811c83bd>] sock_sendmsg+0xa1/0xb6 [ 452.064012] [<ffffffff81254e88>] ? __schedule+0x5c1/0x616 [ 452.064012] [<ffffffff8103c7c6>] ? __dequeue_signal+0xb7/0x10c [ 452.064012] [<ffffffff810bbd21>] ? fget_light+0x75/0x89 [ 452.064012] [<ffffffff811c8444>] ? sockfd_lookup_light+0x20/0x56 [ 452.064012] [<ffffffff811c9b34>] sys_sendto+0x10c/0x13b [ 452.064012] [<ffffffff8125cac2>] system_call_fastpath+0x16/0x1b [ 452.064012] Code: 89 e5 72 0c 31 c0 48 81 ff 45 66 25 81 0f 92 c0 5d c3 55 b8 00 01 00 00 48 89 e5 f0 66 0f c1 07 0f b6 d4 38 d0 74 06 f3 90 8a 07 <eb> f6 5d c3 90 90 55 48 89 e5 9c 58 0f 1f 44 00 00 5d c3 55 48 [ 452.064012] Call Trace: [ 452.064012] <IRQ> [<ffffffff81256559>] _raw_spin_lock+0xe/0x10 [ 452.064012] [<ffffffff8121c64a>] spin_lock+0x9/0xb [ 452.064012] [<ffffffff8121ec7a>] udp_queue_rcv_skb+0x186/0x269 [ 452.064012] [<ffffffff8121f276>] __udp4_lib_rcv+0x297/0x4ae [ 452.064012] [<ffffffff8121c178>] ? raw_rcv+0xe9/0xf0 [ 452.064012] [<ffffffff8121f4a7>] udp_rcv+0x1a/0x1c [ 452.064012] [<ffffffff811fe385>] ip_local_deliver_finish+0x12b/0x1a5 [ 452.064012] [<ffffffff811fe54e>] ip_local_deliver+0x53/0x84 [ 452.064012] [<ffffffff811fe1d0>] ip_rcv_finish+0x2bc/0x2f3 [ 452.064012] [<ffffffff811fe78f>] ip_rcv+0x210/0x269 [ 452.064012] [<ffffffff8101911e>] ? kvm_clock_get_cycles+0x9/0xb [ 452.064012] [<ffffffff811d88cd>] __netif_receive_skb+0x3a5/0x3f7 [ 452.064012] [<ffffffff811d8eba>] netif_receive_skb+0x57/0x5e [ 452.064012] [<ffffffff811cf30f>] ? __netdev_alloc_skb+0x1f/0x3b [ 452.064012] [<ffffffffa0049126>] virtnet_poll+0x4ba/0x5a4 [virtio_net] [ 452.064012] [<ffffffff811d9417>] net_rx_action+0x73/0x184 [ 452.064012] [<ffffffffa01b2cc2>] ? l2tp_xmit_skb+0x27a/0x4ac [l2tp_core] [ 452.064012] [<ffffffff810343b9>] __do_softirq+0xc3/0x1a8 [ 452.064012] [<ffffffff81013b56>] ? ack_APIC_irq+0x10/0x12 [ 452.064012] [<ffffffff81256559>] ? _raw_spin_lock+0xe/0x10 [ 452.064012] [<ffffffff8125e0ac>] call_softirq+0x1c/0x26 [ 452.064012] [<ffffffff81003587>] do_softirq+0x45/0x82 [ 452.064012] [<ffffffff81034667>] irq_exit+0x42/0x9c [ 452.064012] [<ffffffff8125e146>] do_IRQ+0x8e/0xa5 [ 452.064012] [<ffffffff8125676e>] common_interrupt+0x6e/0x6e [ 452.064012] <EOI> [<ffffffff810b82a1>] ? kfree+0x8a/0xa3 [ 452.064012] [<ffffffffa01b2cc2>] ? l2tp_xmit_skb+0x27a/0x4ac [l2tp_core] [ 452.064012] [<ffffffffa01b2c25>] ? l2tp_xmit_skb+0x1dd/0x4ac [l2tp_core] [ 452.064012] [<ffffffffa01c2d36>] pppol2tp_sendmsg+0x15e/0x19c [l2tp_ppp] [ 452.064012] [<ffffffff811c7872>] __sock_sendmsg_nosec+0x22/0x24 [ 452.064012] [<ffffffff811c83bd>] sock_sendmsg+0xa1/0xb6 [ 452.064012] [<ffffffff81254e88>] ? __schedule+0x5c1/0x616 [ 452.064012] [<ffffffff8103c7c6>] ? __dequeue_signal+0xb7/0x10c [ 452.064012] [<ffffffff810bbd21>] ? fget_light+0x75/0x89 [ 452.064012] [<ffffffff811c8444>] ? sockfd_lookup_light+0x20/0x56 [ 452.064012] [<ffffffff811c9b34>] sys_sendto+0x10c/0x13b [ 452.064012] [<ffffffff8125cac2>] system_call_fastpath+0x16/0x1b Reported-by: François Cachereul <f.cachereul@alphalink.fr> Tested-by: François Cachereul <f.cachereul@alphalink.fr> Signed-off-by: Eric Dumazet <edumazet@google.com> Cc: James Chapman <jchapman@katalix.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2013-10-10 20:30:09 +07:00
local_bh_disable();
l2tp_xmit_skb(session, skb, session->hdr_len);
l2tp: must disable bh before calling l2tp_xmit_skb() François Cachereul made a very nice bug report and suspected the bh_lock_sock() / bh_unlok_sock() pair used in l2tp_xmit_skb() from process context was not good. This problem was added by commit 6af88da14ee284aaad6e4326da09a89191ab6165 ("l2tp: Fix locking in l2tp_core.c"). l2tp_eth_dev_xmit() runs from BH context, so we must disable BH from other l2tp_xmit_skb() users. [ 452.060011] BUG: soft lockup - CPU#1 stuck for 23s! [accel-pppd:6662] [ 452.061757] Modules linked in: l2tp_ppp l2tp_netlink l2tp_core pppoe pppox ppp_generic slhc ipv6 ext3 mbcache jbd virtio_balloon xfs exportfs dm_mod virtio_blk ata_generic virtio_net floppy ata_piix libata virtio_pci virtio_ring virtio [last unloaded: scsi_wait_scan] [ 452.064012] CPU 1 [ 452.080015] BUG: soft lockup - CPU#2 stuck for 23s! [accel-pppd:6643] [ 452.080015] CPU 2 [ 452.080015] [ 452.080015] Pid: 6643, comm: accel-pppd Not tainted 3.2.46.mini #1 Bochs Bochs [ 452.080015] RIP: 0010:[<ffffffff81059f6c>] [<ffffffff81059f6c>] do_raw_spin_lock+0x17/0x1f [ 452.080015] RSP: 0018:ffff88007125fc18 EFLAGS: 00000293 [ 452.080015] RAX: 000000000000aba9 RBX: ffffffff811d0703 RCX: 0000000000000000 [ 452.080015] RDX: 00000000000000ab RSI: ffff8800711f6896 RDI: ffff8800745c8110 [ 452.080015] RBP: ffff88007125fc18 R08: 0000000000000020 R09: 0000000000000000 [ 452.080015] R10: 0000000000000000 R11: 0000000000000280 R12: 0000000000000286 [ 452.080015] R13: 0000000000000020 R14: 0000000000000240 R15: 0000000000000000 [ 452.080015] FS: 00007fdc0cc24700(0000) GS:ffff8800b6f00000(0000) knlGS:0000000000000000 [ 452.080015] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 452.080015] CR2: 00007fdb054899b8 CR3: 0000000074404000 CR4: 00000000000006a0 [ 452.080015] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 [ 452.080015] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400 [ 452.080015] Process accel-pppd (pid: 6643, threadinfo ffff88007125e000, task ffff8800b27e6dd0) [ 452.080015] Stack: [ 452.080015] ffff88007125fc28 ffffffff81256559 ffff88007125fc98 ffffffffa01b2bd1 [ 452.080015] ffff88007125fc58 000000000000000c 00000000029490d0 0000009c71dbe25e [ 452.080015] 000000000000005c 000000080000000e 0000000000000000 ffff880071170600 [ 452.080015] Call Trace: [ 452.080015] [<ffffffff81256559>] _raw_spin_lock+0xe/0x10 [ 452.080015] [<ffffffffa01b2bd1>] l2tp_xmit_skb+0x189/0x4ac [l2tp_core] [ 452.080015] [<ffffffffa01c2d36>] pppol2tp_sendmsg+0x15e/0x19c [l2tp_ppp] [ 452.080015] [<ffffffff811c7872>] __sock_sendmsg_nosec+0x22/0x24 [ 452.080015] [<ffffffff811c83bd>] sock_sendmsg+0xa1/0xb6 [ 452.080015] [<ffffffff81254e88>] ? __schedule+0x5c1/0x616 [ 452.080015] [<ffffffff8103c7c6>] ? __dequeue_signal+0xb7/0x10c [ 452.080015] [<ffffffff810bbd21>] ? fget_light+0x75/0x89 [ 452.080015] [<ffffffff811c8444>] ? sockfd_lookup_light+0x20/0x56 [ 452.080015] [<ffffffff811c9b34>] sys_sendto+0x10c/0x13b [ 452.080015] [<ffffffff8125cac2>] system_call_fastpath+0x16/0x1b [ 452.080015] Code: 81 48 89 e5 72 0c 31 c0 48 81 ff 45 66 25 81 0f 92 c0 5d c3 55 b8 00 01 00 00 48 89 e5 f0 66 0f c1 07 0f b6 d4 38 d0 74 06 f3 90 <8a> 07 eb f6 5d c3 90 90 55 48 89 e5 9c 58 0f 1f 44 00 00 5d c3 [ 452.080015] Call Trace: [ 452.080015] [<ffffffff81256559>] _raw_spin_lock+0xe/0x10 [ 452.080015] [<ffffffffa01b2bd1>] l2tp_xmit_skb+0x189/0x4ac [l2tp_core] [ 452.080015] [<ffffffffa01c2d36>] pppol2tp_sendmsg+0x15e/0x19c [l2tp_ppp] [ 452.080015] [<ffffffff811c7872>] __sock_sendmsg_nosec+0x22/0x24 [ 452.080015] [<ffffffff811c83bd>] sock_sendmsg+0xa1/0xb6 [ 452.080015] [<ffffffff81254e88>] ? __schedule+0x5c1/0x616 [ 452.080015] [<ffffffff8103c7c6>] ? __dequeue_signal+0xb7/0x10c [ 452.080015] [<ffffffff810bbd21>] ? fget_light+0x75/0x89 [ 452.080015] [<ffffffff811c8444>] ? sockfd_lookup_light+0x20/0x56 [ 452.080015] [<ffffffff811c9b34>] sys_sendto+0x10c/0x13b [ 452.080015] [<ffffffff8125cac2>] system_call_fastpath+0x16/0x1b [ 452.064012] [ 452.064012] Pid: 6662, comm: accel-pppd Not tainted 3.2.46.mini #1 Bochs Bochs [ 452.064012] RIP: 0010:[<ffffffff81059f6e>] [<ffffffff81059f6e>] do_raw_spin_lock+0x19/0x1f [ 452.064012] RSP: 0018:ffff8800b6e83ba0 EFLAGS: 00000297 [ 452.064012] RAX: 000000000000aaa9 RBX: ffff8800b6e83b40 RCX: 0000000000000002 [ 452.064012] RDX: 00000000000000aa RSI: 000000000000000a RDI: ffff8800745c8110 [ 452.064012] RBP: ffff8800b6e83ba0 R08: 000000000000c802 R09: 000000000000001c [ 452.064012] R10: ffff880071096c4e R11: 0000000000000006 R12: ffff8800b6e83b18 [ 452.064012] R13: ffffffff8125d51e R14: ffff8800b6e83ba0 R15: ffff880072a589c0 [ 452.064012] FS: 00007fdc0b81e700(0000) GS:ffff8800b6e80000(0000) knlGS:0000000000000000 [ 452.064012] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 452.064012] CR2: 0000000000625208 CR3: 0000000074404000 CR4: 00000000000006a0 [ 452.064012] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 [ 452.064012] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400 [ 452.064012] Process accel-pppd (pid: 6662, threadinfo ffff88007129a000, task ffff8800744f7410) [ 452.064012] Stack: [ 452.064012] ffff8800b6e83bb0 ffffffff81256559 ffff8800b6e83bc0 ffffffff8121c64a [ 452.064012] ffff8800b6e83bf0 ffffffff8121ec7a ffff880072a589c0 ffff880071096c62 [ 452.064012] 0000000000000011 ffffffff81430024 ffff8800b6e83c80 ffffffff8121f276 [ 452.064012] Call Trace: [ 452.064012] <IRQ> [ 452.064012] [<ffffffff81256559>] _raw_spin_lock+0xe/0x10 [ 452.064012] [<ffffffff8121c64a>] spin_lock+0x9/0xb [ 452.064012] [<ffffffff8121ec7a>] udp_queue_rcv_skb+0x186/0x269 [ 452.064012] [<ffffffff8121f276>] __udp4_lib_rcv+0x297/0x4ae [ 452.064012] [<ffffffff8121c178>] ? raw_rcv+0xe9/0xf0 [ 452.064012] [<ffffffff8121f4a7>] udp_rcv+0x1a/0x1c [ 452.064012] [<ffffffff811fe385>] ip_local_deliver_finish+0x12b/0x1a5 [ 452.064012] [<ffffffff811fe54e>] ip_local_deliver+0x53/0x84 [ 452.064012] [<ffffffff811fe1d0>] ip_rcv_finish+0x2bc/0x2f3 [ 452.064012] [<ffffffff811fe78f>] ip_rcv+0x210/0x269 [ 452.064012] [<ffffffff8101911e>] ? kvm_clock_get_cycles+0x9/0xb [ 452.064012] [<ffffffff811d88cd>] __netif_receive_skb+0x3a5/0x3f7 [ 452.064012] [<ffffffff811d8eba>] netif_receive_skb+0x57/0x5e [ 452.064012] [<ffffffff811cf30f>] ? __netdev_alloc_skb+0x1f/0x3b [ 452.064012] [<ffffffffa0049126>] virtnet_poll+0x4ba/0x5a4 [virtio_net] [ 452.064012] [<ffffffff811d9417>] net_rx_action+0x73/0x184 [ 452.064012] [<ffffffffa01b2cc2>] ? l2tp_xmit_skb+0x27a/0x4ac [l2tp_core] [ 452.064012] [<ffffffff810343b9>] __do_softirq+0xc3/0x1a8 [ 452.064012] [<ffffffff81013b56>] ? ack_APIC_irq+0x10/0x12 [ 452.064012] [<ffffffff81256559>] ? _raw_spin_lock+0xe/0x10 [ 452.064012] [<ffffffff8125e0ac>] call_softirq+0x1c/0x26 [ 452.064012] [<ffffffff81003587>] do_softirq+0x45/0x82 [ 452.064012] [<ffffffff81034667>] irq_exit+0x42/0x9c [ 452.064012] [<ffffffff8125e146>] do_IRQ+0x8e/0xa5 [ 452.064012] [<ffffffff8125676e>] common_interrupt+0x6e/0x6e [ 452.064012] <EOI> [ 452.064012] [<ffffffff810b82a1>] ? kfree+0x8a/0xa3 [ 452.064012] [<ffffffffa01b2cc2>] ? l2tp_xmit_skb+0x27a/0x4ac [l2tp_core] [ 452.064012] [<ffffffffa01b2c25>] ? l2tp_xmit_skb+0x1dd/0x4ac [l2tp_core] [ 452.064012] [<ffffffffa01c2d36>] pppol2tp_sendmsg+0x15e/0x19c [l2tp_ppp] [ 452.064012] [<ffffffff811c7872>] __sock_sendmsg_nosec+0x22/0x24 [ 452.064012] [<ffffffff811c83bd>] sock_sendmsg+0xa1/0xb6 [ 452.064012] [<ffffffff81254e88>] ? __schedule+0x5c1/0x616 [ 452.064012] [<ffffffff8103c7c6>] ? __dequeue_signal+0xb7/0x10c [ 452.064012] [<ffffffff810bbd21>] ? fget_light+0x75/0x89 [ 452.064012] [<ffffffff811c8444>] ? sockfd_lookup_light+0x20/0x56 [ 452.064012] [<ffffffff811c9b34>] sys_sendto+0x10c/0x13b [ 452.064012] [<ffffffff8125cac2>] system_call_fastpath+0x16/0x1b [ 452.064012] Code: 89 e5 72 0c 31 c0 48 81 ff 45 66 25 81 0f 92 c0 5d c3 55 b8 00 01 00 00 48 89 e5 f0 66 0f c1 07 0f b6 d4 38 d0 74 06 f3 90 8a 07 <eb> f6 5d c3 90 90 55 48 89 e5 9c 58 0f 1f 44 00 00 5d c3 55 48 [ 452.064012] Call Trace: [ 452.064012] <IRQ> [<ffffffff81256559>] _raw_spin_lock+0xe/0x10 [ 452.064012] [<ffffffff8121c64a>] spin_lock+0x9/0xb [ 452.064012] [<ffffffff8121ec7a>] udp_queue_rcv_skb+0x186/0x269 [ 452.064012] [<ffffffff8121f276>] __udp4_lib_rcv+0x297/0x4ae [ 452.064012] [<ffffffff8121c178>] ? raw_rcv+0xe9/0xf0 [ 452.064012] [<ffffffff8121f4a7>] udp_rcv+0x1a/0x1c [ 452.064012] [<ffffffff811fe385>] ip_local_deliver_finish+0x12b/0x1a5 [ 452.064012] [<ffffffff811fe54e>] ip_local_deliver+0x53/0x84 [ 452.064012] [<ffffffff811fe1d0>] ip_rcv_finish+0x2bc/0x2f3 [ 452.064012] [<ffffffff811fe78f>] ip_rcv+0x210/0x269 [ 452.064012] [<ffffffff8101911e>] ? kvm_clock_get_cycles+0x9/0xb [ 452.064012] [<ffffffff811d88cd>] __netif_receive_skb+0x3a5/0x3f7 [ 452.064012] [<ffffffff811d8eba>] netif_receive_skb+0x57/0x5e [ 452.064012] [<ffffffff811cf30f>] ? __netdev_alloc_skb+0x1f/0x3b [ 452.064012] [<ffffffffa0049126>] virtnet_poll+0x4ba/0x5a4 [virtio_net] [ 452.064012] [<ffffffff811d9417>] net_rx_action+0x73/0x184 [ 452.064012] [<ffffffffa01b2cc2>] ? l2tp_xmit_skb+0x27a/0x4ac [l2tp_core] [ 452.064012] [<ffffffff810343b9>] __do_softirq+0xc3/0x1a8 [ 452.064012] [<ffffffff81013b56>] ? ack_APIC_irq+0x10/0x12 [ 452.064012] [<ffffffff81256559>] ? _raw_spin_lock+0xe/0x10 [ 452.064012] [<ffffffff8125e0ac>] call_softirq+0x1c/0x26 [ 452.064012] [<ffffffff81003587>] do_softirq+0x45/0x82 [ 452.064012] [<ffffffff81034667>] irq_exit+0x42/0x9c [ 452.064012] [<ffffffff8125e146>] do_IRQ+0x8e/0xa5 [ 452.064012] [<ffffffff8125676e>] common_interrupt+0x6e/0x6e [ 452.064012] <EOI> [<ffffffff810b82a1>] ? kfree+0x8a/0xa3 [ 452.064012] [<ffffffffa01b2cc2>] ? l2tp_xmit_skb+0x27a/0x4ac [l2tp_core] [ 452.064012] [<ffffffffa01b2c25>] ? l2tp_xmit_skb+0x1dd/0x4ac [l2tp_core] [ 452.064012] [<ffffffffa01c2d36>] pppol2tp_sendmsg+0x15e/0x19c [l2tp_ppp] [ 452.064012] [<ffffffff811c7872>] __sock_sendmsg_nosec+0x22/0x24 [ 452.064012] [<ffffffff811c83bd>] sock_sendmsg+0xa1/0xb6 [ 452.064012] [<ffffffff81254e88>] ? __schedule+0x5c1/0x616 [ 452.064012] [<ffffffff8103c7c6>] ? __dequeue_signal+0xb7/0x10c [ 452.064012] [<ffffffff810bbd21>] ? fget_light+0x75/0x89 [ 452.064012] [<ffffffff811c8444>] ? sockfd_lookup_light+0x20/0x56 [ 452.064012] [<ffffffff811c9b34>] sys_sendto+0x10c/0x13b [ 452.064012] [<ffffffff8125cac2>] system_call_fastpath+0x16/0x1b Reported-by: François Cachereul <f.cachereul@alphalink.fr> Tested-by: François Cachereul <f.cachereul@alphalink.fr> Signed-off-by: Eric Dumazet <edumazet@google.com> Cc: James Chapman <jchapman@katalix.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2013-10-10 20:30:09 +07:00
local_bh_enable();
sock_put(sk);
return total_len;
error_put_sess:
sock_put(sk);
error:
return error;
}
/* Transmit function called by generic PPP driver. Sends PPP frame
* over PPPoL2TP socket.
*
* This is almost the same as pppol2tp_sendmsg(), but rather than
* being called with a msghdr from userspace, it is called with a skb
* from the kernel.
