From 0a1c7a7b0adbf595ce7f218609db53749e966573 Mon Sep 17 00:00:00 2001 From: Johan Jonker Date: Mon, 4 Nov 2024 21:01:39 +0800 Subject: [PATCH 01/16] net: arc: rockchip: fix emac mdio node support The binding emac_rockchip.txt is converted to YAML. Changed against the original binding is an added MDIO subnode. This make the driver failed to find the PHY, and given the 'mdio has invalid PHY address' it is probably looking in the wrong node. Fix emac_mdio.c so that it can handle both old and new device trees. Fixes: 1dabb74971b3 ("ARM: dts: rockchip: restyle emac nodes") Signed-off-by: Johan Jonker Tested-by: Andy Yan Link: https://lore.kernel.org/r/20220603163539.537-3-jbx6244@gmail.com Signed-off-by: Andy Yan Reviewed-by: Andrew Lunn Signed-off-by: Paolo Abeni --- drivers/net/ethernet/arc/emac_mdio.c | 9 ++++++++- 1 file changed, 8 insertions(+), 1 deletion(-) diff --git a/drivers/net/ethernet/arc/emac_mdio.c b/drivers/net/ethernet/arc/emac_mdio.c index 87f40c2ba904..078b1a72c161 100644 --- a/drivers/net/ethernet/arc/emac_mdio.c +++ b/drivers/net/ethernet/arc/emac_mdio.c @@ -133,6 +133,7 @@ int arc_mdio_probe(struct arc_emac_priv *priv) struct arc_emac_mdio_bus_data *data = &priv->bus_data; struct device_node *np = priv->dev->of_node; const char *name = "Synopsys MII Bus"; + struct device_node *mdio_node; struct mii_bus *bus; int error; @@ -164,7 +165,13 @@ int arc_mdio_probe(struct arc_emac_priv *priv) snprintf(bus->id, MII_BUS_ID_SIZE, "%s", bus->name); - error = of_mdiobus_register(bus, priv->dev->of_node); + /* Backwards compatibility for EMAC nodes without MDIO subnode. */ + mdio_node = of_get_child_by_name(np, "mdio"); + if (!mdio_node) + mdio_node = of_node_get(np); + + error = of_mdiobus_register(bus, mdio_node); + of_node_put(mdio_node); if (error) { mdiobus_free(bus); return dev_err_probe(priv->dev, error, -- 2.51.0 From 84b9749a3a704dcc824a88aa8267247c801d51e4 Mon Sep 17 00:00:00 2001 From: David Wang <00107082@163.com> Date: Wed, 6 Nov 2024 10:12:28 +0800 Subject: [PATCH 02/16] proc/softirqs: replace seq_printf with seq_put_decimal_ull_width seq_printf is costy, on a system with n CPUs, reading /proc/softirqs would yield 10*n decimal values, and the extra cost parsing format string grows linearly with number of cpus. Replace seq_printf with seq_put_decimal_ull_width have significant performance improvement. On an 8CPUs system, reading /proc/softirqs show ~40% performance gain with this patch. Signed-off-by: David Wang <00107082@163.com> Signed-off-by: Linus Torvalds --- fs/proc/softirqs.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/fs/proc/softirqs.c b/fs/proc/softirqs.c index f4616083faef..04bb29721419 100644 --- a/fs/proc/softirqs.c +++ b/fs/proc/softirqs.c @@ -20,7 +20,7 @@ static int show_softirqs(struct seq_file *p, void *v) for (i = 0; i < NR_SOFTIRQS; i++) { seq_printf(p, "%12s:", softirq_to_name[i]); for_each_possible_cpu(j) - seq_printf(p, " %10u", kstat_softirqs_cpu(i, j)); + seq_put_decimal_ull_width(p, " ", kstat_softirqs_cpu(i, j), 10); seq_putc(p, '\n'); } return 0; -- 2.51.0 From de88df01796b309903b70888fbdf2b89607e3a6a Mon Sep 17 00:00:00 2001 From: Wenjia Zhang Date: Wed, 6 Nov 2024 09:26:12 +0100 Subject: [PATCH 03/16] net/smc: Fix lookup of netdev by using ib_device_get_netdev() The SMC-R variant of the SMC protocol used direct call to function ib_device_ops.get_netdev() to lookup netdev. As we used mlx5 device driver to run SMC-R, it failed to find a device, because in mlx5_ib the internal net device management for retrieving net devices was replaced by a common interface ib_device_get_netdev() in commit 8d159eb2117b ("RDMA/mlx5: Use IB set_netdev and get_netdev functions"). Since such direct accesses to the internal net device management is not recommended at all, update the SMC-R code to use proper API ib_device_get_netdev(). Fixes: 54903572c23c ("net/smc: allow pnetid-less configuration") Reported-by: Aswin K Reviewed-by: Gerd Bayer Reviewed-by: Halil Pasic Reviewed-by: Simon Horman Reviewed-by: Dust Li Reviewed-by: Wen Gu Reviewed-by: Zhu Yanjun Reviewed-by: D. Wythe Signed-off-by: Wenjia Zhang Reviewed-by: Leon Romanovsky Link: https://patch.msgid.link/20241106082612.57803-1-wenjia@linux.ibm.com Signed-off-by: Jakub Kicinski --- net/smc/smc_ib.c | 8 ++------ net/smc/smc_pnet.c | 4 +--- 2 files changed, 3 insertions(+), 9 deletions(-) diff --git a/net/smc/smc_ib.c b/net/smc/smc_ib.c index 9297dc20bfe2..9c563cdbea90 100644 --- a/net/smc/smc_ib.c +++ b/net/smc/smc_ib.c @@ -899,9 +899,7 @@ static void smc_copy_netdev_ifindex(struct smc_ib_device *smcibdev, int port) struct ib_device *ibdev = smcibdev->ibdev; struct net_device *ndev; - if (!ibdev->ops.get_netdev) - return; - ndev = ibdev->ops.get_netdev(ibdev, port + 1); + ndev = ib_device_get_netdev(ibdev, port + 1); if (ndev) { smcibdev->ndev_ifidx[port] = ndev->ifindex; dev_put(ndev); @@ -921,9 +919,7 @@ void smc_ib_ndev_change(struct net_device *ndev, unsigned long event) port_cnt = smcibdev->ibdev->phys_port_cnt; for (i = 0; i < min_t(size_t, port_cnt, SMC_MAX_PORTS); i++) { libdev = smcibdev->ibdev; - if (!libdev->ops.get_netdev) - continue; - lndev = libdev->ops.get_netdev(libdev, i + 1); + lndev = ib_device_get_netdev(libdev, i + 1); dev_put(lndev); if (lndev != ndev) continue; diff --git a/net/smc/smc_pnet.c b/net/smc/smc_pnet.c index a04aa0e882f8..716808f374a8 100644 --- a/net/smc/smc_pnet.c +++ b/net/smc/smc_pnet.c @@ -1054,9 +1054,7 @@ static