linux_dsm_epyc7002/net/bluetooth/l2cap_sock.c

1754 lines
38 KiB
C
Raw Normal View History

/*
BlueZ - Bluetooth protocol stack for Linux
Copyright (C) 2000-2001 Qualcomm Incorporated
Copyright (C) 2009-2010 Gustavo F. Padovan <gustavo@padovan.org>
Copyright (C) 2010 Google Inc.
Copyright (C) 2011 ProFUSION Embedded Systems
Written 2000,2001 by Maxim Krasnyansky <maxk@qualcomm.com>
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License version 2 as
published by the Free Software Foundation;
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT OF THIRD PARTY RIGHTS.
IN NO EVENT SHALL THE COPYRIGHT HOLDER(S) AND AUTHOR(S) BE LIABLE FOR ANY
CLAIM, OR ANY SPECIAL INDIRECT OR CONSEQUENTIAL DAMAGES, OR ANY DAMAGES
WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
ALL LIABILITY, INCLUDING LIABILITY FOR INFRINGEMENT OF ANY PATENTS,
COPYRIGHTS, TRADEMARKS OR OTHER RIGHTS, RELATING TO USE OF THIS
SOFTWARE IS DISCLAIMED.
*/
/* Bluetooth L2CAP sockets. */
#include <linux/module.h>
#include <linux/export.h>
#include <linux/sched/signal.h>
#include <net/bluetooth/bluetooth.h>
#include <net/bluetooth/hci_core.h>
#include <net/bluetooth/l2cap.h>
#include "smp.h"
static struct bt_sock_list l2cap_sk_list = {
.lock = __RW_LOCK_UNLOCKED(l2cap_sk_list.lock)
};
static const struct proto_ops l2cap_sock_ops;
static void l2cap_sock_init(struct sock *sk, struct sock *parent);
static struct sock *l2cap_sock_alloc(struct net *net, struct socket *sock,
int proto, gfp_t prio, int kern);
bool l2cap_is_socket(struct socket *sock)
{
return sock && sock->ops == &l2cap_sock_ops;
}
EXPORT_SYMBOL(l2cap_is_socket);
static int l2cap_validate_bredr_psm(u16 psm)
{
/* PSM must be odd and lsb of upper byte must be 0 */
if ((psm & 0x0101) != 0x0001)
return -EINVAL;
/* Restrict usage of well-known PSMs */
if (psm < L2CAP_PSM_DYN_START && !capable(CAP_NET_BIND_SERVICE))
return -EACCES;
return 0;
}
static int l2cap_validate_le_psm(u16 psm)
{
/* Valid LE_PSM ranges are defined only until 0x00ff */
if (psm > L2CAP_PSM_LE_DYN_END)
return -EINVAL;
/* Restrict fixed, SIG assigned PSM values to CAP_NET_BIND_SERVICE */
if (psm < L2CAP_PSM_LE_DYN_START && !capable(CAP_NET_BIND_SERVICE))
return -EACCES;
return 0;
}
static int l2cap_sock_bind(struct socket *sock, struct sockaddr *addr, int alen)
{
struct sock *sk = sock->sk;
struct l2cap_chan *chan = l2cap_pi(sk)->chan;
struct sockaddr_l2 la;
int len, err = 0;
BT_DBG("sk %p", sk);
if (!addr || alen < offsetofend(struct sockaddr, sa_family) ||
addr->sa_family != AF_BLUETOOTH)
return -EINVAL;
memset(&la, 0, sizeof(la));
len = min_t(unsigned int, sizeof(la), alen);
memcpy(&la, addr, len);
if (la.l2_cid && la.l2_psm)
return -EINVAL;
if (!bdaddr_type_is_valid(la.l2_bdaddr_type))
return -EINVAL;
if (bdaddr_type_is_le(la.l2_bdaddr_type)) {
/* We only allow ATT user space socket */
if (la.l2_cid &&
la.l2_cid != cpu_to_le16(L2CAP_CID_ATT))
return -EINVAL;
}
lock_sock(sk);
if (sk->sk_state != BT_OPEN) {
err = -EBADFD;
goto done;
}
if (la.l2_psm) {
__u16 psm = __le16_to_cpu(la.l2_psm);
if (la.l2_bdaddr_type == BDADDR_BREDR)
err = l2cap_validate_bredr_psm(psm);
else
err = l2cap_validate_le_psm(psm);
if (err)
goto done;
}
bacpy(&chan->src, &la.l2_bdaddr);
chan->src_type = la.l2_bdaddr_type;
if (la.l2_cid)
err = l2cap_add_scid(chan, __le16_to_cpu(la.l2_cid));
else
err = l2cap_add_psm(chan, &la.l2_bdaddr, la.l2_psm);
if (err < 0)
goto done;
switch (chan->chan_type) {
case L2CAP_CHAN_CONN_LESS:
if (__le16_to_cpu(la.l2_psm) == L2CAP_PSM_3DSP)
chan->sec_level = BT_SECURITY_SDP;
break;
case L2CAP_CHAN_CONN_ORIENTED:
if (__le16_to_cpu(la.l2_psm) == L2CAP_PSM_SDP ||
__le16_to_cpu(la.l2_psm) == L2CAP_PSM_RFCOMM)
chan->sec_level = BT_SECURITY_SDP;
break;
case L2CAP_CHAN_RAW:
chan->sec_level = BT_SECURITY_SDP;
break;
case L2CAP_CHAN_FIXED:
/* Fixed channels default to the L2CAP core not holding a
* hci_conn reference for them. For fixed channels mapping to
* L2CAP sockets we do want to hold a reference so set the
* appropriate flag to request it.
*/
set_bit(FLAG_HOLD_HCI_CONN, &chan->flags);
break;
}
if (chan->psm && bdaddr_type_is_le(chan->src_type))
chan->mode = L2CAP_MODE_LE_FLOWCTL;
chan->state = BT_BOUND;
sk->sk_state = BT_BOUND;
done:
release_sock(sk);
return err;
}
static int l2cap_sock_connect(struct socket *sock, struct sockaddr *addr,
int alen, int flags)
{
struct sock *sk = sock->sk;
struct l2cap_chan *chan = l2cap_pi(sk)->chan;
struct sockaddr_l2 la;
int len, err = 0;
BT_DBG("sk %p", sk);
if (!addr || alen < offsetofend(struct sockaddr, sa_family) ||
addr->sa_family != AF_BLUETOOTH)
return -EINVAL;
memset(&la, 0, sizeof(la));
len = min_t(unsigned int, sizeof(la), alen);
memcpy(&la, addr, len);
if (la.l2_cid && la.l2_psm)
return -EINVAL;
if (!bdaddr_type_is_valid(la.l2_bdaddr_type))
return -EINVAL;
/* Check that the socket wasn't bound to something that
* conflicts with the address given to connect(). If chan->src
* is BDADDR_ANY it means bind() was never used, in which case
* chan->src_type and la.l2_bdaddr_type do not need to match.
*/
if (chan->src_type == BDADDR_BREDR && bacmp(&chan->src, BDADDR_ANY) &&
bdaddr_type_is_le(la.l2_bdaddr_type)) {
/* Old user space versions will try to incorrectly bind
* the ATT socket using BDADDR_BREDR. We need to accept
* this and fix up the source address type only when
* both the source CID and destination CID indicate
* ATT. Anything else is an invalid combination.
*/
if (chan->scid != L2CAP_CID_ATT ||
la.l2_cid != cpu_to_le16(L2CAP_CID_ATT))
return -EINVAL;
/* We don't have the hdev available here to make a
* better decision on random vs public, but since all
* user space versions that exhibit this issue anyway do
* not support random local addresses assuming public
* here is good enough.
*/
chan->src_type = BDADDR_LE_PUBLIC;
}
if (chan->src_type != BDADDR_BREDR && la.l2_bdaddr_type == BDADDR_BREDR)
return -EINVAL;
if (bdaddr_type_is_le(la.l2_bdaddr_type)) {
/* We only allow ATT user space socket */
if (la.l2_cid &&
la.l2_cid != cpu_to_le16(L2CAP_CID_ATT))
return -EINVAL;
}
if (chan->psm && bdaddr_type_is_le(chan->src_type) && !chan->mode)
chan->mode = L2CAP_MODE_LE_FLOWCTL;
err = l2cap_chan_connect(chan, la.l2_psm, __le16_to_cpu(la.l2_cid),
&la.l2_bdaddr, la.l2_bdaddr_type);
if (err)
return err;
lock_sock(sk);
err = bt_sock_wait_state(sk, BT_CONNECTED,
sock_sndtimeo(sk, flags & O_NONBLOCK));
release_sock(sk);
return err;
}
static int l2cap_sock_listen(struct socket *sock, int backlog)
{
struct sock *sk = sock->sk;
struct l2cap_chan *chan = l2cap_pi(sk)->chan;
int err = 0;
BT_DBG("sk %p backlog %d", sk, backlog);
lock_sock(sk);
if (sk->sk_state != BT_BOUND) {
err = -EBADFD;
goto done;
}
if (sk->sk_type != SOCK_SEQPACKET && sk->sk_type != SOCK_STREAM) {
err = -EINVAL;
goto done;
}
switch (chan->mode) {
case L2CAP_MODE_BASIC:
case L2CAP_MODE_LE_FLOWCTL:
case L2CAP_MODE_EXT_FLOWCTL:
break;
case L2CAP_MODE_ERTM:
case L2CAP_MODE_STREAMING:
if (!disable_ertm)
break;
/* fall through */
default:
err = -EOPNOTSUPP;
goto done;
}
sk->sk_max_ack_backlog = backlog;
sk->sk_ack_backlog = 0;
/* Listening channels need to use nested locking in order not to
* cause lockdep warnings when the created child channels end up
* being locked in the same thread as the parent channel.
