Skip to contentSkip to conversation

Spurious lockdep splat in v4.15-rc9

DISCUSSION Tejun Heo 3 participants 5 messages Jan 22, 2018 – Jan 24, 2018

Subsystems (2)
Scheduler Static branch/call
lore
Share & tools
lore

Tejun Heo

author
#0 lore raw

Hello, Peter, Ingo.

I get the below lockdep warning if I try to write a config into cgroup cpu.max file. It's warning about A-A deadlock, but it's obviously spurious - the system doesn't lock up and the warning is about two get_online_cpus() calls nesting.

Thanks.

[   79.106704]
[   79.106886] ============================================
[   79.107319] WARNING: possible recursive locking detected
[   79.107741] 4.15.0-rc9-work+ #61 Not tainted
[   79.108080] --------------------------------------------
[   79.108505] bash/2133 is trying to acquire lock:
[   79.108872]  (cpu_hotplug_lock.rw_sem){++++}, at: [<00000000b3203afd>] static_key_slow_inc+0xe/0xa0
[   79.109593]
[   79.109593] but task is already holding lock:
[   79.110058]  (cpu_hotplug_lock.rw_sem){++++}, at: [<00000000748e6cec>] tg_set_cfs_bandwidth+0x51/0x330
[   79.110801]
[   79.110801] other info that might help us debug this:
[   79.111322]  Possible unsafe locking scenario:
[   79.111322]
[   79.111792]        CPU0
[   79.111992]        ----
[   79.112197]   lock(cpu_hotplug_lock.rw_sem);
[   79.112537]   lock(cpu_hotplug_lock.rw_sem);
[   79.112880]
[   79.112880]  *** DEADLOCK ***
[   79.112880]
[   79.113355]  May be due to missing lock nesting notation
[   79.113355]
[   79.113893] 5 locks held by bash/2133:
[   79.114199]  #0:  (sb_writers#7){.+.+}, at: [<00000000259a9362>] vfs_write+0x18a/0x1c0
[   79.114830]  #1:  (&of->mutex){+.+.}, at: [<00000000b1a2a028>] kernfs_fop_write+0xde/0x1a0
[   79.115492]  #2:  (kn->count#182){.+.+}, at: [<000000008f74a9a4>] kernfs_fop_write+0xe6/0x1a0
[   79.116182]  #3:  (cpu_hotplug_lock.rw_sem){++++}, at: [<00000000748e6cec>] tg_set_cfs_bandwidth+0x51/0x330
[   79.116956]  #4:  (cfs_constraints_mutex){+.+.}, at: [<000000007a63f0e9>] tg_set_cfs_bandwidth+0x5f/0x330
[   79.117717]
[   79.117717] stack backtrace:
[   79.118072] CPU: 13 PID: 2133 Comm: bash Not tainted 4.15.0-rc9-work+ #61
[   79.118616] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.10.2-3.el7_4.1 04/01/2014
[   79.119331] Call Trace:
[   79.119539]  dump_stack+0x5e/0x8f
[   79.119809]  __lock_acquire+0x150e/0x15f0
[   79.120136]  ? tg_set_cfs_bandwidth+0x5f/0x330
[   79.120497]  ? __mutex_lock+0x204/0x930
[   79.120805]  ? tg_set_cfs_bandwidth+0x5f/0x330
[   79.121164]  lock_acquire+0xb0/0x200
[   79.121454]  ? static_key_slow_inc+0xe/0xa0
[   79.121798]  cpus_read_lock+0x43/0xb0
[   79.122095]  ? static_key_slow_inc+0xe/0xa0
[   79.122438]  static_key_slow_inc+0xe/0xa0
[   79.122763]  tg_set_cfs_bandwidth+0x30e/0x330
[   79.123112]  ? tg_set_cfs_bandwidth+0xaa/0x330
[   79.123473]  cpu_max_write+0xb8/0x100
[   79.123773]  cgroup_file_write+0x69/0x200
[   79.124100]  kernfs_fop_write+0x10e/0x1a0
[   79.124430]  __vfs_write+0x23/0x130
[   79.124721]  ? rcu_read_lock_sched_held+0x96/0xa0
[   79.125100]  ? rcu_sync_lockdep_assert+0x2a/0x50
[   79.125475]  ? __sb_start_write+0x194/0x230
[   79.125812]  ? vfs_write+0x18a/0x1c0
[   79.126109]  ? __close_fd+0x66/0xd0
[   79.126392]  vfs_write+0xbf/0x1c0
[   79.126660]  SyS_write+0x45/0xa0
[   79.126928]  entry_SYSCALL_64_fastpath+0x18/0x85
[   79.127306] RIP: 0033:0x7f3b040e21f0
Signature · 1 line
tejun
maintainer
#1 lore raw
Quoted 6 lines · Tejun Heo

