From: Johannes Berg <johannes@sipsolutions.net>
To: Anton Ivanov <anton.ivanov@kot-begemot.co.uk>,
linux-um@lists.infradead.org
Subject: Re: [PATCH 7/7] um: simplify IRQ handling code
Date: Wed, 02 Dec 2020 12:49:07 +0100 [thread overview]
Message-ID: <1dfcecf63871aa26a3828f1e99227f719085fc8d.camel@sipsolutions.net> (raw)
In-Reply-To: <96e31171-965d-c1d7-282e-b1b0bec6b73f@kot-begemot.co.uk>
> run iperf -s in the UML with a vector network driver.
Btw, 'ip link set vec0 up' reports a lockdep issue.
Looks sort of legitimate, though we're not SMP, so ...
======================================================
WARNING: possible circular locking dependency detected
5.10.0-rc4-00392-gedc4ff2eb69b-dirty #147 Not tainted
------------------------------------------------------
swapper/0 is trying to acquire lock:
0000000063701138 (&result->tail_lock){+.-.}-{2:2}, at: vector_send.isra.0+0x1ba/0x280
but task is already holding lock:
00000000637010f0 (&result->head_lock){+.-.}-{2:2}, at: vector_send.isra.0+0x30/0x280
which lock already depends on the new lock.
the existing dependency chain (in reverse order) is:
-> #1 (&result->head_lock){+.-.}-{2:2}:
save_stack_trace+0x2e/0x30
stack_trace_save+0x34/0x39
save_trace+0x8c/0x261
__lock_acquire+0x1480/0x16b8
lock_acquire+0x335/0x414
_raw_spin_lock+0x31/0x85
vector_net_start_xmit+0x164/0x340
netdev_start_xmit+0x1f/0x47
dev_hard_start_xmit+0x14b/0x230
sch_direct_xmit+0xb0/0x248
__qdisc_run+0x4f3/0x532
qdisc_run+0x3d/0x51
__dev_queue_xmit+0x2e0/0x6e8
dev_queue_xmit+0x12/0x14
neigh_resolve_output+0x13f/0x162
ip6_finish_output2+0x642/0x713
__ip6_finish_output+0xac/0xb5
ip6_output+0xcf/0x113
dst_output+0x72/0x7b
NF_HOOK.constprop.0+0x115/0x124
mld_sendpack+0x20e/0x2ba
mld_ifc_timer_expire+0x259/0x2ab
call_timer_fn+0x116/0x243
__run_timers+0x207/0x246
run_timer_softirq+0x1c/0x2c
__do_softirq+0x1b9/0x43e
irq_exit+0xc1/0x113
do_IRQ+0x45/0x54
timer_handler+0xce/0xed
timer_real_alarm_handler+0x5c/0x5e
unblock_signals+0x9c/0xd7
arch_cpu_idle+0x5b/0x62
default_idle_call+0x52/0x5e
do_idle+0xd8/0x14b
cpu_startup_entry+0x1e/0x20
rest_init+0x135/0x141
0x600016d2
0x60001dfe
0x60003845
new_thread_handler+0x81/0xb2
uml_finishsetup+0x54/0x59
-> #0 (&result->tail_lock){+.-.}-{2:2}:
save_stack_trace+0x2e/0x30
stack_trace_save+0x34/0x39
save_trace+0x8c/0x261
print_circular_bug+0x64/0x24b
check_noncircular+0xd3/0xe3
__lock_acquire+0x12ca/0x16b8
lock_acquire+0x335/0x414
_raw_spin_lock+0x31/0x85
vector_send.isra.0+0x1ba/0x280
vector_net_start_xmit+0x32b/0x340
netdev_start_xmit+0x1f/0x47
dev_hard_start_xmit+0x14b/0x230
sch_direct_xmit+0xb0/0x248
__qdisc_run+0x4f3/0x532
qdisc_run+0x3d/0x51
__dev_queue_xmit+0x2e0/0x6e8
dev_queue_xmit+0x12/0x14
neigh_resolve_output+0x13f/0x162