*/
atomic_set(&chan->nesting, L2CAP_NESTING_PARENT);
chan->state = BT_LISTEN;
sk->sk_state = BT_LISTEN;
done:
release_sock(sk);
return err;
}
static int l2cap_sock_accept(struct socket *sock, struct socket *newsock,
net: Work around lockdep limitation in sockets that use sockets Lockdep issues a circular dependency warning when AFS issues an operation through AF_RXRPC from a context in which the VFS/VM holds the mmap_sem. The theory lockdep comes up with is as follows: (1) If the pagefault handler decides it needs to read pages from AFS, it calls AFS with mmap_sem held and AFS begins an AF_RXRPC call, but creating a call requires the socket lock: mmap_sem must be taken before sk_lock-AF_RXRPC (2) afs_open_socket() opens an AF_RXRPC socket and binds it. rxrpc_bind() binds the underlying UDP socket whilst holding its socket lock. inet_bind() takes its own socket lock: sk_lock-AF_RXRPC must be taken before sk_lock-AF_INET (3) Reading from a TCP socket into a userspace buffer might cause a fault and thus cause the kernel to take the mmap_sem, but the TCP socket is locked whilst doing this: sk_lock-AF_INET must be taken before mmap_sem However, lockdep's theory is wrong in this instance because it deals only with lock classes and not individual locks. The AF_INET lock in (2) isn't really equivalent to the AF_INET lock in (3) as the former deals with a socket entirely internal to the kernel that never sees userspace. This is a limitation in the design of lockdep. Fix the general case by: (1) Double up all the locking keys used in sockets so that one set are used if the socket is created by userspace and the other set is used if the socket is created by the kernel. (2) Store the kern parameter passed to sk_alloc() in a variable in the sock struct (sk_kern_sock). This informs sock_lock_init(), sock_init_data() and sk_clone_lock() as to the lock keys to be used. Note that the child created by sk_clone_lock() inherits the parent's kern setting. (3) Add a 'kern' parameter to ->accept() that is analogous to the one passed in to ->create() that distinguishes whether kernel_accept() or sys_accept4() was the caller and can be passed to sk_alloc(). Note that a lot of accept functions merely dequeue an already allocated socket. I haven't touched these as the new socket already exists before we get the parameter. Note also that there are a couple of places where I've made the accepted socket unconditionally kernel-based: irda_accept() rds_rcp_accept_one() tcp_accept_from_sock() because they follow a sock_create_kern() and accept off of that. Whilst creating this, I noticed that lustre and ocfs don't create sockets through sock_create_kern() and thus they aren't marked as for-kernel, though they appear to be internal. I wonder if these should do that so that they use the new set of lock keys. Signed-off-by: David Howells <dhowells@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-03-09 15:09:05 +07:00
int flags, bool kern)
{
DEFINE_WAIT_FUNC(wait, woken_wake_function);
struct sock *sk = sock->sk, *nsk;
long timeo;
int err = 0;
lock_sock_nested(sk, L2CAP_NESTING_PARENT);
timeo = sock_rcvtimeo(sk, flags & O_NONBLOCK);
BT_DBG("sk %p timeo %ld", sk, timeo);
/* Wait for an incoming connection. (wake-one). */
add_wait_queue_exclusive(sk_sleep(sk), &wait);
while (1) {
if (sk->sk_state != BT_LISTEN) {
err = -EBADFD;
break;
}
nsk = bt_accept_dequeue(sk, newsock);
if (nsk)
break;
if (!timeo) {
err = -EAGAIN;
break;
}
if (signal_pending(current)) {
err = sock_intr_errno(timeo);
break;
}
release_sock(sk);
timeo = wait_woken(&wait, TASK_INTERRUPTIBLE, timeo);
lock_sock_nested(sk, L2CAP_NESTING_PARENT);
}
remove_wait_queue(sk_sleep(sk), &wait);
if (err)
goto done;
newsock->state = SS_CONNECTED;
BT_DBG("new socket %p", nsk);
done:
release_sock(sk);
return err;
}
static int l2cap_sock_getname(struct socket *sock, struct sockaddr *addr,
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)
{
struct sockaddr_l2 *la = (struct sockaddr_l2 *) addr;
struct sock *sk = sock->sk;
struct l2cap_chan *chan = l2cap_pi(sk)->chan;
BT_DBG("sock %p, sk %p", sock, sk);
if (peer && sk->sk_state != BT_CONNECTED &&
sk->sk_state != BT_CONNECT && sk->sk_state != BT_CONNECT2 &&
sk->sk_state != BT_CONFIG)
return -ENOTCONN;
memset(la, 0, sizeof(struct sockaddr_l2));
addr->sa_family = AF_BLUETOOTH;
la->l2_psm = chan->psm;
if (peer) {
bacpy(&la->l2_bdaddr, &chan->dst);
la->l2_cid = cpu_to_le16(chan->dcid);
la->l2_bdaddr_type = chan->dst_type;
} else {
bacpy(&la->l2_bdaddr, &chan->src);
la->l2_cid = cpu_to_le16(chan->scid);
la->l2_bdaddr_type = chan->src_type;
}
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
return sizeof(struct sockaddr_l2);
}
static int l2cap_sock_getsockopt_old(struct socket *sock, int optname,
char __user *optval, int __user *optlen)
{
struct sock *sk = sock->sk;
struct l2cap_chan *chan = l2cap_pi(sk)->chan;
struct l2cap_options opts;
struct l2cap_conninfo cinfo;
int len, err = 0;
u32 opt;
BT_DBG("sk %p", sk);
if (get_user(len, optlen))
return -EFAULT;
lock_sock(sk);
switch (optname) {
case L2CAP_OPTIONS:
/* LE sockets should use BT_SNDMTU/BT_RCVMTU, but since
* legacy ATT code depends on getsockopt for
* L2CAP_OPTIONS we need to let this pass.
*/
if (bdaddr_type_is_le(chan->src_type) &&
chan->scid != L2CAP_CID_ATT) {
err = -EINVAL;
break;
}
memset(&opts, 0, sizeof(opts));
opts.imtu = chan->imtu;
opts.omtu = chan->omtu;
opts.flush_to = chan->flush_to;
opts.mode = chan->mode;
opts.fcs = chan->fcs;
opts.max_tx = chan->max_tx;
opts.txwin_size = chan->tx_win;
BT_DBG("mode 0x%2.2x", chan->mode);
len = min_t(unsigned int, len, sizeof(opts));
if (copy_to_user(optval, (char *) &opts, len))
err = -EFAULT;
break;
case L2CAP_LM:
switch (chan->sec_level) {
case BT_SECURITY_LOW:
opt = L2CAP_LM_AUTH;
break;
case BT_SECURITY_MEDIUM:
opt = L2CAP_LM_AUTH | L2CAP_LM_ENCRYPT;
break;
case BT_SECURITY_HIGH:
opt = L2CAP_LM_AUTH | L2CAP_LM_ENCRYPT |
L2CAP_LM_SECURE;
break;
case BT_SECURITY_FIPS:
opt = L2CAP_LM_AUTH | L2CAP_LM_ENCRYPT |
L2CAP_LM_SECURE | L2CAP_LM_FIPS;
break;
default:
opt = 0;
break;
}
if (test_bit(FLAG_ROLE_SWITCH, &chan->flags))
opt |= L2CAP_LM_MASTER;
if (test_bit(FLAG_FORCE_RELIABLE, &chan->flags))
opt |= L2CAP_LM_RELIABLE;
if (put_user(opt, (u32 __user *) optval))
err = -EFAULT;
break;
case L2CAP_CONNINFO:
if (sk->sk_state != BT_CONNECTED &&
!(sk->sk_state == BT_CONNECT2 &&
test_bit(BT_SK_DEFER_SETUP, &bt_sk(sk)->flags))) {
err = -ENOTCONN;
break;
}
memset(&cinfo, 0, sizeof(cinfo));
cinfo.hci_handle = chan->conn->hcon->handle;
memcpy(cinfo.dev_class, chan->conn->hcon->dev_class, 3);
len = min_t(unsigned int, len, sizeof(cinfo));
if (copy_to_user(optval, (char *) &cinfo, len))
err = -EFAULT;
break;
default:
err = -ENOPROTOOPT;
break;
}
release_sock(sk);
return err;
}
static int l2cap_sock_getsockopt(struct socket *sock, int level, int optname,
char __user *optval, int __user *optlen)
{
struct sock *sk = sock->sk;
struct l2cap_chan *chan = l2cap_pi(sk)->chan;
struct bt_security sec;
struct bt_power pwr;
u32 phys;
int len, err = 0;
BT_DBG("sk %p", sk);
if (level == SOL_L2CAP)
return l2cap_sock_getsockopt_old(sock, optname, optval, optlen);
if (level != SOL_BLUETOOTH)
return -ENOPROTOOPT;
if (get_user(len, optlen))
return -EFAULT;
lock_sock(sk);
switch (optname) {
case BT_SECURITY:
if (chan->chan_type != L2CAP_CHAN_CONN_ORIENTED &&
chan->chan_type != L2CAP_CHAN_FIXED &&
chan->chan_type != L2CAP_CHAN_RAW) {
err = -EINVAL;
break;
}
memset(&sec, 0, sizeof(sec));
if (chan->conn) {
sec.level = chan->conn->hcon->sec_level;
if (sk->sk_state == BT_CONNECTED)
sec.key_size = chan->conn->hcon->enc_key_size;
} else {
sec.level = chan->sec_level;
}
len = min_t(unsigned int, len, sizeof(sec));
if (copy_to_user(optval, (char *) &sec, len))
err = -EFAULT;
break;
case BT_DEFER_SETUP:
if (sk->sk_state != BT_BOUND && sk->sk_state != BT_LISTEN) {
err = -EINVAL;
break;
}
if (put_user(test_bit(BT_SK_DEFER_SETUP, &bt_sk(sk)->flags),
(u32 __user *) optval))
err = -EFAULT;
break;
case BT_FLUSHABLE:
if (put_user(test_bit(FLAG_FLUSHABLE, &chan->flags),
(u32 __user *) optval))
err = -EFAULT;
break;
case BT_POWER:
if (sk->sk_type != SOCK_SEQPACKET && sk->sk_type != SOCK_STREAM
&& sk->sk_type != SOCK_RAW) {
err = -EINVAL;
break;
}
pwr.force_active = test_bit(FLAG_FORCE_ACTIVE, &chan->flags);
len = min_t(unsigned int, len, sizeof(pwr));
if (copy_to_user(optval, (char *) &pwr, len))
err = -EFAULT;
break;
case BT_CHANNEL_POLICY:
if (put_user(chan->chan_policy, (u32 __user *) optval))
err = -EFAULT;
break;
case BT_SNDMTU:
if (!bdaddr_type_is_le(chan->src_type)) {
err = -EINVAL;
break;
}
if (sk->sk_state != BT_CONNECTED) {
err = -ENOTCONN;
break;
}
if (put_user(chan->omtu, (u16 __user *) optval))
err = -EFAULT;
break;
case BT_RCVMTU:
if (!bdaddr_type_is_le(chan->src_type)) {
err = -EINVAL;
break;
}
if (put_user(chan->imtu, (u16 __user *) optval))
err = -EFAULT;
break;
case BT_PHY:
if (sk->sk_state != BT_CONNECTED) {
err = -ENOTCONN;
break;
}
phys = hci_conn_get_phy(chan->conn->hcon);
if (put_user(phys, (u32 __user *) optval))
err = -EFAULT;
break;
default:
err = -ENOPROTOOPT;
break;
}
release_sock(sk);
return err;
}
static bool l2cap_valid_mtu(struct l2cap_chan *chan, u16 mtu)
{
switch (chan->scid) {
case L2CAP_CID_ATT:
if (mtu < L2CAP_LE_MIN_MTU)
return false;
break;
default:
if (mtu < L2CAP_DEFAULT_MIN_MTU)
return false;
}
return true;
}
static int l2cap_sock_setsockopt_old(struct socket *sock, int optname,
char __user *optval, unsigned int optlen)
