Skip to contentSkip to conversation

bonding: rate-limit printing a warning message

PATCH In review Mahesh Bandewar 2 participants 3 messages Jan 26, 2022 – Jan 27, 2022

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 bonding: rate-limit printing a warning message

dev.c:get_link_speed() schedules a work-queue aggressively when it fails to get the speed. If the link n question is a bond device which may have multiple links to iterate through to get the link speed. If the underlying link(s) has/have issues, bonding driver prints a link-status message and this doesn't go well with the aggressive work-queue scheduling and results in a rcu stall. This fix just adds a ratelimiter to the message printing to avoid the stall.

Call Trace:
 <IRQ>
 __dump_stack lib/dump_stack.c:17 [inline]
 dump_stack+0x14d/0x20b lib/dump_stack.c:53
 nmi_cpu_backtrace.cold+0x19/0x98 lib/nmi_backtrace.c:103
 nmi_trigger_cpumask_backtrace+0x16a/0x17e lib/nmi_backtrace.c:62
 arch_trigger_cpumask_backtrace+0x14/0x20 arch/x86/kernel/apic/hw_nmi.c:38
 trigger_single_cpu_backtrace include/linux/nmi.h:161 [inline]
 rcu_dump_cpu_stacks+0x183/0x1cf kernel/rcu/tree.c:1210
 print_cpu_stall kernel/rcu/tree.c:1349 [inline]
 check_cpu_stall kernel/rcu/tree.c:1423 [inline]
 rcu_pending kernel/rcu/tree.c:3010 [inline]
 rcu_check_callbacks.cold+0x494/0x7d3 kernel/rcu/tree.c:2551
 update_process_times+0x32/0x80 kernel/time/timer.c:1641
 tick_sched_handle+0xa0/0x180 kernel/time/tick-sched.c:161
 tick_sched_timer+0x44/0x130 kernel/time/tick-sched.c:1193
 __run_hrtimer kernel/time/hrtimer.c:1396 [inline]
 __hrtimer_run_queues+0x304/0xd80 kernel/time/hrtimer.c:1458
 hrtimer_interrupt+0x2ea/0x730 kernel/time/hrtimer.c:1516
 local_apic_timer_interrupt arch/x86/kernel/apic/apic.c:1031 [inline]
 smp_apic_timer_interrupt+0x150/0x5e0 arch/x86/kernel/apic/apic.c:1056
 apic_timer_interrupt+0x93/0xa0 arch/x86/entry/entry_64.S:780
 </IRQ>
RIP: 0010:arch_local_irq_restore arch/x86/include/asm/paravirt.h:783 [inline]
RIP: 0010:console_unlock+0x82b/0xcc0 kernel/printk/printk.c:2302
 RSP: 0018:ffff8801966cb9e8 EFLAGS: 00000293 ORIG_RAX: ffffffffffffff12
RAX: ffff8801968d0040 RBX: 0000000000000000 RCX: 0000000000000006
RDX: 0000000000000000 RSI: ffffffff815a6515 RDI: 0000000000000293
RBP: ffff8801966cba70 R08: ffff8801968d0040 R09: 0000000000000000
R10: 0000000000000000 R11: 0000000000000000 R12: 0000000000000000
R13: ffffffff82cc61b0 R14: dffffc0000000000 R15: 0000000000000001
 vprintk_emit+0x593/0x610 kernel/printk/printk.c:1836
 vprintk_default+0x28/0x30 kernel/printk/printk.c:1876
 vprintk_func+0x7a/0xed kernel/printk/printk_safe.c:379
 printk+0xba/0xed kernel/printk/printk.c:1909
 get_link_speed.cold+0x43/0x144 net/core/dev.c:1493
 get_link_speed_work+0x1e/0x30 net/core/dev.c:1515
 process_one_work+0x881/0x1560 kernel/workqueue.c:2147
 worker_thread+0x653/0x1150 kernel/workqueue.c:2281
 kthread+0x345/0x410 kernel/kthread.c:246
 ret_from_fork+0x3f/0x50 arch/x86/entry/entry_64.S:393
---
 drivers/net/bonding/bond_main.c | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)
drivers/net/bonding/bond_main.c +6 -3
index 238b56d77c36..98b37af3fb6b 100644
@@ -2538,9 +2538,12 @@ static int bond_miimon_inspect(struct bonding *bond)
 				/* recovered before downdelay expired */
 				bond_propose_link_state(slave, BOND_LINK_UP);
 				slave->last_link_up = jiffies;
-				slave_info(bond->dev, slave->dev, "link status up again after %d ms\n",
-					   (bond->params.downdelay - slave->delay) *
-					   bond->params.miimon);
+				if (net_ratelimit())
+					slave_info(bond->dev, slave->dev,
+						   "link status up again after %d ms\n",
+						   (bond->params.downdelay -
+						    slave->delay) *
+						   bond->params.miimon);
 				commit++;
 				continue;
 			}
Signature · 1 line
2.35.0.rc0.227.g00780c9af4-goog
#1 lore raw
Quoted 7 lines · Mahesh Bandewar

dev.c:get_link_speed() schedules a work-queue aggressively when it fails to get the speed. If the link n question is a bond device which may have multiple links to iterate through to get the link speed. If the underlying link(s) has/have issues, bonding driver prints a link-status message and this doesn't go well with the aggressive work-queue scheduling and results in a rcu stall. This fix just adds a ratelimiter to the message printing to avoid the stall.