*
* The supplied skb from ppp doesn't have enough headroom for the
* insertion of L2TP, UDP and IP headers so we need to allocate more
* headroom in the skb. This will create a cloned skb. But we must be
* careful in the error case because the caller will expect to free
* the skb it supplied, not our cloned skb. So we take care to always
* leave the original skb unfreed if we return an error.
*/
static int pppol2tp_xmit(struct ppp_channel *chan, struct sk_buff *skb)
{
struct sock *sk = (struct sock *) chan->private;
struct l2tp_session *session;
struct l2tp_tunnel *tunnel;
int uhlen, headroom;
if (sock_flag(sk, SOCK_DEAD) || !(sk->sk_state & PPPOX_CONNECTED))
goto abort;
/* Get session and tunnel contexts from the socket */
session = pppol2tp_sock_to_session(sk);
if (session == NULL)
goto abort;
tunnel = session->tunnel;
uhlen = (tunnel->encap == L2TP_ENCAPTYPE_UDP) ? sizeof(struct udphdr) : 0;
headroom = NET_SKB_PAD +
sizeof(struct iphdr) + /* IP header */
uhlen + /* UDP header (if L2TP_ENCAPTYPE_UDP) */
session->hdr_len + /* L2TP header */
2; /* 2 bytes for PPP_ALLSTATIONS & PPP_UI */
if (skb_cow_head(skb, headroom))
goto abort_put_sess;
/* Setup PPP header */
__skb_push(skb, 2);
skb->data[0] = PPP_ALLSTATIONS;
skb->data[1] = PPP_UI;
l2tp: must disable bh before calling l2tp_xmit_skb() François Cachereul made a very nice bug report and suspected the bh_lock_sock() / bh_unlok_sock() pair used in l2tp_xmit_skb() from process context was not good. This problem was added by commit 6af88da14ee284aaad6e4326da09a89191ab6165 ("l2tp: Fix locking in l2tp_core.c"). l2tp_eth_dev_xmit() runs from BH context, so we must disable BH from other l2tp_xmit_skb() users. [ 452.060011] BUG: soft lockup - CPU#1 stuck for 23s! [accel-pppd:6662] [ 452.061757] Modules linked in: l2tp_ppp l2tp_netlink l2tp_core pppoe pppox ppp_generic slhc ipv6 ext3 mbcache jbd virtio_balloon xfs exportfs dm_mod virtio_blk ata_generic virtio_net floppy ata_piix libata virtio_pci virtio_ring virtio [last unloaded: scsi_wait_scan] [ 452.064012] CPU 1 [ 452.080015] BUG: soft lockup - CPU#2 stuck for 23s! [accel-pppd:6643] [ 452.080015] CPU 2 [ 452.080015] [ 452.080015] Pid: 6643, comm: accel-pppd Not tainted 3.2.46.mini #1 Bochs Bochs [ 452.080015] RIP: 0010:[<ffffffff81059f6c>] [<ffffffff81059f6c>] do_raw_spin_lock+0x17/0x1f [ 452.080015] RSP: 0018:ffff88007125fc18 EFLAGS: 00000293 [ 452.080015] RAX: 000000000000aba9 RBX: ffffffff811d0703 RCX: 0000000000000000 [ 452.080015] RDX: 00000000000000ab RSI: ffff8800711f6896 RDI: ffff8800745c8110 [ 452.080015] RBP: ffff88007125fc18 R08: 0000000000000020 R09: 0000000000000000 [ 452.080015] R10: 0000000000000000 R11: 0000000000000280 R12: 0000000000000286 [ 452.080015] R13: 0000000000000020 R14: 0000000000000240 R15: 0000000000000000 [ 452.080015] FS: 00007fdc0cc24700(0000) GS:ffff8800b6f00000(0000) knlGS:0000000000000000 [ 452.080015] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 452.080015] CR2: 00007fdb054899b8 CR3: 0000000074404000 CR4: 00000000000006a0 [ 452.080015] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 [ 452.080015] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400 [ 452.080015] Process accel-pppd (pid: 6643, threadinfo ffff88007125e000, task ffff8800b27e6dd0) [ 452.080015] Stack: [ 452.080015] ffff88007125fc28 ffffffff81256559 ffff88007125fc98 ffffffffa01b2bd1 [ 452.080015] ffff88007125fc58 000000000000000c 00000000029490d0 0000009c71dbe25e [ 452.080015] 000000000000005c 000000080000000e 0000000000000000 ffff880071170600 [ 452.080015] Call Trace: [ 452.080015] [<ffffffff81256559>] _raw_spin_lock+0xe/0x10 [ 452.080015] [<ffffffffa01b2bd1>] l2tp_xmit_skb+0x189/0x4ac [l2tp_core] [ 452.080015] [<ffffffffa01c2d36>] pppol2tp_sendmsg+0x15e/0x19c [l2tp_ppp] [ 452.080015] [<ffffffff811c7872>] __sock_sendmsg_nosec+0x22/0x24 [ 452.080015] [<ffffffff811c83bd>] sock_sendmsg+0xa1/0xb6 [ 452.080015] [<ffffffff81254e88>] ? __schedule+0x5c1/0x616 [ 452.080015] [<ffffffff8103c7c6>] ? __dequeue_signal+0xb7/0x10c [ 452.080015] [<ffffffff810bbd21>] ? fget_light+0x75/0x89 [ 452.080015] [<ffffffff811c8444>] ? sockfd_lookup_light+0x20/0x56 [ 452.080015] [<ffffffff811c9b34>] sys_sendto+0x10c/0x13b [ 452.080015] [<ffffffff8125cac2>] system_call_fastpath+0x16/0x1b [ 452.080015] Code: 81 48 89 e5 72 0c 31 c0 48 81 ff 45 66 25 81 0f 92 c0 5d c3 55 b8 00 01 00 00 48 89 e5 f0 66 0f c1 07 0f b6 d4 38 d0 74 06 f3 90 <8a> 07 eb f6 5d c3 90 90 55 48 89 e5 9c 58 0f 1f 44 00 00 5d c3 [ 452.080015] Call Trace: [ 452.080015] [<ffffffff81256559>] _raw_spin_lock+0xe/0x10 [ 452.080015] [<ffffffffa01b2bd1>] l2tp_xmit_skb+0x189/0x4ac [l2tp_core] [ 452.080015] [<ffffffffa01c2d36>] pppol2tp_sendmsg+0x15e/0x19c [l2tp_ppp] [ 452.080015] [<ffffffff811c7872>] __sock_sendmsg_nosec+0x22/0x24 [ 452.080015] [<ffffffff811c83bd>] sock_sendmsg+0xa1/0xb6 [ 452.080015] [<ffffffff81254e88>] ? __schedule+0x5c1/0x616 [ 452.080015] [<ffffffff8103c7c6>] ? __dequeue_signal+0xb7/0x10c [ 452.080015] [<ffffffff810bbd21>] ? fget_light+0x75/0x89 [ 452.080015] [<ffffffff811c8444>] ? sockfd_lookup_light+0x20/0x56 [ 452.080015] [<ffffffff811c9b34>] sys_sendto+0x10c/0x13b [ 452.080015] [<ffffffff8125cac2>] system_call_fastpath+0x16/0x1b [ 452.064012] [ 452.064012] Pid: 6662, comm: accel-pppd Not tainted 3.2.46.mini #1 Bochs Bochs [ 452.064012] RIP: 0010:[<ffffffff81059f6e>] [<ffffffff81059f6e>] do_raw_spin_lock+0x19/0x1f [ 452.064012] RSP: 0018:ffff8800b6e83ba0 EFLAGS: 00000297 [ 452.064012] RAX: 000000000000aaa9 RBX: ffff8800b6e83b40 RCX: 0000000000000002 [ 452.064012] RDX: 00000000000000aa RSI: 000000000000000a RDI: ffff8800745c8110 [ 452.064012] RBP: ffff8800b6e83ba0 R08: 000000000000c802 R09: 000000000000001c [ 452.064012] R10: ffff880071096c4e R11: 0000000000000006 R12: ffff8800b6e83b18 [ 452.064012] R13: ffffffff8125d51e R14: ffff8800b6e83ba0 R15: ffff880072a589c0 [ 452.064012] FS: 00007fdc0b81e700(0000) GS:ffff8800b6e80000(0000) knlGS:0000000000000000 [ 452.064012] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 452.064012] CR2: 0000000000625208 CR3: 0000000074404000 CR4: 00000000000006a0 [ 452.064012] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 [ 452.064012] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400 [ 452.064012] Process accel-pppd (pid: 6662, threadinfo ffff88007129a000, task ffff8800744f7410) [ 452.064012] Stack: [ 452.064012] ffff8800b6e83bb0 ffffffff81256559 ffff8800b6e83bc0 ffffffff8121c64a [ 452.064012] ffff8800b6e83bf0 ffffffff8121ec7a ffff880072a589c0 ffff880071096c62 [ 452.064012] 0000000000000011 ffffffff81430024 ffff8800b6e83c80 ffffffff8121f276 [ 452.064012] Call Trace: [ 452.064012] <IRQ> [ 452.064012] [<ffffffff81256559>] _raw_spin_lock+0xe/0x10 [ 452.064012] [<ffffffff8121c64a>] spin_lock+0x9/0xb [ 452.064012] [<ffffffff8121ec7a>] udp_queue_rcv_skb+0x186/0x269 [ 452.064012] [<ffffffff8121f276>] __udp4_lib_rcv+0x297/0x4ae [ 452.064012] [<ffffffff8121c178>] ? raw_rcv+0xe9/0xf0 [ 452.064012] [<ffffffff8121f4a7>] udp_rcv+0x1a/0x1c [ 452.064012] [<ffffffff811fe385>] ip_local_deliver_finish+0x12b/0x1a5 [ 452.064012] [<ffffffff811fe54e>] ip_local_deliver+0x53/0x84 [ 452.064012] [<ffffffff811fe1d0>] ip_rcv_finish+0x2bc/0x2f3 [ 452.064012] [<ffffffff811fe78f>] ip_rcv+0x210/0x269 [ 452.064012] [<ffffffff8101911e>] ? kvm_clock_get_cycles+0x9/0xb [ 452.064012] [<ffffffff811d88cd>] __netif_receive_skb+0x3a5/0x3f7 [ 452.064012] [<ffffffff811d8eba>] netif_receive_skb+0x57/0x5e [ 452.064012] [<ffffffff811cf30f>] ? __netdev_alloc_skb+0x1f/0x3b [ 452.064012] [<ffffffffa0049126>] virtnet_poll+0x4ba/0x5a4 [virtio_net] [ 452.064012] [<ffffffff811d9417>] net_rx_action+0x73/0x184 [ 452.064012] [<ffffffffa01b2cc2>] ? l2tp_xmit_skb+0x27a/0x4ac [l2tp_core] [ 452.064012] [<ffffffff810343b9>] __do_softirq+0xc3/0x1a8 [ 452.064012] [<ffffffff81013b56>] ? ack_APIC_irq+0x10/0x12 [ 452.064012] [<ffffffff81256559>] ? _raw_spin_lock+0xe/0x10 [ 452.064012] [<ffffffff8125e0ac>] call_softirq+0x1c/0x26 [ 452.064012] [<ffffffff81003587>] do_softirq+0x45/0x82 [ 452.064012] [<ffffffff81034667>] irq_exit+0x42/0x9c [ 452.064012] [<ffffffff8125e146>] do_IRQ+0x8e/0xa5 [ 452.064012] [<ffffffff8125676e>] common_interrupt+0x6e/0x6e [ 452.064012] <EOI> [ 452.064012] [<ffffffff810b82a1>] ? kfree+0x8a/0xa3 [ 452.064012] [<ffffffffa01b2cc2>] ? l2tp_xmit_skb+0x27a/0x4ac [l2tp_core] [ 452.064012] [<ffffffffa01b2c25>] ? l2tp_xmit_skb+0x1dd/0x4ac [l2tp_core] [ 452.064012] [<ffffffffa01c2d36>] pppol2tp_sendmsg+0x15e/0x19c [l2tp_ppp] [ 452.064012] [<ffffffff811c7872>] __sock_sendmsg_nosec+0x22/0x24 [ 452.064012] [<ffffffff811c83bd>] sock_sendmsg+0xa1/0xb6 [ 452.064012] [<ffffffff81254e88>] ? __schedule+0x5c1/0x616 [ 452.064012] [<ffffffff8103c7c6>] ? __dequeue_signal+0xb7/0x10c [ 452.064012] [<ffffffff810bbd21>] ? fget_light+0x75/0x89 [ 452.064012] [<ffffffff811c8444>] ? sockfd_lookup_light+0x20/0x56 [ 452.064012] [<ffffffff811c9b34>] sys_sendto+0x10c/0x13b [ 452.064012] [<ffffffff8125cac2>] system_call_fastpath+0x16/0x1b [ 452.064012] Code: 89 e5 72 0c 31 c0 48 81 ff 45 66 25 81 0f 92 c0 5d c3 55 b8 00 01 00 00 48 89 e5 f0 66 0f c1 07 0f b6 d4 38 d0 74 06 f3 90 8a 07 <eb> f6 5d c3 90 90 55 48 89 e5 9c 58 0f 1f 44 00 00 5d c3 55 48 [ 452.064012] Call Trace: [ 452.064012] <IRQ> [<ffffffff81256559>] _raw_spin_lock+0xe/0x10 [ 452.064012] [<ffffffff8121c64a>] spin_lock+0x9/0xb [ 452.064012] [<ffffffff8121ec7a>] udp_queue_rcv_skb+0x186/0x269 [ 452.064012] [<ffffffff8121f276>] __udp4_lib_rcv+0x297/0x4ae [ 452.064012] [<ffffffff8121c178>] ? raw_rcv+0xe9/0xf0 [ 452.064012] [<ffffffff8121f4a7>] udp_rcv+0x1a/0x1c [ 452.064012] [<ffffffff811fe385>] ip_local_deliver_finish+0x12b/0x1a5 [ 452.064012] [<ffffffff811fe54e>] ip_local_deliver+0x53/0x84 [ 452.064012] [<ffffffff811fe1d0>] ip_rcv_finish+0x2bc/0x2f3 [ 452.064012] [<ffffffff811fe78f>] ip_rcv+0x210/0x269 [ 452.064012] [<ffffffff8101911e>] ? kvm_clock_get_cycles+0x9/0xb [ 452.064012] [<ffffffff811d88cd>] __netif_receive_skb+0x3a5/0x3f7 [ 452.064012] [<ffffffff811d8eba>] netif_receive_skb+0x57/0x5e [ 452.064012] [<ffffffff811cf30f>] ? __netdev_alloc_skb+0x1f/0x3b [ 452.064012] [<ffffffffa0049126>] virtnet_poll+0x4ba/0x5a4 [virtio_net] [ 452.064012] [<ffffffff811d9417>] net_rx_action+0x73/0x184 [ 452.064012] [<ffffffffa01b2cc2>] ? l2tp_xmit_skb+0x27a/0x4ac [l2tp_core] [ 452.064012] [<ffffffff810343b9>] __do_softirq+0xc3/0x1a8 [ 452.064012] [<ffffffff81013b56>] ? ack_APIC_irq+0x10/0x12 [ 452.064012] [<ffffffff81256559>] ? _raw_spin_lock+0xe/0x10 [ 452.064012] [<ffffffff8125e0ac>] call_softirq+0x1c/0x26 [ 452.064012] [<ffffffff81003587>] do_softirq+0x45/0x82 [ 452.064012] [<ffffffff81034667>] irq_exit+0x42/0x9c [ 452.064012] [<ffffffff8125e146>] do_IRQ+0x8e/0xa5 [ 452.064012] [<ffffffff8125676e>] common_interrupt+0x6e/0x6e [ 452.064012] <EOI> [<ffffffff810b82a1>] ? kfree+0x8a/0xa3 [ 452.064012] [<ffffffffa01b2cc2>] ? l2tp_xmit_skb+0x27a/0x4ac [l2tp_core] [ 452.064012] [<ffffffffa01b2c25>] ? l2tp_xmit_skb+0x1dd/0x4ac [l2tp_core] [ 452.064012] [<ffffffffa01c2d36>] pppol2tp_sendmsg+0x15e/0x19c [l2tp_ppp] [ 452.064012] [<ffffffff811c7872>] __sock_sendmsg_nosec+0x22/0x24 [ 452.064012] [<ffffffff811c83bd>] sock_sendmsg+0xa1/0xb6 [ 452.064012] [<ffffffff81254e88>] ? __schedule+0x5c1/0x616 [ 452.064012] [<ffffffff8103c7c6>] ? __dequeue_signal+0xb7/0x10c [ 452.064012] [<ffffffff810bbd21>] ? fget_light+0x75/0x89 [ 452.064012] [<ffffffff811c8444>] ? sockfd_lookup_light+0x20/0x56 [ 452.064012] [<ffffffff811c9b34>] sys_sendto+0x10c/0x13b [ 452.064012] [<ffffffff8125cac2>] system_call_fastpath+0x16/0x1b Reported-by: François Cachereul <f.cachereul@alphalink.fr> Tested-by: François Cachereul <f.cachereul@alphalink.fr> Signed-off-by: Eric Dumazet <edumazet@google.com> Cc: James Chapman <jchapman@katalix.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2013-10-10 20:30:09 +07:00
local_bh_disable();
l2tp_xmit_skb(session, skb, session->hdr_len);