{
struct sock *sk = sock->sk;
struct l2cap_chan *chan = l2cap_pi(sk)->chan;
struct l2cap_options opts;
int len, err = 0;
u32 opt;
BT_DBG("sk %p", sk);
lock_sock(sk);
switch (optname) {
case L2CAP_OPTIONS:
if (bdaddr_type_is_le(chan->src_type)) {
err = -EINVAL;
break;
}
if (sk->sk_state == BT_CONNECTED) {
err = -EINVAL;
break;
}
opts.imtu = chan->imtu;
opts.omtu = chan->omtu;
opts.flush_to = chan->flush_to;
opts.mode = chan->mode;
opts.fcs = chan->fcs;
opts.max_tx = chan->max_tx;
opts.txwin_size = chan->tx_win;
len = min_t(unsigned int, sizeof(opts), optlen);
if (copy_from_user((char *) &opts, optval, len)) {
err = -EFAULT;
break;
}
if (opts.txwin_size > L2CAP_DEFAULT_EXT_WINDOW) {
err = -EINVAL;
break;
}
if (!l2cap_valid_mtu(chan, opts.imtu)) {
err = -EINVAL;
break;
}
chan->mode = opts.mode;
switch (chan->mode) {
case L2CAP_MODE_LE_FLOWCTL:
break;
case L2CAP_MODE_BASIC:
clear_bit(CONF_STATE2_DEVICE, &chan->conf_state);
break;
case L2CAP_MODE_ERTM:
case L2CAP_MODE_STREAMING:
if (!disable_ertm)
break;
/* fall through */
default:
err = -EINVAL;
break;
}
BT_DBG("mode 0x%2.2x", chan->mode);
chan->imtu = opts.imtu;
chan->omtu = opts.omtu;
chan->fcs = opts.fcs;
chan->max_tx = opts.max_tx;
chan->tx_win = opts.txwin_size;
chan->flush_to = opts.flush_to;
break;
case L2CAP_LM:
if (get_user(opt, (u32 __user *) optval)) {
err = -EFAULT;
break;
}
if (opt & L2CAP_LM_FIPS) {
err = -EINVAL;
break;
}
if (opt & L2CAP_LM_AUTH)
chan->sec_level = BT_SECURITY_LOW;
if (opt & L2CAP_LM_ENCRYPT)
chan->sec_level = BT_SECURITY_MEDIUM;
if (opt & L2CAP_LM_SECURE)
chan->sec_level = BT_SECURITY_HIGH;
if (opt & L2CAP_LM_MASTER)
set_bit(FLAG_ROLE_SWITCH, &chan->flags);
else
clear_bit(FLAG_ROLE_SWITCH, &chan->flags);
if (opt & L2CAP_LM_RELIABLE)
set_bit(FLAG_FORCE_RELIABLE, &chan->flags);
else
clear_bit(FLAG_FORCE_RELIABLE, &chan->flags);
break;
default:
err = -ENOPROTOOPT;
break;
}
release_sock(sk);
return err;
}
static int l2cap_sock_setsockopt(struct socket *sock, int level, int optname,
char __user *optval, unsigned int optlen)
{
struct sock *sk = sock->sk;
struct l2cap_chan *chan = l2cap_pi(sk)->chan;
struct bt_security sec;
struct bt_power pwr;
struct l2cap_conn *conn;
int len, err = 0;
u32 opt;
BT_DBG("sk %p", sk);
if (level == SOL_L2CAP)
return l2cap_sock_setsockopt_old(sock, optname, optval, optlen);
if (level != SOL_BLUETOOTH)
return -ENOPROTOOPT;
lock_sock(sk);
switch (optname) {
case BT_SECURITY:
if (chan->chan_type != L2CAP_CHAN_CONN_ORIENTED &&
chan->chan_type != L2CAP_CHAN_FIXED &&
chan->chan_type != L2CAP_CHAN_RAW) {
err = -EINVAL;
break;
}
sec.level = BT_SECURITY_LOW;
len = min_t(unsigned int, sizeof(sec), optlen);
if (copy_from_user((char *) &sec, optval, len)) {
err = -EFAULT;
break;
}
if (sec.level < BT_SECURITY_LOW ||
sec.level > BT_SECURITY_FIPS) {
err = -EINVAL;
break;
}
chan->sec_level = sec.level;
if (!chan->conn)
break;
conn = chan->conn;
/* change security for LE channels */
if (chan->scid == L2CAP_CID_ATT) {
if (smp_conn_security(conn->hcon, sec.level)) {
err = -EINVAL;
break;
}
set_bit(FLAG_PENDING_SECURITY, &chan->flags);
sk->sk_state = BT_CONFIG;
chan->state = BT_CONFIG;
/* or for ACL link */
} else if ((sk->sk_state == BT_CONNECT2 &&
test_bit(BT_SK_DEFER_SETUP, &bt_sk(sk)->flags)) ||
sk->sk_state == BT_CONNECTED) {
if (!l2cap_chan_check_security(chan, true))
set_bit(BT_SK_SUSPEND, &bt_sk(sk)->flags);
else
sk->sk_state_change(sk);
} else {
err = -EINVAL;
}
break;
case BT_DEFER_SETUP:
if (sk->sk_state != BT_BOUND && sk->sk_state != BT_LISTEN) {
err = -EINVAL;
break;
}
if (get_user(opt, (u32 __user *) optval)) {
err = -EFAULT;
break;
}
if (opt) {
set_bit(BT_SK_DEFER_SETUP, &bt_sk(sk)->flags);
set_bit(FLAG_DEFER_SETUP, &chan->flags);
} else {
clear_bit(BT_SK_DEFER_SETUP, &bt_sk(sk)->flags);
clear_bit(FLAG_DEFER_SETUP, &chan->flags);
}
break;
case BT_FLUSHABLE:
if (get_user(opt, (u32 __user *) optval)) {
err = -EFAULT;
break;
}
if (opt > BT_FLUSHABLE_ON) {
err = -EINVAL;
break;
}
if (opt == BT_FLUSHABLE_OFF) {
conn = chan->conn;
/* proceed further only when we have l2cap_conn and
No Flush support in the LM */
if (!conn || !lmp_no_flush_capable(conn->hcon->hdev)) {
err = -EINVAL;
break;
}
}
if (opt)
set_bit(FLAG_FLUSHABLE, &chan->flags);
else
clear_bit(FLAG_FLUSHABLE, &chan->flags);
break;
case BT_POWER:
if (chan->chan_type != L2CAP_CHAN_CONN_ORIENTED &&
chan->chan_type != L2CAP_CHAN_RAW) {
err = -EINVAL;
break;
}
pwr.force_active = BT_POWER_FORCE_ACTIVE_ON;
len = min_t(unsigned int, sizeof(pwr), optlen);
if (copy_from_user((char *) &pwr, optval, len)) {
err = -EFAULT;
break;
}
if (pwr.force_active)
set_bit(FLAG_FORCE_ACTIVE, &chan->flags);
else
clear_bit(FLAG_FORCE_ACTIVE, &chan->flags);
break;
case BT_CHANNEL_POLICY:
if (get_user(opt, (u32 __user *) optval)) {
err = -EFAULT;
break;
}
if (opt > BT_CHANNEL_POLICY_AMP_PREFERRED) {
err = -EINVAL;
break;
}
if (chan->mode != L2CAP_MODE_ERTM &&
chan->mode != L2CAP_MODE_STREAMING) {
err = -EOPNOTSUPP;
break;
}
chan->chan_policy = (u8) opt;
if (sk->sk_state == BT_CONNECTED &&
chan->move_role == L2CAP_MOVE_ROLE_NONE)
l2cap_move_start(chan);
break;
case BT_SNDMTU:
if (!bdaddr_type_is_le(chan->src_type)) {
err = -EINVAL;
break;
}
/* Setting is not supported as it's the remote side that
* decides this.
*/
err = -EPERM;
break;
case BT_RCVMTU:
if (!bdaddr_type_is_le(chan->src_type)) {
err = -EINVAL;
break;
}
if (chan->mode == L2CAP_MODE_LE_FLOWCTL &&
sk->sk_state == BT_CONNECTED) {
err = -EISCONN;
break;
}
if (get_user(opt, (u16 __user *) optval)) {
err = -EFAULT;
break;
}
if (chan->mode == L2CAP_MODE_EXT_FLOWCTL &&
sk->sk_state == BT_CONNECTED)
err = l2cap_chan_reconfigure(chan, opt);
else
chan->imtu = opt;
break;
default:
err = -ENOPROTOOPT;
break;
}
release_sock(sk);
return err;
}
static int l2cap_sock_sendmsg(struct socket *sock, struct msghdr *msg,
size_t len)
{
struct sock *sk = sock->sk;
struct l2cap_chan *chan = l2cap_pi(sk)->chan;
int err;
BT_DBG("sock %p, sk %p", sock, sk);
err = sock_error(sk);
if (err)
return err;
if (msg->msg_flags & MSG_OOB)
return -EOPNOTSUPP;
if (sk->sk_state != BT_CONNECTED)
return -ENOTCONN;
lock_sock(sk);
err = bt_sock_wait_ready(sk, msg->msg_flags);
release_sock(sk);
if (err)
return err;
l2cap_chan_lock(chan);
err = l2cap_chan_send(chan, msg, len);
l2cap_chan_unlock(chan);
return err;
}
static int l2cap_sock_recvmsg(struct socket *sock, struct msghdr *msg,
size_t len, int flags)
{
struct sock *sk = sock->sk;
struct l2cap_pinfo *pi = l2cap_pi(sk);
int err;
lock_sock(sk);
if (sk->sk_state == BT_CONNECT2 && test_bit(BT_SK_DEFER_SETUP,
&bt_sk(sk)->flags)) {
if (pi->chan->mode == L2CAP_MODE_EXT_FLOWCTL) {
sk->sk_state = BT_CONNECTED;
pi->chan->state = BT_CONNECTED;
__l2cap_ecred_conn_rsp_defer(pi->chan);
} if (bdaddr_type_is_le(pi->chan->src_type)) {
sk->sk_state = BT_CONNECTED;
pi->chan->state = BT_CONNECTED;
__l2cap_le_connect_rsp_defer(pi->chan);
} else {
sk->sk_state = BT_CONFIG;
pi->chan->state = BT_CONFIG;
__l2cap_connect_rsp_defer(pi->chan);
}
err = 0;
goto done;
}
release_sock(sk);
if (sock->type == SOCK_STREAM)
err = bt_sock_stream_recvmsg(sock, msg, len, flags);
else
err = bt_sock_recvmsg(sock, msg, len, flags);
if (pi->chan->mode != L2CAP_MODE_ERTM)
return err;
/* Attempt to put pending rx data in the socket buffer */
lock_sock(sk);
if (!test_bit(CONN_LOCAL_BUSY, &pi->chan->conn_state))
goto done;
if (pi->rx_busy_skb) {
Bluetooth: split sk_filter in l2cap_sock_recv_cb During an audit for sk_filter(), we found that rx_busy_skb handling in l2cap_sock_recv_cb() and l2cap_sock_recvmsg() looks not quite as intended. The assumption from commit e328140fdacb ("Bluetooth: Use event-driven approach for handling ERTM receive buffer") is that errors returned from sock_queue_rcv_skb() are due to receive buffer shortage. However, nothing should prevent doing a setsockopt() with SO_ATTACH_FILTER on the socket, that could drop some of the incoming skbs when handled in sock_queue_rcv_skb(). In that case sock_queue_rcv_skb() will return with -EPERM, propagated from sk_filter() and if in L2CAP_MODE_ERTM mode, wrong assumption was that we failed due to receive buffer being full. From that point onwards, due to the to-be-dropped skb being held in rx_busy_skb, we cannot make any forward progress as rx_busy_skb is never cleared from l2cap_sock_recvmsg(), due to the filter drop verdict over and over coming from sk_filter(). Meanwhile, in l2cap_sock_recv_cb() all new incoming skbs are being dropped due to rx_busy_skb being occupied. Instead, just use __sock_queue_rcv_skb() where an error really tells that there's a receive buffer issue. Split the sk_filter() and enable it for non-segmented modes at queuing time since at this point in time the skb has already been through the ERTM state machine and it has been acked, so dropping is not allowed. Instead, for ERTM and streaming mode, call sk_filter() in l2cap_data_rcv() so the packet can be dropped before the state machine sees it. Fixes: e328140fdacb ("Bluetooth: Use event-driven approach for handling ERTM receive buffer") Signed-off-by: Daniel Borkmann <daniel@iogearbox.net> Signed-off-by: Mat Martineau <mathew.j.martineau@linux.intel.com> Acked-by: Willem de Bruijn <willemb@google.com> Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
2016-07-28 01:40:14 +07:00
if (!__sock_queue_rcv_skb(sk, pi->rx_busy_skb))
pi->rx_busy_skb = NULL;
else
goto done;
}
/* Restore data flow when half of the receive buffer is
* available. This avoids resending large numbers of
* frames.