I don't see a get_link_speed() function in net-next
net/core/dev.c; am I missing something?

	-J
Quoted 70 lines
Call Trace:
<IRQ>
__dump_stack lib/dump_stack.c:17 [inline]
dump_stack+0x14d/0x20b lib/dump_stack.c:53
nmi_cpu_backtrace.cold+0x19/0x98 lib/nmi_backtrace.c:103
nmi_trigger_cpumask_backtrace+0x16a/0x17e lib/nmi_backtrace.c:62
arch_trigger_cpumask_backtrace+0x14/0x20 arch/x86/kernel/apic/hw_nmi.c:38
trigger_single_cpu_backtrace include/linux/nmi.h:161 [inline]
rcu_dump_cpu_stacks+0x183/0x1cf kernel/rcu/tree.c:1210
print_cpu_stall kernel/rcu/tree.c:1349 [inline]
check_cpu_stall kernel/rcu/tree.c:1423 [inline]
rcu_pending kernel/rcu/tree.c:3010 [inline]
rcu_check_callbacks.cold+0x494/0x7d3 kernel/rcu/tree.c:2551
update_process_times+0x32/0x80 kernel/time/timer.c:1641
tick_sched_handle+0xa0/0x180 kernel/time/tick-sched.c:161
tick_sched_timer+0x44/0x130 kernel/time/tick-sched.c:1193
__run_hrtimer kernel/time/hrtimer.c:1396 [inline]
__hrtimer_run_queues+0x304/0xd80 kernel/time/hrtimer.c:1458
hrtimer_interrupt+0x2ea/0x730 kernel/time/hrtimer.c:1516
local_apic_timer_interrupt arch/x86/kernel/apic/apic.c:1031 [inline]
smp_apic_timer_interrupt+0x150/0x5e0 arch/x86/kernel/apic/apic.c:1056
apic_timer_interrupt+0x93/0xa0 arch/x86/entry/entry_64.S:780
</IRQ>
RIP: 0010:arch_local_irq_restore arch/x86/include/asm/paravirt.h:783 [inline]
RIP: 0010:console_unlock+0x82b/0xcc0 kernel/printk/printk.c:2302
RSP: 0018:ffff8801966cb9e8 EFLAGS: 00000293 ORIG_RAX: ffffffffffffff12
RAX: ffff8801968d0040 RBX: 0000000000000000 RCX: 0000000000000006
RDX: 0000000000000000 RSI: ffffffff815a6515 RDI: 0000000000000293
RBP: ffff8801966cba70 R08: ffff8801968d0040 R09: 0000000000000000
R10: 0000000000000000 R11: 0000000000000000 R12: 0000000000000000
R13: ffffffff82cc61b0 R14: dffffc0000000000 R15: 0000000000000001
vprintk_emit+0x593/0x610 kernel/printk/printk.c:1836
vprintk_default+0x28/0x30 kernel/printk/printk.c:1876
vprintk_func+0x7a/0xed kernel/printk/printk_safe.c:379
printk+0xba/0xed kernel/printk/printk.c:1909
get_link_speed.cold+0x43/0x144 net/core/dev.c:1493
get_link_speed_work+0x1e/0x30 net/core/dev.c:1515
process_one_work+0x881/0x1560 kernel/workqueue.c:2147
worker_thread+0x653/0x1150 kernel/workqueue.c:2281
kthread+0x345/0x410 kernel/kthread.c:246
ret_from_fork+0x3f/0x50 arch/x86/entry/entry_64.S:393

---
drivers/net/bonding/bond_main.c | 9 ++++++---

1 file changed, 6 insertions(+), 3 deletions(-)
drivers/net/bonding/bond_main.c +6 -3
index 238b56d77c36..98b37af3fb6b 100644
@@ -2538,9 +2538,12 @@ static int bond_miimon_inspect(struct bonding *bond)
 				/* recovered before downdelay expired */
 				bond_propose_link_state(slave, BOND_LINK_UP);
 				slave->last_link_up = jiffies;
-				slave_info(bond->dev, slave->dev, "link status up again after %d ms\n",
-					   (bond->params.downdelay - slave->delay) *
-					   bond->params.miimon);
+				if (net_ratelimit())
+					slave_info(bond->dev, slave->dev,
+						   "link status up again after %d ms\n",
+						   (bond->params.downdelay -
+						    slave->delay) *
+						   bond->params.miimon);
 				commit++;
 				continue;
 			}
Signature · 1 line
2.35.0.rc0.227.g00780c9af4-goog

---
-Jay Vosburgh, [email protected]

#2 lore raw
Quoted 11 lines · Jay Vosburgh
Quoted 7 lines · Mahesh Bandewar

dev.c:get_link_speed() schedules a work-queue aggressively when it fails to get the speed. If the link n question is a bond device which may have multiple links to iterate through to get the link speed. If the underlying link(s) has/have issues, bonding driver prints a link-status message and this doesn't go well with the aggressive work-queue scheduling and results in a rcu stall. This fix just adds a ratelimiter to the message printing to avoid the stall.

