Skip to contentSkip to conversation

sched: Use RCU-sched in core-scheduling balancing logic

PATCH In review Joel Fernandes (Google) 3 participants 9 messages Mar 13, 2020 – Mar 25, 2020

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 sched: Use RCU-sched in core-scheduling balancing logic

rcu_read_unlock() can incur an infrequent deadlock in sched_core_balance(). Fix this by using the RCU-sched flavor instead.

This fixes the following spinlock recursion observed when testing the core scheduling patches on PREEMPT=y kernel on ChromeOS:

[    3.240891] BUG: spinlock recursion on CPU#2, swapper/2/0
[    3.240900]  lock: 0xffff9cd1eeb28e40, .magic: dead4ead, .owner: swapper/2/0, .owner_cpu: 2
[    3.240905] CPU: 2 PID: 0 Comm: swapper/2 Not tainted 5.4.22htcore #4
[    3.240908] Hardware name: Google Eve/Eve, BIOS Google_Eve.9584.174.0 05/29/2018
[    3.240910] Call Trace:
[    3.240919]  dump_stack+0x97/0xdb
[    3.240924]  ? spin_bug+0xa4/0xb1
[    3.240927]  do_raw_spin_lock+0x79/0x98
[    3.240931]  try_to_wake_up+0x367/0x61b
[    3.240935]  rcu_read_unlock_special+0xde/0x169
[    3.240938]  ? sched_core_balance+0xd9/0x11e
[    3.240941]  __rcu_read_unlock+0x48/0x4a
[    3.240945]  __balance_callback+0x50/0xa1
[    3.240949]  __schedule+0x55a/0x61e
[    3.240952]  schedule_idle+0x21/0x2d
[    3.240956]  do_idle+0x1d5/0x1f8
[    3.240960]  cpu_startup_entry+0x1d/0x1f
[    3.240964]  start_secondary+0x159/0x174
[    3.240967]  secondary_startup_64+0xa4/0xb0
[   14.998590] watchdog: BUG: soft lockup - CPU#0 stuck for 11s! [kworker/0:10:965]
---
 kernel/sched/core.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
kernel/sched/core.c +2 -2
index 3045bd50e249..037e8f2e2686 100644
@@ -4735,7 +4735,7 @@ static void sched_core_balance(struct rq *rq)
 	struct sched_domain *sd;
 	int cpu = cpu_of(rq);
 
-	rcu_read_lock();
+	rcu_read_lock_sched();
 	raw_spin_unlock_irq(rq_lockp(rq));
 	for_each_domain(cpu, sd) {
 		if (!(sd->flags & SD_LOAD_BALANCE))
@@ -4748,7 +4748,7 @@ static void sched_core_balance(struct rq *rq)
 			break;
 	}
 	raw_spin_lock_irq(rq_lockp(rq));
-	rcu_read_unlock();
+	rcu_read_unlock_sched();
 }
 
 static DEFINE_PER_CPU(struct callback_head, core_balance_head);
Signature · 1 line
2.25.1.481.gfbce0eb801-goog
maintainer Acked-by on v1 1/1
#1 lore raw
Quoted 27 lines · Joel Fernandes (Google)

rcu_read_unlock() can incur an infrequent deadlock in sched_core_balance(). Fix this by using the RCU-sched flavor instead.

This fixes the following spinlock recursion observed when testing the core scheduling patches on PREEMPT=y kernel on ChromeOS:

[    3.240891] BUG: spinlock recursion on CPU#2, swapper/2/0
[    3.240900]  lock: 0xffff9cd1eeb28e40, .magic: dead4ead, .owner: swapper/2/0, .owner_cpu: 2
[    3.240905] CPU: 2 PID: 0 Comm: swapper/2 Not tainted 5.4.22htcore #4
[    3.240908] Hardware name: Google Eve/Eve, BIOS Google_Eve.9584.174.0 05/29/2018
[    3.240910] Call Trace:
[    3.240919]  dump_stack+0x97/0xdb
[    3.240924]  ? spin_bug+0xa4/0xb1
[    3.240927]  do_raw_spin_lock+0x79/0x98
[    3.240931]  try_to_wake_up+0x367/0x61b
[    3.240935]  rcu_read_unlock_special+0xde/0x169
[    3.240938]  ? sched_core_balance+0xd9/0x11e
[    3.240941]  __rcu_read_unlock+0x48/0x4a
[    3.240945]  __balance_callback+0x50/0xa1
[    3.240949]  __schedule+0x55a/0x61e
[    3.240952]  schedule_idle+0x21/0x2d
[    3.240956]  do_idle+0x1d5/0x1f8
[    3.240960]  cpu_startup_entry+0x1d/0x1f
[    3.240964]  start_secondary+0x159/0x174
[    3.240967]  secondary_startup_64+0xa4/0xb0
[   14.998590] watchdog: BUG: soft lockup - CPU#0 stuck for 11s! [kworker/0:10:965]
Quoted 6 lines