*/
if (atomic_read(&sk->sk_rmem_alloc) <= sk->sk_rcvbuf >> 1)
l2cap_chan_busy(pi->chan, 0);
done:
release_sock(sk);
return err;
}
/* Kill socket (only if zapped and orphan)
Bluetooth: Fix refcount use-after-free issue There is no lock preventing both l2cap_sock_release() and chan->ops->close() from running at the same time. If we consider Thread A running l2cap_chan_timeout() and Thread B running l2cap_sock_release(), expected behavior is: A::l2cap_chan_timeout()->l2cap_chan_close()->l2cap_sock_teardown_cb() A::l2cap_chan_timeout()->l2cap_sock_close_cb()->l2cap_sock_kill() B::l2cap_sock_release()->sock_orphan() B::l2cap_sock_release()->l2cap_sock_kill() where, sock_orphan() clears "sk->sk_socket" and l2cap_sock_teardown_cb() marks socket as SOCK_ZAPPED. In l2cap_sock_kill(), there is an "if-statement" that checks if both sock_orphan() and sock_teardown() has been run i.e. sk->sk_socket is NULL and socket is marked as SOCK_ZAPPED. Socket is killed if the condition is satisfied. In the race condition, following occurs: A::l2cap_chan_timeout()->l2cap_chan_close()->l2cap_sock_teardown_cb() B::l2cap_sock_release()->sock_orphan() B::l2cap_sock_release()->l2cap_sock_kill() A::l2cap_chan_timeout()->l2cap_sock_close_cb()->l2cap_sock_kill() In this scenario, "if-statement" is true in both B::l2cap_sock_kill() and A::l2cap_sock_kill() and we hit "refcount: underflow; use-after-free" bug. Similar condition occurs at other places where teardown/sock_kill is happening: l2cap_disconnect_rsp()->l2cap_chan_del()->l2cap_sock_teardown_cb() l2cap_disconnect_rsp()->l2cap_sock_close_cb()->l2cap_sock_kill() l2cap_conn_del()->l2cap_chan_del()->l2cap_sock_teardown_cb() l2cap_conn_del()->l2cap_sock_close_cb()->l2cap_sock_kill() l2cap_disconnect_req()->l2cap_chan_del()->l2cap_sock_teardown_cb() l2cap_disconnect_req()->l2cap_sock_close_cb()->l2cap_sock_kill() l2cap_sock_cleanup_listen()->l2cap_chan_close()->l2cap_sock_teardown_cb() l2cap_sock_cleanup_listen()->l2cap_sock_kill() Protect teardown/sock_kill and orphan/sock_kill by adding hold_lock on l2cap channel to ensure that the socket is killed only after marked as zapped and orphan. Signed-off-by: Manish Mandlik <mmandlik@google.com> Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
2020-01-29 01:54:14 +07:00
* Must be called on unlocked socket, with l2cap channel lock.
*/
static void l2cap_sock_kill(struct sock *sk)
{
if (!sock_flag(sk, SOCK_ZAPPED) || sk->sk_socket)
return;
BT_DBG("sk %p state %s", sk, state_to_string(sk->sk_state));
/* Kill poor orphan */
l2cap_chan_put(l2cap_pi(sk)->chan);
sock_set_flag(sk, SOCK_DEAD);
sock_put(sk);
}
static int __l2cap_wait_ack(struct sock *sk, struct l2cap_chan *chan)
{
DECLARE_WAITQUEUE(wait, current);
int err = 0;
int timeo = L2CAP_WAIT_ACK_POLL_PERIOD;
/* Timeout to prevent infinite loop */
unsigned long timeout = jiffies + L2CAP_WAIT_ACK_TIMEOUT;
add_wait_queue(sk_sleep(sk), &wait);
set_current_state(TASK_INTERRUPTIBLE);
do {
BT_DBG("Waiting for %d ACKs, timeout %04d ms",
chan->unacked_frames, time_after(jiffies, timeout) ? 0 :
jiffies_to_msecs(timeout - jiffies));
if (!timeo)
timeo = L2CAP_WAIT_ACK_POLL_PERIOD;
if (signal_pending(current)) {
err = sock_intr_errno(timeo);
break;
}
release_sock(sk);
timeo = schedule_timeout(timeo);
lock_sock(sk);
set_current_state(TASK_INTERRUPTIBLE);
err = sock_error(sk);
if (err)
break;
if (time_after(jiffies, timeout)) {
err = -ENOLINK;
break;
}
} while (chan->unacked_frames > 0 &&
chan->state == BT_CONNECTED);
set_current_state(TASK_RUNNING);
remove_wait_queue(sk_sleep(sk), &wait);
return err;
}
static int l2cap_sock_shutdown(struct socket *sock, int how)
{
struct sock *sk = sock->sk;
struct l2cap_chan *chan;
struct l2cap_conn *conn;
int err = 0;
BT_DBG("sock %p, sk %p", sock, sk);
if (!sk)
return 0;
lock_sock(sk);
if (sk->sk_shutdown)
goto shutdown_already;
BT_DBG("Handling sock shutdown");
/* prevent sk structure from being freed whilst unlocked */
sock_hold(sk);
chan = l2cap_pi(sk)->chan;
/* prevent chan structure from being freed whilst unlocked */
l2cap_chan_hold(chan);
BT_DBG("chan %p state %s", chan, state_to_string(chan->state));
if (chan->mode == L2CAP_MODE_ERTM &&
chan->unacked_frames > 0 &&
chan->state == BT_CONNECTED) {
err = __l2cap_wait_ack(sk, chan);
/* After waiting for ACKs, check whether shutdown
* has already been actioned to close the L2CAP
* link such as by l2cap_disconnection_req().
*/
if (sk->sk_shutdown)
goto has_shutdown;
}
sk->sk_shutdown = SHUTDOWN_MASK;
release_sock(sk);
l2cap_chan_lock(chan);
conn = chan->conn;
if (conn)
/* prevent conn structure from being freed */
l2cap_conn_get(conn);
l2cap_chan_unlock(chan);
if (conn)
/* mutex lock must be taken before l2cap_chan_lock() */
mutex_lock(&conn->chan_lock);
l2cap_chan_lock(chan);
l2cap_chan_close(chan, 0);
l2cap_chan_unlock(chan);
if (conn) {
mutex_unlock(&conn->chan_lock);
l2cap_conn_put(conn);
}
lock_sock(sk);
if (sock_flag(sk, SOCK_LINGER) && sk->sk_lingertime &&
!(current->flags & PF_EXITING))
err = bt_sock_wait_state(sk, BT_CLOSED,
sk->sk_lingertime);
has_shutdown:
l2cap_chan_put(chan);
sock_put(sk);
shutdown_already:
if (!err && sk->sk_err)
err = -sk->sk_err;
release_sock(sk);
BT_DBG("Sock shutdown complete err: %d", err);
return err;
}
static int l2cap_sock_release(struct socket *sock)
{
struct sock *sk = sock->sk;
int err;
struct l2cap_chan *chan;
BT_DBG("sock %p, sk %p", sock, sk);
if (!sk)
return 0;
bt_sock_unlink(&l2cap_sk_list, sk);
err = l2cap_sock_shutdown(sock, 2);
chan = l2cap_pi(sk)->chan;
l2cap_chan_hold(chan);
l2cap_chan_lock(chan);
Bluetooth: Fix refcount use-after-free issue There is no lock preventing both l2cap_sock_release() and chan->ops->close() from running at the same time. If we consider Thread A running l2cap_chan_timeout() and Thread B running l2cap_sock_release(), expected behavior is: A::l2cap_chan_timeout()->l2cap_chan_close()->l2cap_sock_teardown_cb() A::l2cap_chan_timeout()->l2cap_sock_close_cb()->l2cap_sock_kill() B::l2cap_sock_release()->sock_orphan() B::l2cap_sock_release()->l2cap_sock_kill() where, sock_orphan() clears "sk->sk_socket" and l2cap_sock_teardown_cb() marks socket as SOCK_ZAPPED. In l2cap_sock_kill(), there is an "if-statement" that checks if both sock_orphan() and sock_teardown() has been run i.e. sk->sk_socket is NULL and socket is marked as SOCK_ZAPPED. Socket is killed if the condition is satisfied. In the race condition, following occurs: A::l2cap_chan_timeout()->l2cap_chan_close()->l2cap_sock_teardown_cb() B::l2cap_sock_release()->sock_orphan() B::l2cap_sock_release()->l2cap_sock_kill() A::l2cap_chan_timeout()->l2cap_sock_close_cb()->l2cap_sock_kill() In this scenario, "if-statement" is true in both B::l2cap_sock_kill() and A::l2cap_sock_kill() and we hit "refcount: underflow; use-after-free" bug. Similar condition occurs at other places where teardown/sock_kill is happening: l2cap_disconnect_rsp()->l2cap_chan_del()->l2cap_sock_teardown_cb() l2cap_disconnect_rsp()->l2cap_sock_close_cb()->l2cap_sock_kill() l2cap_conn_del()->l2cap_chan_del()->l2cap_sock_teardown_cb() l2cap_conn_del()->l2cap_sock_close_cb()->l2cap_sock_kill() l2cap_disconnect_req()->l2cap_chan_del()->l2cap_sock_teardown_cb() l2cap_disconnect_req()->l2cap_sock_close_cb()->l2cap_sock_kill() l2cap_sock_cleanup_listen()->l2cap_chan_close()->l2cap_sock_teardown_cb() l2cap_sock_cleanup_listen()->l2cap_sock_kill() Protect teardown/sock_kill and orphan/sock_kill by adding hold_lock on l2cap channel to ensure that the socket is killed only after marked as zapped and orphan. Signed-off-by: Manish Mandlik <mmandlik@google.com> Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
2020-01-29 01:54:14 +07:00
sock_orphan(sk);
l2cap_sock_kill(sk);
Bluetooth: Fix refcount use-after-free issue There is no lock preventing both l2cap_sock_release() and chan->ops->close() from running at the same time. If we consider Thread A running l2cap_chan_timeout() and Thread B running l2cap_sock_release(), expected behavior is: A::l2cap_chan_timeout()->l2cap_chan_close()->l2cap_sock_teardown_cb() A::l2cap_chan_timeout()->l2cap_sock_close_cb()->l2cap_sock_kill() B::l2cap_sock_release()->sock_orphan() B::l2cap_sock_release()->l2cap_sock_kill() where, sock_orphan() clears "sk->sk_socket" and l2cap_sock_teardown_cb() marks socket as SOCK_ZAPPED. In l2cap_sock_kill(), there is an "if-statement" that checks if both sock_orphan() and sock_teardown() has been run i.e. sk->sk_socket is NULL and socket is marked as SOCK_ZAPPED. Socket is killed if the condition is satisfied. In the race condition, following occurs: A::l2cap_chan_timeout()->l2cap_chan_close()->l2cap_sock_teardown_cb() B::l2cap_sock_release()->sock_orphan() B::l2cap_sock_release()->l2cap_sock_kill() A::l2cap_chan_timeout()->l2cap_sock_close_cb()->l2cap_sock_kill() In this scenario, "if-statement" is true in both B::l2cap_sock_kill() and A::l2cap_sock_kill() and we hit "refcount: underflow; use-after-free" bug. Similar condition occurs at other places where teardown/sock_kill is happening: l2cap_disconnect_rsp()->l2cap_chan_del()->l2cap_sock_teardown_cb() l2cap_disconnect_rsp()->l2cap_sock_close_cb()->l2cap_sock_kill() l2cap_conn_del()->l2cap_chan_del()->l2cap_sock_teardown_cb() l2cap_conn_del()->l2cap_sock_close_cb()->l2cap_sock_kill() l2cap_disconnect_req()->l2cap_chan_del()->l2cap_sock_teardown_cb() l2cap_disconnect_req()->l2cap_sock_close_cb()->l2cap_sock_kill() l2cap_sock_cleanup_listen()->l2cap_chan_close()->l2cap_sock_teardown_cb() l2cap_sock_cleanup_listen()->l2cap_sock_kill() Protect teardown/sock_kill and orphan/sock_kill by adding hold_lock on l2cap channel to ensure that the socket is killed only after marked as zapped and orphan. Signed-off-by: Manish Mandlik <mmandlik@google.com> Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