Hello, Peter, Ingo.

I get the below lockdep warning if I try to write a config into cgroup cpu.max file. It's warning about A-A deadlock, but it's obviously spurious - the system doesn't lock up and the warning is about two get_online_cpus() calls nesting.

Looks real, just not instantly fatal. It would generate an actual deadlock the moment there is a contending hotplug operation around though.

I'll have a wee look.

maintainer
#2 lore raw
In reply to Tejun Heo

Tejun, does the below work for you (compile tested only).

---
Subject: sched: Fix cpu.max vs cpuhotplug deadlock

Tejun reported the following cpu-hotplug lock (percpu-rwsem) read recursion:

tg_set_cfs_bandwidth()
get_online_cpus()
cpus_read_lock()

cfs_bandwidth_usage_inc()
static_key_slow_inc()
cpus_read_lock()

Reported-by Tejun Heo <[email protected]>
Signed-off-by Peter Zijlstra (Intel) <[email protected]>
---
 include/linux/jump_label.h |  7 +++++++
 kernel/jump_label.c        | 12 +++++++++---
 kernel/sched/fair.c        |  4 ++--
 3 files changed, 18 insertions(+), 5 deletions(-)
include/linux/jump_label.h +7 -0
index c7b368c734af..e0340ca08d98 100644
@@ -160,6 +160,8 @@ extern void arch_jump_label_transform_static(struct jump_entry *entry,
 extern int jump_label_text_reserved(void *start, void *end);
 extern void static_key_slow_inc(struct static_key *key);
 extern void static_key_slow_dec(struct static_key *key);
+extern void static_key_slow_inc_cpuslocked(struct static_key *key);
+extern void static_key_slow_dec_cpuslocked(struct static_key *key);
 extern void jump_label_apply_nops(struct module *mod);
 extern int static_key_count(struct static_key *key);
 extern void static_key_enable(struct static_key *key);

@@ -222,6 +224,9 @@ static inline void static_key_slow_dec(struct static_key *key)
 	atomic_dec(&key->enabled);
 }
 
+#define static_key_slow_inc_cpuslocked(key) static_key_slow_inc(key)
+#define static_key_slow_dec_cpuslocked(key) static_key_slow_dec(key)
+
 static inline int jump_label_text_reserved(void *start, void *end)
 {
 	return 0;

@@ -416,6 +421,8 @@ struct static_key_false {
 
 #define static_branch_inc(x)		static_key_slow_inc(&(x)->key)
 #define static_branch_dec(x)		static_key_slow_dec(&(x)->key)
+#define static_branch_inc_cpuslocked(x)	static_key_slow_inc_cpuslocked(&(x)->key)
+#define static_branch_dec_cpuslocked(x)	static_key_slow_dec_cpuslocked(&(x)->key)
 
 /*
  * Normal usage; boolean enable/disable.
kernel/jump_label.c +9 -3
index 8594d24e4adc..b4517095db6a 100644
@@ -79,7 +79,7 @@ int static_key_count(struct static_key *key)
 }
 EXPORT_SYMBOL_GPL(static_key_count);
 
-static void static_key_slow_inc_cpuslocked(struct static_key *key)
+void static_key_slow_inc_cpuslocked(struct static_key *key)
 {
 	int v, v1;
 