The original could indeed deadlock, and this would avoid that deadlock. (The commit to solve this deadlock is sadly not yet in mainline.)

Quoted 29 lines
---
 kernel/sched/core.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
kernel/sched/core.c +2 -2
index 3045bd50e249..037e8f2e2686 100644
@@ -4735,7 +4735,7 @@ static void sched_core_balance(struct rq *rq)
 	struct sched_domain *sd;
 	int cpu = cpu_of(rq);
 
-	rcu_read_lock();
+	rcu_read_lock_sched();
 	raw_spin_unlock_irq(rq_lockp(rq));
 	for_each_domain(cpu, sd) {
 		if (!(sd->flags & SD_LOAD_BALANCE))
@@ -4748,7 +4748,7 @@ static void sched_core_balance(struct rq *rq)
 			break;
 	}
 	raw_spin_lock_irq(rq_lockp(rq));
-	rcu_read_unlock();
+	rcu_read_unlock_sched();
 }
 
 static DEFINE_PER_CPU(struct callback_head, core_balance_head);
Signature · 1 line
2.25.1.481.gfbce0eb801-goog
#2 lore raw
Quoted 7 lines · Paul E. McKenney
Quoted 6 lines · Joel Fernandes (Google)

rcu_read_unlock() can incur an infrequent deadlock in sched_core_balance(). Fix this by using the RCU-sched flavor instead.

This fixes the following spinlock recursion observed when testing the core scheduling patches on PREEMPT=y kernel on ChromeOS:

Quoted 7 lines
Quoted 2 lines
[   14.998590] watchdog: BUG: soft lockup - CPU#0 stuck for 11s! [kworker/0:10:965]

The original could indeed deadlock, and this would avoid that deadlock. (The commit to solve this deadlock is sadly not yet in mainline.)

I saw this in dmesg with this patch, is it expected?

Thanks,
-Aubrey

[  117.000905] =============================
[  117.000907] WARNING: suspicious RCU usage
[  117.000911] 5.5.7+ #160 Not tainted
[  117.000913] -----------------------------
[  117.000916] kernel/sched/core.c:4747 suspicious rcu_dereference_check() usage!
[  117.000918] 
               other info that might help us debug this:
[  117.000921] 
               rcu_scheduler_active = 2, debug_locks = 1
[  117.000923] 1 lock held by swapper/52/0:
[  117.000925]  #0: ffffffff82670960 (rcu_read_lock_sched){....}, at: sched_core_balance+0x5/0x700
[  117.000937] 
               stack backtrace:
[  117.000940] CPU: 52 PID: 0 Comm: swapper/52 Kdump: loaded Not tainted 5.5.7+ #160
[  117.000943] Hardware name: Intel Corporation S2600WFD/S2600WFD, BIOS SE5C620.86B.01.00.0412.020920172159 02/09/2017
[  117.000945] Call Trace:
[  117.000955]  dump_stack+0x86/0xcb
[  117.000962]  sched_core_balance+0x634/0x700
[  117.000982]  __balance_callback+0x49/0xa0
[  117.000990]  __schedule+0x1416/0x1620
[  117.001000]  ? lockdep_hardirqs_off+0xa0/0xe0
[  117.001005]  ? _raw_spin_unlock_irqrestore+0x41/0x70
[  117.001024]  schedule_idle+0x28/0x40
[  117.001030]  do_idle+0x17e/0x2a0
[  117.001041]  cpu_startup_entry+0x19/0x20
[  117.001048]  start_secondary+0x16c/0x1c0
[  117.001055]  secondary_startup_64+0xa4/0xb0
Quoted 30 lines
Quoted 30 lines
---
 kernel/sched/core.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
kernel/sched/core.c +2 -2
index 3045bd50e249..037e8f2e2686 100644
@@ -4735,7 +4735,7 @@ static void sched_core_balance(struct rq *rq)
 	struct sched_domain *sd;
 	int cpu = cpu_of(rq);
 