2020-01-29 01:54:14 +07:00
l2cap_chan_unlock(chan);
l2cap_chan_put(chan);
Bluetooth: Fix refcount use-after-free issue There is no lock preventing both l2cap_sock_release() and chan->ops->close() from running at the same time. If we consider Thread A running l2cap_chan_timeout() and Thread B running l2cap_sock_release(), expected behavior is: A::l2cap_chan_timeout()->l2cap_chan_close()->l2cap_sock_teardown_cb() A::l2cap_chan_timeout()->l2cap_sock_close_cb()->l2cap_sock_kill() B::l2cap_sock_release()->sock_orphan() B::l2cap_sock_release()->l2cap_sock_kill() where, sock_orphan() clears "sk->sk_socket" and l2cap_sock_teardown_cb() marks socket as SOCK_ZAPPED. In l2cap_sock_kill(), there is an "if-statement" that checks if both sock_orphan() and sock_teardown() has been run i.e. sk->sk_socket is NULL and socket is marked as SOCK_ZAPPED. Socket is killed if the condition is satisfied. In the race condition, following occurs: A::l2cap_chan_timeout()->l2cap_chan_close()->l2cap_sock_teardown_cb() B::l2cap_sock_release()->sock_orphan() B::l2cap_sock_release()->l2cap_sock_kill() A::l2cap_chan_timeout()->l2cap_sock_close_cb()->l2cap_sock_kill() In this scenario, "if-statement" is true in both B::l2cap_sock_kill() and A::l2cap_sock_kill() and we hit "refcount: underflow; use-after-free" bug. Similar condition occurs at other places where teardown/sock_kill is happening: l2cap_disconnect_rsp()->l2cap_chan_del()->l2cap_sock_teardown_cb() l2cap_disconnect_rsp()->l2cap_sock_close_cb()->l2cap_sock_kill() l2cap_conn_del()->l2cap_chan_del()->l2cap_sock_teardown_cb() l2cap_conn_del()->l2cap_sock_close_cb()->l2cap_sock_kill() l2cap_disconnect_req()->l2cap_chan_del()->l2cap_sock_teardown_cb() l2cap_disconnect_req()->l2cap_sock_close_cb()->l2cap_sock_kill() l2cap_sock_cleanup_listen()->l2cap_chan_close()->l2cap_sock_teardown_cb() l2cap_sock_cleanup_listen()->l2cap_sock_kill() Protect teardown/sock_kill and orphan/sock_kill by adding hold_lock on l2cap channel to ensure that the socket is killed only after marked as zapped and orphan. Signed-off-by: Manish Mandlik <mmandlik@google.com> Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
2020-01-29 01:54:14 +07:00
return err;
}
static void l2cap_sock_cleanup_listen(struct sock *parent)
{
struct sock *sk;
BT_DBG("parent %p state %s", parent,
state_to_string(parent->sk_state));
/* Close not yet accepted channels */
while ((sk = bt_accept_dequeue(parent, NULL))) {
struct l2cap_chan *chan = l2cap_pi(sk)->chan;
BT_DBG("child chan %p state %s", chan,
state_to_string(chan->state));
Bluetooth: Fix refcount use-after-free issue There is no lock preventing both l2cap_sock_release() and chan->ops->close() from running at the same time. If we consider Thread A running l2cap_chan_timeout() and Thread B running l2cap_sock_release(), expected behavior is: A::l2cap_chan_timeout()->l2cap_chan_close()->l2cap_sock_teardown_cb() A::l2cap_chan_timeout()->l2cap_sock_close_cb()->l2cap_sock_kill() B::l2cap_sock_release()->sock_orphan() B::l2cap_sock_release()->l2cap_sock_kill() where, sock_orphan() clears "sk->sk_socket" and l2cap_sock_teardown_cb() marks socket as SOCK_ZAPPED. In l2cap_sock_kill(), there is an "if-statement" that checks if both sock_orphan() and sock_teardown() has been run i.e. sk->sk_socket is NULL and socket is marked as SOCK_ZAPPED. Socket is killed if the condition is satisfied. In the race condition, following occurs: A::l2cap_chan_timeout()->l2cap_chan_close()->l2cap_sock_teardown_cb() B::l2cap_sock_release()->sock_orphan() B::l2cap_sock_release()->l2cap_sock_kill() A::l2cap_chan_timeout()->l2cap_sock_close_cb()->l2cap_sock_kill() In this scenario, "if-statement" is true in both B::l2cap_sock_kill() and A::l2cap_sock_kill() and we hit "refcount: underflow; use-after-free" bug. Similar condition occurs at other places where teardown/sock_kill is happening: l2cap_disconnect_rsp()->l2cap_chan_del()->l2cap_sock_teardown_cb() l2cap_disconnect_rsp()->l2cap_sock_close_cb()->l2cap_sock_kill() l2cap_conn_del()->l2cap_chan_del()->l2cap_sock_teardown_cb() l2cap_conn_del()->l2cap_sock_close_cb()->l2cap_sock_kill() l2cap_disconnect_req()->l2cap_chan_del()->l2cap_sock_teardown_cb() l2cap_disconnect_req()->l2cap_sock_close_cb()->l2cap_sock_kill() l2cap_sock_cleanup_listen()->l2cap_chan_close()->l2cap_sock_teardown_cb() l2cap_sock_cleanup_listen()->l2cap_sock_kill() Protect teardown/sock_kill and orphan/sock_kill by adding hold_lock on l2cap channel to ensure that the socket is killed only after marked as zapped and orphan. Signed-off-by: Manish Mandlik <mmandlik@google.com> Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
2020-01-29 01:54:14 +07:00
l2cap_chan_hold(chan);
l2cap_chan_lock(chan);
Bluetooth: Fix refcount use-after-free issue There is no lock preventing both l2cap_sock_release() and chan->ops->close() from running at the same time. If we consider Thread A running l2cap_chan_timeout() and Thread B running l2cap_sock_release(), expected behavior is: A::l2cap_chan_timeout()->l2cap_chan_close()->l2cap_sock_teardown_cb() A::l2cap_chan_timeout()->l2cap_sock_close_cb()->l2cap_sock_kill() B::l2cap_sock_release()->sock_orphan() B::l2cap_sock_release()->l2cap_sock_kill() where, sock_orphan() clears "sk->sk_socket" and l2cap_sock_teardown_cb() marks socket as SOCK_ZAPPED. In l2cap_sock_kill(), there is an "if-statement" that checks if both sock_orphan() and sock_teardown() has been run i.e. sk->sk_socket is NULL and socket is marked as SOCK_ZAPPED. Socket is killed if the condition is satisfied. In the race condition, following occurs: A::l2cap_chan_timeout()->l2cap_chan_close()->l2cap_sock_teardown_cb() B::l2cap_sock_release()->sock_orphan() B::l2cap_sock_release()->l2cap_sock_kill() A::l2cap_chan_timeout()->l2cap_sock_close_cb()->l2cap_sock_kill() In this scenario, "if-statement" is true in both B::l2cap_sock_kill() and A::l2cap_sock_kill() and we hit "refcount: underflow; use-after-free" bug. Similar condition occurs at other places where teardown/sock_kill is happening: l2cap_disconnect_rsp()->l2cap_chan_del()->l2cap_sock_teardown_cb() l2cap_disconnect_rsp()->l2cap_sock_close_cb()->l2cap_sock_kill() l2cap_conn_del()->l2cap_chan_del()->l2cap_sock_teardown_cb() l2cap_conn_del()->l2cap_sock_close_cb()->l2cap_sock_kill() l2cap_disconnect_req()->l2cap_chan_del()->l2cap_sock_teardown_cb() l2cap_disconnect_req()->l2cap_sock_close_cb()->l2cap_sock_kill() l2cap_sock_cleanup_listen()->l2cap_chan_close()->l2cap_sock_teardown_cb() l2cap_sock_cleanup_listen()->l2cap_sock_kill() Protect teardown/sock_kill and orphan/sock_kill by adding hold_lock on l2cap channel to ensure that the socket is killed only after marked as zapped and orphan. Signed-off-by: Manish Mandlik <mmandlik@google.com> Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
2020-01-29 01:54:14 +07:00
__clear_chan_timer(chan);
l2cap_chan_close(chan, ECONNRESET);
l2cap_sock_kill(sk);
Bluetooth: Fix refcount use-after-free issue There is no lock preventing both l2cap_sock_release() and chan->ops->close() from running at the same time. If we consider Thread A running l2cap_chan_timeout() and Thread B running l2cap_sock_release(), expected behavior is: A::l2cap_chan_timeout()->l2cap_chan_close()->l2cap_sock_teardown_cb() A::l2cap_chan_timeout()->l2cap_sock_close_cb()->l2cap_sock_kill() B::l2cap_sock_release()->sock_orphan() B::l2cap_sock_release()->l2cap_sock_kill() where, sock_orphan() clears "sk->sk_socket" and l2cap_sock_teardown_cb() marks socket as SOCK_ZAPPED. In l2cap_sock_kill(), there is an "if-statement" that checks if both sock_orphan() and sock_teardown() has been run i.e. sk->sk_socket is NULL and socket is marked as SOCK_ZAPPED. Socket is killed if the condition is satisfied. In the race condition, following occurs: A::l2cap_chan_timeout()->l2cap_chan_close()->l2cap_sock_teardown_cb() B::l2cap_sock_release()->sock_orphan() B::l2cap_sock_release()->l2cap_sock_kill() A::l2cap_chan_timeout()->l2cap_sock_close_cb()->l2cap_sock_kill() In this scenario, "if-statement" is true in both B::l2cap_sock_kill() and A::l2cap_sock_kill() and we hit "refcount: underflow; use-after-free" bug. Similar condition occurs at other places where teardown/sock_kill is happening: l2cap_disconnect_rsp()->l2cap_chan_del()->l2cap_sock_teardown_cb() l2cap_disconnect_rsp()->l2cap_sock_close_cb()->l2cap_sock_kill() l2cap_conn_del()->l2cap_chan_del()->l2cap_sock_teardown_cb() l2cap_conn_del()->l2cap_sock_close_cb()->l2cap_sock_kill() l2cap_disconnect_req()->l2cap_chan_del()->l2cap_sock_teardown_cb() l2cap_disconnect_req()->l2cap_sock_close_cb()->l2cap_sock_kill() l2cap_sock_cleanup_listen()->l2cap_chan_close()->l2cap_sock_teardown_cb() l2cap_sock_cleanup_listen()->l2cap_sock_kill() Protect teardown/sock_kill and orphan/sock_kill by adding hold_lock on l2cap channel to ensure that the socket is killed only after marked as zapped and orphan. Signed-off-by: Manish Mandlik <mmandlik@google.com> Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
