[PATCH 0/2] lockdep/selftests: Restore PREEMPT_RT task state

[PATCH 0/2] lockdep/selftests: Restore PREEMPT_RT task state

[Karl Mehltretter]

Hi,

the lockdep selftests deliberately run broken locking patterns. On
PREEMPT_RT, some of those paths also leave task-local RT state behind, and
dotest() has to restore that state before the next testcase.

Patch 1 restores current->migration_disabled on all PREEMPT_RT builds.
The missing restore is visible with CONFIG_SMP=n as a pinned-lock warning,
"bad: scheduling from the idle thread!" and a panic after the selftests.

Patch 2 restores current->sched_rt_mutex after WW-mutex selftests. Without
that, stale RT-mutex scheduler state produces rt_mutex_pre_schedule()
warnings in later testcases.

Both changes are in dotest()'s PREEMPT_RT cleanup path in
lib/locking-selftest.c.

Tested on x86_64 QEMU with CONFIG_PREEMPT_RT=y,
CONFIG_DEBUG_LOCKING_API_SELFTESTS=y, CONFIG_LOCKDEP=y and
CONFIG_PROVE_LOCKING=y. Both CONFIG_SMP=n and CONFIG_SMP=y configurations
boot to userspace without the pinned-lock panic or rt_mutex_pre_schedule()
warnings.

Karl Mehltretter (2):
  lockdep/selftests: Restore migrate_disable() state on PREEMPT_RT
  lockdep/selftests: Restore sched_rt_mutex state on PREEMPT_RT

 lib/locking-selftest.c | 7 +++----
 1 file changed, 3 insertions(+), 4 deletions(-)

[PATCH 1/2] lockdep/selftests: Restore migrate_disable() state on PREEMPT_RT

[Karl Mehltretter]

The lockdep selftests deliberately run unbalanced locking patterns.
dotest() restores the task state they leave behind before running the
next testcase.

On PREEMPT_RT, spin_lock() uses migrate_disable() instead of disabling
preemption. dotest() cleans up the resulting migration-disabled state, but
that cleanup is still guarded by CONFIG_SMP.

That used to match the scheduler data model, where migration_disabled was
also CONFIG_SMP-only. The commit referenced below made SMP scheduler state
unconditional, so CONFIG_SMP=n PREEMPT_RT kernels with
CONFIG_DEBUG_LOCKING_API_SELFTESTS=y report success from the selftests and
then trip over stale current->migration_disabled state:

  releasing a pinned lock
  bad: scheduling from the idle thread!
  Kernel panic - not syncing: Fatal exception

Save and restore current->migration_disabled for every PREEMPT_RT build.

Fixes: cac5cefbade9 ("sched/smp: Make SMP unconditional")
Assisted-by: Codex:gpt-5
Signed-off-by: Karl Mehltretter <kmehltretter@gmail.com>
---
 lib/locking-selftest.c | 4 ----
 1 file changed, 4 deletions(-)

diff --git a/lib/locking-selftest.c b/lib/locking-selftest.c
index d939403331b5..34554e26d477 100644
--- a/lib/locking-selftest.c
+++ b/lib/locking-selftest.c
@@ -1431,9 +1431,7 @@ static void dotest(void (*testcase_fn)(void), int expected, int lockclass_mask)
 {
 	int saved_preempt_count = preempt_count();
 #ifdef CONFIG_PREEMPT_RT
-#ifdef CONFIG_SMP
 	int saved_mgd_count = current->migration_disabled;
-#endif
 	int saved_rcu_count = current->rcu_read_lock_nesting;
 #endif
 
@@ -1471,10 +1469,8 @@ static void dotest(void (*testcase_fn)(void), int expected, int lockclass_mask)
 	preempt_count_set(saved_preempt_count);
 
 #ifdef CONFIG_PREEMPT_RT
-#ifdef CONFIG_SMP
 	while (current->migration_disabled > saved_mgd_count)
 		migrate_enable();
-#endif
 
 	while (current->rcu_read_lock_nesting > saved_rcu_count)
 		rcu_read_unlock();
-- 
2.39.5 (Apple Git-154)