l2tp: must disable bh before calling l2tp_xmit_skb() François Cachereul made a very nice bug report and suspected the bh_lock_sock() / bh_unlok_sock() pair used in l2tp_xmit_skb() from process context was not good. This problem was added by commit 6af88da14ee284aaad6e4326da09a89191ab6165 ("l2tp: Fix locking in l2tp_core.c"). l2tp_eth_dev_xmit() runs from BH context, so we must disable BH from other l2tp_xmit_skb() users. [ 452.060011] BUG: soft lockup - CPU#1 stuck for 23s! [accel-pppd:6662] [ 452.061757] Modules linked in: l2tp_ppp l2tp_netlink l2tp_core pppoe pppox ppp_generic slhc ipv6 ext3 mbcache jbd virtio_balloon xfs exportfs dm_mod virtio_blk ata_generic virtio_net floppy ata_piix libata virtio_pci virtio_ring virtio [last unloaded: scsi_wait_scan] [ 452.064012] CPU 1 [ 452.080015] BUG: soft lockup - CPU#2 stuck for 23s! [accel-pppd:6643] [ 452.080015] CPU 2 [ 452.080015] [ 452.080015] Pid: 6643, comm: accel-pppd Not tainted 3.2.46.mini #1 Bochs Bochs [ 452.080015] RIP: 0010:[<ffffffff81059f6c>] [<ffffffff81059f6c>] do_raw_spin_lock+0x17/0x1f [ 452.080015] RSP: 0018:ffff88007125fc18 EFLAGS: 00000293 [ 452.080015] RAX: 000000000000aba9 RBX: ffffffff811d0703 RCX: 0000000000000000 [ 452.080015] RDX: 00000000000000ab RSI: ffff8800711f6896 RDI: ffff8800745c8110 [ 452.080015] RBP: ffff88007125fc18 R08: 0000000000000020 R09: 0000000000000000 [ 452.080015] R10: 0000000000000000 R11: 0000000000000280 R12: 0000000000000286 [ 452.080015] R13: 0000000000000020 R14: 0000000000000240 R15: 0000000000000000 [ 452.080015] FS: 00007fdc0cc24700(0000) GS:ffff8800b6f00000(0000) knlGS:0000000000000000 [ 452.080015] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 452.080015] CR2: 00007fdb054899b8 CR3: 0000000074404000 CR4: 00000000000006a0 [ 452.080015] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 [ 452.080015] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400 [ 452.080015] Process accel-pppd (pid: 6643, threadinfo ffff88007125e000, task ffff8800b27e6dd0) [ 452.080015] Stack: [ 452.080015] ffff88007125fc28 ffffffff81256559 ffff88007125fc98 ffffffffa01b2bd1 [ 452.080015] ffff88007125fc58 000000000000000c 00000000029490d0 0000009c71dbe25e [ 452.080015] 000000000000005c 000000080000000e 0000000000000000 ffff880071170600 [ 452.080015] Call Trace: [ 452.080015] [<ffffffff81256559>] _raw_spin_lock+0xe/0x10 [ 452.080015] [<ffffffffa01b2bd1>] l2tp_xmit_skb+0x189/0x4ac [l2tp_core] [ 452.080015] [<ffffffffa01c2d36>] pppol2tp_sendmsg+0x15e/0x19c [l2tp_ppp] [ 452.080015] [<ffffffff811c7872>] __sock_sendmsg_nosec+0x22/0x24 [ 452.080015] [<ffffffff811c83bd>] sock_sendmsg+0xa1/0xb6 [ 452.080015] [<ffffffff81254e88>] ? __schedule+0x5c1/0x616 [ 452.080015] [<ffffffff8103c7c6>] ? __dequeue_signal+0xb7/0x10c [ 452.080015] [<ffffffff810bbd21>] ? fget_light+0x75/0x89 [ 452.080015] [<ffffffff811c8444>] ? sockfd_lookup_light+0x20/0x56 [ 452.080015] [<ffffffff811c9b34>] sys_sendto+0x10c/0x13b [ 452.080015] [<ffffffff8125cac2>] system_call_fastpath+0x16/0x1b [ 452.080015] Code: 81 48 89 e5 72 0c 31 c0 48 81 ff 45 66 25 81 0f 92 c0 5d c3 55 b8 00 01 00 00 48 89 e5 f0 66 0f c1 07 0f b6 d4 38 d0 74 06 f3 90 <8a> 07 eb f6 5d c3 90 90 55 48 89 e5 9c 58 0f 1f 44 00 00 5d c3 [ 452.080015] Call Trace: [ 452.080015] [<ffffffff81256559>] _raw_spin_lock+0xe/0x10 [ 452.080015] [<ffffffffa01b2bd1>] l2tp_xmit_skb+0x189/0x4ac [l2tp_core] [ 452.080015] [<ffffffffa01c2d36>] pppol2tp_sendmsg+0x15e/0x19c [l2tp_ppp] [ 452.080015] [<ffffffff811c7872>] __sock_sendmsg_nosec+0x22/0x24 [ 452.080015] [<ffffffff811c83bd>] sock_sendmsg+0xa1/0xb6 [ 452.080015] [<ffffffff81254e88>] ? __schedule+0x5c1/0x616 [ 452.080015] [<ffffffff8103c7c6>] ? __dequeue_signal+0xb7/0x10c [ 452.080015] [<ffffffff810bbd21>] ? fget_light+0x75/0x89 [ 452.080015] [<ffffffff811c8444>] ? sockfd_lookup_light+0x20/0x56 [ 452.080015] [<ffffffff811c9b34>] sys_sendto+0x10c/0x13b [ 452.080015] [<ffffffff8125cac2>] system_call_fastpath+0x16/0x1b [ 452.064012] [ 452.064012] Pid: 6662, comm: accel-pppd Not tainted 3.2.46.mini #1 Bochs Bochs [ 452.064012] RIP: 0010:[<ffffffff81059f6e>] [<ffffffff81059f6e>] do_raw_spin_lock+0x19/0x1f [ 452.064012] RSP: 0018:ffff8800b6e83ba0 EFLAGS: 00000297 [ 452.064012] RAX: 000000000000aaa9 RBX: ffff8800b6e83b40 RCX: 0000000000000002 [ 452.064012] RDX: 00000000000000aa RSI: 000000000000000a RDI: ffff8800745c8110 [ 452.064012] RBP: ffff8800b6e83ba0 R08: 000000000000c802 R09: 000000000000001c [ 452.064012] R10: ffff880071096c4e R11: 0000000000000006 R12: ffff8800b6e83b18 [ 452.064012] R13: ffffffff8125d51e R14: ffff8800b6e83ba0 R15: ffff880072a589c0 [ 452.064012] FS: 00007fdc0b81e700(0000) GS:ffff8800b6e80000(0000) knlGS:0000000000000000 [ 452.064012] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 452.064012] CR2: 0000000000625208 CR3: 0000000074404000 CR4: 00000000000006a0 [ 452.064012] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 [ 452.064012] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400 [ 452.064012] Process accel-pppd (pid: 6662, threadinfo ffff88007129a000, task ffff8800744f7410) [ 452.064012] Stack: [ 452.064012] ffff8800b6e83bb0 ffffffff81256559 ffff8800b6e83bc0 ffffffff8121c64a [ 452.064012] ffff8800b6e83bf0 ffffffff8121ec7a ffff880072a589c0 ffff880071096c62 [ 452.064012] 0000000000000011 ffffffff81430024 ffff8800b6e83c80 ffffffff8121f276 [ 452.064012] Call Trace: [ 452.064012] <IRQ> [ 452.064012] [<ffffffff81256559>] _raw_spin_lock+0xe/0x10 [ 452.064012] [<ffffffff8121c64a>] spin_lock+0x9/0xb [ 452.064012] [<ffffffff8121ec7a>] udp_queue_rcv_skb+0x186/0x269 [ 452.064012] [<ffffffff8121f276>] __udp4_lib_rcv+0x297/0x4ae [ 452.064012] [<ffffffff8121c178>] ? raw_rcv+0xe9/0xf0 [ 452.064012] [<ffffffff8121f4a7>] udp_rcv+0x1a/0x1c [ 452.064012] [<ffffffff811fe385>] ip_local_deliver_finish+0x12b/0x1a5 [ 452.064012] [<ffffffff811fe54e>] ip_local_deliver+0x53/0x84 [ 452.064012] [<ffffffff811fe1d0>] ip_rcv_finish+0x2bc/0x2f3 [ 452.064012] [<ffffffff811fe78f>] ip_rcv+0x210/0x269 [ 452.064012] [<ffffffff8101911e>] ? kvm_clock_get_cycles+0x9/0xb [ 452.064012] [<ffffffff811d88cd>] __netif_receive_skb+0x3a5/0x3f7 [ 452.064012] [<ffffffff811d8eba>] netif_receive_skb+0x57/0x5e [ 452.064012] [<ffffffff811cf30f>] ? __netdev_alloc_skb+0x1f/0x3b [ 452.064012] [<ffffffffa0049126>] virtnet_poll+0x4ba/0x5a4 [virtio_net] [ 452.064012] [<ffffffff811d9417>] net_rx_action+0x73/0x184 [ 452.064012] [<ffffffffa01b2cc2>] ? l2tp_xmit_skb+0x27a/0x4ac [l2tp_core] [ 452.064012] [<ffffffff810343b9>] __do_softirq+0xc3/0x1a8 [ 452.064012] [<ffffffff81013b56>] ? ack_APIC_irq+0x10/0x12 [ 452.064012] [<ffffffff81256559>] ? _raw_spin_lock+0xe/0x10 [ 452.064012] [<ffffffff8125e0ac>] call_softirq+0x1c/0x26 [ 452.064012] [<ffffffff81003587>] do_softirq+0x45/0x82 [ 452.064012] [<ffffffff81034667>] irq_exit+0x42/0x9c [ 452.064012] [<ffffffff8125e146>] do_IRQ+0x8e/0xa5 [ 452.064012] [<ffffffff8125676e>] common_interrupt+0x6e/0x6e [ 452.064012] <EOI> [ 452.064012] [<ffffffff810b82a1>] ? kfree+0x8a/0xa3 [ 452.064012] [<ffffffffa01b2cc2>] ? l2tp_xmit_skb+0x27a/0x4ac [l2tp_core] [ 452.064012] [<ffffffffa01b2c25>] ? l2tp_xmit_skb+0x1dd/0x4ac [l2tp_core] [ 452.064012] [<ffffffffa01c2d36>] pppol2tp_sendmsg+0x15e/0x19c [l2tp_ppp] [ 452.064012] [<ffffffff811c7872>] __sock_sendmsg_nosec+0x22/0x24 [ 452.064012] [<ffffffff811c83bd>] sock_sendmsg+0xa1/0xb6 [ 452.064012] [<ffffffff81254e88>] ? __schedule+0x5c1/0x616 [ 452.064012] [<ffffffff8103c7c6>] ? __dequeue_signal+0xb7/0x10c [ 452.064012] [<ffffffff810bbd21>] ? fget_light+0x75/0x89 [ 452.064012] [<ffffffff811c8444>] ? sockfd_lookup_light+0x20/0x56 [ 452.064012] [<ffffffff811c9b34>] sys_sendto+0x10c/0x13b [ 452.064012] [<ffffffff8125cac2>] system_call_fastpath+0x16/0x1b [ 452.064012] Code: 89 e5 72 0c 31 c0 48 81 ff 45 66 25 81 0f 92 c0 5d c3 55 b8 00 01 00 00 48 89 e5 f0 66 0f c1 07 0f b6 d4 38 d0 74 06 f3 90 8a 07 <eb> f6 5d c3 90 90 55 48 89 e5 9c 58 0f 1f 44 00 00 5d c3 55 48 [ 452.064012] Call Trace: [ 452.064012] <IRQ> [<ffffffff81256559>] _raw_spin_lock+0xe/0x10 [ 452.064012] [<ffffffff8121c64a>] spin_lock+0x9/0xb [ 452.064012] [<ffffffff8121ec7a>] udp_queue_rcv_skb+0x186/0x269 [ 452.064012] [<ffffffff8121f276>] __udp4_lib_rcv+0x297/0x4ae [ 452.064012] [<ffffffff8121c178>] ? raw_rcv+0xe9/0xf0 [ 452.064012] [<ffffffff8121f4a7>] udp_rcv+0x1a/0x1c [ 452.064012] [<ffffffff811fe385>] ip_local_deliver_finish+0x12b/0x1a5 [ 452.064012] [<ffffffff811fe54e>] ip_local_deliver+0x53/0x84 [ 452.064012] [<ffffffff811fe1d0>] ip_rcv_finish+0x2bc/0x2f3 [ 452.064012] [<ffffffff811fe78f>] ip_rcv+0x210/0x269 [ 452.064012] [<ffffffff8101911e>] ? kvm_clock_get_cycles+0x9/0xb [ 452.064012] [<ffffffff811d88cd>] __netif_receive_skb+0x3a5/0x3f7 [ 452.064012] [<ffffffff811d8eba>] netif_receive_skb+0x57/0x5e [ 452.064012] [<ffffffff811cf30f>] ? __netdev_alloc_skb+0x1f/0x3b [ 452.064012] [<ffffffffa0049126>] virtnet_poll+0x4ba/0x5a4 [virtio_net] [ 452.064012] [<ffffffff811d9417>] net_rx_action+0x73/0x184 [ 452.064012] [<ffffffffa01b2cc2>] ? l2tp_xmit_skb+0x27a/0x4ac [l2tp_core] [ 452.064012] [<ffffffff810343b9>] __do_softirq+0xc3/0x1a8 [ 452.064012] [<ffffffff81013b56>] ? ack_APIC_irq+0x10/0x12 [ 452.064012] [<ffffffff81256559>] ? _raw_spin_lock+0xe/0x10 [ 452.064012] [<ffffffff8125e0ac>] call_softirq+0x1c/0x26 [ 452.064012] [<ffffffff81003587>] do_softirq+0x45/0x82 [ 452.064012] [<ffffffff81034667>] irq_exit+0x42/0x9c [ 452.064012] [<ffffffff8125e146>] do_IRQ+0x8e/0xa5 [ 452.064012] [<ffffffff8125676e>] common_interrupt+0x6e/0x6e [ 452.064012] <EOI> [<ffffffff810b82a1>] ? kfree+0x8a/0xa3 [ 452.064012] [<ffffffffa01b2cc2>] ? l2tp_xmit_skb+0x27a/0x4ac [l2tp_core] [ 452.064012] [<ffffffffa01b2c25>] ? l2tp_xmit_skb+0x1dd/0x4ac [l2tp_core] [ 452.064012] [<ffffffffa01c2d36>] pppol2tp_sendmsg+0x15e/0x19c [l2tp_ppp] [ 452.064012] [<ffffffff811c7872>] __sock_sendmsg_nosec+0x22/0x24 [ 452.064012] [<ffffffff811c83bd>] sock_sendmsg+0xa1/0xb6 [ 452.064012] [<ffffffff81254e88>] ? __schedule+0x5c1/0x616 [ 452.064012] [<ffffffff8103c7c6>] ? __dequeue_signal+0xb7/0x10c [ 452.064012] [<ffffffff810bbd21>] ? fget_light+0x75/0x89 [ 452.064012] [<ffffffff811c8444>] ? sockfd_lookup_light+0x20/0x56 [ 452.064012] [<ffffffff811c9b34>] sys_sendto+0x10c/0x13b [ 452.064012] [<ffffffff8125cac2>] system_call_fastpath+0x16/0x1b Reported-by: François Cachereul <f.cachereul@alphalink.fr> Tested-by: François Cachereul <f.cachereul@alphalink.fr> Signed-off-by: Eric Dumazet <edumazet@google.com> Cc: James Chapman <jchapman@katalix.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2013-10-10 20:30:09 +07:00
local_bh_enable();
sock_put(sk);
return 1;
abort_put_sess:
sock_put(sk);
abort:
/* Free the original skb */
kfree_skb(skb);
return 1;
}
/*****************************************************************************
* Session (and tunnel control) socket create/destroy.
*****************************************************************************/
l2tp: fix race in pppol2tp_release with session object destroy pppol2tp_release uses call_rcu to put the final ref on its socket. But the session object doesn't hold a ref on the session socket so may be freed while the pppol2tp_put_sk RCU callback is scheduled. Fix this by having the session hold a ref on its socket until the session is destroyed. It is this ref that is dropped via call_rcu. Sessions are also deleted via l2tp_tunnel_closeall. This must now also put the final ref via call_rcu. So move the call_rcu call site into pppol2tp_session_close so that this happens in both destroy paths. A common destroy path should really be implemented, perhaps with l2tp_tunnel_closeall calling l2tp_session_delete like pppol2tp_release does, but this will be looked at later. ODEBUG: activate active (active state 1) object type: rcu_head hint: (null) WARNING: CPU: 3 PID: 13407 at lib/debugobjects.c:291 debug_print_object+0x166/0x220 Modules linked in: CPU: 3 PID: 13407 Comm: syzbot_19c09769 Not tainted 4.16.0-rc2+ #38 Hardware name: innotek GmbH VirtualBox/VirtualBox, BIOS VirtualBox 12/01/2006 RIP: 0010:debug_print_object+0x166/0x220 RSP: 0018:ffff880013647a00 EFLAGS: 00010082 RAX: dffffc0000000008 RBX: 0000000000000003 RCX: ffffffff814d3333 RDX: 0000000000000000 RSI: 0000000000000001 RDI: ffff88001a59f6d0 RBP: ffff880013647a40 R08: 0000000000000000 R09: 0000000000000001 R10: ffff8800136479a8 R11: 0000000000000000 R12: 0000000000000001 R13: ffffffff86161420 R14: ffffffff85648b60 R15: 0000000000000000 FS: 0000000000000000(0000) GS:ffff88001a580000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 0000000020e77000 CR3: 0000000006022000 CR4: 00000000000006e0 Call Trace: debug_object_activate+0x38b/0x530 ? debug_object_assert_init+0x3b0/0x3b0 ? __mutex_unlock_slowpath+0x85/0x8b0 ? pppol2tp_session_destruct+0x110/0x110 __call_rcu.constprop.66+0x39/0x890 ? __call_rcu.constprop.66+0x39/0x890 call_rcu_sched+0x17/0x20 pppol2tp_release+0x2c7/0x440 ? fcntl_setlk+0xca0/0xca0 ? sock_alloc_file+0x340/0x340 sock_release+0x92/0x1e0 sock_close+0x1b/0x20 __fput+0x296/0x6e0 ____fput+0x1a/0x20 task_work_run+0x127/0x1a0 do_exit+0x7f9/0x2ce0 ? SYSC_connect+0x212/0x310 ? mm_update_next_owner+0x690/0x690 ? up_read+0x1f/0x40 ? __do_page_fault+0x3c8/0xca0 do_group_exit+0x10d/0x330 ? do_group_exit+0x330/0x330 SyS_exit_group+0x22/0x30 do_syscall_64+0x1e0/0x730 ? trace_hardirqs_off_thunk+0x1a/0x1c entry_SYSCALL_64_after_hwframe+0x42/0xb7 RIP: 0033:0x7f362e471259 RSP: 002b:00007ffe389abe08 EFLAGS: 00000202 ORIG_RAX: 00000000000000e7 RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00007f362e471259 RDX: 00007f362e471259 RSI: 000000000000002e RDI: 0000000000000000 RBP: 00007ffe389abe30 R08: 0000000000000000 R09: 00007f362e944270 R10: 0000000000000000 R11: 0000000000000202 R12: 0000000000400b60 R13: 00007ffe389abf50 R14: 0000000000000000 R15: 0000000000000000 Code: 8d 3c dd a0 8f 64 85 48 89 fa 48 c1 ea 03 80 3c 02 00 75 7b 48 8b 14 dd a0 8f 64 85 4c 89 f6 48 c7 c7 20 85 64 85 e 8 2a 55 14 ff <0f> 0b 83 05 ad 2a 68 04 01 48 83 c4 18 5b 41 5c 41 5d 41 5e 41 Fixes: ee40fb2e1eb5b ("l2tp: protect sock pointer of struct pppol2tp_session with RCU") Signed-off-by: James Chapman <jchapman@katalix.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-02-24 00:45:46 +07:00
static void pppol2tp_put_sk(struct rcu_head *head)
{
struct pppol2tp_session *ps;
ps = container_of(head, typeof(*ps), rcu);
sock_put(ps->__sk);
}
/* Really kill the session socket. (Called from sock_put() if
* refcnt == 0.)
*/
static void pppol2tp_session_destruct(struct sock *sk)
{
struct l2tp_session *session = sk->sk_user_data;
skb_queue_purge(&sk->sk_receive_queue);
skb_queue_purge(&sk->sk_write_queue);
if (session) {
sk->sk_user_data = NULL;
BUG_ON(session->magic != L2TP_SESSION_MAGIC);
l2tp_session_dec_refcount(session);
}
}
/* Called when the PPPoX socket (session) is closed.