ip6_finish_output2+0x642/0x713
__ip6_finish_output+0xac/0xb5
ip6_output+0xcf/0x113
dst_output+0x72/0x7b
NF_HOOK.constprop.0+0x115/0x124
mld_sendpack+0x20e/0x2ba
mld_ifc_timer_expire+0x259/0x2ab
call_timer_fn+0x116/0x243
__run_timers+0x207/0x246
run_timer_softirq+0x1c/0x2c
__do_softirq+0x1b9/0x43e
irq_exit+0xc1/0x113
do_IRQ+0x45/0x54
timer_handler+0xce/0xed
timer_real_alarm_handler+0x5c/0x5e
unblock_signals+0x9c/0xd7
arch_cpu_idle+0x5b/0x62
default_idle_call+0x52/0x5e
do_idle+0xd8/0x14b
cpu_startup_entry+0x1e/0x20
rest_init+0x135/0x141
0x600016d2
0x60001dfe
0x60003845
new_thread_handler+0x81/0xb2
uml_finishsetup+0x54/0x59
other info that might help us debug this:
Possible unsafe locking scenario:
CPU0 CPU1
---- ----
lock(&result->head_lock);
lock(&result->tail_lock);
lock(&result->head_lock);
lock(&result->tail_lock);
*** DEADLOCK ***
8 locks held by swapper/0:
#0: 0000000060687050 ((&idev->mc_ifc_timer)){+.-.}-{0:0}, at: call_timer_fn+0x0/0x243
#1: 0000000060887e40 (rcu_read_lock){....}-{1:2}, at: rcu_lock_acquire.constprop.0+0x0/0x39
#2: 0000000060887e00 (rcu_read_lock_bh){....}-{1:2}, at: rcu_lock_acquire+0x0/0x2f
#3: 0000000060887e00 (rcu_read_lock_bh){....}-{1:2}, at: rcu_lock_acquire+0x0/0x2f
#4: 0000000063fa9258 (&sch->seqlock){+...}-{2:2}, at: qdisc_run_begin+0x22/0x76
#5: 0000000063fa9140 (dev->qdisc_running_key ?: &qdisc_running_key){+...}-{0:0}, at: qdisc_run+0x16/0x51
#6: 0000000063636e88 (_xmit_ETHER#2){+.-.}-{2:2}, at: sch_direct_xmit+0x84/0x248
#7: 00000000637010f0 (&result->head_lock){+.-.}-{2:2}, at: vector_send.isra.0+0x30/0x280
stack backtrace:
CPU: 0 PID: 0 Comm: swapper Not tainted 5.10.0-rc4-00392-gedc4ff2eb69b-dirty #147
Stack:
605d20dd 60c14280 60686830 604765ed
60475ac7 60c14280 60c14180 60c14280
60686840 60477bcd 606868a0 6007f8e4
Call Trace:
[<604765ed>] ? printk+0x0/0x94
[<60025c1b>] show_stack+0x13e/0x14d
[<604765ed>] ? printk+0x0/0x94
[<60475ac7>] ? __print_lock_name+0x0/0x90
[<60477bcd>] dump_stack+0x34/0x36
[<6007f8e4>] print_circular_bug+0x23c/0x24b
[<6007d8de>] ? save_trace+0x8c/0x261
[<6007f9c6>] check_noncircular+0xd3/0xe3
[<6007dda1>] ? hlock_class+0x1e/0x9e
[<60080880>] ? mark_lock.part.0+0x0/0x410
[<6007dd83>] ? hlock_class+0x0/0x9e
[<6008304e>] __lock_acquire+0x12ca/0x16b8
[<6003b4b5>] ? set_signals+0x37/0x3f
[<6007dd83>] ? hlock_class+0x0/0x9e
[<60081515>] lock_acquire+0x335/0x414
[<6002db48>] ? vector_send.isra.0+0x1ba/0x280
[<6047a506>] ? debug_lockdep_rcu_enabled+0x0/0x3b
[<6047f337>] _raw_spin_lock+0x31/0x85
[<6002db48>] ? vector_send.isra.0+0x1ba/0x280
[<6047f306>] ? _raw_spin_lock+0x0/0x85
[<6002db48>] vector_send.isra.0+0x1ba/0x280
[<6047f7cb>] ? _raw_spin_unlock+0x0/0x32
[<6002e06d>] vector_net_start_xmit+0x32b/0x340
