[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <Z-GPFQou5GomWCOo@mini-arch>
Date: Mon, 24 Mar 2025 09:57:57 -0700
From: Stanislav Fomichev <stfomichev@...il.com>
To: Taehee Yoo <ap420073@...il.com>
Cc: Netdev <netdev@...r.kernel.org>, Stanislav Fomichev <sdf@...ichev.me>,
David Miller <davem@...emloft.net>,
Eric Dumazet <edumazet@...gle.com>,
Jakub Kicinski <kuba@...nel.org>, Paolo Abeni <pabeni@...hat.com>,
Simon Horman <horms@...nel.org>
Subject: Re: Report deadlock in the latest net-next
On 03/17, Taehee Yoo wrote:
> Hi Stanislav,
> I found a deadlock in the latest net-next kernel.
> The calltrace indicates your current
> commit ad7c7b2172c3 ("net: hold netdev instance lock during sysfs operations").
> The dev->lock was acquired in do_setlink.constprop.0+0x12a/0x3440,
> which is net/core/rtnetlink.c:3025
> And then dev->lock is acquired in dev_disable_lro+0x81/0x1f0,
> which is /net/core/dev_api.c:255
> dev_disable_lro() is called by netdev notification, but notification
> seems to be called both outside and inside dev->lock context.
> This case is that netdev notification is called inside dev->lock context.
> So deadlock occurs.
> Could you please look into this?
>
> Reproducer:
> modprobe netdevsim
> ip netns add ns_test
> echo 1 > /sys/bus/netdevsim/new_device
> ip link set $interface netns ns_test
>
> ============================================
> WARNING: possible recursive locking detected
> 6.14.0-rc6+ #56 Not tainted
> --------------------------------------------
> ip/1672 is trying to acquire lock:
> ffff888231fbad90 (&dev->lock){+.+.}-{4:4}, at: dev_disable_lro+0x81/0x1f0
>
> but task is already holding lock:
> ffff888231fbad90 (&dev->lock){+.+.}-{4:4}, at:
> do_setlink.constprop.0+0x12a/0x3440
>
> other info that might help us debug this:
> Possible unsafe locking scenario:
>
> CPU0
> ----
> lock(&dev->lock);
> lock(&dev->lock);
>
> *** DEADLOCK ***
>
> May be due to missing lock nesting notation
>
> 3 locks held by ip/1672:
> #0: ffffffff943ba050 (rtnl_mutex){+.+.}-{4:4}, at: rtnl_newlink+0x6b4/0x1c60
> #1: ffff88813abc6170 (&net->rtnl_mutex){+.+.}-{4:4}, at:
> rtnl_newlink+0x6f6/0x1c60
> #2: ffff888231fbad90 (&dev->lock){+.+.}-{4:4}, at:
> do_setlink.constprop.0+0x12a/0x3440
>
> stack backtrace:
> CPU: 2 UID: 0 PID: 1672 Comm: ip Not tainted 6.14.0-rc6+ #56
> 66129e0c5b1b922fef38623168aea99c0593a519
> Hardware name: ASUS System Product Name/PRIME Z690-P D4, BIOS 0603 11/01/2021
> Call Trace:
> <TASK>
> dump_stack_lvl+0x7e/0xc0
> print_deadlock_bug+0x4fd/0x8e0
> __lock_acquire+0x3082/0x4fd0
> ? __pfx___lock_acquire+0x10/0x10
> ? mark_lock.part.0+0xfa/0x2f60
> ? __pfx___lock_acquire+0x10/0x10
> ? check_chain_key+0x1c1/0x520
> lock_acquire+0x1b0/0x570
> ? dev_disable_lro+0x81/0x1f0
> ? __pfx_lock_acquire+0x10/0x10
> __mutex_lock+0x17c/0x17c0
> ? dev_disable_lro+0x81/0x1f0
> ? dev_disable_lro+0x81/0x1f0
> ? __pfx___mutex_lock+0x10/0x10
> ? mark_held_locks+0xa5/0xf0
> ? neigh_parms_alloc+0x36b/0x4f0
> ? __local_bh_enable_ip+0xa5/0x120
> ? lockdep_hardirqs_on+0xbe/0x140
> ? dev_disable_lro+0x81/0x1f0
