From 86c6739eda7d2a03f2db30cbee67a5fb81afa8ba Mon Sep 17 00:00:00 2001 From: Steffen Klassert Date: Wed, 6 Nov 2019 08:13:49 +0100 Subject: xfrm: Fix memleak on xfrm state destroy We leak the page that we use to create skb page fragments when destroying the xfrm_state. Fix this by dropping a page reference if a page was assigned to the xfrm_state. Fixes: cac2661c53f3 ("esp4: Avoid skb_cow_data whenever possible") Reported-by: JD Reported-by: Paul Wouters Signed-off-by: Steffen Klassert --- net/xfrm/xfrm_state.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'net') diff --git a/net/xfrm/xfrm_state.c b/net/xfrm/xfrm_state.c index c6f3c4a1bd99..f3423562d933 100644 --- a/net/xfrm/xfrm_state.c +++ b/net/xfrm/xfrm_state.c @@ -495,6 +495,8 @@ static void ___xfrm_state_destroy(struct xfrm_state *x) x->type->destructor(x); xfrm_put_type(x->type); } + if (x->xfrag.page) + put_page(x->xfrag.page); xfrm_dev_state_free(x); security_xfrm_state_free(x); xfrm_state_free(x); -- cgit v1.2.3 From 5a508a254bed9a2e36a5fb96c9065532a6bf1e9c Mon Sep 17 00:00:00 2001 From: Jiri Pirko Date: Sat, 9 Nov 2019 11:29:46 +0100 Subject: devlink: disallow reload operation during device cleanup There is a race between driver code that does setup/cleanup of device and devlink reload operation that in some drivers works with the same code. Use after free could we easily obtained by running: while true; do echo "0000:00:10.0" >/sys/bus/pci/drivers/mlxsw_spectrum2/bind devlink dev reload pci/0000:00:10.0 & echo "0000:00:10.0" >/sys/bus/pci/drivers/mlxsw_spectrum2/unbind done Fix this by enabling reload only after setup of device is complete and disabling it at the beginning of the cleanup process. Reported-by: Ido Schimmel Fixes: 2d8dc5bbf4e7 ("devlink: Add support for reload") Signed-off-by: Jiri Pirko Signed-off-by: David S. Miller --- drivers/net/ethernet/mellanox/mlx4/main.c | 3 +++ drivers/net/ethernet/mellanox/mlxsw/core.c | 6 ++++- drivers/net/netdevsim/dev.c | 2 ++ include/net/devlink.h | 5 +++- net/core/devlink.c | 39 +++++++++++++++++++++++++++++- 5 files changed, 52 insertions(+), 3 deletions(-) (limited to 'net') diff --git a/drivers/net/ethernet/mellanox/mlx4/main.c b/drivers/net/ethernet/mellanox/mlx4/main.c index 69bb6bb06e76..d44ac666e730 100644 --- a/drivers/net/ethernet/mellanox/mlx4/main.c +++ b/drivers/net/ethernet/mellanox/mlx4/main.c @@ -4010,6 +4010,7 @@ static int mlx4_init_one(struct pci_dev *pdev, const struct pci_device_id *id) goto err_params_unregister; devlink_params_publish(devlink); + devlink_reload_enable(devlink); pci_save_state(pdev); return 0; @@ -4121,6 +4122,8 @@ static void mlx4_remove_one(struct pci_dev *pdev) struct devlink *devlink = priv_to_devlink(priv); int active_vfs = 0; + devlink_reload_disable(devlink); + if (mlx4_is_slave(dev)) persist->interface_state |= MLX4_INTERFACE_STATE_NOWAIT; diff --git a/drivers/net/ethernet/mellanox/mlxsw/core.c b/drivers/net/ethernet/mellanox/mlxsw/core.c index 4421ab22182f..20e9dc46cacd 100644 --- a/drivers/net/ethernet/mellanox/mlxsw/core.c +++ b/drivers/net/ethernet/mellanox/mlxsw/core.c @@ -1186,8 +1186,10 @@ __mlxsw_core_bus_device_register(const struct mlxsw_bus_info *mlxsw_bus_info, if (err) goto err_thermal_init; - if (mlxsw_driver->params_register) + if (mlxsw_driver->params_register) { devlink_params_publish(devlink); + devlink_reload_enable(devlink); + } return 0; @@ -1249,6 +1251,8 @@ void mlxsw_core_bus_device_unregister(struct mlxsw_core *mlxsw_core, { struct devlink *devlink = priv_to_devlink(mlxsw_core); + if (!reload) + devlink_reload_disable(devlink); if (devlink_is_reload_failed(devlink)) { if (!reload) /* Only the parts that were not de-initialized in the diff --git a/drivers/net/netdevsim/dev.c b/drivers/net/netdevsim/dev.c index 54ca6681ba31..44c2d857a7fa 100644 --- a/drivers/net/netdevsim/dev.c +++ b/drivers/net/netdevsim/dev.c @@ -708,6 +708,7 @@ nsim_dev_create(struct nsim_bus_dev *nsim_bus_dev, unsigned int port_count) goto err_debugfs_exit; devlink_params_publish(devlink); + devlink_reload_enable(devlink); return nsim_dev; err_debugfs_exit: @@ -732,6 +733,7 @@ static void nsim_dev_destroy(struct nsim_dev *nsim_dev) { struct devlink *devlink = priv_to_devlink(nsim_dev); + devlink_reload_disable(devlink); nsim_bpf_dev_exit(nsim_dev); nsim_dev_debugfs_exit(nsim_dev); nsim_dev_traps_exit(devlink); diff --git a/include/net/devlink.h b/include/net/devlink.h index 23e4b65ec9df..2116c88663a1 100644 --- a/include/net/devlink.h +++ b/include/net/devlink.h @@ -38,7 +38,8 @@ struct devlink { struct device *dev; possible_net_t _net; struct mutex lock; - bool reload_failed; + u8 reload_failed:1, + reload_enabled:1; char priv[0] __aligned(NETDEV_ALIGN); }; @@ -774,6 +775,8 @@ struct ib_device; struct devlink *devlink_alloc(const struct devlink_ops *ops, size_t priv_size); int devlink_register(struct devlink *devlink, struct device *dev); void devlink_unregister(struct devlink *devlink); +void devlink_reload_enable(struct devlink *devlink); +void devlink_reload_disable(struct devlink *devlink); void devlink_free(struct devlink *devlink); int devlink_port_register(struct devlink *devlink, struct devlink_port *devlink_port, diff --git a/net/core/devlink.c b/net/core/devlink.c index f80151eeaf51..7d64660a72fc 100644 --- a/net/core/devlink.c +++ b/net/core/devlink.c @@ -2699,7 +2699,7 @@ static int devlink_nl_cmd_reload(struct sk_buff *skb, struct genl_info *info) struct devlink *devlink = info->user_ptr[0]; int err; - if (!devlink_reload_supported(devlink)) + if (!devlink_reload_supported(devlink) || !devlink->reload_enabled) return -EOPNOTSUPP; err = devlink_resources_validate(devlink, NULL, info); @@ -6196,12 +6196,49 @@ EXPORT_SYMBOL_GPL(devlink_register); void devlink_unregister(struct devlink *devlink) { mutex_lock(&devlink_mutex); + WARN_ON(devlink_reload_supported(devlink) && + devlink->reload_enabled); devlink_notify(devlink, DEVLINK_CMD_DEL); list_del(&devlink->list); mutex_unlock(&devlink_mutex); } EXPORT_SYMBOL_GPL(devlink_unregister); +/** + * devlink_reload_enable - Enable reload of devlink instance + * + * @devlink: devlink + * + * Should be called at end of device initialization + * process when reload operation is supported. + */ +void devlink_reload_enable(struct devlink *devlink) +{ + mutex_lock(&devlink_mutex); + devlink->reload_enabled = true; + mutex_unlock(&devlink_mutex); +} +EXPORT_SYMBOL_GPL(devlink_reload_enable); + +/** + * devlink_reload_disable - Disable reload of devlink instance + * + * @devlink: devlink + * + * Should be called at the beginning of device cleanup + * process when reload operation is supported. + */ +void devlink_reload_disable(struct devlink *devlink) +{ + mutex_lock(&devlink_mutex); + /* Mutex is taken which ensures that no reload operation is in + * progress while setting up forbidded flag. + */ + devlink->reload_enabled = false; + mutex_unlock(&devlink_mutex); +} +EXPORT_SYMBOL_GPL(devlink_reload_disable); + /** * devlink_free - Free devlink instance resources * -- cgit v1.2.3 From d279505b723cba058b604ed8cf9cd4c854e2a041 Mon Sep 17 00:00:00 2001 From: Aya Levin Date: Sun, 10 Nov 2019 14:11:56 +0200 Subject: devlink: Add method for time-stamp on reporter's dump When setting the dump's time-stamp, use ktime_get_real in addition to jiffies. This simplifies the user space implementation and bypasses some inconsistent behavior with translating jiffies to current time. The time taken is transformed into nsec, to comply with y2038 issue. Fixes: c8e1da0bf923 ("devlink: Add health report functionality") Signed-off-by: Aya Levin Acked-by: Jiri Pirko Acked-by: Arnd Bergmann Signed-off-by: David S. Miller --- include/uapi/linux/devlink.h | 1 + net/core/devlink.c | 6 ++++++ 2 files changed, 7 insertions(+) (limited to 'net') diff --git a/include/uapi/linux/devlink.h b/include/uapi/linux/devlink.h index 580b7a2e40e1..a8a2174db030 100644 --- a/include/uapi/linux/devlink.h +++ b/include/uapi/linux/devlink.h @@ -421,6 +421,7 @@ enum devlink_attr { DEVLINK_ATTR_RELOAD_FAILED, /* u8 0 or 1 */ + DEVLINK_ATTR_HEALTH_REPORTER_DUMP_TS_NS, /* u64 */ /* add new attributes above here, update the policy in devlink.c */ __DEVLINK_ATTR_MAX, diff --git a/net/core/devlink.c b/net/core/devlink.c index 7d64660a72fc..93905dc7c179 100644 --- a/net/core/devlink.c +++ b/net/core/devlink.c @@ -4618,6 +4618,7 @@ struct devlink_health_reporter { bool auto_recover; u8 health_state; u64 dump_ts; + u64 dump_real_ts; u64 error_count; u64 recovery_count; u64 last_recovery_ts; @@ -4790,6 +4791,7 @@ static int devlink_health_do_dump(struct devlink_health_reporter *reporter, goto dump_err; reporter->dump_ts = jiffies; + reporter->dump_real_ts = ktime_get_real_ns(); return 0; @@ -4952,6 +4954,10 @@ devlink_nl_health_reporter_fill(struct sk_buff *msg, jiffies_to_msecs(reporter->dump_ts), DEVLINK_ATTR_PAD)) goto reporter_nest_cancel; + if (reporter->dump_fmsg && + nla_put_u64_64bit(msg, DEVLINK_ATTR_HEALTH_REPORTER_DUMP_TS_NS, + reporter->dump_real_ts, DEVLINK_ATTR_PAD)) + goto reporter_nest_cancel; nla_nest_end(msg, reporter_attr); genlmsg_end(msg, hdr); -- cgit v1.2.3 From 4944a4b1077f74d89073624bd286219d2fcbfce3 Mon Sep 17 00:00:00 2001 From: Xiaodong Xu Date: Mon, 11 Nov 2019 15:05:46 -0800 Subject: xfrm: release device reference for invalid state An ESP packet could be decrypted in async mode if the input handler for this packet returns -EINPROGRESS in xfrm_input(). At this moment the device reference in skb is held. Later xfrm_input() will be invoked again to resume the processing. If the transform state is still valid it would continue to release the device reference and there won't be a problem; however if the transform state is not valid when async resumption happens, the packet will be dropped while the device reference is still being held. When the device is deleted for some reason and the reference to this device is not properly released, the kernel will keep logging like: unregister_netdevice: waiting for ppp2 to become free. Usage count = 1 The issue is observed when running IPsec traffic over a PPPoE device based on a bridge interface. By terminating the PPPoE connection on the server end for multiple times, the PPPoE device on the client side will eventually get stuck on the above warning message. This patch will check the async mode first and continue to release device reference in async resumption, before it is dropped due to invalid state. v2: Do not assign address family from outer_mode in the transform if the state is invalid v3: Release device reference in the error path instead of jumping to resume Fixes: 4ce3dbe397d7b ("xfrm: Fix xfrm_input() to verify state is valid when (encap_type < 0)") Signed-off-by: Xiaodong Xu Reported-by: Bo Chen Tested-by: Bo Chen Signed-off-by: Steffen Klassert --- net/xfrm/xfrm_input.c | 3 +++ 1 file changed, 3 insertions(+) (limited to 'net') diff --git a/net/xfrm/xfrm_input.c b/net/xfrm/xfrm_input.c index 9b599ed66d97..2c86a2fc3915 100644 --- a/net/xfrm/xfrm_input.c +++ b/net/xfrm/xfrm_input.c @@ -480,6 +480,9 @@ int xfrm_input(struct sk_buff *skb, int nexthdr, __be32 spi, int encap_type) else XFRM_INC_STATS(net, LINUX_MIB_XFRMINSTATEINVALID); + + if (encap_type == -1) + dev_put(skb->dev); goto drop; } -- cgit v1.2.3 From 6d6dd528d5af05dc2d0c773951ed68d630a0c3f1 Mon Sep 17 00:00:00 2001 From: Ursula Braun Date: Tue, 12 Nov 2019 16:03:41 +0100 Subject: net/smc: fix refcount non-blocking connect() -part 2 If an SMC socket is immediately terminated after a non-blocking connect() has been called, a memory leak is possible. Due to the sock_hold move in commit 301428ea3708 ("net/smc: fix refcounting for non-blocking connect()") an extra sock_put() is needed in smc_connect_work(), if the internal TCP socket is aborted and cancels the sk_stream_wait_connect() of the connect worker. Reported-by: syzbot+4b73ad6fc767e576e275@syzkaller.appspotmail.com Fixes: 301428ea3708 ("net/smc: fix refcounting for non-blocking connect()") Signed-off-by: Ursula Braun Signed-off-by: Karsten Graul Signed-off-by: David S. Miller --- net/smc/af_smc.c | 1 + 1 file changed, 1 insertion(+) (limited to 'net') diff --git a/net/smc/af_smc.c b/net/smc/af_smc.c index 47946f489fd4..8edf1619f0e4 100644 --- a/net/smc/af_smc.c +++ b/net/smc/af_smc.c @@ -796,6 +796,7 @@ static void smc_connect_work(struct work_struct *work) smc->sk.sk_err = EPIPE; else if (signal_pending(current)) smc->sk.sk_err = -sock_intr_errno(timeo); + sock_put(&smc->sk); /* passive closing */ goto out; } -- cgit v1.2.3 From 975987e7015bb12a482df7f14fd524417d2c8e8f Mon Sep 17 00:00:00 2001 From: Oleksij Rempel Date: Thu, 7 Nov 2019 11:55:42 +0100 Subject: can: af_can: export can_sock_destruct() In j1939 we need our own struct sock::sk_destruct callback. Export the generic af_can can_sock_destruct() that allows us to chain-call it. Fixes: 9d71dd0c7009 ("can: add support of SAE J1939 protocol") Signed-off-by: Oleksij Rempel --- include/linux/can/core.h | 1 + net/can/af_can.c | 3 ++- 2 files changed, 3 insertions(+), 1 deletion(-) (limited to 'net') diff --git a/include/linux/can/core.h b/include/linux/can/core.h index 8339071ab08b..e20a0cd09ba5 100644 --- a/include/linux/can/core.h +++ b/include/linux/can/core.h @@ -65,5 +65,6 @@ extern void can_rx_unregister(struct net *net, struct net_device *dev, void *data); extern int can_send(struct sk_buff *skb, int loop); +void can_sock_destruct(struct sock *sk); #endif /* !_CAN_CORE_H */ diff --git a/net/can/af_can.c b/net/can/af_can.c index 5518a7d9eed9..128d37a4c2e0 100644 --- a/net/can/af_can.c +++ b/net/can/af_can.c @@ -86,11 +86,12 @@ static atomic_t skbcounter = ATOMIC_INIT(0); /* af_can socket functions */ -static void can_sock_destruct(struct sock *sk) +void can_sock_destruct(struct sock *sk) { skb_queue_purge(&sk->sk_receive_queue); skb_queue_purge(&sk->sk_error_queue); } +EXPORT_SYMBOL(can_sock_destruct); static const struct can_proto *can_get_proto(int protocol) { -- cgit v1.2.3 From 25fe97cb7620ef2e6b4f44ef0de4e371adf6c1d0 Mon Sep 17 00:00:00 2001 From: Oleksij Rempel Date: Thu, 7 Nov 2019 11:57:36 +0100 Subject: can: j1939: move j1939_priv_put() into sk_destruct callback This patch delays the j1939_priv_put() until the socket is destroyed via the sk_destruct callback, to avoid use-after-free problems. Fixes: 9d71dd0c7009 ("can: add support of SAE J1939 protocol") Signed-off-by: Oleksij Rempel --- net/can/j1939/socket.c | 37 +++++++++++++++++++++++++++++++++++-- 1 file changed, 35 insertions(+), 2 deletions(-) (limited to 'net') diff --git a/net/can/j1939/socket.c b/net/can/j1939/socket.c index 4d8ba701e15d..aee94b09ef08 100644 --- a/net/can/j1939/socket.c +++ b/net/can/j1939/socket.c @@ -78,7 +78,6 @@ static void j1939_jsk_add(struct j1939_priv *priv, struct j1939_sock *jsk) { jsk->state |= J1939_SOCK_BOUND; j1939_priv_get(priv); - jsk->priv = priv; spin_lock_bh(&priv->j1939_socks_lock); list_add_tail(&jsk->list, &priv->j1939_socks); @@ -91,7 +90,6 @@ static void j1939_jsk_del(struct j1939_priv *priv, struct j1939_sock *jsk) list_del_init(&jsk->list); spin_unlock_bh(&priv->j1939_socks_lock); - jsk->priv = NULL; j1939_priv_put(priv); jsk->state &= ~J1939_SOCK_BOUND; } @@ -349,6 +347,34 @@ void j1939_sk_recv(struct j1939_priv *priv, struct sk_buff *skb) spin_unlock_bh(&priv->j1939_socks_lock); } +static void j1939_sk_sock_destruct(struct sock *sk) +{ + struct j1939_sock *jsk = j1939_sk(sk); + + /* This function will be call by the generic networking code, when then + * the socket is ultimately closed (sk->sk_destruct). + * + * The race between + * - processing a received CAN frame + * (can_receive -> j1939_can_recv) + * and accessing j1939_priv + * ... and ... + * - closing a socket + * (j1939_can_rx_unregister -> can_rx_unregister) + * and calling the final j1939_priv_put() + * + * is avoided by calling the final j1939_priv_put() from this + * RCU deferred cleanup call. + */ + if (jsk->priv) { + j1939_priv_put(jsk->priv); + jsk->priv = NULL; + } + + /* call generic CAN sock destruct */ + can_sock_destruct(sk); +} + static int j1939_sk_init(struct sock *sk) { struct j1939_sock *jsk = j1939_sk(sk); @@ -371,6 +397,7 @@ static int j1939_sk_init(struct sock *sk) atomic_set(&jsk->skb_pending, 0); spin_lock_init(&jsk->sk_session_queue_lock); INIT_LIST_HEAD(&jsk->sk_session_queue); + sk->sk_destruct = j1939_sk_sock_destruct; return 0; } @@ -443,6 +470,12 @@ static int j1939_sk_bind(struct socket *sock, struct sockaddr *uaddr, int len) } jsk->ifindex = addr->can_ifindex; + + /* the corresponding j1939_priv_put() is called via + * sk->sk_destruct, which points to j1939_sk_sock_destruct() + */ + j1939_priv_get(priv); + jsk->priv = priv; } /* set default transmit pgn */ -- cgit v1.2.3 From c48c8c1e2e81e71a0f13b83cc5124333f3750064 Mon Sep 17 00:00:00 2001 From: Oleksij Rempel Date: Tue, 5 Nov 2019 11:07:08 +0100 Subject: can: j1939: main: j1939_ndev_to_priv(): avoid crash if can_ml_priv is NULL This patch avoids a NULL pointer deref crash if ndev->ml_priv is NULL. Reported-by: syzbot+95c8e0d9dffde15b6c5c@syzkaller.appspotmail.com Fixes: 9d71dd0c7009 ("can: add support of SAE J1939 protocol") Signed-off-by: Oleksij Rempel --- net/can/j1939/main.c | 3 +++ 1 file changed, 3 insertions(+) (limited to 'net') diff --git a/net/can/j1939/main.c b/net/can/j1939/main.c index def2f813ffce..8dc935dc2e54 100644 --- a/net/can/j1939/main.c +++ b/net/can/j1939/main.c @@ -207,6 +207,9 @@ static inline struct j1939_priv *j1939_ndev_to_priv(struct net_device *ndev) { struct can_ml_priv *can_ml_priv = ndev->ml_priv; + if (!can_ml_priv) + return NULL; + return can_ml_priv->j1939_priv; } -- cgit v1.2.3 From fd81ebfe7975b9a69494430676d16f7125aac3ee Mon Sep 17 00:00:00 2001 From: Oleksij Rempel Date: Tue, 5 Nov 2019 14:31:58 +0100 Subject: can: j1939: socket: rework socket locking for j1939_sk_release() and j1939_sk_sendmsg() j1939_sk_sendmsg() should be protected by lock_sock() to avoid race with j1939_sk_bind() and j1939_sk_release(). Reported-by: syzbot+afd421337a736d6c1ee6@syzkaller.appspotmail.com Reported-by: syzbot+6d04f6a1b31a0ae12ca9@syzkaller.appspotmail.com Fixes: 9d71dd0c7009 ("can: add support of SAE J1939 protocol") Signed-off-by: Oleksij Rempel --- net/can/j1939/socket.c | 57 ++++++++++++++++++++++++++++++++++---------------- 1 file changed, 39 insertions(+), 18 deletions(-) (limited to 'net') diff --git a/net/can/j1939/socket.c b/net/can/j1939/socket.c index aee94b09ef08..de09b0a65791 100644 --- a/net/can/j1939/socket.c +++ b/net/can/j1939/socket.c @@ -593,8 +593,8 @@ static int j1939_sk_release(struct socket *sock) if (!sk) return 0; - jsk = j1939_sk(sk); lock_sock(sk); + jsk = j1939_sk(sk); if (jsk->state & J1939_SOCK_BOUND) { struct j1939_priv *priv = jsk->priv; @@ -1092,51 +1092,72 @@ static int j1939_sk_sendmsg(struct socket *sock, struct msghdr *msg, { struct sock *sk = sock->sk; struct j1939_sock *jsk = j1939_sk(sk); - struct j1939_priv *priv = jsk->priv; + struct j1939_priv *priv; int ifindex; int ret; + lock_sock(sock->sk); /* various socket state tests */ - if (!(jsk->state & J1939_SOCK_BOUND)) - return -EBADFD; + if (!(jsk->state & J1939_SOCK_BOUND)) { + ret = -EBADFD; + goto sendmsg_done; + } + priv = jsk->priv; ifindex = jsk->ifindex; - if (!jsk->addr.src_name && jsk->addr.sa == J1939_NO_ADDR) + if (!jsk->addr.src_name && jsk->addr.sa == J1939_NO_ADDR) { /* no source address assigned yet */ - return -EBADFD; + ret = -EBADFD; + goto sendmsg_done; + } /* deal with provided destination address info */ if (msg->msg_name) { struct sockaddr_can *addr = msg->msg_name; - if (msg->msg_namelen < J1939_MIN_NAMELEN) - return -EINVAL; + if (msg->msg_namelen < J1939_MIN_NAMELEN) { + ret = -EINVAL; + goto sendmsg_done; + } - if (addr->can_family != AF_CAN) - return -EINVAL; + if (addr->can_family != AF_CAN) { + ret = -EINVAL; + goto sendmsg_done; + } - if (addr->can_ifindex && addr->can_ifindex != ifindex) - return -EBADFD; + if (addr->can_ifindex && addr->can_ifindex != ifindex) { + ret = -EBADFD; + goto sendmsg_done; + } if (j1939_pgn_is_valid(addr->can_addr.j1939.pgn) && - !j1939_pgn_is_clean_pdu(addr->can_addr.j1939.pgn)) - return -EINVAL; + !j1939_pgn_is_clean_pdu(addr->can_addr.j1939.pgn)) { + ret = -EINVAL; + goto sendmsg_done; + } if (!addr->can_addr.j1939.name && addr->can_addr.j1939.addr == J1939_NO_ADDR && - !sock_flag(sk, SOCK_BROADCAST)) + !sock_flag(sk, SOCK_BROADCAST)) { /* broadcast, but SO_BROADCAST not set */ - return -EACCES; + ret = -EACCES; + goto sendmsg_done; + } } else { if (!jsk->addr.dst_name && jsk->addr.da == J1939_NO_ADDR && - !sock_flag(sk, SOCK_BROADCAST)) + !sock_flag(sk, SOCK_BROADCAST)) { /* broadcast, but SO_BROADCAST not set */ - return -EACCES; + ret = -EACCES; + goto sendmsg_done; + } } ret = j1939_sk_send_loop(priv, sk, msg, size); +sendmsg_done: + release_sock(sock->sk); + return ret; } -- cgit v1.2.3 From d966635b384b9571a43bd38c61f280c47eb564ad Mon Sep 17 00:00:00 2001 From: Oleksij Rempel Date: Thu, 7 Nov 2019 18:46:38 +0100 Subject: can: j1939: transport: make sure the aborted session will be deactivated only once j1939_session_cancel() was modifying session->state without protecting it by locks and without checking actual state of the session. This patch moves j1939_tp_set_rxtimeout() into j1939_session_cancel() and adds the missing locking. Fixes: 9d71dd0c7009 ("can: add support of SAE J1939 protocol") Signed-off-by: Oleksij Rempel --- net/can/j1939/transport.c | 25 +++++++++++++++++-------- 1 file changed, 17 insertions(+), 8 deletions(-) (limited to 'net') diff --git a/net/can/j1939/transport.c b/net/can/j1939/transport.c index e5f1a56994c6..ecdedfc0b10c 100644 --- a/net/can/j1939/transport.c +++ b/net/can/j1939/transport.c @@ -1042,12 +1042,13 @@ j1939_session_deactivate_activate_next(struct j1939_session *session) j1939_sk_queue_activate_next(session); } -static void j1939_session_cancel(struct j1939_session *session, +static void __j1939_session_cancel(struct j1939_session *session, enum j1939_xtp_abort err) { struct j1939_priv *priv = session->priv; WARN_ON_ONCE(!err); + lockdep_assert_held(&session->priv->active_session_list_lock); session->err = j1939_xtp_abort_to_errno(priv, err); /* do not send aborts on incoming broadcasts */ @@ -1062,6 +1063,20 @@ static void j1939_session_cancel(struct j1939_session *session, j1939_sk_send_loop_abort(session->sk, session->err); } +static void j1939_session_cancel(struct j1939_session *session, + enum j1939_xtp_abort err) +{ + j1939_session_list_lock(session->priv); + + if (session->state >= J1939_SESSION_ACTIVE && + session->state < J1939_SESSION_WAITING_ABORT) { + j1939_tp_set_rxtimeout(session, J1939_XTP_ABORT_TIMEOUT_MS); + __j1939_session_cancel(session, err); + } + + j1939_session_list_unlock(session->priv); +} + static enum hrtimer_restart j1939_tp_txtimer(struct