*/
static int pppol2tp_release(struct socket *sock)
{
struct sock *sk = sock->sk;
struct l2tp_session *session;
int error;
if (!sk)
return 0;
error = -EBADF;
lock_sock(sk);
if (sock_flag(sk, SOCK_DEAD) != 0)
goto error;
pppox_unbind_sock(sk);
/* Signal the death of the socket. */
sk->sk_state = PPPOX_DEAD;
sock_orphan(sk);
sock->sk = NULL;
session = pppol2tp_sock_to_session(sk);
l2tp: fix race in pppol2tp_release with session object destroy pppol2tp_release uses call_rcu to put the final ref on its socket. But the session object doesn't hold a ref on the session socket so may be freed while the pppol2tp_put_sk RCU callback is scheduled. Fix this by having the session hold a ref on its socket until the session is destroyed. It is this ref that is dropped via call_rcu. Sessions are also deleted via l2tp_tunnel_closeall. This must now also put the final ref via call_rcu. So move the call_rcu call site into pppol2tp_session_close so that this happens in both destroy paths. A common destroy path should really be implemented, perhaps with l2tp_tunnel_closeall calling l2tp_session_delete like pppol2tp_release does, but this will be looked at later. ODEBUG: activate active (active state 1) object type: rcu_head hint: (null) WARNING: CPU: 3 PID: 13407 at lib/debugobjects.c:291 debug_print_object+0x166/0x220 Modules linked in: CPU: 3 PID: 13407 Comm: syzbot_19c09769 Not tainted 4.16.0-rc2+ #38 Hardware name: innotek GmbH VirtualBox/VirtualBox, BIOS VirtualBox 12/01/2006 RIP: 0010:debug_print_object+0x166/0x220 RSP: 0018:ffff880013647a00 EFLAGS: 00010082 RAX: dffffc0000000008 RBX: 0000000000000003 RCX: ffffffff814d3333 RDX: 0000000000000000 RSI: 0000000000000001 RDI: ffff88001a59f6d0 RBP: ffff880013647a40 R08: 0000000000000000 R09: 0000000000000001 R10: ffff8800136479a8 R11: 0000000000000000 R12: 0000000000000001 R13: ffffffff86161420 R14: ffffffff85648b60 R15: 0000000000000000 FS: 0000000000000000(0000) GS:ffff88001a580000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 0000000020e77000 CR3: 0000000006022000 CR4: 00000000000006e0 Call Trace: debug_object_activate+0x38b/0x530 ? debug_object_assert_init+0x3b0/0x3b0 ? __mutex_unlock_slowpath+0x85/0x8b0 ? pppol2tp_session_destruct+0x110/0x110 __call_rcu.constprop.66+0x39/0x890 ? __call_rcu.constprop.66+0x39/0x890 call_rcu_sched+0x17/0x20 pppol2tp_release+0x2c7/0x440 ? fcntl_setlk+0xca0/0xca0 ? sock_alloc_file+0x340/0x340 sock_release+0x92/0x1e0 sock_close+0x1b/0x20 __fput+0x296/0x6e0 ____fput+0x1a/0x20 task_work_run+0x127/0x1a0 do_exit+0x7f9/0x2ce0 ? SYSC_connect+0x212/0x310 ? mm_update_next_owner+0x690/0x690 ? up_read+0x1f/0x40 ? __do_page_fault+0x3c8/0xca0 do_group_exit+0x10d/0x330 ? do_group_exit+0x330/0x330 SyS_exit_group+0x22/0x30 do_syscall_64+0x1e0/0x730 ? trace_hardirqs_off_thunk+0x1a/0x1c entry_SYSCALL_64_after_hwframe+0x42/0xb7 RIP: 0033:0x7f362e471259 RSP: 002b:00007ffe389abe08 EFLAGS: 00000202 ORIG_RAX: 00000000000000e7 RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00007f362e471259 RDX: 00007f362e471259 RSI: 000000000000002e RDI: 0000000000000000 RBP: 00007ffe389abe30 R08: 0000000000000000 R09: 00007f362e944270 R10: 0000000000000000 R11: 0000000000000202 R12: 0000000000400b60 R13: 00007ffe389abf50 R14: 0000000000000000 R15: 0000000000000000 Code: 8d 3c dd a0 8f 64 85 48 89 fa 48 c1 ea 03 80 3c 02 00 75 7b 48 8b 14 dd a0 8f 64 85 4c 89 f6 48 c7 c7 20 85 64 85 e 8 2a 55 14 ff <0f> 0b 83 05 ad 2a 68 04 01 48 83 c4 18 5b 41 5c 41 5d 41 5e 41 Fixes: ee40fb2e1eb5b ("l2tp: protect sock pointer of struct pppol2tp_session with RCU") Signed-off-by: James Chapman <jchapman@katalix.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-02-24 00:45:46 +07:00
if (session) {
l2tp: fix refcount leakage on PPPoL2TP sockets Commit d02ba2a6110c ("l2tp: fix race in pppol2tp_release with session object destroy") tried to fix a race condition where a PPPoL2TP socket would disappear while the L2TP session was still using it. However, it missed the root issue which is that an L2TP session may accept to be reconnected if its associated socket has entered the release process. The tentative fix makes the session hold the socket it is connected to. That saves the kernel from crashing, but introduces refcount leakage, preventing the socket from completing the release process. Once stalled, everything the socket depends on can't be released anymore, including the L2TP session and the l2tp_ppp module. The root issue is that, when releasing a connected PPPoL2TP socket, the session's ->sk pointer (RCU-protected) is reset to NULL and we have to wait for a grace period before destroying the socket. The socket drops the session in its ->sk_destruct callback function, so the session will exist until the last reference on the socket is dropped. Therefore, there is a time frame where pppol2tp_connect() may accept reconnecting a session, as it only checks ->sk to figure out if the session is connected. This time frame is shortened by the fact that pppol2tp_release() calls l2tp_session_delete(), making the session unreachable before resetting ->sk. However, pppol2tp_connect() may grab the session before it gets unhashed by l2tp_session_delete(), but it may test ->sk after the later got reset. The race is not so hard to trigger and syzbot found a pretty reliable reproducer: https://syzkaller.appspot.com/bug?id=418578d2a4389074524e04d641eacb091961b2cf Before d02ba2a6110c, another race could let pppol2tp_release() overwrite the ->__sk pointer of an L2TP session, thus tricking pppol2tp_put_sk() into calling sock_put() on a socket that is different than the one for which pppol2tp_release() was originally called. To get there, we had to trigger the race described above, therefore having one PPPoL2TP socket being released, while the session it is connected to is reconnecting to a different PPPoL2TP socket. When releasing this new socket fast enough, pppol2tp_release() overwrites the session's ->__sk pointer with the address of the new socket, before the first pppol2tp_put_sk() call gets scheduled. Then the pppol2tp_put_sk() call invoked by the original socket will sock_put() the new socket, potentially dropping its last reference. When the second pppol2tp_put_sk() finally runs, its socket has already been freed. With d02ba2a6110c, the session takes a reference on both sockets. Furthermore, the session's ->sk pointer is reset in the pppol2tp_session_close() callback function rather than in pppol2tp_release(). Therefore, ->__sk can't be overwritten and pppol2tp_put_sk() is called only once (l2tp_session_delete() will only run pppol2tp_session_close() once, to protect the session against concurrent deletion requests). Now pppol2tp_put_sk() will properly sock_put() the original socket, but the new socket will remain, as l2tp_session_delete() prevented the release process from completing. Here, we don't depend on the ->__sk race to trigger the bug. Getting into the pppol2tp_connect() race is enough to leak the reference, no matter when new socket is released. So it all boils down to pppol2tp_connect() failing to realise that the session has already been connected. This patch drops the unneeded extra reference counting (mostly reverting d02ba2a6110c) and checks that neither ->sk nor ->__sk is set before allowing a session to be connected. Fixes: d02ba2a6110c ("l2tp: fix race in pppol2tp_release with session object destroy") Signed-off-by: Guillaume Nault <g.nault@alphalink.fr> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-06-04 23:52:19 +07:00
struct pppol2tp_session *ps;
l2tp: initialise PPP sessions before registering them pppol2tp_connect() initialises L2TP sessions after they've been exposed to the rest of the system by l2tp_session_register(). This puts sessions into transient states that are the source of several races, in particular with session's deletion path. This patch centralises the initialisation code into pppol2tp_session_init(), which is called before the registration phase. The only field that can't be set before session registration is the pppol2tp socket pointer, which has already been converted to RCU. So pppol2tp_connect() should now be race-free. The session's .session_close() callback is now set before registration. Therefore, it's always called when l2tp_core deletes the session, even if it was created by pppol2tp_session_create() and hasn't been plugged to a pppol2tp socket yet. That'd prevent session free because the extra reference taken by pppol2tp_session_close() wouldn't be dropped by the socket's ->sk_destruct() callback (pppol2tp_session_destruct()). We could set .session_close() only while connecting a session to its pppol2tp socket, or teach pppol2tp_session_close() to avoid grabbing a reference when the session isn't connected, but that'd require adding some form of synchronisation to be race free. Instead of that, we can just let the pppol2tp socket hold a reference on the session as soon as it starts depending on it (that is, in pppol2tp_connect()). Then we don't need to utilise pppol2tp_session_close() to hold a reference at the last moment to prevent l2tp_core from dropping it. When releasing the socket, pppol2tp_release() now deletes the session using the standard l2tp_session_delete() function, instead of merely removing it from hash tables. l2tp_session_delete() drops the reference the sessions holds on itself, but also makes sure it doesn't remove a session twice. So it can safely be called, even if l2tp_core already tried, or is concurrently trying, to remove the session. Finally, pppol2tp_session_destruct() drops the reference held by the socket. Fixes: fd558d186df2 ("l2tp: Split pppol2tp patch into separate l2tp and ppp parts") Signed-off-by: Guillaume Nault <g.nault@alphalink.fr> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-10-27 21:51:52 +07:00
l2tp_session_delete(session);
l2tp: fix refcount leakage on PPPoL2TP sockets Commit d02ba2a6110c ("l2tp: fix race in pppol2tp_release with session object destroy") tried to fix a race condition where a PPPoL2TP socket would disappear while the L2TP session was still using it. However, it missed the root issue which is that an L2TP session may accept to be reconnected if its associated socket has entered the release process. The tentative fix makes the session hold the socket it is connected to. That saves the kernel from crashing, but introduces refcount leakage, preventing the socket from completing the release process. Once stalled, everything the socket depends on can't be released anymore, including the L2TP session and the l2tp_ppp module. The root issue is that, when releasing a connected PPPoL2TP socket, the session's ->sk pointer (RCU-protected) is reset to NULL and we have to wait for a grace period before destroying the socket. The socket drops the session in its ->sk_destruct callback function, so the session will exist until the last reference on the socket is dropped. Therefore, there is a time frame where pppol2tp_connect() may accept reconnecting a session, as it only checks ->sk to figure out if the session is connected. This time frame is shortened by the fact that pppol2tp_release() calls l2tp_session_delete(), making the session unreachable before resetting ->sk. However, pppol2tp_connect() may grab the session before it gets unhashed by l2tp_session_delete(), but it may test ->sk after the later got reset. The race is not so hard to trigger and syzbot found a pretty reliable reproducer: https://syzkaller.appspot.com/bug?id=418578d2a4389074524e04d641eacb091961b2cf Before d02ba2a6110c, another race could let pppol2tp_release() overwrite the ->__sk pointer of an L2TP session, thus tricking pppol2tp_put_sk() into calling sock_put() on a socket that is different than the one for which pppol2tp_release() was originally called. To get there, we had to trigger the race described above, therefore having one PPPoL2TP socket being released, while the session it is connected to is reconnecting to a different PPPoL2TP socket. When releasing this new socket fast enough, pppol2tp_release() overwrites the session's ->__sk pointer with the address of the new socket, before the first pppol2tp_put_sk() call gets scheduled. Then the pppol2tp_put_sk() call invoked by the original socket will sock_put() the new socket, potentially dropping its last reference. When the second pppol2tp_put_sk() finally runs, its socket has already been freed. With d02ba2a6110c, the session takes a reference on both sockets. Furthermore, the session's ->sk pointer is reset in the pppol2tp_session_close() callback function rather than in pppol2tp_release(). Therefore, ->__sk can't be overwritten and pppol2tp_put_sk() is called only once (l2tp_session_delete() will only run pppol2tp_session_close() once, to protect the session against concurrent deletion requests). Now pppol2tp_put_sk() will properly sock_put() the original socket, but the new socket will remain, as l2tp_session_delete() prevented the release process from completing. Here, we don't depend on the ->__sk race to trigger the bug. Getting into the pppol2tp_connect() race is enough to leak the reference, no matter when new socket is released. So it all boils down to pppol2tp_connect() failing to realise that the session has already been connected. This patch drops the unneeded extra reference counting (mostly reverting d02ba2a6110c) and checks that neither ->sk nor ->__sk is set before allowing a session to be connected. Fixes: d02ba2a6110c ("l2tp: fix race in pppol2tp_release with session object destroy") Signed-off-by: Guillaume Nault <g.nault@alphalink.fr> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-06-04 23:52:19 +07:00
ps = l2tp_session_priv(session);
mutex_lock(&ps->sk_lock);
ps->__sk = rcu_dereference_protected(ps->sk,
lockdep_is_held(&ps->sk_lock));
RCU_INIT_POINTER(ps->sk, NULL);
mutex_unlock(&ps->sk_lock);
call_rcu(&ps->rcu, pppol2tp_put_sk);
/* Rely on the sock_put() call at the end of the function for
* dropping the reference held by pppol2tp_sock_to_session().
* The last reference will be dropped by pppol2tp_put_sk().
*/
}
l2tp: fix race in pppol2tp_release with session object destroy pppol2tp_release uses call_rcu to put the final ref on its socket. But the session object doesn't hold a ref on the session socket so may be freed while the pppol2tp_put_sk RCU callback is scheduled. Fix this by having the session hold a ref on its socket until the session is destroyed. It is this ref that is dropped via call_rcu. Sessions are also deleted via l2tp_tunnel_closeall. This must now also put the final ref via call_rcu. So move the call_rcu call site into pppol2tp_session_close so that this happens in both destroy paths. A common destroy path should really be implemented, perhaps with l2tp_tunnel_closeall calling l2tp_session_delete like pppol2tp_release does, but this will be looked at later. ODEBUG: activate active (active state 1) object type: rcu_head hint: (null) WARNING: CPU: 3 PID: 13407 at lib/debugobjects.c:291 debug_print_object+0x166/0x220 Modules linked in: CPU: 3 PID: 13407 Comm: syzbot_19c09769 Not tainted 4.16.0-rc2+ #38 Hardware name: innotek GmbH VirtualBox/VirtualBox, BIOS VirtualBox 12/01/2006 RIP: 0010:debug_print_object+0x166/0x220 RSP: 0018:ffff880013647a00 EFLAGS: 00010082 RAX: dffffc0000000008 RBX: 0000000000000003 RCX: ffffffff814d3333 RDX: 0000000000000000 RSI: 0000000000000001 RDI: ffff88001a59f6d0 RBP: ffff880013647a40 R08: 0000000000000000 R09: 0000000000000001 R10: ffff8800136479a8 R11: 0000000000000000 R12: 0000000000000001 R13: ffffffff86161420 R14: ffffffff85648b60 R15: 0000000000000000 FS: 0000000000000000(0000) GS:ffff88001a580000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 0000000020e77000 CR3: 0000000006022000 CR4: 00000000000006e0 Call Trace: debug_object_activate+0x38b/0x530 ? debug_object_assert_init+0x3b0/0x3b0 ? __mutex_unlock_slowpath+0x85/0x8b0 ? pppol2tp_session_destruct+0x110/0x110 __call_rcu.constprop.66+0x39/0x890 ? __call_rcu.constprop.66+0x39/0x890 call_rcu_sched+0x17/0x20 pppol2tp_release+0x2c7/0x440 ? fcntl_setlk+0xca0/0xca0 ? sock_alloc_file+0x340/0x340 sock_release+0x92/0x1e0 sock_close+0x1b/0x20 __fput+0x296/0x6e0 ____fput+0x1a/0x20 task_work_run+0x127/0x1a0 do_exit+0x7f9/0x2ce0 ? SYSC_connect+0x212/0x310 ? mm_update_next_owner+0x690/0x690 ? up_read+0x1f/0x40 ? __do_page_fault+0x3c8/0xca0 do_group_exit+0x10d/0x330 ? do_group_exit+0x330/0x330 SyS_exit_group+0x22/0x30 do_syscall_64+0x1e0/0x730 ? trace_hardirqs_off_thunk+0x1a/0x1c entry_SYSCALL_64_after_hwframe+0x42/0xb7 RIP: 0033:0x7f362e471259 RSP: 002b:00007ffe389abe08 EFLAGS: 00000202 ORIG_RAX: 00000000000000e7 RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00007f362e471259 RDX: 00007f362e471259 RSI: 000000000000002e RDI: 0000000000000000 RBP: 00007ffe389abe30 R08: 0000000000000000 R09: 00007f362e944270 R10: 0000000000000000 R11: 0000000000000202 R12: 0000000000400b60 R13: 00007ffe389abf50 R14: 0000000000000000 R15: 0000000000000000 Code: 8d 3c dd a0 8f 64 85 48 89 fa 48 c1 ea 03 80 3c 02 00 75 7b 48 8b 14 dd a0 8f 64 85 4c 89 f6 48 c7 c7 20 85 64 85 e 8 2a 55 14 ff <0f> 0b 83 05 ad 2a 68 04 01 48 83 c4 18 5b 41 5c 41 5d 41 5e 41 Fixes: ee40fb2e1eb5b ("l2tp: protect sock pointer of struct pppol2tp_session with RCU") Signed-off-by: James Chapman <jchapman@katalix.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-02-24 00:45:46 +07:00
release_sock(sk);
/* This will delete the session context via
* pppol2tp_session_destruct() if the socket's refcnt drops to
* zero.
*/
sock_put(sk);
return 0;
error:
release_sock(sk);
return error;
}
static struct proto pppol2tp_sk_proto = {
.name = "PPPOL2TP",
.owner = THIS_MODULE,
.obj_size = sizeof(struct pppox_sock),
};
static int pppol2tp_backlog_recv(struct sock *sk, struct sk_buff *skb)
{
int rc;
rc = l2tp_udp_encap_recv(sk, skb);
if (rc)
kfree_skb(skb);
return NET_RX_SUCCESS;
}
/* socket() handler. Initialize a new struct sock.
*/
static int pppol2tp_create(struct net *net, struct socket *sock, int kern)
{
int error = -ENOMEM;
struct sock *sk;
sk = sk_alloc(net, PF_PPPOX, GFP_KERNEL, &pppol2tp_sk_proto, kern);
if (!sk)
goto out;
sock_init_data(sock, sk);
sock->state = SS_UNCONNECTED;
sock->ops = &pppol2tp_ops;
sk->sk_backlog_rcv = pppol2tp_backlog_recv;
sk->sk_protocol = PX_PROTO_OL2TP;
sk->sk_family = PF_PPPOX;
sk->sk_state = PPPOX_NONE;
sk->sk_type = SOCK_STREAM;
sk->sk_destruct = pppol2tp_session_destruct;
error = 0;
out:
return error;
}
#if IS_ENABLED(CONFIG_L2TP_DEBUGFS)
static void pppol2tp_show(struct seq_file *m, void *arg)
{
struct l2tp_session *session = arg;
l2tp: protect sock pointer of struct pppol2tp_session with RCU pppol2tp_session_create() registers sessions that can't have their corresponding socket initialised. This socket has to be created by userspace, then connected to the session by pppol2tp_connect(). Therefore, we need to protect the pppol2tp socket pointer of L2TP sessions, so that it can safely be updated when userspace is connecting or closing the socket. This will eventually allow pppol2tp_connect() to avoid generating transient states while initialising its parts of the session. To this end, this patch protects the pppol2tp socket pointer using RCU. The pppol2tp socket pointer is still set in pppol2tp_connect(), but only once we know the function isn't going to fail. It's eventually reset by pppol2tp_release(), which now has to wait for a grace period to elapse before it can drop the last reference on the socket. This ensures that pppol2tp_session_get_sock() can safely grab a reference on the socket, even after ps->sk is reset to NULL but before this operation actually gets visible from pppol2tp_session_get_sock(). The rest is standard RCU conversion: pppol2tp_recv(), which already runs in atomic context, is simply enclosed by rcu_read_lock() and rcu_read_unlock(), while other functions are converted to use pppol2tp_session_get_sock() followed by sock_put(). pppol2tp_session_setsockopt() is a special case. It used to retrieve the pppol2tp socket from the L2TP session, which itself was retrieved from the pppol2tp socket. Therefore we can just avoid dereferencing ps->sk and directly use the original socket pointer instead. With all users of ps->sk now handling NULL and concurrent updates, the L2TP ->ref() and ->deref() callbacks aren't needed anymore. Therefore, rather than converting pppol2tp_session_sock_hold() and pppol2tp_session_sock_put(), we can just drop them. Signed-off-by: Guillaume Nault <g.nault@alphalink.fr> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-10-27 21:51:52 +07:00
struct sock *sk;
sk = pppol2tp_session_get_sock(session);
if (sk) {
struct pppox_sock *po = pppox_sk(sk);
l2tp: protect sock pointer of struct pppol2tp_session with RCU pppol2tp_session_create() registers sessions that can't have their corresponding socket initialised. This socket has to be created by userspace, then connected to the session by pppol2tp_connect(). Therefore, we need to protect the pppol2tp socket pointer of L2TP sessions, so that it can safely be updated when userspace is connecting or closing the socket. This will eventually allow pppol2tp_connect() to avoid generating transient states while initialising its parts of the session. To this end, this patch protects the pppol2tp socket pointer using RCU. The pppol2tp socket pointer is still set in pppol2tp_connect(), but only once we know the function isn't going to fail. It's eventually reset by pppol2tp_release(), which now has to wait for a grace period to elapse before it can drop the last reference on the socket. This ensures that pppol2tp_session_get_sock() can safely grab a reference on the socket, even after ps->sk is reset to NULL but before this operation actually gets visible from pppol2tp_session_get_sock(). The rest is standard RCU conversion: pppol2tp_recv(), which already runs in atomic context, is simply enclosed by rcu_read_lock() and rcu_read_unlock(), while other functions are converted to use pppol2tp_session_get_sock() followed by sock_put(). pppol2tp_session_setsockopt() is a special case. It used to retrieve the pppol2tp socket from the L2TP session, which itself was retrieved from the pppol2tp socket. Therefore we can just avoid dereferencing ps->sk and directly use the original socket pointer instead. With all users of ps->sk now handling NULL and concurrent updates, the L2TP ->ref() and ->deref() callbacks aren't needed anymore. Therefore, rather than converting pppol2tp_session_sock_hold() and pppol2tp_session_sock_put(), we can just drop them. Signed-off-by: Guillaume Nault <g.nault@alphalink.fr> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-10-27 21:51:52 +07:00
seq_printf(m, " interface %s\n", ppp_dev_name(&po->chan));
sock_put(sk);
}
}
#endif
l2tp: initialise PPP sessions before registering them pppol2tp_connect() initialises L2TP sessions after they've been exposed to the rest of the system by l2tp_session_register(). This puts sessions into transient states that are the source of several races, in particular with session's deletion path. This patch centralises the initialisation code into pppol2tp_session_init(), which is called before the registration phase. The only field that can't be set before session registration is the pppol2tp socket pointer, which has already been converted to RCU. So pppol2tp_connect() should now be race-free. The session's .session_close() callback is now set before registration. Therefore, it's always called when l2tp_core deletes the session, even if it was created by pppol2tp_session_create() and hasn't been plugged to a pppol2tp socket yet. That'd prevent session free because the extra reference taken by pppol2tp_session_close() wouldn't be dropped by the socket's ->sk_destruct() callback (pppol2tp_session_destruct()). We could set .session_close() only while connecting a session to its pppol2tp socket, or teach pppol2tp_session_close() to avoid grabbing a reference when the session isn't connected, but that'd require adding some form of synchronisation to be race free. Instead of that, we can just let the pppol2tp socket hold a reference on the session as soon as it starts depending on it (that is, in pppol2tp_connect()). Then we don't need to utilise pppol2tp_session_close() to hold a reference at the last moment to prevent l2tp_core from dropping it. When releasing the socket, pppol2tp_release() now deletes the session using the standard l2tp_session_delete() function, instead of merely removing it from hash tables. l2tp_session_delete() drops the reference the sessions holds on itself, but also makes sure it doesn't remove a session twice. So it can safely be called, even if l2tp_core already tried, or is concurrently trying, to remove the session. Finally, pppol2tp_session_destruct() drops the reference held by the socket. Fixes: fd558d186df2 ("l2tp: Split pppol2tp patch into separate l2tp and ppp parts") Signed-off-by: Guillaume Nault <g.nault@alphalink.fr> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-10-27 21:51:52 +07:00
static void pppol2tp_session_init(struct l2tp_session *session)
{
struct pppol2tp_session *ps;
struct dst_entry *dst;
session->recv_skb = pppol2tp_recv;
#if IS_ENABLED(CONFIG_L2TP_DEBUGFS)
session->show = pppol2tp_show;
#endif
ps = l2tp_session_priv(session);
mutex_init(&ps->sk_lock);
ps->owner = current->pid;
/* If PMTU discovery was enabled, use the MTU that was discovered */
dst = sk_dst_get(session->tunnel->sock);
if (dst) {
u32 pmtu = dst_mtu(dst);
if (pmtu) {
session->mtu = pmtu - PPPOL2TP_HEADER_OVERHEAD;
session->mru = pmtu - PPPOL2TP_HEADER_OVERHEAD;
}
dst_release(dst);
}
}
struct l2tp_connect_info {
u8 version;
int fd;
u32 tunnel_id;
u32 peer_tunnel_id;
u32 session_id;
u32 peer_session_id;
};
static int pppol2tp_sockaddr_get_info(const void *sa, int sa_len,
struct l2tp_connect_info *info)
{
switch (sa_len) {
case sizeof(struct sockaddr_pppol2tp):
{
const struct sockaddr_pppol2tp *sa_v2in4 = sa;
if (sa_v2in4->sa_protocol != PX_PROTO_OL2TP)
return -EINVAL;
info->version = 2;
info->fd = sa_v2in4->pppol2tp.fd;
info->tunnel_id = sa_v2in4->pppol2tp.s_tunnel;
info->peer_tunnel_id = sa_v2in4->pppol2tp.d_tunnel;
info->session_id = sa_v2in4->pppol2tp.s_session;
info->peer_session_id = sa_v2in4->pppol2tp.d_session;
break;
}
case sizeof(struct sockaddr_pppol2tpv3):
{
const struct sockaddr_pppol2tpv3 *sa_v3in4 = sa;
if (sa_v3in4->sa_protocol != PX_PROTO_OL2TP)
return -EINVAL;
info->version = 3;
info->fd = sa_v3in4->pppol2tp.fd;
info->tunnel_id = sa_v3in4->pppol2tp.s_tunnel;
info->peer_tunnel_id = sa_v3in4->pppol2tp.d_tunnel;
info->session_id = sa_v3in4->pppol2tp.s_session;
info->peer_session_id = sa_v3in4->pppol2tp.d_session;
break;
}
case sizeof(struct sockaddr_pppol2tpin6):
{
const struct sockaddr_pppol2tpin6 *sa_v2in6 = sa;
if (sa_v2in6->sa_protocol != PX_PROTO_OL2TP)
return -EINVAL;
info->version = 2;
info->fd = sa_v2in6->pppol2tp.fd;
info->tunnel_id = sa_v2in6->pppol2tp.s_tunnel;
info->peer_tunnel_id = sa_v2in6->pppol2tp.d_tunnel;
info->session_id = sa_v2in6->pppol2tp.s_session;
info->peer_session_id = sa_v2in6->pppol2tp.d_session;
break;
}
case sizeof(struct sockaddr_pppol2tpv3in6):
{
const struct sockaddr_pppol2tpv3in6 *sa_v3in6 = sa;
if (sa_v3in6->sa_protocol != PX_PROTO_OL2TP)
return -EINVAL;
info->version = 3;
info->fd = sa_v3in6->pppol2tp.fd;
info->tunnel_id = sa_v3in6->pppol2tp.s_tunnel;
info->peer_tunnel_id = sa_v3in6->pppol2tp.d_tunnel;
info->session_id = sa_v3in6->pppol2tp.s_session;
info->peer_session_id = sa_v3in6->pppol2tp.d_session;
break;
}
default:
return -EINVAL;
}
return 0;
}
/* connect() handler. Attach a PPPoX socket to a tunnel UDP socket
*/
static int pppol2tp_connect(struct socket *sock, struct sockaddr *uservaddr,
int sockaddr_len, int flags)
{
struct sock *sk = sock->sk;
struct pppox_sock *po = pppox_sk(sk);
struct l2tp_session *session = NULL;
struct l2tp_connect_info info;
struct l2tp_tunnel *tunnel;
struct pppol2tp_session *ps;
struct l2tp_session_cfg cfg = { 0, };
bool drop_refcnt = false;
bool drop_tunnel = false;
bool new_session = false;
bool new_tunnel = false;
int error;
error = pppol2tp_sockaddr_get_info(uservaddr, sockaddr_len, &info);
if (error < 0)
return error;
lock_sock(sk);
/* Check for already bound sockets */
error = -EBUSY;
if (sk->sk_state & PPPOX_CONNECTED)
goto end;
/* We don't supporting rebinding anyway */
error = -EALREADY;
if (sk->sk_user_data)
goto end; /* socket is already attached */
/* Don't bind if tunnel_id is 0 */
error = -EINVAL;
if (!info.tunnel_id)
goto end;
tunnel = l2tp_tunnel_get(sock_net(sk), info.tunnel_id);
if (tunnel)
drop_tunnel = true;
/* Special case: create tunnel context if session_id and
* peer_session_id is 0. Otherwise look up tunnel using supplied
* tunnel id.
