Hi Ying,
I sometimes get the following deadlock warning. As I understand it the reason
is that we are calling rhashtable_walk_enter() in a tmer, i.e., in an SW
interrupt, something that is not permitted. Do you agree with this
interpretation? Since you have worked more with these hash tables than I have,
can you see any easy solution to it. I would hate to introduce a work queue to
solve this...
///jon
[346769.617370] ================================
[346769.618331] WARNING: inconsistent lock state
[346769.619187] 4.19.0-rc6+ #27 Tainted: G E
[346769.619651] --------------------------------
[346769.619651] inconsistent {SOFTIRQ-ON-W} -> {IN-SOFTIRQ-W} usage.
[346769.619651] swapper/3/0 [HC0[0]:SC1[3]:HE1:SE0] takes:
[346769.619651] 000000009e3eefe1 (&(&ht->lock)->rlock){+.?.}, at:
rhashtable_walk_enter+0x36/0xb0
[346769.619651] {SOFTIRQ-ON-W} state was registered at:
[346769.619651] _raw_spin_lock+0x29/0x60
[346769.619651] rht_deferred_worker+0x556/0x810
[346769.619651] process_one_work+0x1f5/0x540
[346769.619651] worker_thread+0x64/0x3e0
[346769.619651] kthread+0x112/0x150
[346769.619651] ret_from_fork+0x3a/0x50
[346769.619651] irq event stamp: 8581694
[346769.619651] hardirqs last enabled at (8581694): [<ffffffff9107f290>]
__local_bh_enable_ip+0x80 /0x100
[346769.619651] hardirqs last disabled at (8581693): [<ffffffff9107f257>]
__local_bh_enable_ip+0x47 /0x100
[346769.619651] softirqs last enabled at (8581678): [<ffffffff9107fa1e>]
irq_enter+0x5e/0x60
[346769.619651] softirqs last disabled at (8581679): [<ffffffff9107fadb>]
irq_exit+0xbb/0xc0
[346769.619651]
[346769.619651] other info that might help us debug this:
[346769.619651] Possible unsafe locking scenario:
[346769.619651]
[346769.619651] CPU0
[346769.619651] ----
[346769.619651] lock(&(&ht->lock)->rlock);
[346769.619651] <Interrupt>
[346769.619651] lock(&(&ht->lock)->rlock);
[346769.619651]
[346769.619651] *** DEADLOCK ***
[346769.619651]
[346769.619651] 2 locks held by swapper/3/0:
[346769.619651] #0: 00000000d9e59d74 ((&d->timer)){+.-.}, at:
call_timer_fn+0x5/0x280
[346769.619651] #1: 000000000bf452d8 (&(&d->lock)->rlock){+.-.}, at:
tipc_disc_timeout+0xc8/0x540 [tipc]
[346769.619651]
[346769.619651] stack backtrace:
[346769.619651] CPU: 3 PID: 0 Comm: swapper/3 Tainted: G E
4.19.0-rc6+ #27
[346769.619651] Hardware name: Bochs Bochs, BIOS Bochs 01/01/2011
[346769.619651] Call Trace:
[346769.619651] <IRQ>
[346769.619651] dump_stack+0x78/0xb3
[346769.619651] print_usage_bug+0x1ed/0x20b
[346769.619651] mark_lock+0x5ca/0x640
[346769.619651] __lock_acquire+0x41f/0x1b10
[346769.619651] ? sched_clock_local+0x12/0x80
[346769.619651] ? lock_acquire+0xb3/0x190
[346769.619651] lock_acquire+0xb3/0x190
[346769.619651] ? rhashtable_walk_enter+0x36/0xb0
[346769.651609] _raw_spin_lock+0x29/0x60
[346769.651609] ? rhashtable_walk_enter+0x36/0xb0
[346769.651609] rhashtable_walk_enter+0x36/0xb0
[346769.651609] tipc_sk_reinit+0xb0/0x410 [tipc]
[346769.651609] ? mark_held_locks+0x6f/0x90
[346769.651609] ? __local_bh_enable_ip+0x80/0x100
[346769.651609] ? debug_show_all_locks+0x170/0x190
[346769.651609] tipc_net_finalize+0xbf/0x180 [tipc]
[346769.651609] tipc_disc_timeout+0x509/0x540 [tipc]
_______________________________________________
tipc-discussion mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/tipc-discussion
