From d59b73a66e5e0682442b6d7b4965364e57078b80 Mon Sep 17 00:00:00 2001 From: Moshe Shemesh Date: Wed, 3 Aug 2022 10:49:23 +0300 Subject: net/mlx5: Avoid false positive lockdep warning by adding lock_class_key Add a lock_class_key per mlx5 device to avoid a false positive "possible circular locking dependency" warning by lockdep, on flows which lock more than one mlx5 device, such as adding SF. kernel log: ====================================================== WARNING: possible circular locking dependency detected 5.19.0-rc8+ #2 Not tainted ------------------------------------------------------ kworker/u20:0/8 is trying to acquire lock: ffff88812dfe0d98 (&dev->intf_state_mutex){+.+.}-{3:3}, at: mlx5_init_one+0x2e/0x490 [mlx5_core] but task is already holding lock: ffff888101aa7898 (&(¬ifier->n_head)->rwsem){++++}-{3:3}, at: blocking_notifier_call_chain+0x5a/0x130 which lock already depends on the new lock. the existing dependency chain (in reverse order) is: -> #1 (&(¬ifier->n_head)->rwsem){++++}-{3:3}: down_write+0x90/0x150 blocking_notifier_chain_register+0x53/0xa0 mlx5_sf_table_init+0x369/0x4a0 [mlx5_core] mlx5_init_one+0x261/0x490 [mlx5_core] probe_one+0x430/0x680 [mlx5_core] local_pci_probe+0xd6/0x170 work_for_cpu_fn+0x4e/0xa0 process_one_work+0x7c2/0x1340 worker_thread+0x6f6/0xec0 kthread+0x28f/0x330 ret_from_fork+0x1f/0x30 -> #0 (&dev->intf_state_mutex){+.+.}-{3:3}: __lock_acquire+0x2fc7/0x6720 lock_acquire+0x1c1/0x550 __mutex_lock+0x12c/0x14b0 mlx5_init_one+0x2e/0x490 [mlx5_core] mlx5_sf_dev_probe+0x29c/0x370 [mlx5_core] auxiliary_bus_probe+0x9d/0xe0 really_probe+0x1e0/0xaa0 __driver_probe_device+0x219/0x480 driver_probe_device+0x49/0x130 __device_attach_driver+0x1b8/0x280 bus_for_each_drv+0x123/0x1a0 __device_attach+0x1a3/0x460 bus_probe_device+0x1a2/0x260 device_add+0x9b1/0x1b40 __auxiliary_device_add+0x88/0xc0 mlx5_sf_dev_state_change_handler+0x67e/0x9d0 [mlx5_core] blocking_notifier_call_chain+0xd5/0x130 mlx5_vhca_state_work_handler+0x2b0/0x3f0 [mlx5_core] process_one_work+0x7c2/0x1340 worker_thread+0x59d/0xec0 kthread+0x28f/0x330 ret_from_fork+0x1f/0x30 other info that might help us debug this: Possible unsafe locking scenario: CPU0 CPU1 ---- ---- lock(&(¬ifier->n_head)->rwsem); lock(&dev->intf_state_mutex); lock(&(¬ifier->n_head)->rwsem); lock(&dev->intf_state_mutex); *** DEADLOCK *** 4 locks held by kworker/u20:0/8: #0: ffff888150612938 ((wq_completion)mlx5_events){+.+.}-{0:0}, at: process_one_work+0x6e2/0x1340 #1: ffff888100cafdb8 ((work_completion)(&work->work)#3){+.+.}-{0:0}, at: process_one_work+0x70f/0x1340 #2: ffff888101aa7898 (&(¬ifier->n_head)->rwsem){++++}-{3:3}, at: blocking_notifier_call_chain+0x5a/0x130 #3: ffff88813682d0e8 (&dev->mutex){....