void smc_pnet_find_rdma_dev(struct net_device *netdev, for (i = 1; i <= SMC_MAX_PORTS; i++) { if (!rdma_is_port_valid(ibdev->ibdev, i)) continue; - if (!ibdev->ibdev->ops.get_netdev) - continue; - ndev = ibdev->ibdev->ops.get_netdev(ibdev->ibdev, i); + ndev = ib_device_get_netdev(ibdev->ibdev, i); if (!ndev) continue; dev_put(ndev); -- 2.51.0 From fc9de52de38f656399d2ce40f7349a6b5f86e787 Mon Sep 17 00:00:00 2001 From: David Howells Date: Wed, 6 Nov 2024 13:03:22 +0000 Subject: [PATCH 04/16] rxrpc: Fix missing locking causing hanging calls If a call gets aborted (e.g. because kafs saw a signal) between it being queued for connection and the I/O thread picking up the call, the abort will be prioritised over the connection and it will be removed from local->new_client_calls by rxrpc_disconnect_client_call() without a lock being held. This may cause other calls on the list to disappear if a race occurs. Fix this by taking the client_call_lock when removing a call from whatever list its ->wait_link happens to be on. Signed-off-by: David Howells cc: linux-afs@lists.infradead.org Reported-by: Marc Dionne Fixes: 9d35d880e0e4 ("rxrpc: Move client call connection to the I/O thread") Link: https://patch.msgid.link/726660.1730898202@warthog.procyon.org.uk Signed-off-by: Jakub Kicinski --- include/trace/events/rxrpc.h | 1 + net/rxrpc/conn_client.c | 4 ++++ 2 files changed, 5 insertions(+) diff --git a/include/trace/events/rxrpc.h b/include/trace/events/rxrpc.h index a1b126a6b0d7..cc22596c7250 100644 --- a/include/trace/events/rxrpc.h +++ b/include/trace/events/rxrpc.h @@ -287,6 +287,7 @@ EM(rxrpc_call_see_input, "SEE input ") \ EM(rxrpc_call_see_release, "SEE release ") \ EM(rxrpc_call_see_userid_exists, "SEE u-exists") \ + EM(rxrpc_call_see_waiting_call, "SEE q-conn ") \ E_(rxrpc_call_see_zap, "SEE zap ") #define rxrpc_txqueue_traces \ diff --git a/net/rxrpc/conn_client.c b/net/rxrpc/conn_client.c index d25bf1cf3670..bb11e8289d6d 100644 --- a/net/rxrpc/conn_client.c +++ b/net/rxrpc/conn_client.c @@ -516,6 +516,7 @@ void rxrpc_connect_client_calls(struct rxrpc_local *local) spin_lock(&local->client_call_lock); list_move_tail(&call->wait_link, &bundle->waiting_calls); + rxrpc_see_call(call, rxrpc_call_see_waiting_call); spin_unlock(&local->client_call_lock); if (rxrpc_bundle_has_space(bundle)) @@ -586,7 +587,10 @@ void rxrpc_disconnect_client_call(struct rxrpc_bundle *bundle, struct rxrpc_call _debug("call is waiting"); ASSERTCMP(call->call_id, ==, 0); ASSERT(!test_bit(RXRPC_CALL_EXPOSED, &call->flags)); + /* May still be on ->new_client_calls. */ + spin_lock(&local->client_call_lock); list_del_init(&call->wait_link); + spin_unlock(&local->client_call_lock); return; } -- 2.51.0 From d293958a8595ba566fb90b99da4d6263e14fee15 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Wed, 6 Nov 2024 22:19:22 +0000 Subject: [PATCH 05/16] net/smc: do not leave a dangling sk pointer in __smc_create() Thanks to commit 4bbd360a5084 ("socket: Print pf->create() when it does not clear sock->sk on failure."), syzbot found an issue with AF_SMC: smc_create must clear sock->sk on failure, family: 43, type: 1, protocol: 0 WARNING: CPU: 0 PID: 5827 at net/socket.c:1565 __sock_create+0x96f/0xa30 net/socket.c:1563 Modules linked in: CPU: 0 UID: 0 PID: 5827 Comm: syz-executor259 Not tainted 6.12.0-rc6-next-20241106-syzkaller #0 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 09/13/2024 RIP: 0010:__sock_create+0x96f/0xa30 net/socket.c:1563 Code: 03 00 74 08 4c 89 e7 e8 4f 3b 85 f8 49 8b 34 24 48 c7 c7 40 89 0c 8d 8b 54 24 04 8b 4c 24 0c 44 8b 44 24 08 e8 32 78 db f7 90 <0f> 0b 90 90 e9 d3 fd ff ff 89 e9 80 e1 07 fe c1 38 c1 0f 8c ee f7 RSP: 0018:ffffc90003e4fda0 EFLAGS: 00010246 RAX: 099c6f938c7f4700 RBX: 1ffffffff1a595fd RCX: ffff888034823c00 RDX: 0000000000000000 RSI: 0000000000000000 RDI: 0000000000000000 RBP: 00000000ffffffe9 R08: ffffffff81567052 R09: 1ffff920007c9f50 R10: dffffc0000000000 R11: fffff520007c9f51 R12: ffffffff8d2cafe8 R13: 1ffffffff1a595fe R14: ffffffff9a789c40 R15: ffff8880764298c0 FS: 000055557b518380(0000) GS:ffff8880b8600000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 00007fa62ff43225 CR3: 0000000031628000 CR4: 00000000003526f0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 Call Trace: sock_create net/socket.c:1616 [inline] __sys_socket_create net/socket.c:1653 [inline] __sys_socket+0x150/0x3c0 net/socket.c:1700 __do_sys_socket net/socket.c:1714 [inline] __se_sys_socket net/socket.c:1712 [inline] For reference, see commit 2d859aff775d ("Merge branch 'do-not-leave-dangling-sk-pointers-in-pf-create-functions'") Fixes: d25a92ccae6b ("net/smc: Introduce IPPROTO_SMC") Signed-off-by: Eric Dumazet Cc: Ignat Korchagin Cc: D. Wythe Cc: Dust Li Reviewed-by: Kuniyuki Iwashima Reviewed-by: Wenjia Zhang Link: https://patch.msgid.link/20241106221922.1544045-1-edumazet@google.com Signed-off-by: Jakub Kicinski --- net/smc/af_smc.