2020-01-29 01:54:14 +07:00
l2cap_chan_unlock(chan);
l2cap_chan_put(chan);
}
}
static struct l2cap_chan *l2cap_sock_new_connection_cb(struct l2cap_chan *chan)
{
struct sock *sk, *parent = chan->data;
lock_sock(parent);
/* Check for backlog size */
if (sk_acceptq_is_full(parent)) {
BT_DBG("backlog full %d", parent->sk_ack_backlog);
Bluetooth: Fix L2CAP deadlock -[0x01 Introduction We have found a programming error causing a deadlock in Bluetooth subsystem of Linux kernel. The problem is caused by missing release_sock() call when L2CAP connection creation fails due full accept queue. The issue can be reproduced with 3.15-rc5 kernel and is also present in earlier kernels. -[0x02 Details The problem occurs when multiple L2CAP connections are created to a PSM which contains listening socket (like SDP) and left pending, for example, configuration (the underlying ACL link is not disconnected between connections). When L2CAP connection request is received and listening socket is found the l2cap_sock_new_connection_cb() function (net/bluetooth/l2cap_sock.c) is called. This function locks the 'parent' socket and then checks if the accept queue is full. 1178 lock_sock(parent); 1179 1180 /* Check for backlog size */ 1181 if (sk_acceptq_is_full(parent)) { 1182 BT_DBG("backlog full %d", parent->sk_ack_backlog); 1183 return NULL; 1184 } If case the accept queue is full NULL is returned, but the 'parent' socket is not released. Thus when next L2CAP connection request is received the code blocks on lock_sock() since the parent is still locked. Also note that for connections already established and waiting for configuration to complete a timeout will occur and l2cap_chan_timeout() (net/bluetooth/l2cap_core.c) will be called. All threads calling this function will also be blocked waiting for the channel mutex since the thread which is waiting on lock_sock() alread holds the channel mutex. We were able to reproduce this by sending continuously L2CAP connection request followed by disconnection request containing invalid CID. This left the created connections pending configuration. After the deadlock occurs it is impossible to kill bluetoothd, btmon will not get any more data etc. requiring reboot to recover. -[0x03 Fix Releasing the 'parent' socket when l2cap_sock_new_connection_cb() returns NULL seems to fix the issue. Signed-off-by: Jukka Taimisto <jtt@codenomicon.com> Reported-by: Tommi Mäkilä <tmakila@codenomicon.com> Signed-off-by: Johan Hedberg <johan.hedberg@intel.com> Cc: stable@vger.kernel.org
2014-05-22 17:02:39 +07:00
release_sock(parent);
return NULL;
}
sk = l2cap_sock_alloc(sock_net(parent), NULL, BTPROTO_L2CAP,
GFP_ATOMIC, 0);
Bluetooth: Fix L2CAP deadlock -[0x01 Introduction We have found a programming error causing a deadlock in Bluetooth subsystem of Linux kernel. The problem is caused by missing release_sock() call when L2CAP connection creation fails due full accept queue. The issue can be reproduced with 3.15-rc5 kernel and is also present in earlier kernels. -[0x02 Details The problem occurs when multiple L2CAP connections are created to a PSM which contains listening socket (like SDP) and left pending, for example, configuration (the underlying ACL link is not disconnected between connections). When L2CAP connection request is received and listening socket is found the l2cap_sock_new_connection_cb() function (net/bluetooth/l2cap_sock.c) is called. This function locks the 'parent' socket and then checks if the accept queue is full. 1178 lock_sock(parent); 1179 1180 /* Check for backlog size */ 1181 if (sk_acceptq_is_full(parent)) { 1182 BT_DBG("backlog full %d", parent->sk_ack_backlog); 1183 return NULL; 1184 } If case the accept queue is full NULL is returned, but the 'parent' socket is not released. Thus when next L2CAP connection request is received the code blocks on lock_sock() since the parent is still locked. Also note that for connections already established and waiting for configuration to complete a timeout will occur and l2cap_chan_timeout() (net/bluetooth/l2cap_core.c) will be called. All threads calling this function will also be blocked waiting for the channel mutex since the thread which is waiting on lock_sock() alread holds the channel mutex. We were able to reproduce this by sending continuously L2CAP connection request followed by disconnection request containing invalid CID. This left the created connections pending configuration. After the deadlock occurs it is impossible to kill bluetoothd, btmon will not get any more data etc. requiring reboot to recover. -[0x03 Fix Releasing the 'parent' socket when l2cap_sock_new_connection_cb() returns NULL seems to fix the issue. Signed-off-by: Jukka Taimisto <jtt@codenomicon.com> Reported-by: Tommi Mäkilä <tmakila@codenomicon.com> Signed-off-by: Johan Hedberg <johan.hedberg@intel.com> Cc: stable@vger.kernel.org
2014-05-22 17:02:39 +07:00
if (!sk) {
release_sock(parent);
return NULL;
Bluetooth: Fix L2CAP deadlock -[0x01 Introduction We have found a programming error causing a deadlock in Bluetooth subsystem of Linux kernel. The problem is caused by missing release_sock() call when L2CAP connection creation fails due full accept queue. The issue can be reproduced with 3.15-rc5 kernel and is also present in earlier kernels. -[0x02 Details The problem occurs when multiple L2CAP connections are created to a PSM which contains listening socket (like SDP) and left pending, for example, configuration (the underlying ACL link is not disconnected between connections). When L2CAP connection request is received and listening socket is found the l2cap_sock_new_connection_cb() function (net/bluetooth/l2cap_sock.c) is called. This function locks the 'parent' socket and then checks if the accept queue is full. 1178 lock_sock(parent); 1179 1180 /* Check for backlog size */ 1181 if (sk_acceptq_is_full(parent)) { 1182 BT_DBG("backlog full %d", parent->sk_ack_backlog); 1183 return NULL; 1184 } If case the accept queue is full NULL is returned, but the 'parent' socket is not released. Thus when next L2CAP connection request is received the code blocks on lock_sock() since the parent is still locked. Also note that for connections already established and waiting for configuration to complete a timeout will occur and l2cap_chan_timeout() (net/bluetooth/l2cap_core.c) will be called. All threads calling this function will also be blocked waiting for the channel mutex since the thread which is waiting on lock_sock() alread holds the channel mutex. We were able to reproduce this by sending continuously L2CAP connection request followed by disconnection request containing invalid CID. This left the created connections pending configuration. After the deadlock occurs it is impossible to kill bluetoothd, btmon will not get any more data etc. requiring reboot to recover. -[0x03 Fix Releasing the 'parent' socket when l2cap_sock_new_connection_cb() returns NULL seems to fix the issue. Signed-off-by: Jukka Taimisto <jtt@codenomicon.com> Reported-by: Tommi Mäkilä <tmakila@codenomicon.com> Signed-off-by: Johan Hedberg <johan.hedberg@intel.com> Cc: stable@vger.kernel.org
2014-05-22 17:02:39 +07:00
}
Bluetooth: silence lockdep warning Since bluetooth uses multiple protocols types, to avoid lockdep warnings, we need to use different lockdep classes (one for each protocol type). This is already done in bt_sock_create but it misses a couple of cases when new connections are created. This patch corrects that to fix the following warning: <4>[ 1864.732366] ======================================================= <4>[ 1864.733030] [ INFO: possible circular locking dependency detected ] <4>[ 1864.733544] 3.0.16-mid3-00007-gc9a0f62 #3 <4>[ 1864.733883] ------------------------------------------------------- <4>[ 1864.734408] t.android.btclc/4204 is trying to acquire lock: <4>[ 1864.734869] (rfcomm_mutex){+.+.+.}, at: [<c14970ea>] rfcomm_dlc_close+0x15/0x30 <4>[ 1864.735541] <4>[ 1864.735549] but task is already holding lock: <4>[ 1864.736045] (sk_lock-AF_BLUETOOTH){+.+.+.}, at: [<c1498bf7>] lock_sock+0xa/0xc <4>[ 1864.736732] <4>[ 1864.736740] which lock already depends on the new lock. <4>[ 1864.736750] <4>[ 1864.737428] <4>[ 1864.737437] the existing dependency chain (in reverse order) is: <4>[ 1864.738016] <4>[ 1864.738023] -> #1 (sk_lock-AF_BLUETOOTH){+.+.+.}: <4>[ 1864.738549] [<c1062273>] lock_acquire+0x104/0x140 <4>[ 1864.738977] [<c13d35c1>] lock_sock_nested+0x58/0x68 <4>[ 1864.739411] [<c1493c33>] l2cap_sock_sendmsg+0x3e/0x76 <4>[ 1864.739858] [<c13d06c3>] __sock_sendmsg+0x50/0x59 <4>[ 1864.740279] [<c13d0ea2>] sock_sendmsg+0x94/0xa8 <4>[ 1864.740687] [<c13d0ede>] kernel_sendmsg+0x28/0x37 <4>[ 1864.741106] [<c14969ca>] rfcomm_send_frame+0x30/0x38 <4>[ 1864.741542] [<c1496a2a>] rfcomm_send_ua+0x58/0x5a <4>[ 1864.741959] [<c1498447>] rfcomm_run+0x441/0xb52 <4>[ 1864.742365] [<c104f095>] kthread+0x63/0x68 <4>[ 1864.742742] [<c14d5182>] kernel_thread_helper+0x6/0xd <4>[ 1864.743187] <4>[ 1864.743193] -> #0 (rfcomm_mutex){+.+.+.}: <4>[ 1864.743667] [<c1061ada>] __lock_acquire+0x988/0xc00 <4>[ 1864.744100] [<c1062273>] lock_acquire+0x104/0x140 <4>[ 1864.744519] [<c14d2c70>] __mutex_lock_common+0x3b/0x33f <4>[ 1864.744975] [<c14d303e>] mutex_lock_nested+0x2d/0x36 <4>[ 1864.745412] [<c14970ea>] rfcomm_dlc_close+0x15/0x30 <4>[ 1864.745842] [<c14990d9>] __rfcomm_sock_close+0x5f/0x6b <4>[ 1864.746288] [<c1499114>] rfcomm_sock_shutdown+0x2f/0x62 <4>[ 1864.746737] [<c13d275d>] sys_socketcall+0x1db/0x422 <4>[ 1864.747165] [<c14d42f0>] syscall_call+0x7/0xb Signed-off-by: Octavian Purdila <octavian.purdila@intel.com> Acked-by: Marcel Holtmann <marcel@holtmann.org> Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