hrtimer *hrtimer) { struct j1939_session *session = @@ -1108,8 +1123,6 @@ static enum hrtimer_restart j1939_tp_txtimer(struct hrtimer *hrtimer) netdev_alert(priv->ndev, "%s: 0x%p: tx aborted with unknown reason: %i\n", __func__, session, ret); if (session->skcb.addr.type != J1939_SIMPLE) { - j1939_tp_set_rxtimeout(session, - J1939_XTP_ABORT_TIMEOUT_MS); j1939_session_cancel(session, J1939_XTP_ABORT_OTHER); } else { session->err = ret; @@ -1169,7 +1182,7 @@ static enum hrtimer_restart j1939_tp_rxtimer(struct hrtimer *hrtimer) hrtimer_start(&session->rxtimer, ms_to_ktime(J1939_XTP_ABORT_TIMEOUT_MS), HRTIMER_MODE_REL_SOFT); - j1939_session_cancel(session, J1939_XTP_ABORT_TIMEOUT); + __j1939_session_cancel(session, J1939_XTP_ABORT_TIMEOUT); } j1939_session_list_unlock(session->priv); } @@ -1375,7 +1388,6 @@ j1939_xtp_rx_cts_one(struct j1939_session *session, struct sk_buff *skb) out_session_cancel: j1939_session_timers_cancel(session); - j1939_tp_set_rxtimeout(session, J1939_XTP_ABORT_TIMEOUT_MS); j1939_session_cancel(session, err); } @@ -1572,7 +1584,6 @@ static int j1939_xtp_rx_rts_session_active(struct j1939_session *session, /* RTS on active session */ j1939_session_timers_cancel(session); - j1939_tp_set_rxtimeout(session, J1939_XTP_ABORT_TIMEOUT_MS); j1939_session_cancel(session, J1939_XTP_ABORT_BUSY); } @@ -1583,7 +1594,6 @@ static int j1939_xtp_rx_rts_session_active(struct j1939_session *session, session->last_cmd); j1939_session_timers_cancel(session); - j1939_tp_set_rxtimeout(session, J1939_XTP_ABORT_TIMEOUT_MS); j1939_session_cancel(session, J1939_XTP_ABORT_BUSY); return -EBUSY; @@ -1785,7 +1795,6 @@ static void j1939_xtp_rx_dat_one(struct j1939_session *session, out_session_cancel: j1939_session_timers_cancel(session); - j1939_tp_set_rxtimeout(session, J1939_XTP_ABORT_TIMEOUT_MS); j1939_session_cancel(session, J1939_XTP_ABORT_FAULT); j1939_session_put(session); } -- cgit v1.2.3 From 62ebce1dc1fa649a1c54db02f1a3c409bb0529ec Mon Sep 17 00:00:00 2001 From: Oleksij Rempel Date: Thu, 7 Nov 2019 18:51:40 +0100 Subject: can: j1939: make sure socket is held as long as session exists We link the socket to the session to be able provide socket specific notifications. For example messages over error queue. We need to keep the socket held, while we have a reference to it. Fixes: 9d71dd0c7009 ("can: add support of SAE J1939 protocol") Signed-off-by: Oleksij Rempel --- net/can/j1939/transport.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'net') diff --git a/net/can/j1939/transport.c b/net/can/j1939/transport.c index ecdedfc0b10c..afc2adfd97e4 100644 --- a/net/can/j1939/transport.c +++ b/net/can/j1939/transport.c @@ -255,6 +255,7 @@ static void __j1939_session_drop(struct j1939_session *session) return; j1939_sock_pending_del(session->sk); + sock_put(session->sk); } static void j1939_session_destroy(struct j1939_session *session) @@ -1875,6 +1876,7 @@ struct j1939_session *j1939_tp_send(struct j1939_priv *priv, return ERR_PTR(-ENOMEM); /* skb is recounted in j1939_session_new() */ + sock_hold(skb->sk); session->sk = skb->sk; session->transmission = true; session->pkt.total = (size + 6) / 7; -- cgit v1.2.3 From 8d7a5f000e235f1dfc61862197d4e8e72c18c6fc Mon Sep 17 00:00:00 2001 From: Oleksij Rempel Date: Fri, 8 Nov 2019 14:02:10 +0100 Subject: can: j1939: transport: j1939_cancel_active_session(): use hrtimer_try_to_cancel() instead of hrtimer_cancel() This part of the code protected by lock used in the hrtimer as well. Using hrtimer_cancel() will trigger dead lock. Fixes: 9d71dd0c7009 ("can: add support of SAE J1939 protocol") Signed-off-by: Oleksij Rempel --- net/can/j1939/transport.c | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) (limited to 'net') diff --git a/net/can/j1939/transport.c b/net/can/j1939/transport.c index afc2adfd97e4..0c62b8fc4b20 100644 --- a/net/can/j1939/transport.c +++ b/net/can/j1939/transport.c @@ -2039,7 +2039,11 @@ int j1939_cancel_active_session(struct j1939_priv *priv, struct sock *sk) &priv->active_session_list, active_session_list_entry) { if (!sk || sk == session->sk) { - j1939_session_timers_cancel(session); + if (hrtimer_try_to_cancel(&session->txtimer) == 1) + j1939_session_put(session); + if (hrtimer_try_to_cancel(&session->rxtimer) == 1) + j1939_session_put(session); + session->err = ESHUTDOWN; j1939_session_deactivate_locked(session); } -- cgit v1.2.3 From ddeeb7d4822ed06d79fc15e822b70dce3fa77e39 Mon Sep 17 00:00:00 2001 From: Oleksij Rempel Date: Sat, 9 Nov 2019 16:11:18 +0100 Subject: can: j1939: j1939_can_recv(): add priv refcounting j1939_can_recv() can be called in parallel with socket release. In this case sk_release and sk_destruct can be done earlier than j1939_can_recv() is processed. Reported-by: syzbot+ca172a0ac477ac90f045@syzkaller.appspotmail.com Reported-by: syzbot+07ca5bce8530070a5650@syzkaller.appspotmail.com Reported-by: syzbot+a47537d3964ef6c874e1@syzkaller.appspotmail.com Fixes: 9d71dd0c7009 ("can: add support of SAE J1939 protocol") Signed-off-by: Oleksij Rempel --- net/can/j1939/main.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'net') diff --git a/net/can/j1939/main.c b/net/can/j1939/main.c index 8dc935dc2e54..2afcf27c72c8 100644 --- a/net/can/j1939/main.c +++ b/net/can/j1939/main.c @@ -51,6 +51,7 @@ static void j1939_can_recv(struct sk_buff *iskb, void *data) if (!skb) return; + j1939_priv_get(priv); can_skb_set_owner(skb, iskb->sk); /* get a pointer to the header of the skb @@ -104,6 +105,7 @@ static void j1939_can_recv(struct sk_buff *iskb, void *data) j1939_simple_recv(priv, skb); j1939_sk_recv(priv, skb); done: + j1939_priv_put(priv); kfree_skb(skb); } -- cgit v1.2.3 From 4a15d574e68afffbe8d7265e015cda2ac2a248ec Mon Sep 17 00:00:00 2001 From: Oleksij Rempel Date: Fri, 8 Nov 2019 13:57:14 +0100 Subject: can: j1939: warn if resources are still linked on destroy j1939_session_destroy() and __j1939_priv_release() should be called only if session, ecu or socket are not linked or used by any one else. If at least one of these resources is linked, then the reference counting is broken somewhere. This warning will be triggered before KASAN will do, and will make it easier to debug initial issue. This works on platforms without KASAN support. Signed-off-by: Oleksij Rempel --- net/can/j1939/main.c | 4 ++++ net/can/j1939/transport.c | 3 +++ 2 files changed, 7 insertions(+) (limited to 'net') diff --git a/net/can/j1939/main.c b/net/can/j1939/main.c index 2afcf27c72c8..137054bff9ec 100644 --- a/net/can/j1939/main.c +++ b/net/can/j1939/main.c @@ -152,6 +152,10 @@ static void __j1939_priv_release(struct kref *kref) netdev_dbg(priv->ndev, "%s: 0x%p\n", __func__, priv); + WARN_ON_ONCE(!list_empty(&priv->active_session_list)); + WARN_ON_ONCE(!list_empty(&priv->ecus)); + WARN_ON_ONCE(!list_empty(&priv->j1939_socks)); + dev_put(ndev); kfree(priv); } diff --git a/net/can/j1939/transport.c b/net/can/j1939/transport.c index 0c62b8fc4b20..9f99af5b0b11 100644 --- a/net/can/j1939/transport.c +++ b/net/can/j1939/transport.c @@ -267,6 +267,9 @@ static void j1939_session_destroy(struct j1939_session *session) netdev_dbg(session->priv->ndev, "%s: 0x%p\n", __func__, session); + WARN_ON_ONCE(!list_empty(&session->sk_session_queue_entry)); + WARN_ON_ONCE(!list_empty(&session->active_session_list_entry)); + skb_queue_purge(&session->skb_queue); __j1939_session_drop(session); j1939_priv_put(session->priv); -- cgit v1.2.3 From d7f9f47d4d1243cf960725a755123cfae13a3e51 Mon Sep 17 00:00:00 2001 From: Matt Bennett Date: Thu, 14 Nov 2019 12:20:03 +1300 Subject: tipc: add back tipc prefix to log messages The tipc prefix for log messages generated by tipc was removed in commit 07f6c4bc048a ("tipc: convert tipc reference table to use generic rhashtable"). This is still a useful prefix so add it back. Signed-off-by: Matt Bennett Acked-by: Jon Maloy Signed-off-by: David S. Miller --- net/tipc/core.c | 2 -- net/tipc/core.h | 6 ++++++ 2 files changed, 6 insertions(+), 2 deletions(-) (limited to 'net') diff --git a/net/tipc/core.c b/net/tipc/core.c index 23cb379a93d6..8f35060a24e1 100644 --- a/net/tipc/core.c +++ b/net/tipc/core.c @@ -34,8 +34,6 @@ * POSSIBILITY OF SUCH DAMAGE. */ -#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt - #include "core.h" #include "name_table.h" #include "subscr.h" diff --git a/net/tipc/core.h b/net/tipc/core.h index 60d829581068..3042f654e0af 100644 --- a/net/tipc/core.h +++ b/net/tipc/core.h @@ -60,6 +60,12 @@ #include #include +#ifdef pr_fmt +#undef pr_fmt +#endif + +#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt + struct tipc_node; struct tipc_bearer; struct tipc_bc_base; -- cgit v1.2.3 From 7f91ed8c4f4b6368b0603543bccb6b5fd10804e1 Mon Sep 17 00:00:00 2001 From: Andrea Mayer Date: Sat, 16 Nov 2019 16:05:52 +0100 Subject: seg6: fix srh pointer in get_srh() pskb_may_pull may change pointers in header. For this reason, it is mandatory to reload any pointer that points into skb header. Signed-off-by: Andrea Mayer Signed-off-by: David S. Miller --- net/ipv6/seg6_local.c | 5 +++++ 1 file changed, 5 insertions(+) (limited to 'net') diff --git a/net/ipv6/seg6_local.c b/net/ipv6/seg6_local.c index 9d4f75e0d33a..5e3d7004d431 100644 --- a/net/ipv6/seg6_local.c +++ b/net/ipv6/seg6_local.c @@ -81,6 +81,11 @@ static struct ipv6_sr_hdr *get_srh(struct sk_buff *skb) if (!pskb_may_pull(skb, srhoff + len)) return NULL; + /* note that pskb_may_pull may change pointers in header; + * for this reason it is necessary to reload them when needed. + */ + srh = (struct ipv6_sr_hdr *)(skb->data + srhoff); + if (!seg6_validate_srh(srh, len)) return NULL; -- cgit v1.2.3 From c71644d00f9fdd87e5d54fdd388ba899ae3852fa Mon Sep 17 00:00:00 2001 From: Andrea Mayer Date: Sat, 16 Nov 2019 16:05:53 +0100 Subject: seg6: fix skb transport_header after decap_and_validate() in the receive path (more precisely in ip6_rcv_core()) the skb->transport_header is set to skb->network_header + sizeof(*hdr). As a consequence, after routing operations, destination input expects to find skb->transport_header correctly set to the next protocol (or extension header) that follows the network protocol. However, decap behaviors (DX*, DT*) remove the outer IPv6 and SRH extension and do not set again the skb->transport_header pointer correctly. For this reason, the patch sets the skb->transport_header to the skb->network_header + sizeof(hdr) in each DX* and DT* behavior. Signed-off-by: Andrea Mayer Signed-off-by: David S. Miller --- net/ipv6/seg6_local.c | 6 ++++++ 1 file changed, 6 insertions(+) (limited to 'net') diff --git a/net/ipv6/seg6_local.c b/net/ipv6/seg6_local.c index 5e3d7004d431..e70567446f28 100644 --- a/net/ipv6/seg6_local.c +++ b/net/ipv6/seg6_local.c @@ -341,6 +341,8 @@ static int input_action_end_dx6(struct sk_buff *skb, if (!ipv6_addr_any(&slwt->nh6)) nhaddr = &slwt->nh6; + skb_set_transport_header(skb, sizeof(struct ipv6hdr)); + seg6_lookup_nexthop(skb, nhaddr, 0); return dst_input(skb); @@ -370,6 +372,8 @@ static int input_action_end_dx4(struct sk_buff *skb, skb_dst_drop(skb); + skb_set_transport_header(skb, sizeof(struct iphdr)); + err = ip_route_input(skb, nhaddr, iph->saddr, 0, skb->dev); if (err) goto drop; @@ -390,6 +394,8 @@ static int input_action_end_dt6(struct sk_buff *skb, if (!pskb_may_pull(skb, sizeof(struct ipv6hdr))) goto drop; + skb_set_transport_header(skb, sizeof(struct ipv6hdr)); + seg6_lookup_nexthop(skb, NULL, slwt->table); return