@@ -180,7 +180,7 @@ void static_key_disable(struct static_key *key)
 }
 EXPORT_SYMBOL_GPL(static_key_disable);
 
-static void static_key_slow_dec_cpuslocked(struct static_key *key,
+static void __static_key_slow_dec_cpuslocked(struct static_key *key,
 					   unsigned long rate_limit,
 					   struct delayed_work *work)
 {

@@ -211,7 +211,7 @@ static void __static_key_slow_dec(struct static_key *key,
 				  struct delayed_work *work)
 {
 	cpus_read_lock();
-	static_key_slow_dec_cpuslocked(key, rate_limit, work);
+	__static_key_slow_dec_cpuslocked(key, rate_limit, work);
 	cpus_read_unlock();
 }
 

@@ -229,6 +229,12 @@ void static_key_slow_dec(struct static_key *key)
 }
 EXPORT_SYMBOL_GPL(static_key_slow_dec);
 
+void static_key_slow_dec_cpuslocked(struct static_key *key)
+{
+	STATIC_KEY_CHECK_USE(key);
+	__static_key_slow_dec_cpuslocked(key, 0, NULL);
+}
+
 void static_key_slow_dec_deferred(struct static_key_deferred *key)
 {
 	STATIC_KEY_CHECK_USE(key);
kernel/sched/fair.c +2 -2
index 1070803cb423..7b6535987500 100644
@@ -4361,12 +4361,12 @@ static inline bool cfs_bandwidth_used(void)
 
 void cfs_bandwidth_usage_inc(void)
 {
-	static_key_slow_inc(&__cfs_bandwidth_used);
+	static_key_slow_inc_cpuslocked(&__cfs_bandwidth_used);
 }
 
 void cfs_bandwidth_usage_dec(void)
 {
-	static_key_slow_dec(&__cfs_bandwidth_used);
+	static_key_slow_dec_cpuslocked(&__cfs_bandwidth_used);
 }
 #else /* HAVE_JUMP_LABEL */
 static bool cfs_bandwidth_used(void)
author
#3 lore raw

Hello, Peter.

Quoted 2 lines · Peter Zijlstra

Tejun, does the below work for you (compile tested only).

Yeap, that gets rid of the lockdep warning.

Thanks a lot.

Signature · 1 line
tejun
merge bot
#4 lore raw

[tip:sched/urgent] sched/core: Fix cpu.max vs. cpuhotplug deadlock

In reply to Peter Zijlstra
sched/core: Fix cpu.max vs. cpuhotplug deadlock
Commit-ID:  ce48c146495a1a50e48cdbfbfaba3e708be7c07c
Gitweb:     https://git.kernel.org/tip/ce48c146495a1a50e48cdbfbfaba3e708be7c07c
Author:     Peter Zijlstra <[email protected]>
AuthorDate: Mon, 22 Jan 2018 22:53:28 +0100
Committer:  Ingo Molnar <[email protected]>
CommitDate: Wed, 24 Jan 2018 10:03:44 +0100

sched/core: Fix cpu.max vs. cpuhotplug deadlock

Tejun reported the following cpu-hotplug lock (percpu-rwsem) read recursion:

tg_set_cfs_bandwidth()
get_online_cpus()
cpus_read_lock()

cfs_bandwidth_usage_inc()
static_key_slow_inc()
cpus_read_lock()

Reported-by Tejun Heo <[email protected]>
Tested-by Tejun Heo <[email protected]>
Signed-off-by Peter Zijlstra (Intel) <[email protected]>
Cc Linus Torvalds <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Thomas Gleixner <[email protected]>
Link http://lkml.kernel.org/r/20180122215328.GP3397@worktop open here
Signed-off-by Ingo Molnar <[email protected]>
---
 include/linux/jump_label.h |  7 +++++++
 kernel/jump_label.c        | 12 +++++++++---
 kernel/sched/fair.c        |  4 ++--
 3 files changed, 18 insertions(+), 5 deletions(-)
include/linux/jump_label.h +7 -0
index c7b368c..e0340ca 100644
@@ -160,6 +160,8 @@ extern void arch_jump_label_transform_static(struct jump_entry *entry,
 extern int jump_label_text_reserved(void *start, void *end);
 extern void static_key_slow_inc(struct static_key *key);
 extern void static_key_slow_dec(struct static_key *key);
+extern void static_key_slow_inc_cpuslocked(struct static_key *key);
+extern void static_key_slow_dec_cpuslocked(struct static_key *key);
 extern void jump_label_apply_nops(struct module *mod);
 extern int static_key_count(struct static_key *key);
 extern void static_key_enable(struct static_key *key);