[<6032845a>] netdev_start_xmit+0x1f/0x47
[<60479fa2>] ? match_held_lock+0x0/0x1de
[<6047a506>] ? debug_lockdep_rcu_enabled+0x0/0x3b
[<60330d44>] dev_hard_start_xmit+0x14b/0x230
[<6047a506>] ? debug_lockdep_rcu_enabled+0x0/0x3b
[<60367250>] sch_direct_xmit+0xb0/0x248
[<60365ee5>] ? qdisc_qstats_cpu_qlen_dec+0x2a/0x2e
[<60365ebb>] ? qdisc_qstats_cpu_qlen_dec+0x0/0x2e
[<603678db>] __qdisc_run+0x4f3/0x532
[<6047a506>] ? debug_lockdep_rcu_enabled+0x0/0x3b
[<6032d233>] qdisc_run+0x3d/0x51
[<603311af>] __dev_queue_xmit+0x2e0/0x6e8
[<6047a506>] ? debug_lockdep_rcu_enabled+0x0/0x3b
[<60338c91>] ? read_seqbegin+0x0/0xc9
[<603315c9>] dev_queue_xmit+0x12/0x14
[<6033d0d9>] neigh_resolve_output+0x13f/0x162
[<6047a506>] ? debug_lockdep_rcu_enabled+0x0/0x3b
[<603f8e94>] ip6_finish_output2+0x642/0x713
[<603fa9ef>] __ip6_finish_output+0xac/0xb5
[<603faad9>] ip6_output+0xcf/0x113
[<60424cac>] ? dst_output+0x0/0x7b
[<60424d1e>] dst_output+0x72/0x7b
[<6042571a>] NF_HOOK.constprop.0+0x115/0x124
[<6041058a>] ? icmp6_dst_alloc+0xea/0x103
[<6003b47e>] ? set_signals+0x0/0x3f
[<60425937>] mld_sendpack+0x20e/0x2ba
[<60427902>] ? add_grec+0x0/0x3d0
[<6047f493>] ? _raw_spin_lock_bh+0x0/0x9b
[<60428340>] mld_ifc_timer_expire+0x259/0x2ab
[<604280e7>] ? mld_ifc_timer_expire+0x0/0x2ab
[<6047a506>] ? debug_lockdep_rcu_enabled+0x0/0x3b
[<6009a795>] call_timer_fn+0x116/0x243
[<6009a67f>] ? call_timer_fn+0x0/0x243
[<6009b02e>] __run_timers+0x207/0x246
[<604280e7>] ? mld_ifc_timer_expire+0x0/0x2ab
[<6047f52e>] ? _raw_spin_lock_irq+0x0/0xb2
[<6003b4a7>] ? set_signals+0x29/0x3f
[<6047a4f4>] ? lock_is_held_type+0x135/0x147
[<6047a362>] ? lockdep_hardirqs_on+0x1e2/0x23f
[<6009ae27>] ? __run_timers+0x0/0x246
[<6009b089>] run_timer_softirq+0x1c/0x2c
[<60480331>] __do_softirq+0x1b9/0x43e
[<600953e3>] ? unmask_irq+0x0/0x37
[<6047f7cb>] ? _raw_spin_unlock+0x0/0x32
[<60022c6b>] ? rcu_lock_acquire.constprop.0+0x0/0x39
[<600492c8>] irq_exit+0xc1/0x113
[<60049205>] ? irq_enter+0x10/0x12
[<60024138>] do_IRQ+0x45/0x54
[<6003b47e>] ? set_signals+0x0/0x3f
[<60026ba5>] timer_handler+0xce/0xed
[<600838e7>] ? lock_release+0x0/0x36c
[<60061e0a>] ? find_task_by_pid_ns+0x0/0x97
[<6003b03a>] timer_real_alarm_handler+0x5c/0x5e
[<6003b439>] unblock_signals+0x9c/0xd7
[<60024dac>] arch_cpu_idle+0x5b/0x62
[<60097200>] ? rcu_read_lock_sched_held+0x2d/0x34
[<600709f6>] ? trace_cpu_idle.constprop.0+0x0/0xa1
[<6047f2aa>] default_idle_call+0x52/0x5e
[<600838e7>] ? lock_release+0x0/0x36c
[<60061e0a>] ? find_task_by_pid_ns+0x0/0x97
[<6003b474>] ? get_signals+0x0/0xa
[<60070c21>] do_idle+0xd8/0x14b
[<6047ba87>] ? schedule+0x95/0xd6
[<60070b49>] ? do_idle+0x0/0x14b
[<6007101f>] cpu_startup_entry+0x1e/0x20
[<60071001>] ? cpu_startup_entry+0x0/0x20