dst_input(skb); -- cgit v1.2.3 From c80ed84e76886487703bf04b38ce10e92e2d6e26 Mon Sep 17 00:00:00 2001 From: Vladimir Oltean Date: Sat, 16 Nov 2019 18:08:25 +0200 Subject: net: dsa: tag_8021q: Fix dsa_8021q_restore_pvid for an absent pvid This sequence of operations: ip link set dev br0 type bridge vlan_filtering 1 bridge vlan del dev swp2 vid 1 ip link set dev br0 type bridge vlan_filtering 1 ip link set dev br0 type bridge vlan_filtering 0 apparently fails with the message: [ 31.305716] sja1105 spi0.1: Reset switch and programmed static config. Reason: VLAN filtering [ 31.322161] sja1105 spi0.1: Couldn't determine PVID attributes (pvid 0) [ 31.328939] sja1105 spi0.1: Failed to setup VLAN tagging for port 1: -2 [ 31.335599] ------------[ cut here ]------------ [ 31.340215] WARNING: CPU: 1 PID: 194 at net/switchdev/switchdev.c:157 switchdev_port_attr_set_now+0x9c/0xa4 [ 31.349981] br0: Commit of attribute (id=6) failed. [ 31.354890] Modules linked in: [ 31.357942] CPU: 1 PID: 194 Comm: ip Not tainted 5.4.0-rc6-01792-gf4f632e07665-dirty #2062 [ 31.366167] Hardware name: Freescale LS1021A [ 31.370437] [] (unwind_backtrace) from [] (show_stack+0x10/0x14) [ 31.378153] [] (show_stack) from [] (dump_stack+0xe0/0x10c) [ 31.385437] [] (dump_stack) from [] (__warn+0xf4/0x10c) [ 31.392373] [] (__warn) from [] (warn_slowpath_fmt+0x74/0xb8) [ 31.399827] [] (warn_slowpath_fmt) from [] (switchdev_port_attr_set_now+0x9c/0xa4) [ 31.409097] [] (switchdev_port_attr_set_now) from [] (__br_vlan_filter_toggle+0x6c/0x118) [ 31.418971] [] (__br_vlan_filter_toggle) from [] (br_changelink+0xf8/0x518) [ 31.427637] [] (br_changelink) from [] (__rtnl_newlink+0x3f4/0x76c) [ 31.435613] [] (__rtnl_newlink) from [] (rtnl_newlink+0x44/0x60) [ 31.443329] [] (rtnl_newlink) from [] (rtnetlink_rcv_msg+0x2cc/0x51c) [ 31.451477] [] (rtnetlink_rcv_msg) from [] (netlink_rcv_skb+0xb8/0x110) [ 31.459796] [] (netlink_rcv_skb) from [] (netlink_unicast+0x17c/0x1f8) [ 31.468026] [] (netlink_unicast) from [] (netlink_sendmsg+0x2bc/0x3b4) [ 31.476261] [] (netlink_sendmsg) from [] (___sys_sendmsg+0x230/0x250) [ 31.484408] [] (___sys_sendmsg) from [] (__sys_sendmsg+0x50/0x8c) [ 31.492209] [] (__sys_sendmsg) from [] (ret_fast_syscall+0x0/0x28) [ 31.500090] Exception stack(0xedf47fa8 to 0xedf47ff0) [ 31.505122] 7fa0: 00000002 b6f2e060 00000003 beabd6a4 00000000 00000000 [ 31.513265] 7fc0: 00000002 b6f2e060 5d6e3213 00000128 00000000 00000001 00000006 000619c4 [ 31.521405] 7fe0: 00086078 beabd658 0005edbc b6e7ce68 The reason is the implementation of br_get_pvid: static inline u16 br_get_pvid(const struct net_bridge_vlan_group *vg) { if (!vg) return 0; smp_rmb(); return vg->pvid; } Since VID 0 is an invalid pvid from the bridge's point of view, let's add this check in dsa_8021q_restore_pvid to avoid restoring a pvid that doesn't really exist. Fixes: 5f33183b7fdf ("net: dsa: tag_8021q: Restore bridge VLANs when enabling vlan_filtering") Signed-off-by: Vladimir Oltean Reviewed-by: Andrew Lunn Signed-off-by: David S. Miller --- net/dsa/tag_8021q.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'net') diff --git a/net/dsa/tag_8021q.c b/net/dsa/tag_8021q.c index 9c1cc2482b68..9e5a883a9f0c 100644 --- a/net/dsa/tag_8021q.c +++ b/net/dsa/tag_8021q.c @@ -106,7 +106,7 @@ static int dsa_8021q_restore_pvid(struct dsa_switch *ds, int port) slave = ds->ports[port].slave; err = br_vlan_get_pvid(slave, &pvid); - if (err < 0) + if (!pvid || err < 0) /* There is no pvid on the bridge for this port, which is * perfectly valid. Nothing to restore, bye-bye! */ -- cgit v1.2.3 From a36e629ee77a9df78310c972655092e41ce07ea4 Mon Sep 17 00:00:00 2001 From: Dag Moxnes Date: Fri, 15 Nov 2019 09:56:01 +0100 Subject: rds: ib: update WR sizes when bringing up connection Currently WR sizes are updated from rds_ib_sysctl_max_send_wr and rds_ib_sysctl_max_recv_wr when a connection is shut down. As a result, a connection being down while rds_ib_sysctl_max_send_wr or rds_ib_sysctl_max_recv_wr are updated, will not update the sizes when it comes back up. Move resizing of WRs to rds_ib_setup_qp so that connections will be setup with the most current WR sizes. Signed-off-by: Dag Moxnes Acked-by: Santosh Shilimkar Signed-off-by: David S. Miller --- net/rds/ib_cm.c | 23 +++++++++++++++-------- 1 file changed, 15 insertions(+), 8 deletions(-) (limited to 'net') diff --git a/net/rds/ib_cm.c b/net/rds/ib_cm.c index 233f1368162b..18c6fac6ead9 100644 --- a/net/rds/ib_cm.c +++ b/net/rds/ib_cm.c @@ -450,6 +450,7 @@ static int rds_ib_setup_qp(struct rds_connection *conn) struct ib_qp_init_attr attr; struct ib_cq_init_attr cq_attr = {}; struct rds_ib_device *rds_ibdev; + unsigned long max_wrs; int ret, fr_queue_space; /* @@ -469,10 +470,15 @@ static int rds_ib_setup_qp(struct rds_connection *conn) /* add the conn now so that connection establishment has the dev */ rds_ib_add_conn(rds_ibdev, conn); - if (rds_ibdev->max_wrs < ic->i_send_ring.w_nr + 1) - rds_ib_ring_resize(&ic->i_send_ring, rds_ibdev->max_wrs - 1); - if (rds_ibdev->max_wrs < ic->i_recv_ring.w_nr + 1) - rds_ib_ring_resize(&ic->i_recv_ring, rds_ibdev->max_wrs - 1); + max_wrs = rds_ibdev->max_wrs < rds_ib_sysctl_max_send_wr + 1 ? + rds_ibdev->max_wrs - 1 : rds_ib_sysctl_max_send_wr; + if (ic->i_send_ring.w_nr != max_wrs) + rds_ib_ring_resize(&ic->i_send_ring, max_wrs); + + max_wrs = rds_ibdev->max_wrs < rds_ib_sysctl_max_recv_wr + 1 ? + rds_ibdev->max_wrs - 1 : rds_ib_sysctl_max_recv_wr; + if (ic->i_recv_ring.w_nr != max_wrs) + rds_ib_ring_resize(&ic->i_recv_ring, max_wrs); /* Protection domain and memory range */ ic->i_pd = rds_ibdev->pd; @@ -1099,8 +1105,9 @@ void rds_ib_conn_path_shutdown(struct