}-{3:3}, at:__device_attach+0x76/0x460 stack backtrace: CPU: 6 PID: 8 Comm: kworker/u20:0 Not tainted 5.19.0-rc8+ Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS rel-1.13.0-0-gf21b5a4aeb02-prebuilt.qemu.org 04/01/2014 Workqueue: mlx5_events mlx5_vhca_state_work_handler [mlx5_core] Call Trace: dump_stack_lvl+0x57/0x7d check_noncircular+0x278/0x300 ? print_circular_bug+0x460/0x460 ? lock_chain_count+0x20/0x20 ? register_lock_class+0x1880/0x1880 __lock_acquire+0x2fc7/0x6720 ? register_lock_class+0x1880/0x1880 ? register_lock_class+0x1880/0x1880 lock_acquire+0x1c1/0x550 ? mlx5_init_one+0x2e/0x490 [mlx5_core] ? lockdep_hardirqs_on_prepare+0x400/0x400 __mutex_lock+0x12c/0x14b0 ? mlx5_init_one+0x2e/0x490 [mlx5_core] ? mlx5_init_one+0x2e/0x490 [mlx5_core] ? _raw_read_unlock+0x1f/0x30 ? mutex_lock_io_nested+0x1320/0x1320 ? __ioremap_caller.constprop.0+0x306/0x490 ? mlx5_sf_dev_probe+0x269/0x370 [mlx5_core] ? iounmap+0x160/0x160 mlx5_init_one+0x2e/0x490 [mlx5_core] mlx5_sf_dev_probe+0x29c/0x370 [mlx5_core] ? mlx5_sf_dev_remove+0x130/0x130 [mlx5_core] auxiliary_bus_probe+0x9d/0xe0 really_probe+0x1e0/0xaa0 __driver_probe_device+0x219/0x480 ? auxiliary_match_id+0xe9/0x140 driver_probe_device+0x49/0x130 __device_attach_driver+0x1b8/0x280 ? driver_allows_async_probing+0x140/0x140 bus_for_each_drv+0x123/0x1a0 ? bus_for_each_dev+0x1a0/0x1a0 ? lockdep_hardirqs_on_prepare+0x286/0x400 ? trace_hardirqs_on+0x2d/0x100 __device_attach+0x1a3/0x460 ? device_driver_attach+0x1e0/0x1e0 ? kobject_uevent_env+0x22d/0xf10 bus_probe_device+0x1a2/0x260 device_add+0x9b1/0x1b40 ? dev_set_name+0xab/0xe0 ? __fw_devlink_link_to_suppliers+0x260/0x260 ? memset+0x20/0x40 ? lockdep_init_map_type+0x21a/0x7d0 __auxiliary_device_add+0x88/0xc0 ? auxiliary_device_init+0x86/0xa0 mlx5_sf_dev_state_change_handler+0x67e/0x9d0 [mlx5_core] blocking_notifier_call_chain+0xd5/0x130 mlx5_vhca_state_work_handler+0x2b0/0x3f0 [mlx5_core] ? mlx5_vhca_event_arm+0x100/0x100 [mlx5_core] ? lock_downgrade+0x6e0/0x6e0 ? lockdep_hardirqs_on_prepare+0x286/0x400 process_one_work+0x7c2/0x1340 ? lockdep_hardirqs_on_prepare+0x400/0x400 ? pwq_dec_nr_in_flight+0x230/0x230 ? rwlock_bug.part.0+0x90/0x90 worker_thread+0x59d/0xec0 ? process_one_work+0x1340/0x1340 kthread+0x28f/0x330 ? kthread_complete_and_exit+0x20/0x20 ret_from_fork+0x1f/0x30 Fixes: 6a3273217469 ("net/mlx5: SF, Port function state change support") Signed-off-by: Moshe Shemesh Reviewed-by: Shay Drory Signed-off-by: Saeed Mahameed --- include/linux/mlx5/driver.h | 1 + 1 file changed, 1 insertion(+) (limited to 'include/linux') diff --git a/include/linux/mlx5/driver.h b/include/linux/mlx5/driver.h index 96b16fbe1aa4..7b7ce602c808 100644 --- a/include/linux/mlx5/driver.h +++ b/include/linux/mlx5/driver.h @@ -779,6 +779,7 @@ struct mlx5_core_dev { enum mlx5_device_state state; /* sync interface state */ struct mutex intf_state_mutex; + struct lock_class_key lock_key; unsigned long intf_state; struct mlx5_priv priv; struct mlx5_profile profile; -- cgit v1.2.3 From 7997eff82828304b780dc0a39707e1946d6f1ebf Mon Sep 17 00:00:00 2001 From: Florian Westphal Date: Sat, 20 Aug 2022 17:38:37 +0200 Subject: netfilter: ebtables: reject blobs that don't provide all entry points Harshit Mogalapalli says: In ebt_do_table() function dereferencing 'private->hook_entry[hook]' can lead to NULL pointer dereference. [..] Kernel panic: general protection fault, probably for non-canonical address 0xdffffc0000000005: 