[<6047a506>] ? debug_lockdep_rcu_enabled+0x0/0x3b
[<6047a6c5>] rest_init+0x135/0x141
[<6047a590>] ? rest_init+0x0/0x141
[<6003b474>] ? get_signals+0x0/0xa
[<604765ed>] ? printk+0x0/0x94
[<602b569e>] ? strlen+0x0/0x11
[<600016d2>] 0x600016d2
[<60001dfe>] 0x60001dfe
[<6003b38c>] ? block_signals+0x0/0x11
[<60003845>] 0x60003845
[<60024816>] new_thread_handler+0x81/0xb2
[<600037fa>] ? 0x600037fa
[<600284b7>] uml_finishsetup+0x54/0x59
johannes
_______________________________________________
linux-um mailing list
linux-um@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-um
next prev parent reply other threads:[~2020-12-02 11:49 UTC|newest]
Thread overview: 42+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-11-23 19:56 [PATCH 0/7 um: IRQ handling cleanups Johannes Berg
2020-11-23 19:56 ` [PATCH 1/7] um: support dynamic IRQ allocation Johannes Berg
2020-11-30 11:26 ` Anton Ivanov
2020-11-23 19:56 ` [PATCH 2/7] um: virtio: use " Johannes Berg
2020-11-30 13:45 ` Anton Ivanov
2020-11-23 19:56 ` [PATCH 3/7] um: clean up alarm IRQ chip name Johannes Berg
2020-11-30 13:54 ` Anton Ivanov
2020-11-23 19:56 ` [PATCH 4/7] um: irq: clean up and rename struct irq_fd Johannes Berg
2020-11-30 14:01 ` Anton Ivanov
2020-11-23 19:56 ` [PATCH 5/7] um: irq: reduce irq_reg allocation Johannes Berg
2020-11-23 19:56 ` [PATCH 6/7] um: remove IRQ_NONE type Johannes Berg
2020-11-30 14:31 ` Anton Ivanov
2020-11-23 19:56 ` [PATCH 7/7] um: simplify IRQ handling code Johannes Berg
2020-11-24 21:50 ` Anton Ivanov
2020-11-24 21:58 ` Johannes Berg
2020-11-24 22:36 ` Anton Ivanov
2020-11-30 12:00 ` Johannes Berg
2020-11-30 13:40 ` Anton Ivanov
2020-11-30 16:30 ` Anton Ivanov
2020-12-01 20:15 ` Johannes Berg
2020-12-02 9:03 ` Anton Ivanov
2020-12-02 11:29 ` Johannes Berg
2020-12-02 11:31 ` Anton Ivanov
2020-12-02 11:32 ` Johannes Berg
2020-12-02 11:56 ` Anton Ivanov
2020-12-02 11:58 ` Johannes Berg
2020-12-02 11:43 ` Johannes Berg
2020-12-02 11:49 ` Johannes Berg [this message]
2020-12-02 11:54 ` Johannes Berg
2020-12-02 11:58 ` Anton Ivanov
2020-11-24 8:14 ` [PATCH 0/7 um: IRQ handling cleanups Anton Ivanov
2020-11-24 8:55 ` Johannes Berg
2020-11-24 8:58 ` Johannes Berg
2020-11-24 9:06 ` Anton Ivanov
2020-11-24 16:34 ` Anton Ivanov
2020-11-24 16:39 ` Johannes Berg
2020-11-24 16:42 ` Anton Ivanov
2020-11-24 16:45 ` Anton Ivanov
2020-11-24 16:46 ` Johannes Berg
2020-11-24 17:07 ` Anton Ivanov
2020-11-24 18:32 ` Johannes Berg
2020-11-24 21:11 ` Johannes Berg
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1dfcecf63871aa26a3828f1e99227f719085fc8d.camel@sipsolutions.net \
--to=johannes@sipsolutions.net \
--cc=anton.ivanov@kot-begemot.co.uk \
--cc=linux-um@lists.infradead.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox