lockdep: Fix wait context check on softirq for PREEMPT_RT
PATCH v3 1/1 lockdep: Fix wait context check on softirq for PREEMPT_RT
Since commit 0c1d7a2c2d32 ("lockdep: Remove softirq accounting on PREEMPT_RT."), the wait context test for mutex usage within "in softirq context" fails as it references @softirq_context.
[ 0.184549] | wait context tests | [ 0.184549] -------------------------------------------------------------------------- [ 0.184549] | rcu | raw | spin |mutex | [ 0.184549] -------------------------------------------------------------------------- [ 0.184550] in hardirq context: ok | ok | ok | ok | [ 0.185083] in hardirq context (not threaded): ok | ok | ok | ok | [ 0.185606] in softirq context: ok | ok | ok |FAILED|
As a fix, add lockdep map for BH disabled section. This fixes the issue by letting us catch cases when local_bh_disable() gets called with preemption disabled where local_lock doesn't get acquired. In the case of "in softirq context" selftest, local_bh_disable() was being called with preemption disable as it's early in the boot.
---
Hi!
This version uses CONFIG_DEBUG_LOCK_ALLOC instead of CONFIG_LOCKDEP. Thanks Waiman for the advise!
v1:
https://lore.kernel.org/lkml/[email protected]/
v2:
https://lore.kernel.org/lkml/[email protected]/
Boqun, let me know in case if I should add you as <suggested-by> tag. (Please forgive me and ignore if not what the tag intends for... I thought its appropriate reading the document[0].)
Sincerely,
Ryo Takakura
--- include/linux/bottom_half.h | 8 ++++++++ kernel/softirq.c | 12 ++++++++++++ 2 files changed, 20 insertions(+)
include/linux/bottom_half.h +8 -0
@@ -4,6 +4,7 @@#include <linux/instruction_pointer.h> #include <linux/preempt.h> +#include <linux/lockdep.h> #if defined(CONFIG_PREEMPT_RT ) || defined(CONFIG_TRACE_IRQFLAGS )extern void __local_bh_disable_ip (unsigned long ip,unsigned int cnt);
@@ -15,8 +16,13 @@ static __always_inline void __local_bh_disable_ip(unsigned long ip, unsigned int }#endif +#ifdef CONFIG_DEBUG_LOCK_ALLOC +extern struct lockdep_map bh_lock_map; +#endif +static inline void local_bh_disable (void ) { +lock_map_acquire_read (&bh_lock_map);__local_bh_disable_ip (_THIS_IP_,SOFTIRQ_DISABLE_OFFSET ); }
@@ -25,11 +31,13 @@ extern void __local_bh_enable_ip(unsigned long ip, unsigned int cnt);static inline void local_bh_enable_ip (unsigned long ip) { +lock_map_release (&bh_lock_map);__local_bh_enable_ip (ip,SOFTIRQ_DISABLE_OFFSET ); }static inline void local_bh_enable (void ) { +lock_map_release (&bh_lock_map);__local_bh_enable_ip (_THIS_IP_,SOFTIRQ_DISABLE_OFFSET ); }
kernel/softirq.c +12 -0
@@ -1012,3 +1012,15 @@ unsigned int __weak arch_dynirq_lower_bound(unsigned int from) {return from; } + +#ifdef CONFIG_DEBUG_LOCK_ALLOC +static struct lock_class_key bh_lock_key; +struct lockdep_map bh_lock_map = { + .name ="local_bh" , + .key = &bh_lock_key, + .wait_type_outer =LD_WAIT_FREE , + .wait_type_inner =LD_WAIT_CONFIG ,/* PREEMPT_RT makes BH preemptible. */ + .lock_type =LD_LOCK_PERCPU , +}; +EXPORT_SYMBOL_GPL (bh_lock_map); +#endif
Signature · 1 line
2.34.1
Quoted 12 lines · Ryo Takakura
Since commit
0c1d7a2c2d32("lockdep: Remove softirq accounting onPREEMPT_RT."), the wait context test for mutex usage within "in softirq context" fails as it references @softirq_context.[ 0.184549] | wait context tests | [ 0.184549] -------------------------------------------------------------------------- [ 0.184549] | rcu | raw | spin |mutex | [ 0.184549] -------------------------------------------------------------------------- [ 0.184550] in hardirq context: ok | ok | ok | ok | [ 0.185083] in hardirq context (not threaded): ok | ok | ok | ok | [ 0.185606] in softirq context: ok | ok | ok |FAILED|
Quoted 7 lines
As a fix, add lockdep map for BH disabled section. This fixes the issue by letting us catch cases when
local_bh_disable()gets called with preemption disabled wherelocal_lockdoesn't get acquired. In the case of "in softirq context" selftest,local_bh_disable()was being called with preemption disable as it's early in the boot.Signed-off-by Ryo Takakura <[email protected]>
Queued for future tests and reviews, thanks!