*/
if (!info.session_id && !info.peer_session_id) {
if (tunnel == NULL) {
struct l2tp_tunnel_cfg tcfg = {
.encap = L2TP_ENCAPTYPE_UDP,
.debug = 0,
};
/* Prevent l2tp_tunnel_register() from trying to set up
* a kernel socket.
*/
if (info.fd < 0) {
error = -EBADF;
goto end;
}
error = l2tp_tunnel_create(sock_net(sk), info.fd,
info.version,
info.tunnel_id,
info.peer_tunnel_id, &tcfg,
&tunnel);
if (error < 0)
goto end;
l2tp: fix races in tunnel creation l2tp_tunnel_create() inserts the new tunnel into the namespace's tunnel list and sets the socket's ->sk_user_data field, before returning it to the caller. Therefore, there are two ways the tunnel can be accessed and freed, before the caller even had the opportunity to take a reference. In practice, syzbot could crash the module by closing the socket right after a new tunnel was returned to pppol2tp_create(). This patch moves tunnel registration out of l2tp_tunnel_create(), so that the caller can safely hold a reference before publishing the tunnel. This second step is done with the new l2tp_tunnel_register() function, which is now responsible for associating the tunnel to its socket and for inserting it into the namespace's list. While moving the code to l2tp_tunnel_register(), a few modifications have been done. First, the socket validation tests are done in a helper function, for clarity. Also, modifying the socket is now done after having inserted the tunnel to the namespace's tunnels list. This will allow insertion to fail, without having to revert theses modifications in the error path (a followup patch will check for duplicate tunnels before insertion). Either the socket is a kernel socket which we control, or it is a user-space socket for which we have a reference on the file descriptor. In any case, the socket isn't going to be closed from under us. Reported-by: syzbot+fbeeb5c3b538e8545644@syzkaller.appspotmail.com Fixes: fd558d186df2 ("l2tp: Split pppol2tp patch into separate l2tp and ppp parts") Signed-off-by: Guillaume Nault <g.nault@alphalink.fr> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-04-11 02:01:12 +07:00
l2tp_tunnel_inc_refcount(tunnel);
error = l2tp_tunnel_register(tunnel, sock_net(sk),
&tcfg);
if (error < 0) {
kfree(tunnel);
goto end;
}
drop_tunnel = true;
new_tunnel = true;
}
} else {
/* Error if we can't find the tunnel */
error = -ENOENT;
if (tunnel == NULL)
goto end;
/* Error if socket is not prepped */
if (tunnel->sock == NULL)
goto end;
}
if (tunnel->peer_tunnel_id == 0)
tunnel->peer_tunnel_id = info.peer_tunnel_id;
session = l2tp_session_get(sock_net(sk), tunnel, info.session_id);
if (session) {
drop_refcnt = true;
if (session->pwtype != L2TP_PWTYPE_PPP) {
error = -EPROTOTYPE;
goto end;
}
ps = l2tp_session_priv(session);
/* Using a pre-existing session is fine as long as it hasn't
* been connected yet.
*/
l2tp: protect sock pointer of struct pppol2tp_session with RCU pppol2tp_session_create() registers sessions that can't have their corresponding socket initialised. This socket has to be created by userspace, then connected to the session by pppol2tp_connect(). Therefore, we need to protect the pppol2tp socket pointer of L2TP sessions, so that it can safely be updated when userspace is connecting or closing the socket. This will eventually allow pppol2tp_connect() to avoid generating transient states while initialising its parts of the session. To this end, this patch protects the pppol2tp socket pointer using RCU. The pppol2tp socket pointer is still set in pppol2tp_connect(), but only once we know the function isn't going to fail. It's eventually reset by pppol2tp_release(), which now has to wait for a grace period to elapse before it can drop the last reference on the socket. This ensures that pppol2tp_session_get_sock() can safely grab a reference on the socket, even after ps->sk is reset to NULL but before this operation actually gets visible from pppol2tp_session_get_sock(). The rest is standard RCU conversion: pppol2tp_recv(), which already runs in atomic context, is simply enclosed by rcu_read_lock() and rcu_read_unlock(), while other functions are converted to use pppol2tp_session_get_sock() followed by sock_put(). pppol2tp_session_setsockopt() is a special case. It used to retrieve the pppol2tp socket from the L2TP session, which itself was retrieved from the pppol2tp socket. Therefore we can just avoid dereferencing ps->sk and directly use the original socket pointer instead. With all users of ps->sk now handling NULL and concurrent updates, the L2TP ->ref() and ->deref() callbacks aren't needed anymore. Therefore, rather than converting pppol2tp_session_sock_hold() and pppol2tp_session_sock_put(), we can just drop them. Signed-off-by: Guillaume Nault <g.nault@alphalink.fr> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-10-27 21:51:52 +07:00
mutex_lock(&ps->sk_lock);
if (rcu_dereference_protected(ps->sk,
l2tp: fix refcount leakage on PPPoL2TP sockets Commit d02ba2a6110c ("l2tp: fix race in pppol2tp_release with session object destroy") tried to fix a race condition where a PPPoL2TP socket would disappear while the L2TP session was still using it. However, it missed the root issue which is that an L2TP session may accept to be reconnected if its associated socket has entered the release process. The tentative fix makes the session hold the socket it is connected to. That saves the kernel from crashing, but introduces refcount leakage, preventing the socket from completing the release process. Once stalled, everything the socket depends on can't be released anymore, including the L2TP session and the l2tp_ppp module. The root issue is that, when releasing a connected PPPoL2TP socket, the session's ->sk pointer (RCU-protected) is reset to NULL and we have to wait for a grace period before destroying the socket. The socket drops the session in its ->sk_destruct callback function, so the session will exist until the last reference on the socket is dropped. Therefore, there is a time frame where pppol2tp_connect() may accept reconnecting a session, as it only checks ->sk to figure out if the session is connected. This time frame is shortened by the fact that pppol2tp_release() calls l2tp_session_delete(), making the session unreachable before resetting ->sk. However, pppol2tp_connect() may grab the session before it gets unhashed by l2tp_session_delete(), but it may test ->sk after the later got reset. The race is not so hard to trigger and syzbot found a pretty reliable reproducer: https://syzkaller.appspot.com/bug?id=418578d2a4389074524e04d641eacb091961b2cf Before d02ba2a6110c, another race could let pppol2tp_release() overwrite the ->__sk pointer of an L2TP session, thus tricking pppol2tp_put_sk() into calling sock_put() on a socket that is different than the one for which pppol2tp_release() was originally called. To get there, we had to trigger the race described above, therefore having one PPPoL2TP socket being released, while the session it is connected to is reconnecting to a different PPPoL2TP socket. When releasing this new socket fast enough, pppol2tp_release() overwrites the session's ->__sk pointer with the address of the new socket, before the first pppol2tp_put_sk() call gets scheduled. Then the pppol2tp_put_sk() call invoked by the original socket will sock_put() the new socket, potentially dropping its last reference. When the second pppol2tp_put_sk() finally runs, its socket has already been freed. With d02ba2a6110c, the session takes a reference on both sockets. Furthermore, the session's ->sk pointer is reset in the pppol2tp_session_close() callback function rather than in pppol2tp_release(). Therefore, ->__sk can't be overwritten and pppol2tp_put_sk() is called only once (l2tp_session_delete() will only run pppol2tp_session_close() once, to protect the session against concurrent deletion requests). Now pppol2tp_put_sk() will properly sock_put() the original socket, but the new socket will remain, as l2tp_session_delete() prevented the release process from completing. Here, we don't depend on the ->__sk race to trigger the bug. Getting into the pppol2tp_connect() race is enough to leak the reference, no matter when new socket is released. So it all boils down to pppol2tp_connect() failing to realise that the session has already been connected. This patch drops the unneeded extra reference counting (mostly reverting d02ba2a6110c) and checks that neither ->sk nor ->__sk is set before allowing a session to be connected. Fixes: d02ba2a6110c ("l2tp: fix race in pppol2tp_release with session object destroy") Signed-off-by: Guillaume Nault <g.nault@alphalink.fr> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-06-04 23:52:19 +07:00
lockdep_is_held(&ps->sk_lock)) ||
ps->__sk) {
l2tp: protect sock pointer of struct pppol2tp_session with RCU pppol2tp_session_create() registers sessions that can't have their corresponding socket initialised. This socket has to be created by userspace, then connected to the session by pppol2tp_connect(). Therefore, we need to protect the pppol2tp socket pointer of L2TP sessions, so that it can safely be updated when userspace is connecting or closing the socket. This will eventually allow pppol2tp_connect() to avoid generating transient states while initialising its parts of the session. To this end, this patch protects the pppol2tp socket pointer using RCU. The pppol2tp socket pointer is still set in pppol2tp_connect(), but only once we know the function isn't going to fail. It's eventually reset by pppol2tp_release(), which now has to wait for a grace period to elapse before it can drop the last reference on the socket. This ensures that pppol2tp_session_get_sock() can safely grab a reference on the socket, even after ps->sk is reset to NULL but before this operation actually gets visible from pppol2tp_session_get_sock(). The rest is standard RCU conversion: pppol2tp_recv(), which already runs in atomic context, is simply enclosed by rcu_read_lock() and rcu_read_unlock(), while other functions are converted to use pppol2tp_session_get_sock() followed by sock_put(). pppol2tp_session_setsockopt() is a special case. It used to retrieve the pppol2tp socket from the L2TP session, which itself was retrieved from the pppol2tp socket. Therefore we can just avoid dereferencing ps->sk and directly use the original socket pointer instead. With all users of ps->sk now handling NULL and concurrent updates, the L2TP ->ref() and ->deref() callbacks aren't needed anymore. Therefore, rather than converting pppol2tp_session_sock_hold() and pppol2tp_session_sock_put(), we can just drop them. Signed-off-by: Guillaume Nault <g.nault@alphalink.fr> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-10-27 21:51:52 +07:00
mutex_unlock(&ps->sk_lock);
error = -EEXIST;
goto end;
}
} else {
/* Default MTU must allow space for UDP/L2TP/PPP headers */
cfg.mtu = 1500 - PPPOL2TP_HEADER_OVERHEAD;
cfg.mru = cfg.mtu;
cfg.pw_type = L2TP_PWTYPE_PPP;
session = l2tp_session_create(sizeof(struct pppol2tp_session),
tunnel, info.session_id,
info.peer_session_id, &cfg);
if (IS_ERR(session)) {
error = PTR_ERR(session);
goto end;
}
l2tp: initialise PPP sessions before registering them pppol2tp_connect() initialises L2TP sessions after they've been exposed to the rest of the system by l2tp_session_register(). This puts sessions into transient states that are the source of several races, in particular with session's deletion path. This patch centralises the initialisation code into pppol2tp_session_init(), which is called before the registration phase. The only field that can't be set before session registration is the pppol2tp socket pointer, which has already been converted to RCU. So pppol2tp_connect() should now be race-free. The session's .session_close() callback is now set before registration. Therefore, it's always called when l2tp_core deletes the session, even if it was created by pppol2tp_session_create() and hasn't been plugged to a pppol2tp socket yet. That'd prevent session free because the extra reference taken by pppol2tp_session_close() wouldn't be dropped by the socket's ->sk_destruct() callback (pppol2tp_session_destruct()). We could set .session_close() only while connecting a session to its pppol2tp socket, or teach pppol2tp_session_close() to avoid grabbing a reference when the session isn't connected, but that'd require adding some form of synchronisation to be race free. Instead of that, we can just let the pppol2tp socket hold a reference on the session as soon as it starts depending on it (that is, in pppol2tp_connect()). Then we don't need to utilise pppol2tp_session_close() to hold a reference at the last moment to prevent l2tp_core from dropping it. When releasing the socket, pppol2tp_release() now deletes the session using the standard l2tp_session_delete() function, instead of merely removing it from hash tables. l2tp_session_delete() drops the reference the sessions holds on itself, but also makes sure it doesn't remove a session twice. So it can safely be called, even if l2tp_core already tried, or is concurrently trying, to remove the session. Finally, pppol2tp_session_destruct() drops the reference held by the socket. Fixes: fd558d186df2 ("l2tp: Split pppol2tp patch into separate l2tp and ppp parts") Signed-off-by: Guillaume Nault <g.nault@alphalink.fr> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-10-27 21:51:52 +07:00
pppol2tp_session_init(session);
l2tp: protect sock pointer of struct pppol2tp_session with RCU pppol2tp_session_create() registers sessions that can't have their corresponding socket initialised. This socket has to be created by userspace, then connected to the session by pppol2tp_connect(). Therefore, we need to protect the pppol2tp socket pointer of L2TP sessions, so that it can safely be updated when userspace is connecting or closing the socket. This will eventually allow pppol2tp_connect() to avoid generating transient states while initialising its parts of the session. To this end, this patch protects the pppol2tp socket pointer using RCU. The pppol2tp socket pointer is still set in pppol2tp_connect(), but only once we know the function isn't going to fail. It's eventually reset by pppol2tp_release(), which now has to wait for a grace period to elapse before it can drop the last reference on the socket. This ensures that pppol2tp_session_get_sock() can safely grab a reference on the socket, even after ps->sk is reset to NULL but before this operation actually gets visible from pppol2tp_session_get_sock(). The rest is standard RCU conversion: pppol2tp_recv(), which already runs in atomic context, is simply enclosed by rcu_read_lock() and rcu_read_unlock(), while other functions are converted to use pppol2tp_session_get_sock() followed by sock_put(). pppol2tp_session_setsockopt() is a special case. It used to retrieve the pppol2tp socket from the L2TP session, which itself was retrieved from the pppol2tp socket. Therefore we can just avoid dereferencing ps->sk and directly use the original socket pointer instead. With all users of ps->sk now handling NULL and concurrent updates, the L2TP ->ref() and ->deref() callbacks aren't needed anymore. Therefore, rather than converting pppol2tp_session_sock_hold() and pppol2tp_session_sock_put(), we can just drop them. Signed-off-by: Guillaume Nault <g.nault@alphalink.fr> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-10-27 21:51:52 +07:00
ps = l2tp_session_priv(session);
l2tp_session_inc_refcount(session);
l2tp: protect sock pointer of struct pppol2tp_session with RCU pppol2tp_session_create() registers sessions that can't have their corresponding socket initialised. This socket has to be created by userspace, then connected to the session by pppol2tp_connect(). Therefore, we need to protect the pppol2tp socket pointer of L2TP sessions, so that it can safely be updated when userspace is connecting or closing the socket. This will eventually allow pppol2tp_connect() to avoid generating transient states while initialising its parts of the session. To this end, this patch protects the pppol2tp socket pointer using RCU. The pppol2tp socket pointer is still set in pppol2tp_connect(), but only once we know the function isn't going to fail. It's eventually reset by pppol2tp_release(), which now has to wait for a grace period to elapse before it can drop the last reference on the socket. This ensures that pppol2tp_session_get_sock() can safely grab a reference on the socket, even after ps->sk is reset to NULL but before this operation actually gets visible from pppol2tp_session_get_sock(). The rest is standard RCU conversion: pppol2tp_recv(), which already runs in atomic context, is simply enclosed by rcu_read_lock() and rcu_read_unlock(), while other functions are converted to use pppol2tp_session_get_sock() followed by sock_put(). pppol2tp_session_setsockopt() is a special case. It used to retrieve the pppol2tp socket from the L2TP session, which itself was retrieved from the pppol2tp socket. Therefore we can just avoid dereferencing ps->sk and directly use the original socket pointer instead. With all users of ps->sk now handling NULL and concurrent updates, the L2TP ->ref() and ->deref() callbacks aren't needed anymore. Therefore, rather than converting pppol2tp_session_sock_hold() and pppol2tp_session_sock_put(), we can just drop them. Signed-off-by: Guillaume Nault <g.nault@alphalink.fr> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-10-27 21:51:52 +07:00
mutex_lock(&ps->sk_lock);
error = l2tp_session_register(session, tunnel);
if (error < 0) {
l2tp: protect sock pointer of struct pppol2tp_session with RCU pppol2tp_session_create() registers sessions that can't have their corresponding socket initialised. This socket has to be created by userspace, then connected to the session by pppol2tp_connect(). Therefore, we need to protect the pppol2tp socket pointer of L2TP sessions, so that it can safely be updated when userspace is connecting or closing the socket. This will eventually allow pppol2tp_connect() to avoid generating transient states while initialising its parts of the session. To this end, this patch protects the pppol2tp socket pointer using RCU. The pppol2tp socket pointer is still set in pppol2tp_connect(), but only once we know the function isn't going to fail. It's eventually reset by pppol2tp_release(), which now has to wait for a grace period to elapse before it can drop the last reference on the socket. This ensures that pppol2tp_session_get_sock() can safely grab a reference on the socket, even after ps->sk is reset to NULL but before this operation actually gets visible from pppol2tp_session_get_sock(). The rest is standard RCU conversion: pppol2tp_recv(), which already runs in atomic context, is simply enclosed by rcu_read_lock() and rcu_read_unlock(), while other functions are converted to use pppol2tp_session_get_sock() followed by sock_put(). pppol2tp_session_setsockopt() is a special case. It used to retrieve the pppol2tp socket from the L2TP session, which itself was retrieved from the pppol2tp socket. Therefore we can just avoid dereferencing ps->sk and directly use the original socket pointer instead. With all users of ps->sk now handling NULL and concurrent updates, the L2TP ->ref() and ->deref() callbacks aren't needed anymore. Therefore, rather than converting pppol2tp_session_sock_hold() and pppol2tp_session_sock_put(), we can just drop them. Signed-off-by: Guillaume Nault <g.nault@alphalink.fr> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-10-27 21:51:52 +07:00
mutex_unlock(&ps->sk_lock);
kfree(session);
goto end;
}
drop_refcnt = true;
new_session = true;
}
/* Special case: if source & dest session_id == 0x0000, this
* socket is being created to manage the tunnel. Just set up
* the internal context for use by ioctl() and sockopt()
* handlers.