2012-01-22 05:28:34 +07:00
bt_sock_reclassify_lock(sk, BTPROTO_L2CAP);
l2cap_sock_init(sk, parent);
Bluetooth: Fix locking in bt_accept_enqueue() for BH context With commit e16337622016 ("Bluetooth: Handle bt_accept_enqueue() socket atomically") lock_sock[_nested]() is used to acquire the socket lock before manipulating the socket. lock_sock[_nested]() may block, which is problematic since bt_accept_enqueue() can be called in bottom half context (e.g. from rfcomm_connect_ind()): [<ffffff80080d81ec>] __might_sleep+0x4c/0x80 [<ffffff800876c7b0>] lock_sock_nested+0x24/0x58 [<ffffff8000d7c27c>] bt_accept_enqueue+0x48/0xd4 [bluetooth] [<ffffff8000e67d8c>] rfcomm_connect_ind+0x190/0x218 [rfcomm] Add a parameter to bt_accept_enqueue() to indicate whether the function is called from BH context, and acquire the socket lock with bh_lock_sock_nested() if that's the case. Also adapt all callers of bt_accept_enqueue() to pass the new parameter: - l2cap_sock_new_connection_cb() - uses lock_sock() to lock the parent socket => process context - rfcomm_connect_ind() - acquires the parent socket lock with bh_lock_sock() => BH context - __sco_chan_add() - called from sco_chan_add(), which is called from sco_connect(). parent is NULL, hence bt_accept_enqueue() isn't called in this code path and we can ignore it - also called from sco_conn_ready(). uses bh_lock_sock() to acquire the parent lock => BH context Fixes: e16337622016 ("Bluetooth: Handle bt_accept_enqueue() socket atomically") Signed-off-by: Matthias Kaehlcke <mka@chromium.org> Reviewed-by: Douglas Anderson <dianders@chromium.org> Signed-off-by: Marcel Holtmann <marcel@holtmann.org> Cc: stable@vger.kernel.org
2019-01-03 07:11:20 +07:00
bt_accept_enqueue(parent, sk, false);
release_sock(parent);
return l2cap_pi(sk)->chan;
}
static int l2cap_sock_recv_cb(struct l2cap_chan *chan, struct sk_buff *skb)
{
struct sock *sk = chan->data;
int err;
lock_sock(sk);
if (l2cap_pi(sk)->rx_busy_skb) {
err = -ENOMEM;
goto done;
}
Bluetooth: split sk_filter in l2cap_sock_recv_cb During an audit for sk_filter(), we found that rx_busy_skb handling in l2cap_sock_recv_cb() and l2cap_sock_recvmsg() looks not quite as intended. The assumption from commit e328140fdacb ("Bluetooth: Use event-driven approach for handling ERTM receive buffer") is that errors returned from sock_queue_rcv_skb() are due to receive buffer shortage. However, nothing should prevent doing a setsockopt() with SO_ATTACH_FILTER on the socket, that could drop some of the incoming skbs when handled in sock_queue_rcv_skb(). In that case sock_queue_rcv_skb() will return with -EPERM, propagated from sk_filter() and if in L2CAP_MODE_ERTM mode, wrong assumption was that we failed due to receive buffer being full. From that point onwards, due to the to-be-dropped skb being held in rx_busy_skb, we cannot make any forward progress as rx_busy_skb is never cleared from l2cap_sock_recvmsg(), due to the filter drop verdict over and over coming from sk_filter(). Meanwhile, in l2cap_sock_recv_cb() all new incoming skbs are being dropped due to rx_busy_skb being occupied. Instead, just use __sock_queue_rcv_skb() where an error really tells that there's a receive buffer issue. Split the sk_filter() and enable it for non-segmented modes at queuing time since at this point in time the skb has already been through the ERTM state machine and it has been acked, so dropping is not allowed. Instead, for ERTM and streaming mode, call sk_filter() in l2cap_data_rcv() so the packet can be dropped before the state machine sees it. Fixes: e328140fdacb ("Bluetooth: Use event-driven approach for handling ERTM receive buffer") Signed-off-by: Daniel Borkmann <daniel@iogearbox.net> Signed-off-by: Mat Martineau <mathew.j.martineau@linux.intel.com> Acked-by: Willem de Bruijn <willemb@google.com> Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
2016-07-28 01:40:14 +07:00
if (chan->mode != L2CAP_MODE_ERTM &&
chan->mode != L2CAP_MODE_STREAMING) {
/* Even if no filter is attached, we could potentially
* get errors from security modules, etc.
*/
err = sk_filter(sk, skb);
if (err)
goto done;
}
err = __sock_queue_rcv_skb(sk, skb);
/* For ERTM, handle one skb that doesn't fit into the recv
* buffer. This is important to do because the data frames
* have already been acked, so the skb cannot be discarded.
*
* Notify the l2cap core that the buffer is full, so the
* LOCAL_BUSY state is entered and no more frames are
* acked and reassembled until there is buffer space
* available.
*/
if (err < 0 && chan->mode == L2CAP_MODE_ERTM) {
l2cap_pi(sk)->rx_busy_skb = skb;
l2cap_chan_busy(chan, 1);
err = 0;
}
done:
release_sock(sk);
return err;
}
static void l2cap_sock_close_cb(struct l2cap_chan *chan)
{
struct sock *sk = chan->data;
l2cap_sock_kill(sk);
}
static void l2cap_sock_teardown_cb(struct l2cap_chan *chan, int err)
{
struct sock *sk = chan->data;
struct sock *parent;
BT_DBG("chan %p state %s", chan, state_to_string(chan->state));
/* This callback can be called both for server (BT_LISTEN)
* sockets as well as "normal" ones. To avoid lockdep warnings
* with child socket locking (through l2cap_sock_cleanup_listen)
* we need separation into separate nesting levels. The simplest
* way to accomplish this is to inherit the nesting level used
* for the channel.
*/
lock_sock_nested(sk, atomic_read(&chan->nesting));
parent = bt_sk(sk)->parent;
sock_set_flag(sk, SOCK_ZAPPED);
switch (chan->state) {
case BT_OPEN:
case BT_BOUND:
case BT_CLOSED:
break;
case BT_LISTEN:
l2cap_sock_cleanup_listen(sk);
sk->sk_state = BT_CLOSED;
chan->state = BT_CLOSED;
break;
default:
sk->sk_state = BT_CLOSED;
chan->state = BT_CLOSED;
sk->sk_err = err;
if (parent) {
bt_accept_unlink(sk);
parent->sk_data_ready(parent);
} else {
sk->sk_state_change(sk);
}
break;
}
release_sock(sk);
}
static void l2cap_sock_state_change_cb(struct l2cap_chan *chan, int state,
int err)
{
struct sock *sk = chan->data;
sk->sk_state = state;
if (err)
sk->sk_err = err;
}
static struct sk_buff *l2cap_sock_alloc_skb_cb(struct l2cap_chan *chan,
unsigned long hdr_len,
unsigned long len, int nb)
{
struct sock *sk = chan->data;
struct sk_buff *skb;
int err;
l2cap_chan_unlock(chan);
skb = bt_skb_send_alloc(sk, hdr_len + len, nb, &err);
l2cap_chan_lock(chan);
if (!skb)
return ERR_PTR(err);
skb->priority = sk->sk_priority;
bt_cb(skb)->l2cap.chan = chan;
return skb;
}
static void l2cap_sock_ready_cb(struct l2cap_chan *chan)
{
struct sock *sk = chan->data;
struct sock *parent;
lock_sock(sk);
parent = bt_sk(sk)->parent;
BT_DBG("sk %p, parent %p", sk, parent);
sk->sk_state = BT_CONNECTED;
sk->sk_state_change(sk);
if (parent)
parent->sk_data_ready(parent);
release_sock(sk);
}
static void l2cap_sock_defer_cb(struct l2cap_chan *chan)
{
struct sock *parent, *sk = chan->data;
lock_sock(sk);
parent = bt_sk(sk)->parent;
if (parent)
parent->sk_data_ready(parent);
release_sock(sk);
}
static void l2cap_sock_resume_cb(struct l2cap_chan *chan)
{
struct sock *sk = chan->data;
if (test_and_clear_bit(FLAG_PENDING_SECURITY, &chan->flags)) {
sk->sk_state = BT_CONNECTED;
chan->state = BT_CONNECTED;
}
clear_bit(BT_SK_SUSPEND, &bt_sk(sk)->flags);
sk->sk_state_change(sk);
}
static void l2cap_sock_set_shutdown_cb(struct l2cap_chan *chan)
{
struct sock *sk = chan->data;
lock_sock(sk);
sk->sk_shutdown = SHUTDOWN_MASK;
release_sock(sk);
}
static long l2cap_sock_get_sndtimeo_cb(struct l2cap_chan *chan)
{
struct sock *sk = chan->data;
return sk->sk_sndtimeo;
}
static void l2cap_sock_suspend_cb(struct l2cap_chan *chan)
{
struct sock *sk = chan->data;
set_bit(BT_SK_SUSPEND, &bt_sk(sk)->flags);
sk->sk_state_change(sk);
}
static const struct l2cap_ops l2cap_chan_ops = {
.name = "L2CAP Socket Interface",
.new_connection = l2cap_sock_new_connection_cb,
.recv = l2cap_sock_recv_cb,
.close = l2cap_sock_close_cb,
.teardown = l2cap_sock_teardown_cb,
.state_change = l2cap_sock_state_change_cb,