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/net/smc/af_smc.c b/net/smc/af_smc.c index 0316217b7687..9d76e902fd77 100644 --- a/net/smc/af_smc.c +++ b/net/smc/af_smc.c @@ -3359,8 +3359,10 @@ static int __smc_create(struct net *net, struct socket *sock, int protocol, else rc = smc_create_clcsk(net, sk, family); - if (rc) + if (rc) { sk_common_release(sk); + sock->sk = NULL; + } out: return rc; } -- 2.51.0 From 71712cf519faeed529549a79559c06c7fc250a15 Mon Sep 17 00:00:00 2001 From: Wentao Liang Date: Thu, 7 Nov 2024 10:17:56 +0800 Subject: [PATCH 06/16] drivers: net: ionic: add missed debugfs cleanup to ionic_probe() error path The ionic_setup_one() creates a debugfs entry for ionic upon successful execution. However, the ionic_probe() does not release the dentry before returning, resulting in a memory leak. To fix this bug, we add the ionic_debugfs_del_dev() to release the resources in a timely manner before returning. Fixes: 0de38d9f1dba ("ionic: extract common bits from ionic_probe") Signed-off-by: Wentao Liang Acked-by: Shannon Nelson Link: https://patch.msgid.link/20241107021756.1677-1-liangwentao@iscas.ac.cn Signed-off-by: Jakub Kicinski --- drivers/net/ethernet/pensando/ionic/ionic_bus_pci.c | 1 + 1 file changed, 1 insertion(+) diff --git a/drivers/net/ethernet/pensando/ionic/ionic_bus_pci.c b/drivers/net/ethernet/pensando/ionic/ionic_bus_pci.c index b93791d6b593..f5dc876eb500 100644 --- a/drivers/net/ethernet/pensando/ionic/ionic_bus_pci.c +++ b/drivers/net/ethernet/pensando/ionic/ionic_bus_pci.c @@ -394,6 +394,7 @@ err_out_free_irqs: err_out_pci: ionic_dev_teardown(ionic); ionic_clear_pci(ionic); + ionic_debugfs_del_dev(ionic); err_out: mutex_destroy(&ionic->dev_cmd_lock); ionic_devlink_free(ionic); -- 2.51.0 From 1904fb9ebf911441f90a68e96b22aa73e4410505 Mon Sep 17 00:00:00 2001 From: Jakub Kicinski Date: Tue, 5 Nov 2024 17:52:34 -0800 Subject: [PATCH 07/16] netlink: terminate outstanding dump on socket close Netlink supports iterative dumping of data. It provides the families the following ops: - start - (optional) kicks off the dumping process - dump - actual dump helper, keeps getting called until it returns 0 - done - (optional) pairs with .start, can be used for cleanup The whole process is asynchronous and the repeated calls to .dump don't actually happen in a tight loop, but rather are triggered in response to recvmsg() on the socket. This gives the user full control over the dump, but also means that the user can close the socket without getting to the end of the dump. To make sure .start is always paired with .done we check if there is an ongoing dump before freeing the socket, and if so call .done. The complication is that sockets can get freed from BH and .done is allowed to sleep. So we use a workqueue to defer the call, when needed. Unfortunately this does not work correctly. What we defer is not the cleanup but rather releasing a reference on the socket. We have no guarantee that we own the last reference, if someone else holds the socket they may release it in BH and we're back to square one. The whole dance, however, appears to be unnecessary. Only the user can interact with dumps, so we can clean up when socket is closed. And close always happens in process context. Some async code may still access the socket after close, queue notification skbs to it etc. but no dumps can start, end or otherwise make progress. Delete the workqueue and flush the dump state directly from the release handler. Note that further cleanup is possible in -next, for instance we now always call .done before releasing the main module reference, so dump doesn't have to take a reference of its own. Reported-by: syzkaller Fixes: ed5d7788a934 ("netlink: Do not schedule work from sk_destruct") Reviewed-by: Kuniyuki Iwashima Reviewed-by: Eric Dumazet Link: https://patch.msgid.link/20241106015235.2458807-1-kuba@kernel.org Signed-off-by: Jakub Kicinski --- net/netlink/af_netlink.c | 31 ++++++++----------------------- net/netlink/af_netlink.h | 2 -- 2 files changed, 8 insertions(+), 25 deletions(-) diff --git a/net/netlink/af_netlink.c b/net/netlink/af_netlink.c index 0a9287fadb47..f84aad420d44 100644 --- a/net/netlink/af_netlink.c +++ b/net/netlink/af_netlink.c @@ -393,15 +393,6 @@ static void netlink_skb_set_owner_r(struct sk_buff *skb, struct sock *sk) static void netlink_sock_destruct(struct sock *sk) { - struct netlink_sock *nlk = nlk_sk(sk); - - if (nlk->cb_running) { - if (nlk->cb.done) - nlk->cb.done(&nlk->cb); - module_put(nlk->cb.module); - kfree_skb(nlk->cb.skb); - } - skb_queue_purge(&sk->sk_receive_queue); if (!sock_flag(sk, SOCK_DEAD)) { @@ -414,14 +405,6 @@ static void netlink_sock_destruct(struct sock *sk) WARN_ON(nlk_sk(sk)->groups); } -static void netlink_sock_destruct_work(struct work_struct *work) -{ - struct netlink_sock *nlk = container_of(work, struct netlink_sock, - work); - - sk_free(&nlk->sk); -} - /* This lock without WQ_FLAG_EXCLUSIVE is good on UP and it is _very_ bad on * SMP. Look, when several writers sleep and reader wakes them up, all but one * immediately hit write lock and grab all the cpus. Exclusive sleep solves @@ -731,12 +714,6 @@ static void deferred_put_nlk_sk(struct rcu_head *head) if (!refcount_dec_and_test(&sk->sk_refcnt)) return; - if (nlk->cb_running && nlk->cb.done) { - INIT_WORK(&nlk->work, netlink_sock_destruct_work); - schedule_work(&nlk->work); - return; - } - sk_free(sk); } @@ -788,6 +765,14 @@ static int netlink_release(struct socket *sock) NETLINK_URELEASE, &n); } + /* Terminate any outstanding dump */ + if (nlk->cb_running) { + if (nlk->cb.done) + nlk->cb.done(&nlk->cb); + module_put(nlk->cb.module); + kfree_skb(nlk->cb.skb); + } + module_put(nlk->module); if (netlink_is_kernel(sk)) { diff --git a/net/netlink/af_netlink.h b/net/netlink/af_netlink.h index 5b0e4e62ab8b..778a3809361f 100644 --- a/net/netlink/af_netlink.h +++ b/net/netlink/af_netlink.h @@ -4,7 +4,6 @@ #include #include -#include #include /* flags */ @@ -50,7 +49,6 @@ struct netlink_sock { struct rhash_head node; struct rcu_head rcu; - struct work_struct work; }; static inline struct netlink_sock *nlk_sk(struct sock *sk) -- 2.51.0 From 55d42a0c3f9ccd07c199e0ddbe1ba87572d30074 Mon Sep 17 00:00:00 2001 From: Jakub Kicinski Date: Tue, 5 Nov 2024 17:52:35 -0800 Subject: [PATCH 08/16] selftests: net: add a test for closing a netlink socket ith dump in progress Close a socket with dump in progress. We need a dump which generates enough info not to fit into a single skb. Policy dump fits the bill. Use the trick discovered by syzbot for keeping a ref on the socket longer than just close, with mqueue. TAP version 13 1..3 # Starting 3 tests from 1 test cases. # RUN global.test_sanity ... # OK global.test_sanity ok 1 global.test_sanity # RUN global.close_in_progress ... # OK global.close_in_progress ok 2 global.close_in_progress # RUN global.close_with_ref ... # OK global.close_with_ref ok 3 global.close_with_ref # PASSED: 3 / 3 tests passed. # Totals: pass:3 fail:0 xfail:0 xpass:0 skip:0 error:0 Note that this test is not expected to fail but rather crash the kernel if we get the cleanup wrong. Reviewed-by: Kuniyuki Iwashima Link: https://patch.msgid.link/20241106015235.2458807-2-kuba@kernel.org Signed-off-by: Jakub Kicinski --- tools/testing/selftests/net/Makefile | 1 + tools/testing/selftests/net/netlink-dumps.c | 110 ++++++++++++++++++++ 2 files changed, 111 insertions(+) create mode 100644 tools/testing/selftests/net/netlink-dumps.c diff --git a/tools/testing/selftests/net/Makefile b/tools/testing/selftests/net/Makefile index 649f1fe0dc46..5e86f7a51b43 100644 --- a/tools/testing/selftests/net/Makefile +++ b/tools/testing/selftests/net/Makefile @@ -78,6 +78,7 @@ TEST_PROGS += test_vxlan_vnifiltering.sh TEST_GEN_FILES += io_uring_zerocopy_tx TEST_PROGS += io_uring_zerocopy_tx.sh TEST_GEN_FILES += bind_bhash +TEST_GEN_PROGS += netlink-dumps TEST_GEN_PROGS += sk_bind_sendto_listen TEST_GEN_PROGS += sk_connect_zero_addr TEST_GEN_PROGS += sk_so_peek_off diff --git a/tools/testing/selftests/net/netlink-dumps.c b/tools/testing/selftests/net/netlink-dumps.c new file mode 100644 index 000000000000..7ee6dcd334df --- /dev/null +++ b/tools/testing/selftests/net/netlink-dumps.c @@ -0,0 +1,110 @@ +// SPDX-License-Identifier: GPL-2.0 + +#define _GNU_SOURCE + +#include +#include +#include +#include +#include +#include +#include +#include + +#include +#include +#include + +#include "../kselftest_harness.h" + +static const struct { + struct nlmsghdr nlhdr; + struct genlmsghdr genlhdr; + struct nlattr ahdr; + __u16 val; + __u16 pad; +} dump_policies = { + .nlhdr = { + .nlmsg_len = sizeof(dump_policies), + .nlmsg_type = GENL_ID_CTRL, + .nlmsg_flags = NLM_F_REQUEST | NLM_F_ACK | NLM_F_DUMP, + .nlmsg_seq = 1, + }, + .genlhdr = { + .cmd = CTRL_CMD_GETPOLICY, + .version = 2, + }, + .ahdr = { + .nla_len = 6, + .nla_type = CTRL_ATTR_FAMILY_ID, + }, + .val = GENL_ID_CTRL, + .pad = 0, +}; + +// Sanity check for the test itself, make sure the dump doesn't fit in one msg +TEST(test_sanity) +{ + int netlink_sock; + char buf[8192]; + ssize_t n; + + netlink_sock = socket(AF_NETLINK, SOCK_RAW, NETLINK_GENERIC); + ASSERT_GE(netlink_sock, 0); + + n = send(netlink_sock, &dump_policies, sizeof(dump_policies), 0); + ASSERT_EQ(n, sizeof(dump_policies)); + + n = recv(netlink_sock, buf, sizeof(buf), MSG_DONTWAIT); + ASSERT_GE(n, sizeof(struct nlmsghdr)); + + n = recv(netlink_sock, buf, sizeof(buf), MSG_DONTWAIT); + ASSERT_GE(n, sizeof(struct nlmsghdr)); + + close(netlink_sock); +} + +TEST(close_in_progress) +{ + int netlink_sock; + ssize_t n; + + netlink_sock = socket(AF_NETLINK, SOCK_RAW, NETLINK_GENERIC); + ASSERT_GE(netlink_sock, 0); + + n = send(netlink_sock, &dump_policies, sizeof(dump_policies), 0); + ASSERT_EQ(n, sizeof(dump_policies)); + + close(netlink_sock); +} + +TEST(close_with_ref) +{ + char cookie[NOTIFY_COOKIE_LEN] = {}; + int netlink_sock, mq_fd; + struct sigevent sigev; + ssize_t n; + + netlink_sock = socket(AF_NETLINK, SOCK_RAW, NETLINK_GENERIC); + ASSERT_GE(netlink_sock, 0); + + n = send(netlink_sock, &dump_policies, sizeof(dump_policies), 0); + ASSERT_EQ(n, sizeof(dump_policies)); + + mq_fd = syscall(__NR_mq_open, "sed", O_CREAT | O_WRONLY, 0600, 0); + ASSERT_GE(mq_fd, 0); + + memset(&sigev, 0, sizeof(sigev)); + sigev.sigev_notify = SIGEV_THREAD; + sigev.sigev_value.sival_ptr = cookie; + sigev.sigev_signo = netlink_sock; + + syscall(__NR_mq_notify, mq_fd, &sigev); + + close(netlink_sock); + + // give mqueue time to fire + usleep(100 * 1000); +} + +TEST_HARNESS_MAIN -- 2.51.0 From eb72e7fcc83987d5d5595b43222f23b295d5de7f Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Thu, 7 Nov 2024 19:20:21 +0000 Subject: [PATCH 09/16] sctp: fix possible UAF in sctp_v6_available() A lockdep report [1] with CONFIG_PROVE_RCU_LIST=y hints that sctp_v6_available() is calling dev_get_by_index_rcu() and ipv6_chk_addr() without holding rcu. [1] ============================= WARNING: suspicious RCU usage 6.12.0-rc5-virtme #1216 Tainted: G W ----------------------------- net/core/dev.c:876 RCU-list traversed in non-reader section!! other info that might help us debug this: rcu_scheduler_active = 2, debug_locks = 1 1 lock held by sctp_hello/31495: #0: ffff9f1ebbdb7418 (sk_lock-AF_INET6){+.+.