*/
if ((session->session_id == 0) &&
(session->peer_session_id == 0)) {
error = 0;
goto out_no_ppp;
}
/* The only header we need to worry about is the L2TP
* header. This size is different depending on whether
* sequence numbers are enabled for the data channel.
*/
po->chan.hdrlen = PPPOL2TP_L2TP_HDR_SIZE_NOSEQ;
po->chan.private = sk;
po->chan.ops = &pppol2tp_chan_ops;
po->chan.mtu = session->mtu;
error = ppp_register_net_channel(sock_net(sk), &po->chan);
l2tp: protect sock pointer of struct pppol2tp_session with RCU pppol2tp_session_create() registers sessions that can't have their corresponding socket initialised. This socket has to be created by userspace, then connected to the session by pppol2tp_connect(). Therefore, we need to protect the pppol2tp socket pointer of L2TP sessions, so that it can safely be updated when userspace is connecting or closing the socket. This will eventually allow pppol2tp_connect() to avoid generating transient states while initialising its parts of the session. To this end, this patch protects the pppol2tp socket pointer using RCU. The pppol2tp socket pointer is still set in pppol2tp_connect(), but only once we know the function isn't going to fail. It's eventually reset by pppol2tp_release(), which now has to wait for a grace period to elapse before it can drop the last reference on the socket. This ensures that pppol2tp_session_get_sock() can safely grab a reference on the socket, even after ps->sk is reset to NULL but before this operation actually gets visible from pppol2tp_session_get_sock(). The rest is standard RCU conversion: pppol2tp_recv(), which already runs in atomic context, is simply enclosed by rcu_read_lock() and rcu_read_unlock(), while other functions are converted to use pppol2tp_session_get_sock() followed by sock_put(). pppol2tp_session_setsockopt() is a special case. It used to retrieve the pppol2tp socket from the L2TP session, which itself was retrieved from the pppol2tp socket. Therefore we can just avoid dereferencing ps->sk and directly use the original socket pointer instead. With all users of ps->sk now handling NULL and concurrent updates, the L2TP ->ref() and ->deref() callbacks aren't needed anymore. Therefore, rather than converting pppol2tp_session_sock_hold() and pppol2tp_session_sock_put(), we can just drop them. Signed-off-by: Guillaume Nault <g.nault@alphalink.fr> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-10-27 21:51:52 +07:00
if (error) {
mutex_unlock(&ps->sk_lock);
goto end;
l2tp: protect sock pointer of struct pppol2tp_session with RCU pppol2tp_session_create() registers sessions that can't have their corresponding socket initialised. This socket has to be created by userspace, then connected to the session by pppol2tp_connect(). Therefore, we need to protect the pppol2tp socket pointer of L2TP sessions, so that it can safely be updated when userspace is connecting or closing the socket. This will eventually allow pppol2tp_connect() to avoid generating transient states while initialising its parts of the session. To this end, this patch protects the pppol2tp socket pointer using RCU. The pppol2tp socket pointer is still set in pppol2tp_connect(), but only once we know the function isn't going to fail. It's eventually reset by pppol2tp_release(), which now has to wait for a grace period to elapse before it can drop the last reference on the socket. This ensures that pppol2tp_session_get_sock() can safely grab a reference on the socket, even after ps->sk is reset to NULL but before this operation actually gets visible from pppol2tp_session_get_sock(). The rest is standard RCU conversion: pppol2tp_recv(), which already runs in atomic context, is simply enclosed by rcu_read_lock() and rcu_read_unlock(), while other functions are converted to use pppol2tp_session_get_sock() followed by sock_put(). pppol2tp_session_setsockopt() is a special case. It used to retrieve the pppol2tp socket from the L2TP session, which itself was retrieved from the pppol2tp socket. Therefore we can just avoid dereferencing ps->sk and directly use the original socket pointer instead. With all users of ps->sk now handling NULL and concurrent updates, the L2TP ->ref() and ->deref() callbacks aren't needed anymore. Therefore, rather than converting pppol2tp_session_sock_hold() and pppol2tp_session_sock_put(), we can just drop them. Signed-off-by: Guillaume Nault <g.nault@alphalink.fr> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-10-27 21:51:52 +07:00
}
out_no_ppp:
/* This is how we get the session context from the socket. */
sk->sk_user_data = session;
l2tp: protect sock pointer of struct pppol2tp_session with RCU pppol2tp_session_create() registers sessions that can't have their corresponding socket initialised. This socket has to be created by userspace, then connected to the session by pppol2tp_connect(). Therefore, we need to protect the pppol2tp socket pointer of L2TP sessions, so that it can safely be updated when userspace is connecting or closing the socket. This will eventually allow pppol2tp_connect() to avoid generating transient states while initialising its parts of the session. To this end, this patch protects the pppol2tp socket pointer using RCU. The pppol2tp socket pointer is still set in pppol2tp_connect(), but only once we know the function isn't going to fail. It's eventually reset by pppol2tp_release(), which now has to wait for a grace period to elapse before it can drop the last reference on the socket. This ensures that pppol2tp_session_get_sock() can safely grab a reference on the socket, even after ps->sk is reset to NULL but before this operation actually gets visible from pppol2tp_session_get_sock(). The rest is standard RCU conversion: pppol2tp_recv(), which already runs in atomic context, is simply enclosed by rcu_read_lock() and rcu_read_unlock(), while other functions are converted to use pppol2tp_session_get_sock() followed by sock_put(). pppol2tp_session_setsockopt() is a special case. It used to retrieve the pppol2tp socket from the L2TP session, which itself was retrieved from the pppol2tp socket. Therefore we can just avoid dereferencing ps->sk and directly use the original socket pointer instead. With all users of ps->sk now handling NULL and concurrent updates, the L2TP ->ref() and ->deref() callbacks aren't needed anymore. Therefore, rather than converting pppol2tp_session_sock_hold() and pppol2tp_session_sock_put(), we can just drop them. Signed-off-by: Guillaume Nault <g.nault@alphalink.fr> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-10-27 21:51:52 +07:00
rcu_assign_pointer(ps->sk, sk);
mutex_unlock(&ps->sk_lock);
l2tp: initialise PPP sessions before registering them pppol2tp_connect() initialises L2TP sessions after they've been exposed to the rest of the system by l2tp_session_register(). This puts sessions into transient states that are the source of several races, in particular with session's deletion path. This patch centralises the initialisation code into pppol2tp_session_init(), which is called before the registration phase. The only field that can't be set before session registration is the pppol2tp socket pointer, which has already been converted to RCU. So pppol2tp_connect() should now be race-free. The session's .session_close() callback is now set before registration. Therefore, it's always called when l2tp_core deletes the session, even if it was created by pppol2tp_session_create() and hasn't been plugged to a pppol2tp socket yet. That'd prevent session free because the extra reference taken by pppol2tp_session_close() wouldn't be dropped by the socket's ->sk_destruct() callback (pppol2tp_session_destruct()). We could set .session_close() only while connecting a session to its pppol2tp socket, or teach pppol2tp_session_close() to avoid grabbing a reference when the session isn't connected, but that'd require adding some form of synchronisation to be race free. Instead of that, we can just let the pppol2tp socket hold a reference on the session as soon as it starts depending on it (that is, in pppol2tp_connect()). Then we don't need to utilise pppol2tp_session_close() to hold a reference at the last moment to prevent l2tp_core from dropping it. When releasing the socket, pppol2tp_release() now deletes the session using the standard l2tp_session_delete() function, instead of merely removing it from hash tables. l2tp_session_delete() drops the reference the sessions holds on itself, but also makes sure it doesn't remove a session twice. So it can safely be called, even if l2tp_core already tried, or is concurrently trying, to remove the session. Finally, pppol2tp_session_destruct() drops the reference held by the socket. Fixes: fd558d186df2 ("l2tp: Split pppol2tp patch into separate l2tp and ppp parts") Signed-off-by: Guillaume Nault <g.nault@alphalink.fr> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-10-27 21:51:52 +07:00
/* Keep the reference we've grabbed on the session: sk doesn't expect
* the session to disappear. pppol2tp_session_destruct() is responsible
* for dropping it.
*/
drop_refcnt = false;
sk->sk_state = PPPOX_CONNECTED;
l2tp_info(session, L2TP_MSG_CONTROL, "%s: created\n",
session->name);
end:
if (error) {
if (new_session)
l2tp_session_delete(session);
if (new_tunnel)
l2tp_tunnel_delete(tunnel);
}
if (drop_refcnt)
l2tp_session_dec_refcount(session);
if (drop_tunnel)
l2tp_tunnel_dec_refcount(tunnel);
release_sock(sk);
return error;
}
#ifdef CONFIG_L2TP_V3
l2tp: pass tunnel pointer to ->session_create() Using l2tp_tunnel_find() in pppol2tp_session_create() and l2tp_eth_create() is racy, because no reference is held on the returned session. These functions are only used to implement the ->session_create callback which is run by l2tp_nl_cmd_session_create(). Therefore searching for the parent tunnel isn't necessary because l2tp_nl_cmd_session_create() already has a pointer to it and holds a reference. This patch modifies ->session_create()'s prototype to directly pass the the parent tunnel as parameter, thus avoiding searching for it in pppol2tp_session_create() and l2tp_eth_create(). Since we have to touch the ->session_create() call in l2tp_nl_cmd_session_create(), let's also remove the useless conditional: we know that ->session_create isn't NULL at this point because it's already been checked earlier in this same function. Finally, one might be tempted to think that the removed l2tp_tunnel_find() calls were harmless because they would return the same tunnel as the one held by l2tp_nl_cmd_session_create() anyway. But that tunnel might be removed and a new one created with same tunnel Id before the l2tp_tunnel_find() call. In this case l2tp_tunnel_find() would return the new tunnel which wouldn't be protected by the reference held by l2tp_nl_cmd_session_create(). Fixes: 309795f4bec2 ("l2tp: Add netlink control API for L2TP") Fixes: d9e31d17ceba ("l2tp: Add L2TP ethernet pseudowire support") Signed-off-by: Guillaume Nault <g.nault@alphalink.fr> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-09-01 22:58:51 +07:00
/* Called when creating sessions via the netlink interface. */
static int pppol2tp_session_create(struct net *net, struct l2tp_tunnel *tunnel,
u32 session_id, u32 peer_session_id,
struct l2tp_session_cfg *cfg)
{
int error;
struct l2tp_session *session;
/* Error if tunnel socket is not prepped */
l2tp: pass tunnel pointer to ->session_create() Using l2tp_tunnel_find() in pppol2tp_session_create() and l2tp_eth_create() is racy, because no reference is held on the returned session. These functions are only used to implement the ->session_create callback which is run by l2tp_nl_cmd_session_create(). Therefore searching for the parent tunnel isn't necessary because l2tp_nl_cmd_session_create() already has a pointer to it and holds a reference. This patch modifies ->session_create()'s prototype to directly pass the the parent tunnel as parameter, thus avoiding searching for it in pppol2tp_session_create() and l2tp_eth_create(). Since we have to touch the ->session_create() call in l2tp_nl_cmd_session_create(), let's also remove the useless conditional: we know that ->session_create isn't NULL at this point because it's already been checked earlier in this same function. Finally, one might be tempted to think that the removed l2tp_tunnel_find() calls were harmless because they would return the same tunnel as the one held by l2tp_nl_cmd_session_create() anyway. But that tunnel might be removed and a new one created with same tunnel Id before the l2tp_tunnel_find() call. In this case l2tp_tunnel_find() would return the new tunnel which wouldn't be protected by the reference held by l2tp_nl_cmd_session_create(). Fixes: 309795f4bec2 ("l2tp: Add netlink control API for L2TP") Fixes: d9e31d17ceba ("l2tp: Add L2TP ethernet pseudowire support") Signed-off-by: Guillaume Nault <g.nault@alphalink.fr> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-09-01 22:58:51 +07:00
if (!tunnel->sock) {
error = -ENOENT;
goto err;
l2tp: pass tunnel pointer to ->session_create() Using l2tp_tunnel_find() in pppol2tp_session_create() and l2tp_eth_create() is racy, because no reference is held on the returned session. These functions are only used to implement the ->session_create callback which is run by l2tp_nl_cmd_session_create(). Therefore searching for the parent tunnel isn't necessary because l2tp_nl_cmd_session_create() already has a pointer to it and holds a reference. This patch modifies ->session_create()'s prototype to directly pass the the parent tunnel as parameter, thus avoiding searching for it in pppol2tp_session_create() and l2tp_eth_create(). Since we have to touch the ->session_create() call in l2tp_nl_cmd_session_create(), let's also remove the useless conditional: we know that ->session_create isn't NULL at this point because it's already been checked earlier in this same function. Finally, one might be tempted to think that the removed l2tp_tunnel_find() calls were harmless because they would return the same tunnel as the one held by l2tp_nl_cmd_session_create() anyway. But that tunnel might be removed and a new one created with same tunnel Id before the l2tp_tunnel_find() call. In this case l2tp_tunnel_find() would return the new tunnel which wouldn't be protected by the reference held by l2tp_nl_cmd_session_create(). Fixes: 309795f4bec2 ("l2tp: Add netlink control API for L2TP") Fixes: d9e31d17ceba ("l2tp: Add L2TP ethernet pseudowire support") Signed-off-by: Guillaume Nault <g.nault@alphalink.fr> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-09-01 22:58:51 +07:00
}
/* Default MTU values. */
if (cfg->mtu == 0)
cfg->mtu = 1500 - PPPOL2TP_HEADER_OVERHEAD;
if (cfg->mru == 0)
cfg->mru = cfg->mtu;
/* Allocate and initialize a new session context. */
session = l2tp_session_create(sizeof(struct pppol2tp_session),
tunnel, session_id,
peer_session_id, cfg);
if (IS_ERR(session)) {
error = PTR_ERR(session);
goto err;
}
l2tp: initialise PPP sessions before registering them pppol2tp_connect() initialises L2TP sessions after they've been exposed to the rest of the system by l2tp_session_register(). This puts sessions into transient states that are the source of several races, in particular with session's deletion path. This patch centralises the initialisation code into pppol2tp_session_init(), which is called before the registration phase. The only field that can't be set before session registration is the pppol2tp socket pointer, which has already been converted to RCU. So pppol2tp_connect() should now be race-free. The session's .session_close() callback is now set before registration. Therefore, it's always called when l2tp_core deletes the session, even if it was created by pppol2tp_session_create() and hasn't been plugged to a pppol2tp socket yet. That'd prevent session free because the extra reference taken by pppol2tp_session_close() wouldn't be dropped by the socket's ->sk_destruct() callback (pppol2tp_session_destruct()). We could set .session_close() only while connecting a session to its pppol2tp socket, or teach pppol2tp_session_close() to avoid grabbing a reference when the session isn't connected, but that'd require adding some form of synchronisation to be race free. Instead of that, we can just let the pppol2tp socket hold a reference on the session as soon as it starts depending on it (that is, in pppol2tp_connect()). Then we don't need to utilise pppol2tp_session_close() to hold a reference at the last moment to prevent l2tp_core from dropping it. When releasing the socket, pppol2tp_release() now deletes the session using the standard l2tp_session_delete() function, instead of merely removing it from hash tables. l2tp_session_delete() drops the reference the sessions holds on itself, but also makes sure it doesn't remove a session twice. So it can safely be called, even if l2tp_core already tried, or is concurrently trying, to remove the session. Finally, pppol2tp_session_destruct() drops the reference held by the socket. Fixes: fd558d186df2 ("l2tp: Split pppol2tp patch into separate l2tp and ppp parts") Signed-off-by: Guillaume Nault <g.nault@alphalink.fr> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-10-27 21:51:52 +07:00
pppol2tp_session_init(session);
error = l2tp_session_register(session, tunnel);
if (error < 0)
goto err_sess;
return 0;
err_sess:
kfree(session);
err:
return error;
}
#endif /* CONFIG_L2TP_V3 */
/* getname() support.