-	rcu_read_lock();
+	rcu_read_lock_sched();
 	raw_spin_unlock_irq(rq_lockp(rq));
 	for_each_domain(cpu, sd) {
 		if (!(sd->flags & SD_LOAD_BALANCE))
@@ -4748,7 +4748,7 @@ static void sched_core_balance(struct rq *rq)
 			break;
 	}
 	raw_spin_lock_irq(rq_lockp(rq));
-	rcu_read_unlock();
+	rcu_read_unlock_sched();
 }
 
 static DEFINE_PER_CPU(struct callback_head, core_balance_head);
Signature · 1 line
2.25.1.481.gfbce0eb801-goog
#3 lore raw
Quoted 18 lines · Aubrey
Quoted 7 lines · Paul E. McKenney
Quoted 6 lines · Joel Fernandes (Google)

rcu_read_unlock() can incur an infrequent deadlock in sched_core_balance(). Fix this by using the RCU-sched flavor instead.

This fixes the following spinlock recursion observed when testing the core scheduling patches on PREEMPT=y kernel on ChromeOS:

Quoted 7 lines
Quoted 2 lines
[   14.998590] watchdog: BUG: soft lockup - CPU#0 stuck for 11s! [kworker/0:10:965]

The original could indeed deadlock, and this would avoid that deadlock. (The commit to solve this deadlock is sadly not yet in mainline.)

I saw this in dmesg with this patch, is it expected?

Quoted 7 lines
[  117.000905] =============================
[  117.000907] WARNING: suspicious RCU usage
[  117.000911] 5.5.7+ #160 Not tainted
[  117.000913] -----------------------------
[  117.000916] kernel/sched/core.c:4747 suspicious rcu_dereference_check() usage!
[  117.000918] 
               other info that might help us debug this:

Sigh, this is because for_each_domain() expects rcu_read_lock(). From an RCU PoV, the code is correct (warning doesn't cause any issue).

To silence warning, we could replace the rcu_read_lock_sched() in my patch with:
preempt_disable();
rcu_read_lock();

and replace the unlock with:

rcu_read_unlock();
preempt_enable();

That should both take care of both the warning and the scheduler-related deadlock. Thoughts?

Does that fix the warning for you?

thanks,

- Joel

Quoted 53 lines
[  117.000921] 
               rcu_scheduler_active = 2, debug_locks = 1
[  117.000923] 1 lock held by swapper/52/0:
[  117.000925]  #0: ffffffff82670960 (rcu_read_lock_sched){....}, at: sched_core_balance+0x5/0x700
[  117.000937] 
               stack backtrace:
[  117.000940] CPU: 52 PID: 0 Comm: swapper/52 Kdump: loaded Not tainted 5.5.7+ #160
[  117.000943] Hardware name: Intel Corporation S2600WFD/S2600WFD, BIOS SE5C620.86B.01.00.0412.020920172159 02/09/2017
[  117.000945] Call Trace:
[  117.000955]  dump_stack+0x86/0xcb
[  117.000962]  sched_core_balance+0x634/0x700
[  117.000982]  __balance_callback+0x49/0xa0
[  117.000990]  __schedule+0x1416/0x1620
[  117.001000]  ? lockdep_hardirqs_off+0xa0/0xe0
[  117.001005]  ? _raw_spin_unlock_irqrestore+0x41/0x70
[  117.001024]  schedule_idle+0x28/0x40
[  117.001030]  do_idle+0x17e/0x2a0
[  117.001041]  cpu_startup_entry+0x19/0x20
[  117.001048]  start_secondary+0x16c/0x1c0
[  117.001055]  secondary_startup_64+0xa4/0xb0
Quoted 30 lines
Quoted 30 lines
---
 kernel/sched/core.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
kernel/sched/core.c +2 -2
index 3045bd50e249..037e8f2e2686 100644
@@ -4735,7 +4735,7 @@ static void sched_core_balance(struct rq *rq)
 	struct sched_domain *sd;
 	int cpu = cpu_of(rq);
 