}-{0:0}, at: sctp_bind (./arch/x86/include/asm/jump_label.h:27 net/sctp/socket.c:315) sctp stack backtrace: CPU: 7 UID: 0 PID: 31495 Comm: sctp_hello Tainted: G W 6.12.0-rc5-virtme #1216 Tainted: [W]=WARN Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.16.3-debian-1.16.3-2 04/01/2014 Call Trace: dump_stack_lvl (lib/dump_stack.c:123) lockdep_rcu_suspicious (kernel/locking/lockdep.c:6822) dev_get_by_index_rcu (net/core/dev.c:876 (discriminator 7)) sctp_v6_available (net/sctp/ipv6.c:701) sctp sctp_do_bind (net/sctp/socket.c:400 (discriminator 1)) sctp sctp_bind (net/sctp/socket.c:320) sctp inet6_bind_sk (net/ipv6/af_inet6.c:465) ? security_socket_bind (security/security.c:4581 (discriminator 1)) __sys_bind (net/socket.c:1848 net/socket.c:1869) ? do_user_addr_fault (./include/linux/rcupdate.h:347 ./include/linux/rcupdate.h:880 ./include/linux/mm.h:729 arch/x86/mm/fault.c:1340) ? do_user_addr_fault (./arch/x86/include/asm/preempt.h:84 (discriminator 13) ./include/linux/rcupdate.h:98 (discriminator 13) ./include/linux/rcupdate.h:882 (discriminator 13) ./include/linux/mm.h:729 (discriminator 13) arch/x86/mm/fault.c:1340 (discriminator 13)) __x64_sys_bind (net/socket.c:1877 (discriminator 1) net/socket.c:1875 (discriminator 1) net/socket.c:1875 (discriminator 1)) do_syscall_64 (arch/x86/entry/common.c:52 (discriminator 1) arch/x86/entry/common.c:83 (discriminator 1)) entry_SYSCALL_64_after_hwframe (arch/x86/entry/entry_64.S:130) RIP: 0033:0x7f59b934a1e7 Code: 44 00 00 48 8b 15 39 8c 0c 00 f7 d8 64 89 02 b8 ff ff ff ff eb bd 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 00 b8 31 00 00 00 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 8b 0d 09 8c 0c 00 f7 d8 64 89 01 48 All code ======== 0: 44 00 00 add %r8b,(%rax) 3: 48 8b 15 39 8c 0c 00 mov 0xc8c39(%rip),%rdx # 0xc8c43 a: f7 d8 neg %eax c: 64 89 02 mov %eax,%fs:(%rdx) f: b8 ff ff ff ff mov $0xffffffff,%eax 14: eb bd jmp 0xffffffffffffffd3 16: 66 2e 0f 1f 84 00 00 cs nopw 0x0(%rax,%rax,1) 1d: 00 00 00 20: 0f 1f 00 nopl (%rax) 23: b8 31 00 00 00 mov $0x31,%eax 28: 0f 05 syscall 2a:* 48 3d 01 f0 ff ff cmp $0xfffffffffffff001,%rax <-- trapping instruction 30: 73 01 jae 0x33 32: c3 ret 33: 48 8b 0d 09 8c 0c 00 mov 0xc8c09(%rip),%rcx # 0xc8c43 3a: f7 d8 neg %eax 3c: 64 89 01 mov %eax,%fs:(%rcx) 3f: 48 rex.W Code starting with the faulting instruction =========================================== 0: 48 3d 01 f0 ff ff cmp $0xfffffffffffff001,%rax 6: 73 01 jae 0x9 8: c3 ret 9: 48 8b 0d 09 8c 0c 00 mov 0xc8c09(%rip),%rcx # 0xc8c19 10: f7 d8 neg %eax 12: 64 89 01 mov %eax,%fs:(%rcx) 15: 48 rex.W RSP: 002b:00007ffe2d0ad398 EFLAGS: 00000202 ORIG_RAX: 0000000000000031 RAX: ffffffffffffffda RBX: 00007ffe2d0ad3d0 RCX: 00007f59b934a1e7 RDX: 000000000000001c RSI: 00007ffe2d0ad3d0 RDI: 0000000000000005 RBP: 0000000000000005 R08: 1999999999999999 R09: 0000000000000000 R10: 00007f59b9253298 R11: 0000000000000202 R12: 00007ffe2d0ada61 R13: 0000000000000000 R14: 0000562926516dd8 R15: 00007f59b9479000 Fixes: 6fe1e52490a9 ("sctp: check ipv6 addr with sk_bound_dev if set") Signed-off-by: Eric Dumazet Cc: Marcelo Ricardo Leitner Acked-by: Xin Long Link: https://patch.msgid.link/20241107192021.2579789-1-edumazet@google.com Signed-off-by: Jakub Kicinski --- net/sctp/ipv6.c | 19 +++++++++++++------ 1 file changed, 13 insertions(+), 6 deletions(-) diff --git a/net/sctp/ipv6.c b/net/sctp/ipv6.c index f7b809c0d142..38e2fbdcbeac 100644 --- a/net/sctp/ipv6.c +++ b/net/sctp/ipv6.c @@ -683,7 +683,7 @@ static int sctp_v6_available(union sctp_addr *addr, struct sctp_sock *sp) struct sock *sk = &sp->inet.sk; struct net *net = sock_net(sk); struct net_device *dev = NULL; - int type; + int type, res, bound_dev_if; type = ipv6_addr_type(in6); if (IPV6_ADDR_ANY == type) @@ -697,14 +697,21 @@ static int sctp_v6_available(union sctp_addr *addr, struct sctp_sock *sp) if (!(type & IPV6_ADDR_UNICAST)) return 0; - if (sk->sk_bound_dev_if) { - dev = dev_get_by_index_rcu(net, sk->sk_bound_dev_if); + rcu_read_lock(); + bound_dev_if = READ_ONCE(sk->sk_bound_dev_if); + if (bound_dev_if) { + res = 0; + dev = dev_get_by_index_rcu(net, bound_dev_if); if (!dev) - return 0; + goto out; } - return ipv6_can_nonlocal_bind(net, &sp->inet) || - ipv6_chk_addr(net, in6, dev, 0); + res = ipv6_can_nonlocal_bind(net, &sp->inet) || + ipv6_chk_addr(net, in6, dev, 0); + +out: + rcu_read_unlock(); + return res; } /* This function checks if the address is a valid address to be used for -- 2.51.0 From e68da664d379f352d41d7955712c44e0a738e4ab Mon Sep 17 00:00:00 2001 From: Stefan Wahren Date: Fri, 8 Nov 2024 12:43:43 +0100 Subject: [PATCH 10/16] net: vertexcom: mse102x: Fix tx_bytes calculation The tx_bytes should consider the actual size of the Ethernet frames without the SPI encapsulation. But we still need to take care of Ethernet padding. Fixes: 2f207cbf0dd4 ("net: vertexcom: Add MSE102x SPI support") Signed-off-by: Stefan Wahren Link: https://patch.msgid.link/20241108114343.6174-3-wahrenst@gmx.net Signed-off-by: Jakub Kicinski --- drivers/net/ethernet/vertexcom/mse102x.