*/
static int pppol2tp_getname(struct socket *sock, struct sockaddr *uaddr,
net: make getname() functions return length rather than use int* parameter Changes since v1: Added changes in these files: drivers/infiniband/hw/usnic/usnic_transport.c drivers/staging/lustre/lnet/lnet/lib-socket.c drivers/target/iscsi/iscsi_target_login.c drivers/vhost/net.c fs/dlm/lowcomms.c fs/ocfs2/cluster/tcp.c security/tomoyo/network.c Before: All these functions either return a negative error indicator, or store length of sockaddr into "int *socklen" parameter and return zero on success. "int *socklen" parameter is awkward. For example, if caller does not care, it still needs to provide on-stack storage for the value it does not need. None of the many FOO_getname() functions of various protocols ever used old value of *socklen. They always just overwrite it. This change drops this parameter, and makes all these functions, on success, return length of sockaddr. It's always >= 0 and can be differentiated from an error. Tests in callers are changed from "if (err)" to "if (err < 0)", where needed. rpc_sockname() lost "int buflen" parameter, since its only use was to be passed to kernel_getsockname() as &buflen and subsequently not used in any way. Userspace API is not changed. text data bss dec hex filename 30108430 2633624 873672 33615726 200ef6e vmlinux.before.o 30108109 2633612 873672 33615393 200ee21 vmlinux.o Signed-off-by: Denys Vlasenko <dvlasenk@redhat.com> CC: David S. Miller <davem@davemloft.net> CC: linux-kernel@vger.kernel.org CC: netdev@vger.kernel.org CC: linux-bluetooth@vger.kernel.org CC: linux-decnet-user@lists.sourceforge.net CC: linux-wireless@vger.kernel.org CC: linux-rdma@vger.kernel.org CC: linux-sctp@vger.kernel.org CC: linux-nfs@vger.kernel.org CC: linux-x25@vger.kernel.org Signed-off-by: David S. Miller <davem@davemloft.net>
2018-02-13 02:00:20 +07:00
int peer)
{
int len = 0;
int error = 0;
struct l2tp_session *session;
struct l2tp_tunnel *tunnel;
struct sock *sk = sock->sk;
struct inet_sock *inet;
struct pppol2tp_session *pls;
error = -ENOTCONN;
if (sk == NULL)
goto end;
if (!(sk->sk_state & PPPOX_CONNECTED))
goto end;
error = -EBADF;
session = pppol2tp_sock_to_session(sk);
if (session == NULL)
goto end;
pls = l2tp_session_priv(session);
tunnel = session->tunnel;
inet = inet_sk(tunnel->sock);
if ((tunnel->version == 2) && (tunnel->sock->sk_family == AF_INET)) {
struct sockaddr_pppol2tp sp;
len = sizeof(sp);
memset(&sp, 0, len);
sp.sa_family = AF_PPPOX;
sp.sa_protocol = PX_PROTO_OL2TP;
sp.pppol2tp.fd = tunnel->fd;
sp.pppol2tp.pid = pls->owner;
sp.pppol2tp.s_tunnel = tunnel->tunnel_id;
sp.pppol2tp.d_tunnel = tunnel->peer_tunnel_id;
sp.pppol2tp.s_session = session->session_id;
sp.pppol2tp.d_session = session->peer_session_id;
sp.pppol2tp.addr.sin_family = AF_INET;
sp.pppol2tp.addr.sin_port = inet->inet_dport;
sp.pppol2tp.addr.sin_addr.s_addr = inet->inet_daddr;
memcpy(uaddr, &sp, len);
#if IS_ENABLED(CONFIG_IPV6)
} else if ((tunnel->version == 2) &&
(tunnel->sock->sk_family == AF_INET6)) {
struct sockaddr_pppol2tpin6 sp;
len = sizeof(sp);
memset(&sp, 0, len);
sp.sa_family = AF_PPPOX;
sp.sa_protocol = PX_PROTO_OL2TP;
sp.pppol2tp.fd = tunnel->fd;
sp.pppol2tp.pid = pls->owner;
sp.pppol2tp.s_tunnel = tunnel->tunnel_id;
sp.pppol2tp.d_tunnel = tunnel->peer_tunnel_id;
sp.pppol2tp.s_session = session->session_id;
sp.pppol2tp.d_session = session->peer_session_id;
sp.pppol2tp.addr.sin6_family = AF_INET6;
sp.pppol2tp.addr.sin6_port = inet->inet_dport;
memcpy(&sp.pppol2tp.addr.sin6_addr, &tunnel->sock->sk_v6_daddr,
sizeof(tunnel->sock->sk_v6_daddr));
memcpy(uaddr, &sp, len);
} else if ((tunnel->version == 3) &&
(tunnel->sock->sk_family == AF_INET6)) {
struct sockaddr_pppol2tpv3in6 sp;
len = sizeof(sp);
memset(&sp, 0, len);
sp.sa_family = AF_PPPOX;
sp.sa_protocol = PX_PROTO_OL2TP;
sp.pppol2tp.fd = tunnel->fd;
sp.pppol2tp.pid = pls->owner;
sp.pppol2tp.s_tunnel = tunnel->tunnel_id;
sp.pppol2tp.d_tunnel = tunnel->peer_tunnel_id;
sp.pppol2tp.s_session = session->session_id;
sp.pppol2tp.d_session = session->peer_session_id;
sp.pppol2tp.addr.sin6_family = AF_INET6;
sp.pppol2tp.addr.sin6_port = inet->inet_dport;
memcpy(&sp.pppol2tp.addr.sin6_addr, &tunnel->sock->sk_v6_daddr,
sizeof(tunnel->sock->sk_v6_daddr));
memcpy(uaddr, &sp, len);
#endif
} else if (tunnel->version == 3) {
struct sockaddr_pppol2tpv3 sp;
len = sizeof(sp);
memset(&sp, 0, len);
sp.sa_family = AF_PPPOX;
sp.sa_protocol = PX_PROTO_OL2TP;
sp.pppol2tp.fd = tunnel->fd;
sp.pppol2tp.pid = pls->owner;
sp.pppol2tp.s_tunnel = tunnel->tunnel_id;
sp.pppol2tp.d_tunnel = tunnel->peer_tunnel_id;
sp.pppol2tp.s_session = session->session_id;
sp.pppol2tp.d_session = session->peer_session_id;
sp.pppol2tp.addr.sin_family = AF_INET;
sp.pppol2tp.addr.sin_port = inet->inet_dport;
sp.pppol2tp.addr.sin_addr.s_addr = inet->inet_daddr;
memcpy(uaddr, &sp, len);
}
net: make getname() functions return length rather than use int* parameter Changes since v1: Added changes in these files: drivers/infiniband/hw/usnic/usnic_transport.c drivers/staging/lustre/lnet/lnet/lib-socket.c drivers/target/iscsi/iscsi_target_login.c drivers/vhost/net.c fs/dlm/lowcomms.c fs/ocfs2/cluster/tcp.c security/tomoyo/network.c Before: All these functions either return a negative error indicator, or store length of sockaddr into "int *socklen" parameter and return zero on success. "int *socklen" parameter is awkward. For example, if caller does not care, it still needs to provide on-stack storage for the value it does not need. None of the many FOO_getname() functions of various protocols ever used old value of *socklen. They always just overwrite it. This change drops this parameter, and makes all these functions, on success, return length of sockaddr. It's always >= 0 and can be differentiated from an error. Tests in callers are changed from "if (err)" to "if (err < 0)", where needed. rpc_sockname() lost "int buflen" parameter, since its only use was to be passed to kernel_getsockname() as &buflen and subsequently not used in any way. Userspace API is not changed. text data bss dec hex filename 30108430 2633624 873672 33615726 200ef6e vmlinux.before.o 30108109 2633612 873672 33615393 200ee21 vmlinux.o Signed-off-by: Denys Vlasenko <dvlasenk@redhat.com> CC: David S. Miller <davem@davemloft.net> CC: linux-kernel@vger.kernel.org CC: netdev@vger.kernel.org CC: linux-bluetooth@vger.kernel.org CC: linux-decnet-user@lists.sourceforge.net CC: linux-wireless@vger.kernel.org CC: linux-rdma@vger.kernel.org CC: linux-sctp@vger.kernel.org CC: linux-nfs@vger.kernel.org CC: linux-x25@vger.kernel.org Signed-off-by: David S. Miller <davem@davemloft.net>
2018-02-13 02:00:20 +07:00
error = len;
sock_put(sk);
end:
return error;
}
/****************************************************************************
* ioctl() handlers.
*
* The PPPoX socket is created for L2TP sessions: tunnels have their own UDP
* sockets. However, in order to control kernel tunnel features, we allow
* userspace to create a special "tunnel" PPPoX socket which is used for
* control only. Tunnel PPPoX sockets have session_id == 0 and simply allow
* the user application to issue L2TP setsockopt(), getsockopt() and ioctl()
* calls.
****************************************************************************/
static void pppol2tp_copy_stats(struct pppol2tp_ioc_stats *dest,
struct l2tp_stats *stats)
{
dest->tx_packets = atomic_long_read(&stats->tx_packets);
dest->tx_bytes = atomic_long_read(&stats->tx_bytes);
dest->tx_errors = atomic_long_read(&stats->tx_errors);
dest->rx_packets = atomic_long_read(&stats->rx_packets);
dest->rx_bytes = atomic_long_read(&stats->rx_bytes);
dest->rx_seq_discards = atomic_long_read(&stats->rx_seq_discards);
dest->rx_oos_packets = atomic_long_read(&stats->rx_oos_packets);
dest->rx_errors = atomic_long_read(&stats->rx_errors);
}
/* Session ioctl helper.
*/
static int pppol2tp_session_ioctl(struct l2tp_session *session,
unsigned int cmd, unsigned long arg)
{
struct ifreq ifr;
int err = 0;
struct sock *sk;
int val = (int) arg;
struct pppol2tp_session *ps = l2tp_session_priv(session);
struct l2tp_tunnel *tunnel = session->tunnel;
struct pppol2tp_ioc_stats stats;
l2tp_dbg(session, L2TP_MSG_CONTROL,
"%s: pppol2tp_session_ioctl(cmd=%#x, arg=%#lx)\n",
session->name, cmd, arg);
l2tp: protect sock pointer of struct pppol2tp_session with RCU pppol2tp_session_create() registers sessions that can't have their corresponding socket initialised. This socket has to be created by userspace, then connected to the session by pppol2tp_connect(). Therefore, we need to protect the pppol2tp socket pointer of L2TP sessions, so that it can safely be updated when userspace is connecting or closing the socket. This will eventually allow pppol2tp_connect() to avoid generating transient states while initialising its parts of the session. To this end, this patch protects the pppol2tp socket pointer using RCU. The pppol2tp socket pointer is still set in pppol2tp_connect(), but only once we know the function isn't going to fail. It's eventually reset by pppol2tp_release(), which now has to wait for a grace period to elapse before it can drop the last reference on the socket. This ensures that pppol2tp_session_get_sock() can safely grab a reference on the socket, even after ps->sk is reset to NULL but before this operation actually gets visible from pppol2tp_session_get_sock(). The rest is standard RCU conversion: pppol2tp_recv(), which already runs in atomic context, is simply enclosed by rcu_read_lock() and rcu_read_unlock(), while other functions are converted to use pppol2tp_session_get_sock() followed by sock_put(). pppol2tp_session_setsockopt() is a special case. It used to retrieve the pppol2tp socket from the L2TP session, which itself was retrieved from the pppol2tp socket. Therefore we can just avoid dereferencing ps->sk and directly use the original socket pointer instead. With all users of ps->sk now handling NULL and concurrent updates, the L2TP ->ref() and ->deref() callbacks aren't needed anymore. Therefore, rather than converting pppol2tp_session_sock_hold() and pppol2tp_session_sock_put(), we can just drop them. Signed-off-by: Guillaume Nault <g.nault@alphalink.fr> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-10-27 21:51:52 +07:00
sk = pppol2tp_session_get_sock(session);
if (!sk)
return -EBADR;
switch (cmd) {
case SIOCGIFMTU:
err = -ENXIO;
if (!(sk->sk_state & PPPOX_CONNECTED))
break;
err = -EFAULT;
if (copy_from_user(&ifr, (void __user *) arg, sizeof(struct ifreq)))
break;
ifr.ifr_mtu = session->mtu;
if (copy_to_user((void __user *) arg, &ifr, sizeof(struct ifreq)))
break;
l2tp_info(session, L2TP_MSG_CONTROL, "%s: get mtu=%d\n",
session->name, session->mtu);
err = 0;
break;
case SIOCSIFMTU:
err = -ENXIO;
if (!(sk->sk_state & PPPOX_CONNECTED))
break;
err = -EFAULT;
if (copy_from_user(&ifr, (void __user *) arg, sizeof(struct ifreq)))
break;
session->mtu = ifr.ifr_mtu;
l2tp_info(session, L2TP_MSG_CONTROL, "%s: set mtu=%d\n",
session->name, session->mtu);
err = 0;
break;
case PPPIOCGMRU:
err = -ENXIO;
if (!(sk->sk_state & PPPOX_CONNECTED))
break;
err = -EFAULT;
if (put_user(session->mru, (int __user *) arg))
break;
l2tp_info(session, L2TP_MSG_CONTROL, "%s: get mru=%d\n",
session->name, session->mru);
err = 0;
break;
case PPPIOCSMRU:
err = -ENXIO;
if (!(sk->sk_state & PPPOX_CONNECTED))
break;
err = -EFAULT;
if (get_user(val, (int __user *) arg))
break;
session->mru = val;
l2tp_info(session, L2TP_MSG_CONTROL, "%s: set mru=%d\n",
session->name, session->mru);
err = 0;
break;
case PPPIOCGFLAGS:
err = -EFAULT;
if (put_user(ps->flags, (int __user *) arg))
break;
l2tp_info(session, L2TP_MSG_CONTROL, "%s: get flags=%d\n",
session->name, ps->flags);
err = 0;
break;
case PPPIOCSFLAGS:
err = -EFAULT;
if (get_user(val, (int __user *) arg))
break;
ps->flags = val;
l2tp_info(session, L2TP_MSG_CONTROL, "%s: set flags=%d\n",
session->name, ps->flags);
err = 0;
break;
case PPPIOCGL2TPSTATS:
err = -ENXIO;
if (!(sk->sk_state & PPPOX_CONNECTED))
break;
memset(&stats, 0, sizeof(stats));
stats.tunnel_id = tunnel->tunnel_id;
stats.session_id = session->session_id;
pppol2tp_copy_stats(&stats, &session->stats);
if (copy_to_user((void __user *) arg, &stats,
sizeof(stats)))
break;
l2tp_info(session, L2TP_MSG_CONTROL, "%s: get L2TP stats\n",
session->name);
err = 0;
break;
default:
err = -ENOSYS;
break;
}
sock_put(sk);
return err;
}
/* Tunnel ioctl helper.
*
* Note the special handling for PPPIOCGL2TPSTATS below. If the ioctl data
* specifies a session_id, the session ioctl handler is called. This allows an
* application to retrieve session stats via a tunnel socket.
*/
static int pppol2tp_tunnel_ioctl(struct l2tp_tunnel *tunnel,
unsigned int cmd, unsigned long arg)
{
int err = 0;
struct sock *sk;
struct pppol2tp_ioc_stats stats;
l2tp_dbg(tunnel, L2TP_MSG_CONTROL,
"%s: pppol2tp_tunnel_ioctl(cmd=%#x, arg=%#lx)\n",
tunnel->name, cmd, arg);
sk = tunnel->sock;
sock_hold(sk);
switch (cmd) {
case PPPIOCGL2TPSTATS:
err = -ENXIO;
if (!(sk->sk_state & PPPOX_CONNECTED))
break;
if (copy_from_user(&stats, (void __user *) arg,
sizeof(stats))) {
err = -EFAULT;
break;
}
if (stats.session_id != 0) {
/* resend to session ioctl handler */
struct l2tp_session *session =
l2tp_session_get(sock_net(sk), tunnel,
stats.session_id);
if (session && session->pwtype == L2TP_PWTYPE_PPP) {
err = pppol2tp_session_ioctl(session, cmd,
arg);
l2tp_session_dec_refcount(session);
} else {
err = -EBADR;
}
break;
}
#ifdef CONFIG_XFRM
stats.using_ipsec = (sk->sk_policy[0] || sk->sk_policy[1]) ? 1 : 0;
#endif
pppol2tp_copy_stats(&stats, &tunnel->stats);
if (copy_to_user((void __user *) arg, &stats, sizeof(stats))) {
err = -EFAULT;
break;
}
l2tp_info(tunnel, L2TP_MSG_CONTROL, "%s: get L2TP stats\n",
tunnel->name);
err = 0;
break;
default:
err = -ENOSYS;
break;
}
sock_put(sk);
return err;
}
/* Main ioctl() handler.
* Dispatch to tunnel or session helpers depending on the socket.
*/
static int pppol2tp_ioctl(struct socket *sock, unsigned int cmd,
unsigned long arg)
{
struct sock *sk = sock->sk;
struct l2tp_session *session;
struct l2tp_tunnel *tunnel;
int err;
if (!sk)
return 0;
err = -EBADF;
if (sock_flag(sk, SOCK_DEAD) != 0)
goto end;
err = -ENOTCONN;
if ((sk->sk_user_data == NULL) ||
(!(sk->sk_state & (PPPOX_CONNECTED | PPPOX_BOUND))))
goto end;
/* Get session context from the socket */
err = -EBADF;
session = pppol2tp_sock_to_session(sk);
if (session == NULL)
goto end;
/* Special case: if session's session_id is zero, treat ioctl as a
* tunnel ioctl
*/
if ((session->session_id == 0) &&
(session->peer_session_id == 0)) {
tunnel = session->tunnel;
err = pppol2tp_tunnel_ioctl(tunnel, cmd, arg);
goto end_put_sess;
}
err = pppol2tp_session_ioctl(session, cmd, arg);
end_put_sess:
sock_put(sk);
end:
return err;
}
/*****************************************************************************
* setsockopt() / getsockopt() support.
*
* The PPPoX socket is created for L2TP sessions: tunnels have their own UDP
* sockets. In order to control kernel tunnel features, we allow userspace to
* create a special "tunnel" PPPoX socket which is used for control only.
* Tunnel PPPoX sockets have session_id == 0 and simply allow the user
* application to issue L2TP setsockopt(), getsockopt() and ioctl() calls.
*****************************************************************************/
/* Tunnel setsockopt() helper.
*/
static int pppol2tp_tunnel_setsockopt(struct sock *sk,
struct l2tp_tunnel *tunnel,
int optname, int val)
{
int err = 0;
switch (optname) {
case PPPOL2TP_SO_DEBUG:
tunnel->debug = val;
l2tp_info(tunnel, L2TP_MSG_CONTROL, "%s: set debug=%x\n",
tunnel->name, tunnel->debug);
break;
default:
err = -ENOPROTOOPT;
break;
}
return err;
}
/* Session setsockopt helper.
*/
static int pppol2tp_session_setsockopt(struct sock *sk,
struct l2tp_session *session,
int optname, int val)
{
int err = 0;
switch (optname) {
case PPPOL2TP_SO_RECVSEQ:
if ((val != 0) && (val != 1)) {
err = -EINVAL;
break;
}
session->recv_seq = !!val;
l2tp_info(session, L2TP_MSG_CONTROL,
"%s: set recv_seq=%d\n",
session->name, session->recv_seq);
break;
case PPPOL2TP_SO_SENDSEQ:
if ((val != 0) && (val != 1)) {
err = -EINVAL;
break;
}
session->send_seq = !!val;
{
l2tp: protect sock pointer of struct pppol2tp_session with RCU pppol2tp_session_create() registers sessions that can't have their corresponding socket initialised. This socket has to be created by userspace, then connected to the session by pppol2tp_connect(). Therefore, we need to protect the pppol2tp socket pointer of L2TP sessions, so that it can safely be updated when userspace is connecting or closing the socket. This will eventually allow pppol2tp_connect() to avoid generating transient states while initialising its parts of the session. To this end, this patch protects the pppol2tp socket pointer using RCU. The pppol2tp socket pointer is still set in pppol2tp_connect(), but only once we know the function isn't going to fail. It's eventually reset by pppol2tp_release(), which now has to wait for a grace period to elapse before it can drop the last reference on the socket. This ensures that pppol2tp_session_get_sock() can safely grab a reference on the socket, even after ps->sk is reset to NULL but before this operation actually gets visible from pppol2tp_session_get_sock(). The rest is standard RCU conversion: pppol2tp_recv(), which already runs in atomic context, is simply enclosed by rcu_read_lock() and rcu_read_unlock(), while other functions are converted to use pppol2tp_session_get_sock() followed by sock_put(). pppol2tp_session_setsockopt() is a special case. It used to retrieve the pppol2tp socket from the L2TP session, which itself was retrieved from the pppol2tp socket. Therefore we can just avoid dereferencing ps->sk and directly use the original socket pointer instead. With all users of ps->sk now handling NULL and concurrent updates, the L2TP ->ref() and ->deref() callbacks aren't needed anymore. Therefore, rather than converting pppol2tp_session_sock_hold() and pppol2tp_session_sock_put(), we can just drop them. Signed-off-by: Guillaume Nault <g.nault@alphalink.fr> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-10-27 21:51:52 +07:00
struct pppox_sock *po = pppox_sk(sk);
po->chan.hdrlen = val ? PPPOL2TP_L2TP_HDR_SIZE_SEQ :
PPPOL2TP_L2TP_HDR_SIZE_NOSEQ;
}
l2tp_session_set_header_len(session, session->tunnel->version);
l2tp_info(session, L2TP_MSG_CONTROL,
"%s: set send_seq=%d\n",
session->name, session->send_seq);
break;
case PPPOL2TP_SO_LNSMODE:
if ((val != 0) && (val != 1)) {
err = -EINVAL;
break;
}
session->lns_mode = !!val;
l2tp_info(session, L2TP_MSG_CONTROL,
"%s: set lns_mode=%d\n",
session->name, session->lns_mode);
break;
case PPPOL2TP_SO_DEBUG:
session->debug = val;
l2tp_info(session, L2TP_MSG_CONTROL, "%s: set debug=%x\n",
session->name, session->debug);
break;
case PPPOL2TP_SO_REORDERTO:
session->reorder_timeout = msecs_to_jiffies(val);
l2tp_info(session, L2TP_MSG_CONTROL,
"%s: set reorder_timeout=%d\n",
session->name, session->reorder_timeout);
break;
default:
err = -ENOPROTOOPT;
break;
}
return err;
}
/* Main setsockopt() entry point.
* Does API checks, then calls either the tunnel or session setsockopt
* handler, according to whether the PPPoL2TP socket is a for a regular
* session or the special tunnel type.
*/
static int pppol2tp_setsockopt(struct socket *sock, int level, int optname,
char __user *optval, unsigned int optlen)
{
struct sock *sk = sock->sk;
struct l2tp_session *session;
struct l2tp_tunnel *tunnel;
int val;
int err;
if (level != SOL_PPPOL2TP)
return -EINVAL;
if (optlen < sizeof(int))
return -EINVAL;
if (get_user(val, (int __user *)optval))
return -EFAULT;
err = -ENOTCONN;
if (sk->sk_user_data == NULL)
goto end;
/* Get session context from the socket */
err = -EBADF;
session = pppol2tp_sock_to_session(sk);
if (session == NULL)
goto end;
/* Special case: if session_id == 0x0000, treat as operation on tunnel
*/
if ((session->session_id == 0) &&
(session->peer_session_id == 0)) {
tunnel = session->tunnel;
err = pppol2tp_tunnel_setsockopt(sk, tunnel, optname, val);
} else {
err = pppol2tp_session_setsockopt(sk, session, optname, val);
}
sock_put(sk);
end:
return err;
}
/* Tunnel getsockopt helper. Called with sock locked.
*/
static int pppol2tp_tunnel_getsockopt(struct sock *sk,
struct l2tp_tunnel *tunnel,
int optname, int *val)
{
int err = 0;
switch (optname) {
case PPPOL2TP_SO_DEBUG:
*val = tunnel->debug;
l2tp_info(tunnel, L2TP_MSG_CONTROL, "%s: get debug=%x\n",
tunnel->name, tunnel->debug);
break;
default:
err = -ENOPROTOOPT;
break;
}
return err;
}
/* Session getsockopt helper. Called with sock locked.
*/
static int pppol2tp_session_getsockopt(struct sock *sk,
struct l2tp_session *session,
int optname, int *val)
{
int err = 0;
switch (optname) {
case PPPOL2TP_SO_RECVSEQ:
*val = session->recv_seq;
l2tp_info(session, L2TP_MSG_CONTROL,
"%s: get recv_seq=%d\n", session->name, *val);
break;
case PPPOL2TP_SO_SENDSEQ:
*val = session->send_seq;
l2tp_info(session, L2TP_MSG_CONTROL,
"%s: get send_seq=%d\n", session->name, *val);
break;
case PPPOL2TP_SO_LNSMODE:
*val = session->lns_mode;
l2tp_info(session, L2TP_MSG_CONTROL,
"%s: get lns_mode=%d\n", session->name, *val);
break;
case PPPOL2TP_SO_DEBUG:
*val = session->debug;
l2tp_info(session, L2TP_MSG_CONTROL, "%s: get debug=%d\n",
session->name, *val);
break;
case PPPOL2TP_SO_REORDERTO:
*val = (int) jiffies_to_msecs(session->reorder_timeout);
l2tp_info(session, L2TP_MSG_CONTROL,
"%s: get reorder_timeout=%d\n", session->name, *val);
break;
default:
err = -ENOPROTOOPT;
}
return err;
}
/* Main getsockopt() entry point.