-	rcu_read_lock();
+	rcu_read_lock_sched();
 	raw_spin_unlock_irq(rq_lockp(rq));
 	for_each_domain(cpu, sd) {
 		if (!(sd->flags & SD_LOAD_BALANCE))
@@ -4748,7 +4748,7 @@ static void sched_core_balance(struct rq *rq)
 			break;
 	}
 	raw_spin_lock_irq(rq_lockp(rq));
-	rcu_read_unlock();
+	rcu_read_unlock_sched();
 }
 
 static DEFINE_PER_CPU(struct callback_head, core_balance_head);
Signature · 1 line
2.25.1.481.gfbce0eb801-goog
#4 lore raw
Quoted 34 lines · Joel Fernandes
Quoted 18 lines · Aubrey
Quoted 7 lines · Paul E. McKenney
Quoted 6 lines · Joel Fernandes (Google)

rcu_read_unlock() can incur an infrequent deadlock in sched_core_balance(). Fix this by using the RCU-sched flavor instead.

This fixes the following spinlock recursion observed when testing the core scheduling patches on PREEMPT=y kernel on ChromeOS:

Quoted 7 lines
Quoted 2 lines
[   14.998590] watchdog: BUG: soft lockup - CPU#0 stuck for 11s! [kworker/0:10:965]

The original could indeed deadlock, and this would avoid that deadlock. (The commit to solve this deadlock is sadly not yet in mainline.)

I saw this in dmesg with this patch, is it expected?

Quoted 7 lines
[  117.000905] =============================
[  117.000907] WARNING: suspicious RCU usage
[  117.000911] 5.5.7+ #160 Not tainted
[  117.000913] -----------------------------
[  117.000916] kernel/sched/core.c:4747 suspicious rcu_dereference_check() usage!
[  117.000918] 
               other info that might help us debug this:

Sigh, this is because for_each_domain() expects rcu_read_lock(). From an RCU PoV, the code is correct (warning doesn't cause any issue).

To silence warning, we could replace the rcu_read_lock_sched() in my patch with:
preempt_disable();
rcu_read_lock();

Quoted 8 lines

and replace the unlock with:

rcu_read_unlock();
preempt_enable();

That should both take care of both the warning and the scheduler-related deadlock. Thoughts?

How about this?

kernel/sched/core.c +0 -2
index a01df3e..7ff694e 100644
@@ -4743,7 +4743,6 @@ static void sched_core_balance(struct rq *rq)
 	int cpu = cpu_of(rq);
 
 	rcu_read_lock();
-	raw_spin_unlock_irq(rq_lockp(rq));
 	for_each_domain(cpu, sd) {
 		if (!(sd->flags & SD_LOAD_BALANCE))
 			break;
@@ -4754,7 +4753,6 @@ static void sched_core_balance(struct rq *rq)
 		if (steal_cookie_task(cpu, sd))
 			break;
 	}
-	raw_spin_lock_irq(rq_lockp(rq));
 	rcu_read_unlock();
 }
 

Thanks,
-Aubrey

maintainer
#5 lore raw
Quoted 59 lines · Aubrey
Quoted 34 lines · Joel Fernandes
Quoted 18 lines · Aubrey
Quoted 7 lines · Paul E. McKenney
Quoted 6 lines · Joel Fernandes (Google)

rcu_read_unlock() can incur an infrequent deadlock in sched_core_balance(). Fix this by using the RCU-sched flavor instead.

This fixes the following spinlock recursion observed when testing the core scheduling patches on PREEMPT=y kernel on ChromeOS:

Quoted 7 lines
Quoted 2 lines
[   14.998590] watchdog: BUG: soft lockup - CPU#0 stuck for 11s! [kworker/0:10:965]

The original could indeed deadlock, and this would avoid that deadlock. (The commit to solve this deadlock is sadly not yet in mainline.)

I saw this in dmesg with this patch, is it expected?

Quoted 7 lines
[  117.000905] =============================
[  117.000907] WARNING: suspicious RCU usage
[  117.000911] 5.5.7+ #160 Not tainted
[  117.000913] -----------------------------
[  117.000916] kernel/sched/core.c:4747 suspicious rcu_dereference_check() usage!
[  117.000918] 
               other info that might help us debug this:

Sigh, this is because for_each_domain() expects rcu_read_lock(). From an RCU PoV, the code is correct (warning doesn't cause any issue).