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/drivers/net/ethernet/vertexcom/mse102x.c b/drivers/net/ethernet/vertexcom/mse102x.c index 2c37957478fb..89dc4c401a8d 100644 --- a/drivers/net/ethernet/vertexcom/mse102x.c +++ b/drivers/net/ethernet/vertexcom/mse102x.c @@ -437,13 +437,15 @@ static void mse102x_tx_work(struct work_struct *work) mse = &mses->mse102x; while ((txb = skb_dequeue(&mse->txq))) { + unsigned int len = max_t(unsigned int, txb->len, ETH_ZLEN); + mutex_lock(&mses->lock); ret = mse102x_tx_pkt_spi(mse, txb, work_timeout); mutex_unlock(&mses->lock); if (ret) { mse->ndev->stats.tx_dropped++; } else { - mse->ndev->stats.tx_bytes += txb->len; + mse->ndev->stats.tx_bytes += len; mse->ndev->stats.tx_packets++; } -- 2.51.0 From 252e01e68241d33bfe0ed1fc333220d9bd8b06df Mon Sep 17 00:00:00 2001 From: Jakub Kicinski Date: Thu, 7 Nov 2024 16:47:31 -0800 Subject: [PATCH 11/16] selftests: net: add netlink-dumps to .gitignore Commit 55d42a0c3f9c ("selftests: net: add a test for closing a netlink socket ith dump in progress") added a new test but did not add it to gitignore. Reviewed-by: Joe Damato Link: https://patch.msgid.link/20241108004731.2979878-1-kuba@kernel.org Signed-off-by: Jakub Kicinski --- tools/testing/selftests/net/.gitignore | 1 + 1 file changed, 1 insertion(+) diff --git a/tools/testing/selftests/net/.gitignore b/tools/testing/selftests/net/.gitignore index 217d8b7a7365..59fe07ee2df9 100644 --- a/tools/testing/selftests/net/.gitignore +++ b/tools/testing/selftests/net/.gitignore @@ -19,6 +19,7 @@ log.txt msg_oob msg_zerocopy ncdevmem +netlink-dumps nettest psock_fanout psock_snd -- 2.51.0 From 073d89808c065ac4c672c0a613a71b27a80691cb Mon Sep 17 00:00:00 2001 From: Wang Liang Date: Thu, 7 Nov 2024 10:34:05 +0800 Subject: [PATCH 12/16] net: fix data-races around sk->sk_forward_alloc Syzkaller reported this warning: ------------[ cut here ]------------ WARNING: CPU: 0 PID: 16 at net/ipv4/af_inet.c:156 inet_sock_destruct+0x1c5/0x1e0 Modules linked in: CPU: 0 UID: 0 PID: 16 Comm: ksoftirqd/0 Not tainted 6.12.0-rc5 #26 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014 RIP: 0010:inet_sock_destruct+0x1c5/0x1e0 Code: 24 12 4c 89 e2 5b 48 c7 c7 98 ec bb 82 41 5c e9 d1 18 17 ff 4c 89 e6 5b 48 c7 c7 d0 ec bb 82 41 5c e9 bf 18 17 ff 0f 0b eb 83 <0f> 0b eb 97 0f 0b eb 87 0f 0b e9 68 ff ff ff 66 66 2e 0f 1f 84 00 RSP: 0018:ffffc9000008bd90 EFLAGS: 00010206 RAX: 0000000000000300 RBX: ffff88810b172a90 RCX: 0000000000000007 RDX: 0000000000000002 RSI: 0000000000000300 RDI: ffff88810b172a00 RBP: ffff88810b172a00 R08: ffff888104273c00 R09: 0000000000100007 R10: 0000000000020000 R11: 0000000000000006 R12: ffff88810b172a00 R13: 0000000000000004 R14: 0000000000000000 R15: ffff888237c31f78 FS: 0000000000000000(0000) GS:ffff888237c00000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 00007ffc63fecac8 CR3: 000000000342e000 CR4: 00000000000006f0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 Call Trace: ? __warn+0x88/0x130 ? inet_sock_destruct+0x1c5/0x1e0 ? report_bug+0x18e/0x1a0 ? handle_bug+0x53/0x90 ? exc_invalid_op+0x18/0x70 ? asm_exc_invalid_op+0x1a/0x20 ? inet_sock_destruct+0x1c5/0x1e0 __sk_destruct+0x2a/0x200 rcu_do_batch+0x1aa/0x530 ? rcu_do_batch+0x13b/0x530 rcu_core+0x159/0x2f0 handle_softirqs+0xd3/0x2b0 ? __pfx_smpboot_thread_fn+0x10/0x10 run_ksoftirqd+0x25/0x30 smpboot_thread_fn+0xdd/0x1d0 kthread+0xd3/0x100 ? __pfx_kthread+0x10/0x10 ret_from_fork+0x34/0x50 ? __pfx_kthread+0x10/0x10 ret_from_fork_asm+0x1a/0x30 ---[ end trace 0000000000000000 ]--- Its possible that two threads call tcp_v6_do_rcv()/sk_forward_alloc_add() concurrently when sk->sk_state == TCP_LISTEN with sk->sk_lock unlocked, which triggers a data-race around sk->sk_forward_alloc: tcp_v6_rcv tcp_v6_do_rcv skb_clone_and_charge_r sk_rmem_schedule __sk_mem_schedule sk_forward_alloc_add() skb_set_owner_r sk_mem_charge sk_forward_alloc_add() __kfree_skb skb_release_all skb_release_head_state sock_rfree sk_mem_uncharge sk_forward_alloc_add() sk_mem_reclaim // set local var reclaimable __sk_mem_reclaim sk_forward_alloc_add() In this syzkaller testcase, two threads call tcp_v6_do_rcv() with skb->truesize=768, the sk_forward_alloc changes like this: (cpu 1) | (cpu 2) | sk_forward_alloc ... | ... | 0 __sk_mem_schedule() | | +4096 = 4096 | __sk_mem_schedule() | +4096 = 8192 sk_mem_charge() | | -768 = 7424 | sk_mem_charge() | -768 = 6656 ... | ... | sk_mem_uncharge() | | +768 = 7424 reclaimable=7424 | | | sk_mem_uncharge() | +768 = 8192 | reclaimable=8192 | __sk_mem_reclaim() | | -4096 = 4096 | __sk_mem_reclaim() | -8192 = -4096 != 0 The skb_clone_and_charge_r() should not be called in tcp_v6_do_rcv() when sk->sk_state is TCP_LISTEN, it happens later in tcp_v6_syn_recv_sock(). Fix the same issue in dccp_v6_do_rcv(). Suggested-by: Eric Dumazet Reviewed-by: Eric Dumazet Fixes: e994b2f0fb92 ("tcp: do not lock listener to process SYN packets") Signed-off-by: Wang Liang Link: https://patch.msgid.link/20241107023405.889239-1-wangliang74@huawei.com Signed-off-by: Jakub Kicinski --- net/dccp/ipv6.c | 2 +- net/ipv6/tcp_ipv6.c | 4 +--- 2 files changed, 2 insertions(+), 4 deletions(-) diff --git a/net/dccp/ipv6.c b/net/dccp/ipv6.c index da5dba120bc9..d6649246188d 100644 --- a/net/dccp/ipv6.c +++ b/net/dccp/ipv6.c @@ -618,7 +618,7 @@ static int dccp_v6_do_rcv(struct sock *sk, struct sk_buff *skb) by tcp. Feel free to propose better solution. --ANK (980728) */ - if (np->rxopt.all) + if (np->rxopt.all && sk->sk_state != DCCP_LISTEN) opt_skb = skb_clone_and_charge_r(skb, sk); if (sk->sk_state == DCCP_OPEN) { /* Fast path */ diff --git a/net/ipv6/tcp_ipv6.c b/net/ipv6/tcp_ipv6.c index d71ab4e1efe1..c9de5ef8f267 100644 --- a/net/ipv6/tcp_ipv6.c +++ b/net/ipv6/tcp_ipv6.c @@ -1618,7 +1618,7 @@ int tcp_v6_do_rcv(struct sock *sk, struct sk_buff *skb) by tcp. Feel free to propose better solution. --ANK (980728) */ - if (np->rxopt.all) + if (np->rxopt.all && sk->sk_state != TCP_LISTEN) opt_skb = skb_clone_and_charge_r(skb, sk); if (sk->sk_state == TCP_ESTABLISHED) { /* Fast path */ @@ -1656,8 +1656,6 @@ int tcp_v6_do_rcv(struct sock *sk, struct sk_buff *skb) if (reason) goto reset; } - if (opt_skb) - __kfree_skb(opt_skb); return 0; } } else -- 2.51.0 From f2685c00c3222305f5b6740a8b16ea044640283a Mon Sep 17 00:00:00 2001 From: Mina Almasry Date: Thu, 7 Nov 2024 21:03:30 +0000 Subject: [PATCH 13/16] net: fix SO_DEVMEM_DONTNEED looping too long Exit early if we're freeing more than 1024 frags, to prevent looping too long. Also minor code cleanups: - Flip checks to reduce indentation. - Use sizeof(*tokens) everywhere for consistentcy. Cc: Yi Lai Signed-off-by: Mina Almasry Acked-by: Stanislav Fomichev Link: https://patch.msgid.link/20241107210331.3044434-1-almasrymina@google.com Signed-off-by: Jakub Kicinski --- net/core/sock.c | 42 ++++++++++++++++++++++++------------------ 1 file changed, 24 insertions(+), 18 deletions(-) diff --git a/net/core/sock.c b/net/core/sock.c index 039be95c40cf..da50df485090 100644 --- a/net/core/sock.c +++ b/net/core/sock.c @@ -1052,32 +1052,34 @@ static int sock_reserve_memory(struct sock *sk, int bytes) #ifdef CONFIG_PAGE_POOL -/* This is the number of tokens that the user can SO_DEVMEM_DONTNEED in - * 1 syscall. The limit exists to limit the amount of memory the kernel - * allocates to copy these tokens. +/* This is the number of tokens and frags that the user can SO_DEVMEM_DONTNEED + * in 1 syscall. The limit exists to limit the amount of memory the kernel + * allocates to copy these tokens, and to prevent looping over the frags for + * too long. */ #define MAX_DONTNEED_TOKENS 128 +#define MAX_DONTNEED_FRAGS 1024 static noinline_for_stack int sock_devmem_dontneed(struct sock *sk, sockptr_t optval, unsigned int optlen) { unsigned int num_tokens, i, j, k, netmem_num = 0; struct dmabuf_token *tokens; + int ret = 0, num_frags = 0; netmem_ref netmems[16]; - int ret = 0; if (!sk_is_tcp(sk)) return -EBADF; - if (optlen % sizeof(struct dmabuf_token) || + if (optlen % sizeof(*tokens) || optlen > sizeof(*tokens) * MAX_DONTNEED_TOKENS) return -EINVAL; - tokens = kvmalloc_array(optlen, sizeof(*tokens), GFP_KERNEL); + num_tokens = optlen / sizeof(*tokens); + tokens = kvmalloc_array(num_tokens, sizeof(*tokens), GFP_KERNEL); if (!tokens) return -ENOMEM; - num_tokens = optlen / sizeof(struct dmabuf_token); if (copy_from_sockptr(tokens, optval, optlen)) { kvfree(tokens); return -EFAULT; @@ -1086,24 +1088,28 @@ sock_devmem_dontneed(struct sock *sk, sockptr_t optval, unsigned int optlen) xa_lock_bh(&sk->sk_user_frags); for (i = 0; i < num_tokens; i++) { for (j = 0; j < tokens[i].token_count; j++) { + if (++num_frags > MAX_DONTNEED_FRAGS) + goto frag_limit_reached; + netmem_ref netmem = (__force netmem_ref)__xa_erase( &sk->sk_user_frags, tokens[i].token_start + j); - if (netmem && - !WARN_ON_ONCE(!netmem_is_net_iov(netmem))) { - netmems[netmem_num++] = netmem; - if (netmem_num == ARRAY_SIZE(netmems)) { - xa_unlock_bh(&sk->sk_user_frags); - for (k = 0; k < netmem_num; k++) - WARN_ON_ONCE(!napi_pp_put_page(netmems[k])); - netmem_num = 0; - xa_lock_bh(&sk->sk_user_frags); - } - ret++; + if (!netmem || WARN_ON_ONCE(!netmem_is_net_iov(netmem))) + continue; + + netmems[netmem_num++] = netmem; + if (netmem_num == ARRAY_SIZE(netmems)) { + xa_unlock_bh(&sk->sk_user_frags); + for (k = 0; k < netmem_num; k++) + WARN_ON_ONCE(!napi_pp_put_page(netmems[k])); + netmem_num = 0; + xa_lock_bh(&sk->sk_user_frags); } + ret++; } } +frag_limit_reached: xa_unlock_bh(&sk->sk_user_frags); for (k = 0; k < netmem_num; k++) WARN_ON_ONCE(!napi_pp_put_page(netmems[k])); -- 2.51.0 From 102d1404c385611c574498b1e0d1f3762e253359 Mon Sep 17 00:00:00 2001 From: Mina Almasry Date: Thu, 7 Nov 2024 21:03:31 +0000 Subject: [PATCH 14/16] net: clarify SO_DEVMEM_DONTNEED behavior in documentation Document new behavior when the number of frags passed is too big. Signed-off-by: Mina Almasry Link: https://patch.msgid.link/20241107210331.3044434-2-almasrymina@google.com Signed-off-by: Jakub Kicinski --- Documentation/networking/devmem.rst | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/Documentation/networking/devmem.rst b/Documentation/networking/devmem.rst index a55bf21f671c..d95363645331 100644 --- a/Documentation/networking/devmem.rst +++ b/Documentation/networking/devmem.rst @@ -225,6 +225,15 @@ The user must ensure the tokens are returned to the kernel in a timely manner. Failure to do so will exhaust the limited dmabuf that is bound to the RX queue and will lead to packet drops. +The user must pass no more than 128 tokens, with no more than 1024 total frags +among the token->token_count across all the tokens. If the user provides more +than 1024 frags, the kernel will free up to 1024 frags and return early. + +The kernel returns the number of actual frags freed. The number of frags freed +can be less than the tokens provided by the user in case of: + +(a) an internal kernel leak bug. +(b) the user passed more than 1024 frags. Implementation & Caveats ======================== -- 2.51.0 From 581302298524e9d77c4c44ff5156a6cd112227ae Mon Sep 17 00:00:00 2001 From: Paolo Abeni Date: Fri, 8 Nov 2024 11:58:16 +0100 Subject: [PATCH 15/16] mptcp: error out earlier on disconnect Eric reported a division by zero splat in the MPTCP protocol: Oops: divide error: 0000 [#1] PREEMPT SMP KASAN PTI CPU: 1 UID: 0 PID: 6094 Comm: syz-executor317 Not tainted 6.12.0-rc5-syzkaller-00291-g05b92660cdfe #0 