0000 [#1] PREEMPT SMP KASAN KASAN: null-ptr-deref in range [0x0000000000000028-0x000000000000002f] [..] RIP: 0010:ebt_do_table+0x1dc/0x1ce0 Code: 89 fa 48 c1 ea 03 80 3c 02 00 0f 85 5c 16 00 00 48 b8 00 00 00 00 00 fc ff df 49 8b 6c df 08 48 8d 7d 2c 48 89 fa 48 c1 ea 03 <0f> b6 14 02 48 89 f8 83 e0 07 83 c0 03 38 d0 7c 08 84 d2 0f 85 88 [..] Call Trace: nf_hook_slow+0xb1/0x170 __br_forward+0x289/0x730 maybe_deliver+0x24b/0x380 br_flood+0xc6/0x390 br_dev_xmit+0xa2e/0x12c0 For some reason ebtables rejects blobs that provide entry points that are not supported by the table, but what it should instead reject is the opposite: blobs that DO NOT provide an entry point supported by the table. t->valid_hooks is the bitmask of hooks (input, forward ...) that will see packets. Providing an entry point that is not support is harmless (never called/used), but the inverse isn't: it results in a crash because the ebtables traverser doesn't expect a NULL blob for a location its receiving packets for. Instead of fixing all the individual checks, do what iptables is doing and reject all blobs that differ from the expected hooks. Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2") Reported-by: Harshit Mogalapalli Reported-by: syzkaller Signed-off-by: Florian Westphal --- include/linux/netfilter_bridge/ebtables.h | 4 ---- net/bridge/netfilter/ebtable_broute.c | 8 -------- net/bridge/netfilter/ebtable_filter.c | 8 -------- net/bridge/netfilter/ebtable_nat.c | 8 -------- net/bridge/netfilter/ebtables.c | 8 +------- 5 files changed, 1 insertion(+), 35 deletions(-) (limited to 'include/linux') diff --git a/include/linux/netfilter_bridge/ebtables.h b/include/linux/netfilter_bridge/ebtables.h index a13296d6c7ce..fd533552a062 100644 --- a/include/linux/netfilter_bridge/ebtables.h +++ b/include/linux/netfilter_bridge/ebtables.h @@ -94,10 +94,6 @@ struct ebt_table { struct ebt_replace_kernel *table; unsigned int valid_hooks; rwlock_t lock; - /* e.g. could be the table explicitly only allows certain - * matches, targets, ... 0 == let it in */ - int (*check)(const struct ebt_table_info *info, - unsigned int valid_hooks); /* the data used by the kernel */ struct ebt_table_info *private; struct nf_hook_ops *ops; diff --git a/net/bridge/netfilter/ebtable_broute.c b/net/bridge/netfilter/ebtable_broute.c index 1a11064f9990..8f19253024b0 100644 --- a/net/bridge/netfilter/ebtable_broute.c +++ b/net/bridge/netfilter/ebtable_broute.c @@ -36,18 +36,10 @@ static struct ebt_replace_kernel initial_table = { .entries = (char *)&initial_chain, }; -static int check(const struct ebt_table_info *info, unsigned int valid_hooks) -{ - if (valid_hooks & ~(1 << NF_BR_BROUTING)) - return -EINVAL; - return 0; -} - static const struct ebt_table broute_table = { .name = "broute", .table = &initial_table, .valid_hooks = 1 << NF_BR_BROUTING, - .check = check, .me = THIS_MODULE, }; diff --git a/net/bridge/netfilter/ebtable_filter.c b/net/bridge/netfilter/ebtable_filter.c index cb949436bc0e..278f324e6752 100644 --- a/net/bridge/netfilter/ebtable_filter.c +++ b/net/bridge/netfilter/ebtable_filter.c @@ -43,18 +43,10 @@ static struct ebt_replace_kernel initial_table = { .entries = (char *)initial_chains, }; -static int check(const struct ebt_table_info *info, unsigned int valid_hooks) -{ - if (valid_hooks & ~FILTER_VALID_HOOKS) - return -EINVAL; - return 0; -} - static const struct ebt_table frame_filter = { .name = "filter", .table = &initial_table, .valid_hooks = FILTER_VALID_HOOKS, - .check = check, .me = THIS_MODULE, }; diff --git a/net/bridge/netfilter/ebtable_nat.c b/net/bridge/netfilter/ebtable_nat.c index 5ee0531ae506..9066f7f376d5 100644 --- a/net/bridge/netfilter/ebtable_nat.c +++ b/net/bridge/netfilter/ebtable_nat.c @@ -43,18 +43,10 @@ static struct ebt_replace_kernel initial_table = { .entries = (char *)initial_chains, }; -static int check(const struct ebt_table_info *info, unsigned int valid_hooks) -{ - if (valid_hooks & ~NAT_VALID_HOOKS) - return -EINVAL; - return 0; -} - static const struct ebt_table frame_nat = { .name = "nat", .table = &initial_table, .valid_hooks = NAT_VALID_HOOKS, - .check = check, .me = THIS_MODULE, }; diff --git a/net/bridge/netfilter/ebtables.c b/net/bridge/netfilter/ebtables.c index f2dbefb61ce8..9a0ae59cdc50 100644 --- a/net/bridge/netfilter/ebtables.c +++ b/net/bridge/netfilter/ebtables.c @@ -1040,8 +1040,7 @@ static int do_replace_finish(struct net *net, struct ebt_replace *repl, goto free_iterate; } - /* the table doesn't like it */ - if (t->check && (ret = t->check(newinfo, repl->valid_hooks))) + if (repl->valid_hooks != t->valid_hooks) goto free_unlock; if (repl->num_counters && repl->num_counters != t->private->nentries) { @@ -1231,11 +1230,6 @@ int ebt_register_table(struct net *net, const struct ebt_table *input_table, if (ret != 0) goto free_chainstack; - if (table->check && table->check(newinfo, table->valid_hooks)) { - ret = -EINVAL; - goto free_chainstack; - } - table->private = newinfo; rwlock_init(&table->lock); mutex_lock(&ebt_mutex); -- cgit v1.2.3 From af67508ea6cbf0e4ea27f8120056fa2efce127dd Mon Sep 17 00:00:00 2001 From: Kuniyuki Iwashima Date: Tue, 23 Aug 2022 10:46:56 -0700 Subject: net: Fix data-races around sysctl_fb_tunnels_only_for_init_net. While reading sysctl_fb_tunnels_only_for_init_net, it can be changed concurrently. Thus, we need to add READ_ONCE() to its readers. Fixes: 79134e6ce2c9 ("net: do not create fallback tunnels for non-default namespaces") Signed-off-by: Kuniyuki Iwashima Signed-off-by: David S. Miller --- include/linux/netdevice.h | 11 ++++++++--- 1 file changed, 8 insertions(+), 3 deletions(-) (limited to 'include/linux') diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h index 1a3cb93c3dcc..6d3a33fd0cdb 100644 --- a/include/linux/netdevice.h +++ b/include/linux/netdevice.h @@ -640,9 +640,14 @@ extern int sysctl_devconf_inherit_init_net; */ static inline bool net_has_fallback_tunnels(const struct net *net) { - return !IS_ENABLED(CONFIG_SYSCTL) || - !sysctl_fb_tunnels_only_for_init_net || - (net == &init_net && sysctl_fb_tunnels_only_for_init_net == 1); +#if IS_ENABLED(CONFIG_SYSCTL) + int fb_tunnels_only_for_init_net = READ_ONCE(sysctl_fb_tunnels_only_for_init_net); + + return !fb_tunnels_only_for_init_net || + (net_eq(net, &init_net) && fb_tunnels_only_for_init_net == 1); +#else + return true; +#endif } static inline int netdev_queue_numa_node_read(const struct netdev_queue *q) -- cgit v1.2.3 From a5612ca10d1aa05624ebe72633e0c8c792970833 Mon Sep 17 00:00:00 2001 From: Kuniyuki Iwashima Date: Tue, 23 Aug 2022 10:46:57 -0700 Subject: net: Fix data-races around sysctl_devconf_inherit_init_net. While reading sysctl_devconf_inherit_init_net, it can be changed concurrently. Thus, we need to add READ_ONCE() to its readers. Fixes: 856c395cfa63 ("net: introduce a knob to control whether to inherit devconf config") Signed-off-by: Kuniyuki Iwashima Signed-off-by: David S. Miller --- include/linux/netdevice.h | 9 +++++++++ net/ipv4/devinet.c | 16 ++++++++++------ net/ipv6/addrconf.c | 5 ++--- 3 files changed, 21 insertions(+), 9 deletions(-) (limited to 'include/linux') diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h index 6d3a33fd0cdb..05d6f3facd5a 100644 --- a/include/linux/netdevice.h +++ b/include/linux/netdevice.h @@ -650,6 +650,15 @@ static inline bool net_has_fallback_tunnels(const struct net *net) #endif } +static inline int net_inherit_devconf(void) +{ +#if IS_ENABLED(CONFIG_SYSCTL) + return READ_ONCE(sysctl_devconf_inherit_init_net); +#else + return 0; +#endif +} + static inline int netdev_queue_numa_node_read(const struct netdev_queue *q) { #if defined(CONFIG_XPS) && defined(CONFIG_NUMA) diff --git a/net/ipv4/devinet.c b/net/ipv4/devinet.c index 92b778e423df..e8b9a9202fec 100644 --- a/net/ipv4/devinet.c +++ b/net/ipv4/devinet.c @@ -2682,23 +2682,27 @@ static __net_init int devinet_init_net(struct net *net) #endif if (!net_eq(net, &init_net)) { - if (IS_ENABLED(CONFIG_SYSCTL) && - sysctl_devconf_inherit_init_net == 3) { + switch (net_inherit_devconf()) { + case 3: /* copy from the current netns */ memcpy(all, current->nsproxy->net_ns->ipv4.devconf_all, sizeof(ipv4_devconf)); memcpy(dflt, current->nsproxy->net_ns->ipv4.devconf_dflt, sizeof(ipv4_devconf_dflt)); - } else if (!IS_ENABLED(CONFIG_SYSCTL) || - sysctl_devconf_inherit_init_net != 2) { - /* inherit == 0 or 1: copy from init_net */ + break; + case 0: + case 1: + /* copy from init_net */ memcpy(all, init_net.ipv4.devconf_all, sizeof(ipv4_devconf)); memcpy(dflt, init_net.ipv4.devconf_dflt, sizeof(ipv4_devconf_dflt)); + break; + case 2: + /* use compiled values */ + break; } - /* else inherit == 2: use compiled values */ } #ifdef CONFIG_SYSCTL diff --git a/net/ipv6/addrconf.c b/net/ipv6/addrconf.c index b624e3d8c5f0..e15f64f22fa8 100644 --- a/net/ipv6/addrconf.c +++ b/net/ipv6/addrconf.c @@ -7162,9 +7162,8 @@ static int __net_init addrconf_init_net(struct net *net) if (!dflt) goto err_alloc_dflt; - if (IS_ENABLED(CONFIG_SYSCTL) && - !net_eq(net, &init_net)) { - switch (sysctl_devconf_inherit_init_net) { + if (!net_eq(net, &init_net)) { + switch (net_inherit_devconf()) { case 1: /* copy from init_net */ memcpy(all, init_net.ipv6.devconf_all, sizeof(ipv6_devconf)); -- cgit v1.2.3