Regards,
Boqun
Quoted 90 lines
---
Hi!
This version uses
CONFIG_DEBUG_LOCK_ALLOCinstead ofCONFIG_LOCKDEP. Thanks Waiman for the advise!v1:
https://lore.kernel.org/lkml/[email protected]/v2:
https://lore.kernel.org/lkml/[email protected]/Boqun, let me know in case if I should add you as <suggested-by> tag. (Please forgive me and ignore if not what the tag intends for... I thought its appropriate reading the document[0].)
Sincerely,
Ryo Takakura--- include/linux/bottom_half.h | 8 ++++++++ kernel/softirq.c | 12 ++++++++++++ 2 files changed, 20 insertions(+)include/linux/bottom_half.h +8
-0@@ -4,6 +4,7 @@#include <linux/instruction_pointer.h> #include <linux/preempt.h> +#include <linux/lockdep.h> #if defined(CONFIG_PREEMPT_RT ) || defined(CONFIG_TRACE_IRQFLAGS )extern void __local_bh_disable_ip (unsigned long ip,unsigned int cnt);@@ -15,8 +16,13 @@ static __always_inline void __local_bh_disable_ip(unsigned long ip, unsigned int }#endif +#ifdef CONFIG_DEBUG_LOCK_ALLOC +extern struct lockdep_map bh_lock_map; +#endif +static inline void local_bh_disable (void ) { +lock_map_acquire_read (&bh_lock_map);__local_bh_disable_ip (_THIS_IP_,SOFTIRQ_DISABLE_OFFSET ); }@@ -25,11 +31,13 @@ extern void __local_bh_enable_ip(unsigned long ip, unsigned int cnt);static inline void local_bh_enable_ip (unsigned long ip) { +lock_map_release (&bh_lock_map);__local_bh_enable_ip (ip,SOFTIRQ_DISABLE_OFFSET ); }static inline void local_bh_enable (void ) { +lock_map_release (&bh_lock_map);__local_bh_enable_ip (_THIS_IP_,SOFTIRQ_DISABLE_OFFSET ); }kernel/softirq.c +12
-0@@ -1012,3 +1012,15 @@ unsigned int __weak arch_dynirq_lower_bound(unsigned int from) {return from; } + +#ifdef CONFIG_DEBUG_LOCK_ALLOC +static struct lock_class_key bh_lock_key; +struct lockdep_map bh_lock_map = { + .name ="local_bh" , + .key = &bh_lock_key, + .wait_type_outer =LD_WAIT_FREE , + .wait_type_inner =LD_WAIT_CONFIG ,/* PREEMPT_RT makes BH preemptible. */ + .lock_type =LD_LOCK_PERCPU , +}; +EXPORT_SYMBOL_GPL (bh_lock_map); +#endif Signature · 1 line
2.34.1
The following commit has been merged into the locking/core branch of tip:
The following commit has been merged into the locking/core branch of tip:
Commit-ID: 8a9d677a395703ef9075c91dd04066be8a553405 Gitweb: https://git.kernel.org/tip/8a9d677a395703ef9075c91dd04066be8a553405 Author: Ryo Takakura <[email protected]> AuthorDate: Sat, 18 Jan 2025 14:49:00 +09:00 Committer: Boqun Feng <[email protected]> CommitterDate: Sun, 23 Feb 2025 18:24:46 -08:00
lockdep: Fix wait context check on softirq for PREEMPT_RT
Since commit 0c1d7a2c2d32 ("lockdep: Remove softirq accounting on PREEMPT_RT."), the wait context test for mutex usage within "in softirq context" fails as it references @softirq_context.