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 09/13/2024 RIP: 0010:__tcp_select_window+0x5b4/0x1310 net/ipv4/tcp_output.c:3163 Code: f6 44 01 e3 89 df e8 9b 75 09 f8 44 39 f3 0f 8d 11 ff ff ff e8 0d 74 09 f8 45 89 f4 e9 04 ff ff ff e8 00 74 09 f8 44 89 f0 99 7c 24 14 41 29 d6 45 89 f4 e9 ec fe ff ff e8 e8 73 09 f8 48 89 RSP: 0018:ffffc900041f7930 EFLAGS: 00010293 RAX: 0000000000017e67 RBX: 0000000000017e67 RCX: ffffffff8983314b RDX: 0000000000000000 RSI: ffffffff898331b0 RDI: 0000000000000004 RBP: 00000000005d6000 R08: 0000000000000004 R09: 0000000000017e67 R10: 0000000000003e80 R11: 0000000000000000 R12: 0000000000003e80 R13: ffff888031d9b440 R14: 0000000000017e67 R15: 00000000002eb000 FS: 00007feb5d7f16c0(0000) GS:ffff8880b8700000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 00007feb5d8adbb8 CR3: 0000000074e4c000 CR4: 00000000003526f0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 Call Trace: __tcp_cleanup_rbuf+0x3e7/0x4b0 net/ipv4/tcp.c:1493 mptcp_rcv_space_adjust net/mptcp/protocol.c:2085 [inline] mptcp_recvmsg+0x2156/0x2600 net/mptcp/protocol.c:2289 inet_recvmsg+0x469/0x6a0 net/ipv4/af_inet.c:885 sock_recvmsg_nosec net/socket.c:1051 [inline] sock_recvmsg+0x1b2/0x250 net/socket.c:1073 __sys_recvfrom+0x1a5/0x2e0 net/socket.c:2265 __do_sys_recvfrom net/socket.c:2283 [inline] __se_sys_recvfrom net/socket.c:2279 [inline] __x64_sys_recvfrom+0xe0/0x1c0 net/socket.c:2279 do_syscall_x64 arch/x86/entry/common.c:52 [inline] do_syscall_64+0xcd/0x250 arch/x86/entry/common.c:83 entry_SYSCALL_64_after_hwframe+0x77/0x7f RIP: 0033:0x7feb5d857559 Code: 28 00 00 00 75 05 48 83 c4 28 c3 e8 51 18 00 00 90 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 c7 c1 b0 ff ff ff f7 d8 64 89 01 48 RSP: 002b:00007feb5d7f1208 EFLAGS: 00000246 ORIG_RAX: 000000000000002d RAX: ffffffffffffffda RBX: 00007feb5d8e1318 RCX: 00007feb5d857559 RDX: 000000800000000e RSI: 0000000000000000 RDI: 0000000000000003 RBP: 00007feb5d8e1310 R08: 0000000000000000 R09: ffffffff81000000 R10: 0000000000000100 R11: 0000000000000246 R12: 00007feb5d8e131c R13: 00007feb5d8ae074 R14: 000000800000000e R15: 00000000fffffdef and provided a nice reproducer. The root cause is the current bad handling of racing disconnect. After the blamed commit below, sk_wait_data() can return (with error) with the underlying socket disconnected and a zero rcv_mss. Catch the error and return without performing any additional operations on the current socket. Reported-by: Eric Dumazet Fixes: 419ce133ab92 ("tcp: allow again tcp_disconnect() when threads are waiting") Signed-off-by: Paolo Abeni Reviewed-by: Matthieu Baerts (NGI0) Link: https://patch.msgid.link/8c82ecf71662ecbc47bf390f9905de70884c9f2d.1731060874.git.pabeni@redhat.com Signed-off-by: Jakub Kicinski --- net/mptcp/protocol.c | 13 +++++++++---- 1 file changed, 9 insertions(+), 4 deletions(-) diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c index d263091659e0..95a5a3da3944 100644 --- a/net/mptcp/protocol.c +++ b/net/mptcp/protocol.c @@ -2205,7 +2205,7 @@ static int mptcp_recvmsg(struct sock *sk, struct msghdr *msg, size_t len, cmsg_flags = MPTCP_CMSG_INQ; while (copied < len) { - int bytes_read; + int err, bytes_read; bytes_read = __mptcp_recvmsg_mskq(msk, msg, len - copied, flags, &tss, &cmsg_flags); if (unlikely(bytes_read < 0)) { @@ -2267,9 +2267,16 @@ static int mptcp_recvmsg(struct sock *sk, struct msghdr *msg, size_t len, } pr_debug("block timeout %ld\n", timeo); - sk_wait_data(sk, &timeo, NULL); + mptcp_rcv_space_adjust(msk, copied); + err = sk_wait_data(sk, &timeo, NULL); + if (err < 0) { + err = copied ? : err; + goto out_err; + } } + mptcp_rcv_space_adjust(msk, copied); + out_err: if (cmsg_flags && copied >= 0) { if (cmsg_flags & MPTCP_CMSG_TS) @@ -2285,8 +2292,6 @@ out_err: pr_debug("msk=%p rx queue empty=%d:%d copied=%d\n", msk, skb_queue_empty_lockless(&sk->sk_receive_queue), skb_queue_empty(&msk->receive_queue), copied); - if (!(flags & MSG_PEEK)) - mptcp_rcv_space_adjust(msk, copied); release_sock(sk); return copied; -- 2.51.0 From ce7356ae35943cc6494cc692e62d51a734062b7d Mon Sep 17 00:00:00 2001 From: Paolo Abeni Date: Fri, 8 Nov 2024 11:58:17 +0100 Subject: [PATCH 16/16] mptcp: cope racing subflow creation in mptcp_rcv_space_adjust Additional active subflows - i.e. created by the in kernel path manager - are included into the subflow list before starting the 3whs. A racing recvmsg() spooling data received on an already established subflow would unconditionally call tcp_cleanup_rbuf() on all the current subflows, potentially hitting a divide by zero error on the newly created ones. Explicitly check that the subflow is in a suitable state before invoking tcp_cleanup_rbuf(). Fixes: c76c6956566f ("mptcp: call tcp_cleanup_rbuf on subflows") Signed-off-by: Paolo Abeni Reviewed-by: Matthieu Baerts (NGI0) Link: https://patch.msgid.link/02374660836e1b52afc91966b7535c8c5f7bafb0.1731060874.git.pabeni@redhat.com Signed-off-by: Jakub Kicinski --- net/mptcp/protocol.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c index 95a5a3da3944..48d480982b78 100644 --- a/net/mptcp/protocol.c +++ b/net/mptcp/protocol.c @@ -2082,7 +2082,8 @@ static void mptcp_rcv_space_adjust(struct mptcp_sock *msk, int copied) slow = lock_sock_fast(ssk); WRITE_ONCE(ssk->sk_rcvbuf, rcvbuf); WRITE_ONCE(tcp_sk(ssk)->window_clamp, window_clamp); - tcp_cleanup_rbuf(ssk, 1); + if (tcp_can_send_ack(ssk)) + tcp_cleanup_rbuf(ssk, 1); unlock_sock_fast(ssk, slow); } } -- 2.51.0