> dev_disable_lro+0x81/0x1f0
> inetdev_init+0x2d1/0x4a0
> inetdev_event+0x9b3/0x1590
> ? __pfx_lock_release+0x10/0x10
> ? __pfx_inetdev_event+0x10/0x10
> ? notifier_call_chain+0x9b/0x300
> notifier_call_chain+0x9b/0x300
> netif_change_net_namespace+0xdfe/0x1390
> ? __pfx_netif_change_net_namespace+0x10/0x10
> ? __pfx_validate_linkmsg+0x10/0x10
> ? __pfx___lock_acquire+0x10/0x10
> do_setlink.constprop.0+0x241/0x3440
> ? lock_acquire+0x1b0/0x570
> ? __pfx_do_setlink.constprop.0+0x10/0x10
> ? rtnl_newlink+0x6f6/0x1c60
> ? __pfx_lock_acquired+0x10/0x10
> ? netlink_sendmsg+0x712/0xbc0
> ? rcu_is_watching+0x11/0xb0
> ? trace_contention_end+0xef/0x140
> ? __mutex_lock+0x935/0x17c0
> ? __create_object+0x36/0x90
> ? __pfx_lock_release+0x10/0x10
> ? rtnl_newlink+0x6f6/0x1c60
> ? __nla_validate_parse+0xb9/0x2830
> ? __pfx___mutex_lock+0x10/0x10
> ? lockdep_hardirqs_on+0xbe/0x140
> ? __pfx___nla_validate_parse+0x10/0x10
> ? rcu_is_watching+0x11/0xb0
> ? cap_capable+0x17d/0x360
> ? fdget+0x4e/0x1d0
> rtnl_newlink+0x108d/0x1c60
> ? __pfx_rtnl_newlink+0x10/0x10
> ? mark_lock.part.0+0xfa/0x2f60
> ? __pfx___lock_acquire+0x10/0x10
> ? __pfx_mark_lock.part.0+0x10/0x10
> ? __pfx_lock_release+0x10/0x10
> ? __pfx_rtnl_newlink+0x10/0x10
> rtnetlink_rcv_msg+0x71c/0xc10
> ? __pfx_rtnetlink_rcv_msg+0x10/0x10
> ? check_chain_key+0x1c1/0x520
> ? __pfx___lock_acquire+0x10/0x10
> netlink_rcv_skb+0x12c/0x360
> ? __pfx_rtnetlink_rcv_msg+0x10/0x10
> ? __pfx_netlink_rcv_skb+0x10/0x10
> ? netlink_deliver_tap+0xcb/0x9e0
> ? netlink_deliver_tap+0x14b/0x9e0
> netlink_unicast+0x447/0x710
> ? __pfx_netlink_unicast+0x10/0x10
> netlink_sendmsg+0x712/0xbc0
> ? __pfx_netlink_sendmsg+0x10/0x10
> ? _copy_from_user+0x3e/0xa0
> ____sys_sendmsg+0x7ab/0xa10
> ? __pfx_____sys_sendmsg+0x10/0x10
> ? __pfx_copy_msghdr_from_user+0x10/0x10
> ___sys_sendmsg+0xee/0x170
> ? __pfx___lock_acquire+0x10/0x10
> ? kasan_save_stack+0x20/0x40
> ? __pfx____sys_sendmsg+0x10/0x10
> ? entry_SYSCALL_64_after_hwframe+0x76/0x7e
> ? kasan_save_stack+0x30/0x40
> ? __pfx_lock_release+0x10/0x10
> ? __might_fault+0xbf/0x170
> __sys_sendmsg+0x105/0x190
> ? __pfx___sys_sendmsg+0x10/0x10
> ? rseq_syscall+0xc3/0x130
> do_syscall_64+0x64/0x140
> entry_SYSCALL_64_after_hwframe+0x76/0x7e
> RIP: 0033:0x7fd20f92c004
> Code: 15 19 6e 0d 00 f7 d8 64 89 02 b8 ff ff ff ff eb bf 0f 1f 44 00
> 00 f3 0f 1e fa 80 3d 45 f0 0d 00 00 74 13 b8 2e 00 00 00 0f 05 <48> 3d
> 005
> RSP: 002b:00007fff40636e68 EFLAGS: 00000202 ORIG_RAX: 000000000000002e
> RAX: ffffffffffffffda RBX: 0000000000000003 RCX: 00007fd20f92c004
> RDX: 0000000000000000 RSI: 00007fff40636ee0 RDI: 0000000000000003
> RBP: 00007fff40636f50 R08: 0000000067d7b7e9 R09: 0000000000000050
> R10: 0000000000000001 R11: 0000000000000202 R12: 0000000000000003
> R13: 0000000067d7b7ea R14: 000055d14b9e4040 R15: 0000000000000000
>
> Thanks a lot!
> Taehee Yoo
Sorry, I completely missed that, I think this is similar to:
https://lore.kernel.org/netdev/Z-GDBlDsnPyc21RM@mini-arch/T/#u
?
Can you give it a quick test with the patches from that link?
Powered by blists - more mailing lists