.ready = l2cap_sock_ready_cb,
.defer = l2cap_sock_defer_cb,
.resume = l2cap_sock_resume_cb,
.suspend = l2cap_sock_suspend_cb,
.set_shutdown = l2cap_sock_set_shutdown_cb,
.get_sndtimeo = l2cap_sock_get_sndtimeo_cb,
.alloc_skb = l2cap_sock_alloc_skb_cb,
};
static void l2cap_sock_destruct(struct sock *sk)
{
BT_DBG("sk %p", sk);
Bluetooth: don't attempt to free a channel that wasn't created We may currently attempt to free a channel which wasn't created due to an error in the initialization path, this would cause a NULL ptr deref. This would cause the following oops: [ 12.919073] BUG: unable to handle kernel NULL pointer dereference at 0000000000000010 [ 12.919131] IP: [<ffffffff836645c4>] l2cap_chan_put+0x34/0x50 [ 12.919135] PGD 0 [ 12.919138] Oops: 0002 [#1] PREEMPT SMP DEBUG_PAGEALLOC [ 12.919193] Dumping ftrace buffer: [ 12.919242] (ftrace buffer empty) [ 12.919314] Modules linked in: [ 12.919318] CPU 1 [ 12.919319] Pid: 6210, comm: krfcommd Tainted: G W 3.6.0-next-20121004-sasha-00005-gb010653-dirty #30 [ 12.919374] RIP: 0010:[<ffffffff836645c4>] [<ffffffff836645c4>] l2cap_chan_put+0x34/0x50 [ 12.919377] RSP: 0000:ffff880066933c38 EFLAGS: 00010246 [ 12.919378] RAX: ffffffff8366c780 RBX: 0000000000000000 RCX: 6666666666666667 [ 12.919379] RDX: 0000000000000fa0 RSI: ffffffff84d3f79e RDI: 0000000000000010 [ 12.919381] RBP: ffff880066933c48 R08: ffffffff859989f8 R09: 0000000000000001 [ 12.919382] R10: 0000000000000000 R11: 7fffffffffffffff R12: 0000000000000000 [ 12.919383] R13: ffff88009b00a200 R14: ffff88009b00a200 R15: 0000000000000001 [ 12.919385] FS: 0000000000000000(0000) GS:ffff880033600000(0000) knlGS:0000000000000000 [ 12.919437] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 12.919440] CR2: 0000000000000010 CR3: 0000000005026000 CR4: 00000000000406e0 [ 12.919446] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 [ 12.919451] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400 [ 12.919504] Process krfcommd (pid: 6210, threadinfo ffff880066932000, task ffff880065c4b000) [ 12.919506] Stack: [ 12.919510] ffff88009b00a200 ffff880032084000 ffff880066933c68 ffffffff8366c7bc [ 12.919513] 7fffffffffffffff ffff880032084000 ffff880066933c98 ffffffff833ae0ae [ 12.919516] ffff880066933ca8 0000000000000000 0000000000000000 ffff88009b00a200 [ 12.919517] Call Trace: [ 12.919522] [<ffffffff8366c7bc>] l2cap_sock_destruct+0x3c/0x80 [ 12.919527] [<ffffffff833ae0ae>] __sk_free+0x1e/0x1f0 [ 12.919530] [<ffffffff833ae2f7>] sk_free+0x17/0x20 [ 12.919585] [<ffffffff8366ca4e>] l2cap_sock_alloc.constprop.5+0x9e/0xd0 [ 12.919591] [<ffffffff8366cb9e>] l2cap_sock_create+0x7e/0x100 [ 12.919652] [<ffffffff83a4f32a>] ? _raw_read_lock+0x6a/0x80 [ 12.919658] [<ffffffff836402c4>] ? bt_sock_create+0x74/0x110 [ 12.919660] [<ffffffff83640308>] bt_sock_create+0xb8/0x110 [ 12.919664] [<ffffffff833aa232>] __sock_create+0x282/0x3b0 [ 12.919720] [<ffffffff833aa0b0>] ? __sock_create+0x100/0x3b0 [ 12.919725] [<ffffffff836785b0>] ? rfcomm_process_sessions+0x17e0/0x17e0 [ 12.919779] [<ffffffff833aa37f>] sock_create_kern+0x1f/0x30 [ 12.919784] [<ffffffff83675714>] rfcomm_l2sock_create+0x44/0x70 [ 12.919787] [<ffffffff836785b0>] ? rfcomm_process_sessions+0x17e0/0x17e0 [ 12.919790] [<ffffffff836785fe>] rfcomm_run+0x4e/0x1f0 [ 12.919846] [<ffffffff836785b0>] ? rfcomm_process_sessions+0x17e0/0x17e0 [ 12.919852] [<ffffffff81138ee3>] kthread+0xe3/0xf0 [ 12.919908] [<ffffffff8117b12e>] ? put_lock_stats.isra.14+0xe/0x40 [ 12.919914] [<ffffffff81138e00>] ? flush_kthread_work+0x1f0/0x1f0 [ 12.919968] [<ffffffff83a5077c>] ret_from_fork+0x7c/0x90 [ 12.919973] [<ffffffff81138e00>] ? flush_kthread_work+0x1f0/0x1f0 [ 12.920161] Code: 83 ec 08 f6 05 ff 58 44 02 04 74 1b 8b 4f 10 48 89 fa 48 c7 c6 d9 d7 d4 84 48 c7 c7 80 9e aa 85 31 c0 e8 80 ac 3a fe 48 8d 7b 10 <f0> 83 6b 10 01 0f 94 c0 84 c0 74 05 e8 8b e0 ff ff 48 83 c4 08 [ 12.920165] RIP [<ffffffff836645c4>] l2cap_chan_put+0x34/0x50 [ 12.920166] RSP <ffff880066933c38> [ 12.920167] CR2: 0000000000000010 [ 12.920417] ---[ end trace 5a9114e8a158ab84 ]--- Introduced in commit 61d6ef3e ("Bluetooth: Make better use of l2cap_chan reference counting"). Signed-off-by: Sasha Levin <sasha.levin@oracle.com> Signed-off-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>
2012-10-09 03:48:32 +07:00
if (l2cap_pi(sk)->chan)
l2cap_chan_put(l2cap_pi(sk)->chan);
if (l2cap_pi(sk)->rx_busy_skb) {
kfree_skb(l2cap_pi(sk)->rx_busy_skb);
l2cap_pi(sk)->rx_busy_skb = NULL;
}
skb_queue_purge(&sk->sk_receive_queue);
skb_queue_purge(&sk->sk_write_queue);
}
static void l2cap_skb_msg_name(struct sk_buff *skb, void *msg_name,
int *msg_namelen)
{
DECLARE_SOCKADDR(struct sockaddr_l2 *, la, msg_name);
memset(la, 0, sizeof(struct sockaddr_l2));
la->l2_family = AF_BLUETOOTH;
la->l2_psm = bt_cb(skb)->l2cap.psm;
bacpy(&la->l2_bdaddr, &bt_cb(skb)->l2cap.bdaddr);
*msg_namelen = sizeof(struct sockaddr_l2);
}
static void l2cap_sock_init(struct sock *sk, struct sock *parent)
{
struct l2cap_chan *chan = l2cap_pi(sk)->chan;
BT_DBG("sk %p", sk);
if (parent) {
struct l2cap_chan *pchan = l2cap_pi(parent)->chan;
sk->sk_type = parent->sk_type;
bt_sk(sk)->flags = bt_sk(parent)->flags;
chan->chan_type = pchan->chan_type;
chan->imtu = pchan->imtu;
chan->omtu = pchan->omtu;
chan->conf_state = pchan->conf_state;
chan->mode = pchan->mode;
chan->fcs = pchan->fcs;
chan->max_tx = pchan->max_tx;
chan->tx_win = pchan->tx_win;
chan->tx_win_max = pchan->tx_win_max;
chan->sec_level = pchan->sec_level;
chan->flags = pchan->flags;
chan->tx_credits = pchan->tx_credits;
chan->rx_credits = pchan->rx_credits;
if (chan->chan_type == L2CAP_CHAN_FIXED) {
chan->scid = pchan->scid;
chan->dcid = pchan->scid;
}
security_sk_clone(parent, sk);
} else {
switch (sk->sk_type) {
case SOCK_RAW:
chan->chan_type = L2CAP_CHAN_RAW;
break;
case SOCK_DGRAM:
chan->chan_type = L2CAP_CHAN_CONN_LESS;
bt_sk(sk)->skb_msg_name = l2cap_skb_msg_name;
break;
case SOCK_SEQPACKET:
case SOCK_STREAM:
chan->chan_type = L2CAP_CHAN_CONN_ORIENTED;
break;
}
chan->imtu = L2CAP_DEFAULT_MTU;
chan->omtu = 0;
if (!disable_ertm && sk->sk_type == SOCK_STREAM) {
chan->mode = L2CAP_MODE_ERTM;
set_bit(CONF_STATE2_DEVICE, &chan->conf_state);
} else {
chan->mode = L2CAP_MODE_BASIC;
}
l2cap_chan_set_defaults(chan);
}
/* Default config options */
chan->flush_to = L2CAP_DEFAULT_FLUSH_TO;
chan->data = sk;
chan->ops = &l2cap_chan_ops;
}
static struct proto l2cap_proto = {
.name = "L2CAP",
.owner = THIS_MODULE,
.obj_size = sizeof(struct l2cap_pinfo)
};
static struct sock *l2cap_sock_alloc(struct net *net, struct socket *sock,
int proto, gfp_t prio, int kern)
{
struct sock *sk;
struct l2cap_chan *chan;
sk = sk_alloc(net, PF_BLUETOOTH, prio, &l2cap_proto, kern);
if (!sk)
return NULL;
sock_init_data(sock, sk);
INIT_LIST_HEAD(&bt_sk(sk)->accept_q);
sk->sk_destruct = l2cap_sock_destruct;
sk->sk_sndtimeo = L2CAP_CONN_TIMEOUT;
sock_reset_flag(sk, SOCK_ZAPPED);
sk->sk_protocol = proto;
sk->sk_state = BT_OPEN;
chan = l2cap_chan_create();
if (!chan) {
sk_free(sk);
return NULL;
}
l2cap_chan_hold(chan);
l2cap_pi(sk)->chan = chan;
return sk;
}
static int l2cap_sock_create(struct net *net, struct socket *sock, int protocol,
int kern)
{
struct sock *sk;
BT_DBG("sock %p", sock);
sock->state = SS_UNCONNECTED;
if (sock->type != SOCK_SEQPACKET && sock->type != SOCK_STREAM &&
sock->type != SOCK_DGRAM && sock->type != SOCK_RAW)
return -ESOCKTNOSUPPORT;
if (sock->type == SOCK_RAW && !kern && !capable(CAP_NET_RAW))
return -EPERM;
sock->ops = &l2cap_sock_ops;
sk = l2cap_sock_alloc(net, sock, protocol, GFP_ATOMIC, kern);
if (!sk)
return -ENOMEM;
l2cap_sock_init(sk, NULL);
bt_sock_link(&l2cap_sk_list, sk);
return 0;
}
static const struct proto_ops l2cap_sock_ops = {
.family = PF_BLUETOOTH,
.owner = THIS_MODULE,
.release = l2cap_sock_release,
.bind = l2cap_sock_bind,
.connect = l2cap_sock_connect,
.listen = l2cap_sock_listen,
.accept = l2cap_sock_accept,
.getname = l2cap_sock_getname,
.sendmsg = l2cap_sock_sendmsg,
.recvmsg = l2cap_sock_recvmsg,
.poll = bt_sock_poll,
.ioctl = bt_sock_ioctl,
.gettstamp = sock_gettstamp,
.mmap = sock_no_mmap,
.socketpair = sock_no_socketpair,
.shutdown = l2cap_sock_shutdown,
.setsockopt = l2cap_sock_setsockopt,
.getsockopt = l2cap_sock_getsockopt
};
static const struct net_proto_family l2cap_sock_family_ops = {
.family = PF_BLUETOOTH,
.owner = THIS_MODULE,
.create = l2cap_sock_create,
};
int __init l2cap_init_sockets(void)
{
int err;
BUILD_BUG_ON(sizeof(struct sockaddr_l2) > sizeof(struct sockaddr));
err = proto_register(&l2cap_proto, 0);
if (err < 0)
return err;
err = bt_sock_register(BTPROTO_L2CAP, &l2cap_sock_family_ops);
if (err < 0) {
BT_ERR("L2CAP socket registration failed");
goto error;
}
err = bt_procfs_init(&init_net, "l2cap", &l2cap_sk_list,
NULL);
if (err < 0) {
BT_ERR("Failed to create L2CAP proc file");
bt_sock_unregister(BTPROTO_L2CAP);
goto error;
}
BT_INFO("L2CAP socket layer initialized");
return 0;
error:
proto_unregister(&l2cap_proto);
return err;
}
void l2cap_cleanup_sockets(void)
{
bt_procfs_cleanup(&init_net, "l2cap");
bt_sock_unregister(BTPROTO_L2CAP);
proto_unregister(&l2cap_proto);
}