rds_conn_path *cp) ic->i_flowctl = 0; atomic_set(&ic->i_credits, 0); - rds_ib_ring_init(&ic->i_send_ring, rds_ib_sysctl_max_send_wr); - rds_ib_ring_init(&ic->i_recv_ring, rds_ib_sysctl_max_recv_wr); + /* Re-init rings, but retain sizes. */ + rds_ib_ring_init(&ic->i_send_ring, ic->i_send_ring.w_nr); + rds_ib_ring_init(&ic->i_recv_ring, ic->i_recv_ring.w_nr); if (ic->i_ibinc) { rds_inc_put(&ic->i_ibinc->ii_inc); @@ -1147,8 +1154,8 @@ int rds_ib_conn_alloc(struct rds_connection *conn, gfp_t gfp) * rds_ib_conn_shutdown() waits for these to be emptied so they * must be initialized before it can be called. */ - rds_ib_ring_init(&ic->i_send_ring, rds_ib_sysctl_max_send_wr); - rds_ib_ring_init(&ic->i_recv_ring, rds_ib_sysctl_max_recv_wr); + rds_ib_ring_init(&ic->i_send_ring, 0); + rds_ib_ring_init(&ic->i_recv_ring, 0); ic->conn = conn; conn->c_transport_data = ic; -- cgit v1.2.3 From 8204df72bea1a7d83d0777add6da98a41dfbdc34 Mon Sep 17 00:00:00 2001 From: Ursula Braun Date: Fri, 15 Nov 2019 12:39:30 +0100 Subject: net/smc: fix fastopen for non-blocking connect() FASTOPEN does not work with SMC-sockets. Since SMC allows fallback to TCP native during connection start, the FASTOPEN setsockopts trigger this fallback, if the SMC-socket is still in state SMC_INIT. But if a FASTOPEN setsockopt is called after a non-blocking connect(), this is broken, and fallback does not make sense. This change complements commit cd2063604ea6 ("net/smc: avoid fallback in case of non-blocking connect") and fixes the syzbot reported problem "WARNING in smc_unhash_sk". Reported-by: syzbot+8488cc4cf1c9e09b8b86@syzkaller.appspotmail.com Fixes: e1bbdd570474 ("net/smc: reduce sock_put() for fallback sockets") Signed-off-by: Ursula Braun Signed-off-by: Karsten Graul Signed-off-by: David S. Miller --- net/smc/af_smc.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'net') diff --git a/net/smc/af_smc.c b/net/smc/af_smc.c index 8edf1619f0e4..737b49909a7a 100644 --- a/net/smc/af_smc.c +++ b/net/smc/af_smc.c @@ -1732,7 +1732,7 @@ static int smc_setsockopt(struct socket *sock, int level, int optname, case TCP_FASTOPEN_KEY: case TCP_FASTOPEN_NO_COOKIE: /* option not supported by SMC */ - if (sk->sk_state == SMC_INIT) { + if (sk->sk_state == SMC_INIT && !smc->connect_nonblock) { smc_switch_to_fallback(smc); smc->fallback_rsn = SMC_CLC_DECL_OPTUNSUPP; } else { -- cgit v1.2.3 From 7901cd97963d6cbde88fa25a4a446db3554c16c6 Mon Sep 17 00:00:00 2001 From: Guillaume Nault Date: Fri, 15 Nov 2019 18:29:52 +0100 Subject: ipmr: Fix skb headroom in ipmr_get_route(). In route.c, inet_rtm_getroute_build_skb() creates an skb with no headroom. This skb is then used by inet_rtm_getroute() which may pass it to rt_fill_info() and, from there, to ipmr_get_route(). The later might try to reuse this skb by cloning it and prepending an IPv4 header. But since the original skb has no headroom, skb_push() triggers skb_under_panic(): skbuff: skb_under_panic: text:00000000ca46ad8a len:80 put:20 head:00000000cd28494e data:000000009366fd6b tail:0x3c end:0xec0 dev:veth0 ------------[ cut here ]------------ kernel BUG at net/core/skbuff.c:108! invalid opcode: 0000 [#1] SMP KASAN PTI CPU: 6 PID: 587 Comm: ip Not tainted 5.4.0-rc6+ #1 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.12.0-2.fc30 04/01/2014 RIP: 0010:skb_panic+0xbf/0xd0 Code: 41 a2 ff 8b 4b 70 4c 8b 4d d0 48 c7 c7 20 76 f5 8b 44 8b 45 bc 48 8b 55 c0 48 8b 75 c8 41 54 41 57 41 56 41 55 e8 75 dc 7a ff <0f> 0b 0f 1f 44 00 00 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 44 00 00 RSP: 0018:ffff888059ddf0b0 EFLAGS: 00010286 RAX: 0000000000000086 RBX: ffff888060a315c0 RCX: ffffffff8abe4822 RDX: 0000000000000000 RSI: 0000000000000008 RDI: ffff88806c9a79cc RBP: ffff888059ddf118 R08: ffffed100d9361b1 R09: ffffed100d9361b0 R10: ffff88805c68aee3 R11: ffffed100d9361b1 R12: ffff88805d218000 R13: ffff88805c689fec R14: 000000000000003c R15: 0000000000000ec0 FS: 00007f6af184b700(0000) GS:ffff88806c980000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 00007ffc8204a000 CR3: 0000000057b40006 CR4: 0000000000360ee0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 Call Trace: skb_push+0x7e/0x80 ipmr_get_route+0x459/0x6fa rt_fill_info+0x692/0x9f0 inet_rtm_getroute+0xd26/0xf20 rtnetlink_rcv_msg+0x45d/0x630 netlink_rcv_skb+0x1a5/0x220 rtnetlink_rcv+0x15/0x20 netlink_unicast+0x305/0x3a0 netlink_sendmsg+0x575/0x730 sock_sendmsg+0xb5/0xc0 ___sys_sendmsg+0x497/0x4f0 __sys_sendmsg+0xcb/0x150 __x64_sys_sendmsg+0x48/0x50 do_syscall_64+0xd2/0xac0 entry_SYSCALL_64_after_hwframe+0x49/0xbe Actually the original skb used to have enough headroom, but the reserve_skb() call was lost with the introduction of inet_rtm_getroute_build_skb() by commit 404eb77ea766 ("ipv4: support sport, dport and ip_proto in RTM_GETROUTE"). We could reserve some headroom again in inet_rtm_getroute_build_skb(), but this function shouldn't be responsible for handling the special case of ipmr_get_route(). Let's handle that directly in ipmr_get_route() by calling skb_realloc_headroom() instead of skb_clone(). Fixes: 404eb77ea766 ("ipv4: support sport, dport and ip_proto in RTM_GETROUTE") Signed-off-by: Guillaume Nault Signed-off-by: David S. Miller --- net/ipv4/ipmr.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'net') diff --git a/net/ipv4/ipmr.c b/net/ipv4/ipmr.c index 716d5472c022..58007439cffd 100644 --- a/net/ipv4/ipmr.c +++ b/net/ipv4/ipmr.c @@ -2289,7 +2289,8 @@ int ipmr_get_route(struct net *net, struct sk_buff *skb, rcu_read_unlock(); return -ENODEV; } - skb2 = skb_clone(skb, GFP_ATOMIC); + + skb2 = skb_realloc_headroom(skb, sizeof(struct iphdr)); if (!skb2) { read_unlock(&mrt_lock); rcu_read_unlock(); -- cgit v1.2.3