[ 0.184549] | wait context tests | [ 0.184549] -------------------------------------------------------------------------- [ 0.184549] | rcu | raw | spin |mutex | [ 0.184549] -------------------------------------------------------------------------- [ 0.184550] in hardirq context: ok | ok | ok | ok | [ 0.185083] in hardirq context (not threaded): ok | ok | ok | ok | [ 0.185606] in softirq context: ok | ok | ok |FAILED|
As a fix, add lockdep map for BH disabled section. This fixes the issue by letting us catch cases when local_bh_disable() gets called with preemption disabled where local_lock doesn't get acquired. In the case of "in softirq context" selftest, local_bh_disable() was being called with preemption disable as it's early in the boot.
--- include/linux/bottom_half.h | 8 ++++++++ kernel/softirq.c | 12 ++++++++++++ 2 files changed, 20 insertions(+)
include/linux/bottom_half.h +8 -0
@@ -4,6 +4,7 @@#include <linux/instruction_pointer.h> #include <linux/preempt.h> +#include <linux/lockdep.h> #if defined(CONFIG_PREEMPT_RT ) || defined(CONFIG_TRACE_IRQFLAGS )extern void __local_bh_disable_ip (unsigned long ip,unsigned int cnt);
@@ -15,8 +16,13 @@ static __always_inline void __local_bh_disable_ip(unsigned long ip, unsigned int }#endif +#ifdef CONFIG_DEBUG_LOCK_ALLOC +extern struct lockdep_map bh_lock_map; +#endif +static inline void local_bh_disable (void ) { +lock_map_acquire_read (&bh_lock_map);__local_bh_disable_ip (_THIS_IP_,SOFTIRQ_DISABLE_OFFSET ); }
@@ -25,11 +31,13 @@ extern void __local_bh_enable_ip(unsigned long ip, unsigned int cnt);static inline void local_bh_enable_ip (unsigned long ip) { +lock_map_release (&bh_lock_map);__local_bh_enable_ip (ip,SOFTIRQ_DISABLE_OFFSET ); }static inline void local_bh_enable (void ) { +lock_map_release (&bh_lock_map);__local_bh_enable_ip (_THIS_IP_,SOFTIRQ_DISABLE_OFFSET ); }
kernel/softirq.c +12 -0
@@ -1073,3 +1073,15 @@ unsigned int __weak arch_dynirq_lower_bound(unsigned int from) {return from; } + +#ifdef CONFIG_DEBUG_LOCK_ALLOC +static struct lock_class_key bh_lock_key; +struct lockdep_map bh_lock_map = { + .name ="local_bh" , + .key = &bh_lock_key, + .wait_type_outer =LD_WAIT_FREE , + .wait_type_inner =LD_WAIT_CONFIG ,/* PREEMPT_RT makes BH preemptible. */ + .lock_type =LD_LOCK_PERCPU , +}; +EXPORT_SYMBOL_GPL (bh_lock_map); +#endif
Quoted 29 lines · On Mon, Mar 03, 2025 at 06:51:59PM -0000, tip-bot2 for Ryo Takakura wrote:
The following commit has been merged into the locking/core branch of tip:
Commit-ID: 8a9d677a395703ef9075c91dd04066be8a553405 Gitweb: https://git.kernel.org/tip/8a9d677a395703ef9075c91dd04066be8a553405 Author: Ryo Takakura <[email protected]> AuthorDate: Sat, 18 Jan 2025 14:49:00 +09:00 Committer: Boqun Feng <[email protected]> CommitterDate: Sun, 23 Feb 2025 18:24:46 -08:00lockdep: Fix wait context check on softirq for
PREEMPT_RTSince commit
0c1d7a2c2d32("lockdep: Remove softirq accounting onPREEMPT_RT."), the wait context test for mutex usage within "in softirq context" fails as it references @softirq_context.[ 0.184549] | wait context tests | [ 0.184549] -------------------------------------------------------------------------- [ 0.184549] | rcu | raw | spin |mutex | [ 0.184549] -------------------------------------------------------------------------- [ 0.184550] in hardirq context: ok | ok | ok | ok | [ 0.185083] in hardirq context (not threaded): ok | ok | ok | ok | [ 0.185606] in softirq context: ok | ok | ok |FAILED|As a fix, add lockdep map for BH disabled section. This fixes the issue by letting us catch cases when
local_bh_disable()gets called with preemption disabled wherelocal_lockdoesn't get acquired. In the case of "in softirq context" selftest,local_bh_disable()was being called with preemption disable as it's early in the boot.
Quoted 3 lines
Signed-off-by Ryo Takakura <[email protected]>Signed-off-by Boqun Feng <[email protected]>
This commit is causing:
[ 7.184373] NET: Registered PF_INET6 protocol family
[ 7.196129] =============================
[ 7.196129] [ BUG: Invalid wait context ]
[ 7.196129] 6.14.0-rc5-00547-g67de62470d82-dirty #629 Not tainted
[ 7.196129] -----------------------------
[ 7.196129] swapper/0/1 is trying to lock:
[ 7.196129] ffffffff83312108 (pcpu_alloc_mutex){+.+.}-{4:4}, at: pcpu_alloc_noprof+0x818/0xc20
[ 7.196129] other info that might help us debug this:
[ 7.196129] context-{5:5}
[ 7.238009] ata7.00: ATA-8: ST91000640NS, SN03, max UDMA/133
[ 7.196129] 3 locks held by swapper/0/1:
[ 7.196129] #0: ffffffff834414d0 (pernet_ops_rwsem){+.+.}-{4:4}, at: register_netdevice_notifier+0x1a/0x120
[ 7.196129] #1: ffffffff83442988 (rtnl_mutex){+.+.}-{4:4}, at: register_netdevice_notifier+0x1f/0x120
[ 7.196129] #2: ffffffff8324c740 (local_bh){.+.+}-{1:3}, at: dev_mc_add+0x39/0xb0
[ 7.196129] stack backtrace:
[ 7.196129] CPU: 31 UID: 0 PID: 1 Comm: swapper/0 Not tainted 6.14.0-rc5-00547-g67de62470d82-dirty #629
[ 7.196129] Hardware name: Intel Corporation S2600GZ/S2600GZ, BIOS SE5C600.86B.02.02.0002.122320131210 12/23/2013
[ 7.196129] Call Trace:
[ 7.196129] <TASK>
[ 7.196129] dump_stack_lvl+0x57/0x80
[ 7.196129] __lock_acquire+0xd72/0x17d0
[ 7.196129] ? ret_from_fork_asm+0x1a/0x30
[ 7.196129] lock_acquire+0xcd/0x2f0
[ 7.196129] ? pcpu_alloc_noprof+0x818/0xc20
[ 7.196129] __mutex_lock+0xa4/0x820
[ 7.196129] ? pcpu_alloc_noprof+0x818/0xc20
[ 7.196129] ? pcpu_alloc_noprof+0x818/0xc20
[ 7.196129] ? lock_acquire+0xcd/0x2f0
[ 7.239989] ? pcpu_alloc_noprof+0x818/0xc20
[ 7.239990] pcpu_alloc_noprof+0x818/0xc20
[ 7.239993] ? lockdep_hardirqs_on+0x74/0x110
[ 7.239996] ? neigh_parms_alloc+0xed/0x160
[ 7.240001] ? neigh_parms_alloc+0xed/0x160
[ 7.240005] ipv6_add_dev+0x154/0x520
[ 7.240005] addrconf_notify+0x2de/0x8b0
[ 7.240005] ? register_netdevice_notifier+0x1f/0x120
[ 7.240005] ? lock_acquire+0xdd/0x2f0
[ 7.240005] call_netdevice_register_net_notifiers+0x5b/0x100
[ 7.240005] register_netdevice_notifier+0x87/0x120
[ 7.240005] addrconf_init+0xa5/0x150
[ 7.240005] inet6_init+0x1f3/0x3b0
[ 7.240005] ? __pfx_inet6_init+0x10/0x10
[ 7.240005] do_one_initcall+0x53/0x2b0
[ 7.240005] ? rcu_is_watching+0xd/0x40
[ 7.240005] kernel_init_freeable+0x23f/0x280
[ 7.240005] ? __pfx_kernel_init+0x10/0x10
[ 7.240005] kernel_init+0x16/0x130
[ 7.240005] ret_from_fork+0x2d/0x50
[ 7.240005] ? __pfx_kernel_init+0x10/0x10
[ 7.240005] ret_from_fork_asm+0x1a/0x30
[ 7.240005] </TASK>And some other weirdness for bpetkov:
https://lkml.kernel.org/r/20250306122413.GBZ8mT7Z61Tmgnh5Y9@fat_crate.local
I suspect you missed a release somewhere.