[tip: locking/core] lockdep/selftests: Restore migrate_disable() state on PREEMPT_RT

[tip-bot2 for Karl Mehltretter]

The following commit has been merged into the locking/core branch of tip:

Commit-ID:     d8c897b20bf4d4cbb1e935a8ceb666bcc0f82580
Gitweb:        https://git.kernel.org/tip/d8c897b20bf4d4cbb1e935a8ceb666bcc0f82580
Author:        Karl Mehltretter <kmehltretter@gmail.com>
AuthorDate:    Sat, 23 May 2026 20:51:22 +02:00
Committer:     Peter Zijlstra <peterz@infradead.org>
CommitterDate: Wed, 03 Jun 2026 11:38:47 +02:00

lockdep/selftests: Restore migrate_disable() state on PREEMPT_RT

The lockdep selftests deliberately run unbalanced locking patterns.
dotest() restores the task state they leave behind before running the
next testcase.

On PREEMPT_RT, spin_lock() uses migrate_disable() instead of disabling
preemption. dotest() cleans up the resulting migration-disabled state, but
that cleanup is still guarded by CONFIG_SMP.

That used to match the scheduler data model, where migration_disabled was
also CONFIG_SMP-only. The commit referenced below made SMP scheduler state
unconditional, so CONFIG_SMP=n PREEMPT_RT kernels with
CONFIG_DEBUG_LOCKING_API_SELFTESTS=y report success from the selftests and
then trip over stale current->migration_disabled state:

  releasing a pinned lock
  bad: scheduling from the idle thread!
  Kernel panic - not syncing: Fatal exception

Save and restore current->migration_disabled for every PREEMPT_RT build.

Fixes: cac5cefbade9 ("sched/smp: Make SMP unconditional")
Assisted-by: Codex:gpt-5
Signed-off-by: Karl Mehltretter <kmehltretter@gmail.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Link: https://patch.msgid.link/20260523185123.17482-2-kmehltretter@gmail.com
---
 lib/locking-selftest.c | 4 ----
 1 file changed, 4 deletions(-)

diff --git a/lib/locking-selftest.c b/lib/locking-selftest.c
index d939403..cb806b6 100644
--- a/lib/locking-selftest.c
+++ b/lib/locking-selftest.c
@@ -1431,9 +1431,7 @@ static void dotest(void (*testcase_fn)(void), int expected, int lockclass_mask)
 {
 	int saved_preempt_count = preempt_count();
 #ifdef CONFIG_PREEMPT_RT
-#ifdef CONFIG_SMP
 	int saved_mgd_count = current->migration_disabled;
-#endif
 	int saved_rcu_count = current->rcu_read_lock_nesting;
 #endif
 
@@ -1471,10 +1469,8 @@ static void dotest(void (*testcase_fn)(void), int expected, int lockclass_mask)
 	preempt_count_set(saved_preempt_count);
 
 #ifdef CONFIG_PREEMPT_RT
-#ifdef CONFIG_SMP
 	while (current->migration_disabled > saved_mgd_count)
 		migrate_enable();
-#endif
 
 	while (current->rcu_read_lock_nesting > saved_rcu_count)
 		rcu_read_unlock();

[PATCH 2/2] lockdep/selftests: Restore sched_rt_mutex state on PREEMPT_RT

[Karl Mehltretter]

The WW-mutex selftests deliberately exercise failing lock paths. On
PREEMPT_RT, some of those paths enter the RT-mutex scheduler helpers.

The change referenced by the Fixes tag made those helpers track RT-mutex
scheduling state in current->sched_rt_mutex. The bit is normally cleared by
the matching post-schedule helper, but some WW-mutex selftests disable
the runtime debug_locks flag before that happens. With debug_locks cleared,
lockdep_assert() does not evaluate the expression that clears the bit,
leaving stale state for the next testcase.