Quoted 3 lines

I don't see a get_link_speed() function in net-next
net/core/dev.c; am I missing something?

Oops, looks like this case is relevant to our kernel only, please ignore the patch.

Quoted 75 lines
        -J
Quoted 68 lines
Call Trace:
<IRQ>
__dump_stack lib/dump_stack.c:17 [inline]
dump_stack+0x14d/0x20b lib/dump_stack.c:53
nmi_cpu_backtrace.cold+0x19/0x98 lib/nmi_backtrace.c:103
nmi_trigger_cpumask_backtrace+0x16a/0x17e lib/nmi_backtrace.c:62
arch_trigger_cpumask_backtrace+0x14/0x20 arch/x86/kernel/apic/hw_nmi.c:38
trigger_single_cpu_backtrace include/linux/nmi.h:161 [inline]
rcu_dump_cpu_stacks+0x183/0x1cf kernel/rcu/tree.c:1210
print_cpu_stall kernel/rcu/tree.c:1349 [inline]
check_cpu_stall kernel/rcu/tree.c:1423 [inline]
rcu_pending kernel/rcu/tree.c:3010 [inline]
rcu_check_callbacks.cold+0x494/0x7d3 kernel/rcu/tree.c:2551
update_process_times+0x32/0x80 kernel/time/timer.c:1641
tick_sched_handle+0xa0/0x180 kernel/time/tick-sched.c:161
tick_sched_timer+0x44/0x130 kernel/time/tick-sched.c:1193
__run_hrtimer kernel/time/hrtimer.c:1396 [inline]
__hrtimer_run_queues+0x304/0xd80 kernel/time/hrtimer.c:1458
hrtimer_interrupt+0x2ea/0x730 kernel/time/hrtimer.c:1516
local_apic_timer_interrupt arch/x86/kernel/apic/apic.c:1031 [inline]
smp_apic_timer_interrupt+0x150/0x5e0 arch/x86/kernel/apic/apic.c:1056
apic_timer_interrupt+0x93/0xa0 arch/x86/entry/entry_64.S:780
</IRQ>
RIP: 0010:arch_local_irq_restore arch/x86/include/asm/paravirt.h:783 [inline]
RIP: 0010:console_unlock+0x82b/0xcc0 kernel/printk/printk.c:2302
RSP: 0018:ffff8801966cb9e8 EFLAGS: 00000293 ORIG_RAX: ffffffffffffff12
RAX: ffff8801968d0040 RBX: 0000000000000000 RCX: 0000000000000006
RDX: 0000000000000000 RSI: ffffffff815a6515 RDI: 0000000000000293
RBP: ffff8801966cba70 R08: ffff8801968d0040 R09: 0000000000000000
R10: 0000000000000000 R11: 0000000000000000 R12: 0000000000000000
R13: ffffffff82cc61b0 R14: dffffc0000000000 R15: 0000000000000001
vprintk_emit+0x593/0x610 kernel/printk/printk.c:1836
vprintk_default+0x28/0x30 kernel/printk/printk.c:1876
vprintk_func+0x7a/0xed kernel/printk/printk_safe.c:379
printk+0xba/0xed kernel/printk/printk.c:1909
get_link_speed.cold+0x43/0x144 net/core/dev.c:1493
get_link_speed_work+0x1e/0x30 net/core/dev.c:1515
process_one_work+0x881/0x1560 kernel/workqueue.c:2147
worker_thread+0x653/0x1150 kernel/workqueue.c:2281
kthread+0x345/0x410 kernel/kthread.c:246
ret_from_fork+0x3f/0x50 arch/x86/entry/entry_64.S:393

---
drivers/net/bonding/bond_main.c | 9 ++++++---

1 file changed, 6 insertions(+), 3 deletions(-)
drivers/net/bonding/bond_main.c +6 -4
index 238b56d77c36..98b37af3fb6b 100644
@@ -2538,9 +2538,12 @@ static int bond_miimon_inspect(struct bonding *bond)
                              /* recovered before downdelay expired */
                              bond_propose_link_state(slave, BOND_LINK_UP);
                              slave->last_link_up = jiffies;
-                              slave_info(bond->dev, slave->dev, "link status up again after %d ms\n",
-                                         (bond->params.downdelay - slave->delay) *
-                                         bond->params.miimon);
+                              if (net_ratelimit())
+                                      slave_info(bond->dev, slave->dev,
+                                                 "link status up again after %d ms\n",
+                                                 (bond->params.downdelay -
+                                                  slave->delay) *
+                                                 bond->params.miimon);
                              commit++;
                              continue;
                      }
--
Quoted 2 lines

2.35.0.rc0.227.g00780c9af4-goog

---
-Jay Vosburgh, [email protected]