To silence warning, we could replace the rcu_read_lock_sched() in my patch with:
preempt_disable();
rcu_read_lock();

Quoted 8 lines

and replace the unlock with:

rcu_read_unlock();
preempt_enable();

That should both take care of both the warning and the scheduler-related deadlock. Thoughts?

How about this?

kernel/sched/core.c +0 -1
index a01df3e..7ff694e 100644
@@ -4743,7 +4743,6 @@ static void sched_core_balance(struct rq *rq)
 	int cpu = cpu_of(rq);
 
 	rcu_read_lock();
-	raw_spin_unlock_irq(rq_lockp(rq));
 	for_each_domain(cpu, sd) {
 		if (!(sd->flags & SD_LOAD_BALANCE))
 			break;

kernel/sched/core.c static void sched_core_balance(struct rq *rq) · line 4757

Quoted 6 lines
kernel/sched/core.c +0 -1
@@ -4754,7 +4753,6 @@ static void sched_core_balance(struct rq *rq)
 		if (steal_cookie_task(cpu, sd))
 			break;
 	}
-	raw_spin_lock_irq(rq_lockp(rq));
 	rcu_read_unlock();
 }

As an alternative, I am backporting the -rcu commit 2b5e19e20fc2 ("rcu:
Make rcu_read_unlock_special() safe for rq/pi locks") to v5.6-rc6 and testing it. The other support for this is already in mainline. I just now started testing it, and will let you know how it goes.

If that works for you, and if the bug you are looking to fix is also in v5.5 or earlier, please let me know so that we can work out how to deal with the older releases.

Thanx, Paul

maintainer
#6 lore raw
Quoted 72 lines · Paul E. McKenney
Quoted 59 lines · Aubrey
Quoted 34 lines · Joel Fernandes
Quoted 18 lines · Aubrey
Quoted 7 lines · Paul E. McKenney
Quoted 6 lines · Joel Fernandes (Google)

rcu_read_unlock() can incur an infrequent deadlock in sched_core_balance(). Fix this by using the RCU-sched flavor instead.

This fixes the following spinlock recursion observed when testing the core scheduling patches on PREEMPT=y kernel on ChromeOS:

Quoted 7 lines
Quoted 2 lines
[   14.998590] watchdog: BUG: soft lockup - CPU#0 stuck for 11s! [kworker/0:10:965]

The original could indeed deadlock, and this would avoid that deadlock. (The commit to solve this deadlock is sadly not yet in mainline.)

I saw this in dmesg with this patch, is it expected?

Quoted 7 lines
[  117.000905] =============================
[  117.000907] WARNING: suspicious RCU usage
[  117.000911] 5.5.7+ #160 Not tainted
[  117.000913] -----------------------------
[  117.000916] kernel/sched/core.c:4747 suspicious rcu_dereference_check() usage!
[  117.000918] 
               other info that might help us debug this:

Sigh, this is because for_each_domain() expects rcu_read_lock(). From an RCU PoV, the code is correct (warning doesn't cause any issue).

To silence warning, we could replace the rcu_read_lock_sched() in my patch with:
preempt_disable();
rcu_read_lock();

Quoted 8 lines

and replace the unlock with:

rcu_read_unlock();
preempt_enable();

That should both take care of both the warning and the scheduler-related deadlock. Thoughts?

How about this?

kernel/sched/core.c +0 -1
index a01df3e..7ff694e 100644
@@ -4743,7 +4743,6 @@ static void sched_core_balance(struct rq *rq)
 	int cpu = cpu_of(rq);
 
 	rcu_read_lock();
-	raw_spin_unlock_irq(rq_lockp(rq));
 	for_each_domain(cpu, sd) {
 		if (!(sd->flags & SD_LOAD_BALANCE))
 			break;

kernel/sched/core.c static void sched_core_balance(struct rq *rq) · line 4757

Quoted 6 lines
kernel/sched/core.c +0 -1
@@ -4754,7 +4753,6 @@ static void sched_core_balance(struct rq *rq)
 		if (steal_cookie_task(cpu, sd))
 			break;
 	}
-	raw_spin_lock_irq(rq_lockp(rq));
 	rcu_read_unlock();
 }

As an alternative, I am backporting the -rcu commit 2b5e19e20fc2 ("rcu:
Make rcu_read_unlock_special() safe for rq/pi locks") to v5.6-rc6 and testing it. The other support for this is already in mainline. I just now started testing it, and will let you know how it goes.