@@ -222,6 +224,9 @@ static inline void static_key_slow_dec(struct static_key *key)
 	atomic_dec(&key->enabled);
 }
 
+#define static_key_slow_inc_cpuslocked(key) static_key_slow_inc(key)
+#define static_key_slow_dec_cpuslocked(key) static_key_slow_dec(key)
+
 static inline int jump_label_text_reserved(void *start, void *end)
 {
 	return 0;

@@ -416,6 +421,8 @@ extern bool ____wrong_branch_error(void);
 
 #define static_branch_inc(x)		static_key_slow_inc(&(x)->key)
 #define static_branch_dec(x)		static_key_slow_dec(&(x)->key)
+#define static_branch_inc_cpuslocked(x)	static_key_slow_inc_cpuslocked(&(x)->key)
+#define static_branch_dec_cpuslocked(x)	static_key_slow_dec_cpuslocked(&(x)->key)
 
 /*
  * Normal usage; boolean enable/disable.
kernel/jump_label.c +9 -3
index 8594d24..b451709 100644
@@ -79,7 +79,7 @@ int static_key_count(struct static_key *key)
 }
 EXPORT_SYMBOL_GPL(static_key_count);
 
-static void static_key_slow_inc_cpuslocked(struct static_key *key)
+void static_key_slow_inc_cpuslocked(struct static_key *key)
 {
 	int v, v1;
 

@@ -180,7 +180,7 @@ void static_key_disable(struct static_key *key)
 }
 EXPORT_SYMBOL_GPL(static_key_disable);
 
-static void static_key_slow_dec_cpuslocked(struct static_key *key,
+static void __static_key_slow_dec_cpuslocked(struct static_key *key,
 					   unsigned long rate_limit,
 					   struct delayed_work *work)
 {

@@ -211,7 +211,7 @@ static void __static_key_slow_dec(struct static_key *key,
 				  struct delayed_work *work)
 {
 	cpus_read_lock();
-	static_key_slow_dec_cpuslocked(key, rate_limit, work);
+	__static_key_slow_dec_cpuslocked(key, rate_limit, work);
 	cpus_read_unlock();
 }
 

@@ -229,6 +229,12 @@ void static_key_slow_dec(struct static_key *key)
 }
 EXPORT_SYMBOL_GPL(static_key_slow_dec);
 
+void static_key_slow_dec_cpuslocked(struct static_key *key)
+{
+	STATIC_KEY_CHECK_USE(key);
+	__static_key_slow_dec_cpuslocked(key, 0, NULL);
+}
+
 void static_key_slow_dec_deferred(struct static_key_deferred *key)
 {
 	STATIC_KEY_CHECK_USE(key);
kernel/sched/fair.c +2 -2
index 2fe3aa8..26a71eb 100644
@@ -4365,12 +4365,12 @@ static inline bool cfs_bandwidth_used(void)
 
 void cfs_bandwidth_usage_inc(void)
 {
-	static_key_slow_inc(&__cfs_bandwidth_used);
+	static_key_slow_inc_cpuslocked(&__cfs_bandwidth_used);
 }
 
 void cfs_bandwidth_usage_dec(void)
 {
-	static_key_slow_dec(&__cfs_bandwidth_used);
+	static_key_slow_dec_cpuslocked(&__cfs_bandwidth_used);
 }
 #else /* HAVE_JUMP_LABEL */
 static bool cfs_bandwidth_used(void)