* Does API checks, then calls either the tunnel or session getsockopt
* handler, according to whether the PPPoX socket is a for a regular session
* or the special tunnel type.
*/
static int pppol2tp_getsockopt(struct socket *sock, int level, int optname,
char __user *optval, int __user *optlen)
{
struct sock *sk = sock->sk;
struct l2tp_session *session;
struct l2tp_tunnel *tunnel;
int val, len;
int err;
if (level != SOL_PPPOL2TP)
return -EINVAL;
if (get_user(len, optlen))
return -EFAULT;
len = min_t(unsigned int, len, sizeof(int));
if (len < 0)
return -EINVAL;
err = -ENOTCONN;
if (sk->sk_user_data == NULL)
goto end;
/* Get the session context */
err = -EBADF;
session = pppol2tp_sock_to_session(sk);
if (session == NULL)
goto end;
/* Special case: if session_id == 0x0000, treat as operation on tunnel */
if ((session->session_id == 0) &&
(session->peer_session_id == 0)) {
tunnel = session->tunnel;
err = pppol2tp_tunnel_getsockopt(sk, tunnel, optname, &val);
if (err)
goto end_put_sess;
} else {
err = pppol2tp_session_getsockopt(sk, session, optname, &val);
if (err)
goto end_put_sess;
}
err = -EFAULT;
if (put_user(len, optlen))
goto end_put_sess;
if (copy_to_user((void __user *) optval, &val, len))
goto end_put_sess;
err = 0;
end_put_sess:
sock_put(sk);
end:
return err;
}
/*****************************************************************************
* /proc filesystem for debug
* Since the original pppol2tp driver provided /proc/net/pppol2tp for
* L2TPv2, we dump only L2TPv2 tunnels and sessions here.
*****************************************************************************/
static unsigned int pppol2tp_net_id;
#ifdef CONFIG_PROC_FS
struct pppol2tp_seq_data {
struct seq_net_private p;
int tunnel_idx; /* current tunnel */
int session_idx; /* index of session within current tunnel */
struct l2tp_tunnel *tunnel;
struct l2tp_session *session; /* NULL means get next tunnel */
};
static void pppol2tp_next_tunnel(struct net *net, struct pppol2tp_seq_data *pd)
{
/* Drop reference taken during previous invocation */
if (pd->tunnel)
l2tp_tunnel_dec_refcount(pd->tunnel);
for (;;) {
pd->tunnel = l2tp_tunnel_get_nth(net, pd->tunnel_idx);
pd->tunnel_idx++;
/* Only accept L2TPv2 tunnels */
if (!pd->tunnel || pd->tunnel->version == 2)
return;
l2tp_tunnel_dec_refcount(pd->tunnel);
}
}
static void pppol2tp_next_session(struct net *net, struct pppol2tp_seq_data *pd)
{
/* Drop reference taken during previous invocation */
if (pd->session)
l2tp_session_dec_refcount(pd->session);
pd->session = l2tp_session_get_nth(pd->tunnel, pd->session_idx);
pd->session_idx++;
if (pd->session == NULL) {
pd->session_idx = 0;
pppol2tp_next_tunnel(net, pd);
}
}
static void *pppol2tp_seq_start(struct seq_file *m, loff_t *offs)
{
struct pppol2tp_seq_data *pd = SEQ_START_TOKEN;
loff_t pos = *offs;
struct net *net;
if (!pos)
goto out;
BUG_ON(m->private == NULL);
pd = m->private;
net = seq_file_net(m);
if (pd->tunnel == NULL)
pppol2tp_next_tunnel(net, pd);
else
pppol2tp_next_session(net, pd);
/* NULL tunnel and session indicates end of list */
if ((pd->tunnel == NULL) && (pd->session == NULL))
pd = NULL;
out:
return pd;
}
static void *pppol2tp_seq_next(struct seq_file *m, void *v, loff_t *pos)
{
(*pos)++;
return NULL;
}
static void pppol2tp_seq_stop(struct seq_file *p, void *v)
{
struct pppol2tp_seq_data *pd = v;
if (!pd || pd == SEQ_START_TOKEN)
return;
/* Drop reference taken by last invocation of pppol2tp_next_session()
* or pppol2tp_next_tunnel().
*/
if (pd->session) {
l2tp_session_dec_refcount(pd->session);
pd->session = NULL;
}
if (pd->tunnel) {
l2tp_tunnel_dec_refcount(pd->tunnel);
pd->tunnel = NULL;
}
}
static void pppol2tp_seq_tunnel_show(struct seq_file *m, void *v)
{
struct l2tp_tunnel *tunnel = v;
seq_printf(m, "\nTUNNEL '%s', %c %d\n",
tunnel->name,
(tunnel == tunnel->sock->sk_user_data) ? 'Y' : 'N',
refcount_read(&tunnel->ref_count) - 1);
seq_printf(m, " %08x %ld/%ld/%ld %ld/%ld/%ld\n",
tunnel->debug,
atomic_long_read(&tunnel->stats.tx_packets),
atomic_long_read(&tunnel->stats.tx_bytes),
atomic_long_read(&tunnel->stats.tx_errors),
atomic_long_read(&tunnel->stats.rx_packets),
atomic_long_read(&tunnel->stats.rx_bytes),
atomic_long_read(&tunnel->stats.rx_errors));
}
static void pppol2tp_seq_session_show(struct seq_file *m, void *v)
{
struct l2tp_session *session = v;
struct l2tp_tunnel *tunnel = session->tunnel;
l2tp: protect sock pointer of struct pppol2tp_session with RCU pppol2tp_session_create() registers sessions that can't have their corresponding socket initialised. This socket has to be created by userspace, then connected to the session by pppol2tp_connect(). Therefore, we need to protect the pppol2tp socket pointer of L2TP sessions, so that it can safely be updated when userspace is connecting or closing the socket. This will eventually allow pppol2tp_connect() to avoid generating transient states while initialising its parts of the session. To this end, this patch protects the pppol2tp socket pointer using RCU. The pppol2tp socket pointer is still set in pppol2tp_connect(), but only once we know the function isn't going to fail. It's eventually reset by pppol2tp_release(), which now has to wait for a grace period to elapse before it can drop the last reference on the socket. This ensures that pppol2tp_session_get_sock() can safely grab a reference on the socket, even after ps->sk is reset to NULL but before this operation actually gets visible from pppol2tp_session_get_sock(). The rest is standard RCU conversion: pppol2tp_recv(), which already runs in atomic context, is simply enclosed by rcu_read_lock() and rcu_read_unlock(), while other functions are converted to use pppol2tp_session_get_sock() followed by sock_put(). pppol2tp_session_setsockopt() is a special case. It used to retrieve the pppol2tp socket from the L2TP session, which itself was retrieved from the pppol2tp socket. Therefore we can just avoid dereferencing ps->sk and directly use the original socket pointer instead. With all users of ps->sk now handling NULL and concurrent updates, the L2TP ->ref() and ->deref() callbacks aren't needed anymore. Therefore, rather than converting pppol2tp_session_sock_hold() and pppol2tp_session_sock_put(), we can just drop them. Signed-off-by: Guillaume Nault <g.nault@alphalink.fr> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-10-27 21:51:52 +07:00
unsigned char state;
char user_data_ok;
struct sock *sk;
u32 ip = 0;
u16 port = 0;
if (tunnel->sock) {
struct inet_sock *inet = inet_sk(tunnel->sock);
ip = ntohl(inet->inet_saddr);
port = ntohs(inet->inet_sport);
}
l2tp: protect sock pointer of struct pppol2tp_session with RCU pppol2tp_session_create() registers sessions that can't have their corresponding socket initialised. This socket has to be created by userspace, then connected to the session by pppol2tp_connect(). Therefore, we need to protect the pppol2tp socket pointer of L2TP sessions, so that it can safely be updated when userspace is connecting or closing the socket. This will eventually allow pppol2tp_connect() to avoid generating transient states while initialising its parts of the session. To this end, this patch protects the pppol2tp socket pointer using RCU. The pppol2tp socket pointer is still set in pppol2tp_connect(), but only once we know the function isn't going to fail. It's eventually reset by pppol2tp_release(), which now has to wait for a grace period to elapse before it can drop the last reference on the socket. This ensures that pppol2tp_session_get_sock() can safely grab a reference on the socket, even after ps->sk is reset to NULL but before this operation actually gets visible from pppol2tp_session_get_sock(). The rest is standard RCU conversion: pppol2tp_recv(), which already runs in atomic context, is simply enclosed by rcu_read_lock() and rcu_read_unlock(), while other functions are converted to use pppol2tp_session_get_sock() followed by sock_put(). pppol2tp_session_setsockopt() is a special case. It used to retrieve the pppol2tp socket from the L2TP session, which itself was retrieved from the pppol2tp socket. Therefore we can just avoid dereferencing ps->sk and directly use the original socket pointer instead. With all users of ps->sk now handling NULL and concurrent updates, the L2TP ->ref() and ->deref() callbacks aren't needed anymore. Therefore, rather than converting pppol2tp_session_sock_hold() and pppol2tp_session_sock_put(), we can just drop them. Signed-off-by: Guillaume Nault <g.nault@alphalink.fr> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-10-27 21:51:52 +07:00
sk = pppol2tp_session_get_sock(session);
if (sk) {
state = sk->sk_state;
user_data_ok = (session == sk->sk_user_data) ? 'Y' : 'N';
} else {
state = 0;
user_data_ok = 'N';
}
seq_printf(m, " SESSION '%s' %08X/%d %04X/%04X -> "
"%04X/%04X %d %c\n",
session->name, ip, port,
tunnel->tunnel_id,
session->session_id,
tunnel->peer_tunnel_id,
session->peer_session_id,
l2tp: protect sock pointer of struct pppol2tp_session with RCU pppol2tp_session_create() registers sessions that can't have their corresponding socket initialised. This socket has to be created by userspace, then connected to the session by pppol2tp_connect(). Therefore, we need to protect the pppol2tp socket pointer of L2TP sessions, so that it can safely be updated when userspace is connecting or closing the socket. This will eventually allow pppol2tp_connect() to avoid generating transient states while initialising its parts of the session. To this end, this patch protects the pppol2tp socket pointer using RCU. The pppol2tp socket pointer is still set in pppol2tp_connect(), but only once we know the function isn't going to fail. It's eventually reset by pppol2tp_release(), which now has to wait for a grace period to elapse before it can drop the last reference on the socket. This ensures that pppol2tp_session_get_sock() can safely grab a reference on the socket, even after ps->sk is reset to NULL but before this operation actually gets visible from pppol2tp_session_get_sock(). The rest is standard RCU conversion: pppol2tp_recv(), which already runs in atomic context, is simply enclosed by rcu_read_lock() and rcu_read_unlock(), while other functions are converted to use pppol2tp_session_get_sock() followed by sock_put(). pppol2tp_session_setsockopt() is a special case. It used to retrieve the pppol2tp socket from the L2TP session, which itself was retrieved from the pppol2tp socket. Therefore we can just avoid dereferencing ps->sk and directly use the original socket pointer instead. With all users of ps->sk now handling NULL and concurrent updates, the L2TP ->ref() and ->deref() callbacks aren't needed anymore. Therefore, rather than converting pppol2tp_session_sock_hold() and pppol2tp_session_sock_put(), we can just drop them. Signed-off-by: Guillaume Nault <g.nault@alphalink.fr> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-10-27 21:51:52 +07:00
state, user_data_ok);
seq_printf(m, " %d/%d/%c/%c/%s %08x %u\n",
session->mtu, session->mru,
session->recv_seq ? 'R' : '-',
session->send_seq ? 'S' : '-',
session->lns_mode ? "LNS" : "LAC",
session->debug,
jiffies_to_msecs(session->reorder_timeout));
seq_printf(m, " %hu/%hu %ld/%ld/%ld %ld/%ld/%ld\n",
session->nr, session->ns,
atomic_long_read(&session->stats.tx_packets),
atomic_long_read(&session->stats.tx_bytes),
atomic_long_read(&session->stats.tx_errors),
atomic_long_read(&session->stats.rx_packets),
atomic_long_read(&session->stats.rx_bytes),
atomic_long_read(&session->stats.rx_errors));
l2tp: protect sock pointer of struct pppol2tp_session with RCU pppol2tp_session_create() registers sessions that can't have their corresponding socket initialised. This socket has to be created by userspace, then connected to the session by pppol2tp_connect(). Therefore, we need to protect the pppol2tp socket pointer of L2TP sessions, so that it can safely be updated when userspace is connecting or closing the socket. This will eventually allow pppol2tp_connect() to avoid generating transient states while initialising its parts of the session. To this end, this patch protects the pppol2tp socket pointer using RCU. The pppol2tp socket pointer is still set in pppol2tp_connect(), but only once we know the function isn't going to fail. It's eventually reset by pppol2tp_release(), which now has to wait for a grace period to elapse before it can drop the last reference on the socket. This ensures that pppol2tp_session_get_sock() can safely grab a reference on the socket, even after ps->sk is reset to NULL but before this operation actually gets visible from pppol2tp_session_get_sock(). The rest is standard RCU conversion: pppol2tp_recv(), which already runs in atomic context, is simply enclosed by rcu_read_lock() and rcu_read_unlock(), while other functions are converted to use pppol2tp_session_get_sock() followed by sock_put(). pppol2tp_session_setsockopt() is a special case. It used to retrieve the pppol2tp socket from the L2TP session, which itself was retrieved from the pppol2tp socket. Therefore we can just avoid dereferencing ps->sk and directly use the original socket pointer instead. With all users of ps->sk now handling NULL and concurrent updates, the L2TP ->ref() and ->deref() callbacks aren't needed anymore. Therefore, rather than converting pppol2tp_session_sock_hold() and pppol2tp_session_sock_put(), we can just drop them. Signed-off-by: Guillaume Nault <g.nault@alphalink.fr> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-10-27 21:51:52 +07:00
if (sk) {
struct pppox_sock *po = pppox_sk(sk);
seq_printf(m, " interface %s\n", ppp_dev_name(&po->chan));
l2tp: protect sock pointer of struct pppol2tp_session with RCU pppol2tp_session_create() registers sessions that can't have their corresponding socket initialised. This socket has to be created by userspace, then connected to the session by pppol2tp_connect(). Therefore, we need to protect the pppol2tp socket pointer of L2TP sessions, so that it can safely be updated when userspace is connecting or closing the socket. This will eventually allow pppol2tp_connect() to avoid generating transient states while initialising its parts of the session. To this end, this patch protects the pppol2tp socket pointer using RCU. The pppol2tp socket pointer is still set in pppol2tp_connect(), but only once we know the function isn't going to fail. It's eventually reset by pppol2tp_release(), which now has to wait for a grace period to elapse before it can drop the last reference on the socket. This ensures that pppol2tp_session_get_sock() can safely grab a reference on the socket, even after ps->sk is reset to NULL but before this operation actually gets visible from pppol2tp_session_get_sock(). The rest is standard RCU conversion: pppol2tp_recv(), which already runs in atomic context, is simply enclosed by rcu_read_lock() and rcu_read_unlock(), while other functions are converted to use pppol2tp_session_get_sock() followed by sock_put(). pppol2tp_session_setsockopt() is a special case. It used to retrieve the pppol2tp socket from the L2TP session, which itself was retrieved from the pppol2tp socket. Therefore we can just avoid dereferencing ps->sk and directly use the original socket pointer instead. With all users of ps->sk now handling NULL and concurrent updates, the L2TP ->ref() and ->deref() callbacks aren't needed anymore. Therefore, rather than converting pppol2tp_session_sock_hold() and pppol2tp_session_sock_put(), we can just drop them. Signed-off-by: Guillaume Nault <g.nault@alphalink.fr> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-10-27 21:51:52 +07:00
sock_put(sk);
}
}
static int pppol2tp_seq_show(struct seq_file *m, void *v)
{
struct pppol2tp_seq_data *pd = v;
/* display header on line 1 */
if (v == SEQ_START_TOKEN) {
seq_puts(m, "PPPoL2TP driver info, " PPPOL2TP_DRV_VERSION "\n");
seq_puts(m, "TUNNEL name, user-data-ok session-count\n");
seq_puts(m, " debug tx-pkts/bytes/errs rx-pkts/bytes/errs\n");
seq_puts(m, " SESSION name, addr/port src-tid/sid "
"dest-tid/sid state user-data-ok\n");
seq_puts(m, " mtu/mru/rcvseq/sendseq/lns debug reorderto\n");
seq_puts(m, " nr/ns tx-pkts/bytes/errs rx-pkts/bytes/errs\n");
goto out;
}
if (!pd->session)
pppol2tp_seq_tunnel_show(m, pd->tunnel);
else
pppol2tp_seq_session_show(m, pd->session);
out:
return 0;
}
static const struct seq_operations pppol2tp_seq_ops = {
.start = pppol2tp_seq_start,
.next = pppol2tp_seq_next,
.stop = pppol2tp_seq_stop,
.show = pppol2tp_seq_show,
};
#endif /* CONFIG_PROC_FS */
/*****************************************************************************
* Network namespace
*****************************************************************************/
static __net_init int pppol2tp_init_net(struct net *net)
{
struct proc_dir_entry *pde;
int err = 0;
pde = proc_create_net("pppol2tp", 0444, net->proc_net,
&pppol2tp_seq_ops, sizeof(struct pppol2tp_seq_data));
if (!pde) {
err = -ENOMEM;
goto out;
}
out:
return err;
}
static __net_exit void pppol2tp_exit_net(struct net *net)
{
remove_proc_entry("pppol2tp", net->proc_net);
}
static struct pernet_operations pppol2tp_net_ops = {
.init = pppol2tp_init_net,
.exit = pppol2tp_exit_net,
.id = &pppol2tp_net_id,
};
/*****************************************************************************
* Init and cleanup
*****************************************************************************/
static const struct proto_ops pppol2tp_ops = {
.family = AF_PPPOX,
.owner = THIS_MODULE,
.release = pppol2tp_release,
.bind = sock_no_bind,
.connect = pppol2tp_connect,
.socketpair = sock_no_socketpair,
.accept = sock_no_accept,
.getname = pppol2tp_getname,
.poll = datagram_poll,
.listen = sock_no_listen,
.shutdown = sock_no_shutdown,
.setsockopt = pppol2tp_setsockopt,
.getsockopt = pppol2tp_getsockopt,
.sendmsg = pppol2tp_sendmsg,
.recvmsg = pppol2tp_recvmsg,
.mmap = sock_no_mmap,
.ioctl = pppox_ioctl,
};
static const struct pppox_proto pppol2tp_proto = {
.create = pppol2tp_create,
.ioctl = pppol2tp_ioctl,
.owner = THIS_MODULE,
};
#ifdef CONFIG_L2TP_V3
static const struct l2tp_nl_cmd_ops pppol2tp_nl_cmd_ops = {
.session_create = pppol2tp_session_create,
.session_delete = l2tp_session_delete,
};
#endif /* CONFIG_L2TP_V3 */
static int __init pppol2tp_init(void)
{
int err;
err = register_pernet_device(&pppol2tp_net_ops);
if (err)
goto out;
err = proto_register(&pppol2tp_sk_proto, 0);
if (err)
goto out_unregister_pppol2tp_pernet;
err = register_pppox_proto(PX_PROTO_OL2TP, &pppol2tp_proto);
if (err)
goto out_unregister_pppol2tp_proto;
#ifdef CONFIG_L2TP_V3
err = l2tp_nl_register_ops(L2TP_PWTYPE_PPP, &pppol2tp_nl_cmd_ops);
if (err)
goto out_unregister_pppox;
#endif
pr_info("PPPoL2TP kernel driver, %s\n", PPPOL2TP_DRV_VERSION);
out:
return err;
#ifdef CONFIG_L2TP_V3
out_unregister_pppox:
unregister_pppox_proto(PX_PROTO_OL2TP);
#endif
out_unregister_pppol2tp_proto:
proto_unregister(&pppol2tp_sk_proto);
out_unregister_pppol2tp_pernet:
unregister_pernet_device(&pppol2tp_net_ops);
goto out;
}
static void __exit pppol2tp_exit(void)
{
#ifdef CONFIG_L2TP_V3
l2tp_nl_unregister_ops(L2TP_PWTYPE_PPP);
#endif
unregister_pppox_proto(PX_PROTO_OL2TP);
proto_unregister(&pppol2tp_sk_proto);
unregister_pernet_device(&pppol2tp_net_ops);
}
module_init(pppol2tp_init);
module_exit(pppol2tp_exit);
MODULE_AUTHOR("James Chapman <jchapman@katalix.com>");
MODULE_DESCRIPTION("PPP over L2TP over UDP");
MODULE_LICENSE("GPL");
MODULE_VERSION(PPPOL2TP_DRV_VERSION);
MODULE_ALIAS_NET_PF_PROTO(PF_PPPOX, PX_PROTO_OL2TP);
MODULE_ALIAS_L2TP_PWTYPE(7);