Quoted 3 lines

If that works for you, and if the bug you are looking to fix is also in v5.5 or earlier, please let me know so that we can work out how to deal with the older releases.

And the backported patch below passes moderate rcutorture testing. But the big question... Does it work for you?

Thanx, Paul

------------------------------------------------------------------------

commit 0ed0648e23f6aca2a6543fee011d74ab674e08e8

Author: Paul E. McKenney <[email protected]>
Date:   Tue Mar 24 06:20:14 2020 -0700

rcu: Make rcu_read_unlock_special() safe for rq/pi locks

The scheduler is currently required to hold rq/pi locks across the entire RCU read-side critical section or not at all. This is inconvenient and leaves traps for the unwary, including the author of this commit.

But now that excessively ong grace periods enable scheduling-clock interrupts for holdout nohz_full CPUs, the nohz_full rescue logic in rcu_read_unlock_special() can be dispensed with. In other words, the rcu_read_unlock_special() function can refrain from doing wakeups unless such wakeups are guaranteed safe.

This commit therefore avoids unsafe wakeups, freeing the scheduler to hold rq/pi locks across rcu_read_unlock() even if the corresponding RCU read-side critical section might have been preempted.

This commit is inspired by a patch from Lai Jiangshan:
https://lore.kernel.org/lkml/[email protected]
This commit is further intended to be a step towards his goal of permitting the inlining of RCU-preempt's rcu_read_lock() and rcu_read_unlock().

Cc Lai Jiangshan <[email protected]>
[ paulmck: Backported to v5.6-rc6. ]
kernel/rcu/tree_plugin.h +6 -8
index c6ea81cd4189..2e1bfa0cc393 100644
@@ -614,18 +614,16 @@ static void rcu_read_unlock_special(struct task_struct *t)
 		struct rcu_node *rnp = rdp->mynode;
 
 		exp = (t->rcu_blocked_node && t->rcu_blocked_node->exp_tasks) ||
-		      (rdp->grpmask & READ_ONCE(rnp->expmask)) ||
-		      tick_nohz_full_cpu(rdp->cpu);
+		      (rdp->grpmask & READ_ONCE(rnp->expmask));
 		// Need to defer quiescent state until everything is enabled.
-		if (irqs_were_disabled && use_softirq &&
-		    (in_interrupt() ||
-		     (exp && !t->rcu_read_unlock_special.b.deferred_qs))) {
-			// Using softirq, safe to awaken, and we get
-			// no help from enabling irqs, unlike bh/preempt.
+		if (use_softirq && (in_irq() || (exp && !irqs_were_disabled))) {
+			// Using softirq, safe to awaken, and either the
+			// wakeup is free or there is an expedited GP.
 			raise_softirq_irqoff(RCU_SOFTIRQ);
 		} else {
 			// Enabling BH or preempt does reschedule, so...
-			// Also if no expediting or NO_HZ_FULL, slow is OK.
+			// Also if no expediting, slow is OK.
+			// Plus nohz_full CPUs eventually get tick enabled.
 			set_tsk_need_resched(current);
 			set_preempt_need_resched();
 			if (IS_ENABLED(CONFIG_IRQ_WORK) && irqs_were_disabled &&
#7 lore raw
In reply to Li, Aubrey
Quoted 59 lines · Aubrey
Quoted 34 lines · Joel Fernandes
Quoted 18 lines · Aubrey
Quoted 7 lines · Paul E. McKenney
Quoted 6 lines · Joel Fernandes (Google)

rcu_read_unlock() can incur an infrequent deadlock in sched_core_balance(). Fix this by using the RCU-sched flavor instead.

This fixes the following spinlock recursion observed when testing the core scheduling patches on PREEMPT=y kernel on ChromeOS:

Quoted 7 lines
Quoted 2 lines
[   14.998590] watchdog: BUG: soft lockup - CPU#0 stuck for 11s! [kworker/0:10:965]

The original could indeed deadlock, and this would avoid that deadlock. (The commit to solve this deadlock is sadly not yet in mainline.)

I saw this in dmesg with this patch, is it expected?

Quoted 7 lines
[  117.000905] =============================
[  117.000907] WARNING: suspicious RCU usage
[  117.000911] 5.5.7+ #160 Not tainted
[  117.000913] -----------------------------
[  117.000916] kernel/sched/core.c:4747 suspicious rcu_dereference_check() usage!
[  117.000918] 
               other info that might help us debug this:

Sigh, this is because for_each_domain() expects rcu_read_lock(). From an RCU PoV, the code is correct (warning doesn't cause any issue).

To silence warning, we could replace the rcu_read_lock_sched() in my patch with:
preempt_disable();
rcu_read_lock();

Quoted 8 lines

and replace the unlock with:

rcu_read_unlock();
preempt_enable();

That should both take care of both the warning and the scheduler-related deadlock. Thoughts?

How about this?

kernel/sched/core.c +0 -1
index a01df3e..7ff694e 100644
@@ -4743,7 +4743,6 @@ static void sched_core_balance(struct rq *rq)
 	int cpu = cpu_of(rq);
 
 	rcu_read_lock();
-	raw_spin_unlock_irq(rq_lockp(rq));
 	for_each_domain(cpu, sd) {
 		if (!(sd->flags & SD_LOAD_BALANCE))
 			break;

kernel/sched/core.c static void sched_core_balance(struct rq *rq) · line 4757

Quoted 4 lines
kernel/sched/core.c +0 -1
@@ -4754,7 +4753,6 @@ static void sched_core_balance(struct rq *rq)
 		if (steal_cookie_task(cpu, sd))
 			break;
 	}
-	raw_spin_lock_irq(rq_lockp(rq));

try_steal_cookie() does a double_rq_lock(). Would this change not deadlock with that?

thanks,

- Joel

#8 lore raw
Quoted 54 lines · Joel Fernandes
Quoted 53 lines · Aubrey
Quoted 34 lines · Joel Fernandes
Quoted 18 lines · Aubrey
Quoted 7 lines · Paul E. McKenney
Quoted 6 lines · Joel Fernandes (Google)

rcu_read_unlock() can incur an infrequent deadlock in sched_core_balance(). Fix this by using the RCU-sched flavor instead.

This fixes the following spinlock recursion observed when testing the core scheduling patches on PREEMPT=y kernel on ChromeOS:

Quoted 7 lines
Quoted 2 lines
[   14.998590] watchdog: BUG: soft lockup - CPU#0 stuck for 11s! [kworker/0:10:965]

The original could indeed deadlock, and this would avoid that deadlock. (The commit to solve this deadlock is sadly not yet in mainline.)

I saw this in dmesg with this patch, is it expected?

Quoted 7 lines
[  117.000905] =============================
[  117.000907] WARNING: suspicious RCU usage
[  117.000911] 5.5.7+ #160 Not tainted
[  117.000913] -----------------------------
[  117.000916] kernel/sched/core.c:4747 suspicious rcu_dereference_check() usage!
[  117.000918] 
               other info that might help us debug this:

Sigh, this is because for_each_domain() expects rcu_read_lock(). From an RCU PoV, the code is correct (warning doesn't cause any issue).

To silence warning, we could replace the rcu_read_lock_sched() in my patch with:
preempt_disable();
rcu_read_lock();

Quoted 8 lines

and replace the unlock with:

rcu_read_unlock();
preempt_enable();

That should both take care of both the warning and the scheduler-related deadlock. Thoughts?

How about this?

kernel/sched/core.c +0 -0
index a01df3e..7ff694e 100644
@@ -4743,7 +4743,6 @@ static void sched_core_balance(struct rq *rq)
 	int cpu = cpu_of(rq);
 
Quoted 14 lines

kernel/sched/core.c static void sched_core_balance(struct rq *rq) · line 4757

Quoted 10 lines
kernel/sched/core.c +0 -2
 	rcu_read_lock();
-	raw_spin_unlock_irq(rq_lockp(rq));
 	for_each_domain(cpu, sd) {
 		if (!(sd->flags & SD_LOAD_BALANCE))
 			break;
@@ -4754,7 +4753,6 @@ static void sched_core_balance(struct rq *rq)
 		if (steal_cookie_task(cpu, sd))
 			break;
 	}
-	raw_spin_lock_irq(rq_lockp(rq));

try_steal_cookie() does a double_rq_lock(). Would this change not deadlock with that?

Oh yes, missed double_rq_lock() inside, just want to keep local intr disabled to avoid preemption. will try Paul's patch and report back.

Thanks,
-Aubrey