With CONFIG_PREEMPT_RT=y and CONFIG_DEBUG_LOCKING_API_SELFTESTS=y, that
stale state produces warnings such as:

  WARNING: kernel/sched/core.c:7557 at rt_mutex_pre_schedule+0x26/0x2d
  RIP: 0010:rt_mutex_pre_schedule+0x26/0x2d

Save and restore current->sched_rt_mutex around each testcase, matching the
existing PREEMPT_RT cleanup for task-local migration and RCU state.

Fixes: d14f9e930b90 ("locking/rtmutex: Use rt_mutex specific scheduler helpers")
Assisted-by: Codex:gpt-5
Signed-off-by: Karl Mehltretter <kmehltretter@gmail.com>
---
 lib/locking-selftest.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/lib/locking-selftest.c b/lib/locking-selftest.c
index 34554e26d477..e7c3ccb23b58 100644
--- a/lib/locking-selftest.c
+++ b/lib/locking-selftest.c
@@ -1433,6 +1433,7 @@ static void dotest(void (*testcase_fn)(void), int expected, int lockclass_mask)
 #ifdef CONFIG_PREEMPT_RT
 	int saved_mgd_count = current->migration_disabled;
 	int saved_rcu_count = current->rcu_read_lock_nesting;
+	int saved_sched_rt_mutex = current->sched_rt_mutex;
 #endif
 
 	WARN_ON(irqs_disabled());
@@ -1469,6 +1470,8 @@ static void dotest(void (*testcase_fn)(void), int expected, int lockclass_mask)
 	preempt_count_set(saved_preempt_count);
 
 #ifdef CONFIG_PREEMPT_RT
+	current->sched_rt_mutex = saved_sched_rt_mutex;
+
 	while (current->migration_disabled > saved_mgd_count)
 		migrate_enable();
 
-- 
2.39.5 (Apple Git-154)

[tip: locking/core] lockdep/selftests: Restore sched_rt_mutex state on PREEMPT_RT

[tip-bot2 for Karl Mehltretter]

The following commit has been merged into the locking/core branch of tip:

Commit-ID:     06961d60a0e410bf8df69ccff7eb1bd824912b8f
Gitweb:        https://git.kernel.org/tip/06961d60a0e410bf8df69ccff7eb1bd824912b8f
Author:        Karl Mehltretter <kmehltretter@gmail.com>
AuthorDate:    Sat, 23 May 2026 20:51:23 +02:00
Committer:     Peter Zijlstra <peterz@infradead.org>
CommitterDate: Wed, 03 Jun 2026 11:38:47 +02:00

lockdep/selftests: Restore sched_rt_mutex state on PREEMPT_RT

The WW-mutex selftests deliberately exercise failing lock paths. On
PREEMPT_RT, some of those paths enter the RT-mutex scheduler helpers.

The change referenced by the Fixes tag made those helpers track RT-mutex
scheduling state in current->sched_rt_mutex. The bit is normally cleared by
the matching post-schedule helper, but some WW-mutex selftests disable
the runtime debug_locks flag before that happens. With debug_locks cleared,
lockdep_assert() does not evaluate the expression that clears the bit,
leaving stale state for the next testcase.

With CONFIG_PREEMPT_RT=y and CONFIG_DEBUG_LOCKING_API_SELFTESTS=y, that
stale state produces warnings such as:

  WARNING: kernel/sched/core.c:7557 at rt_mutex_pre_schedule+0x26/0x2d
  RIP: 0010:rt_mutex_pre_schedule+0x26/0x2d

Save and restore current->sched_rt_mutex around each testcase, matching the
existing PREEMPT_RT cleanup for task-local migration and RCU state.

Fixes: d14f9e930b90 ("locking/rtmutex: Use rt_mutex specific scheduler helpers")
Assisted-by: Codex:gpt-5
Signed-off-by: Karl Mehltretter <kmehltretter@gmail.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Link: https://patch.msgid.link/20260523185123.17482-3-kmehltretter@gmail.com
---
 lib/locking-selftest.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/lib/locking-selftest.c b/lib/locking-selftest.c
index cb806b6..bfafe12 100644
--- a/lib/locking-selftest.c
+++ b/lib/locking-selftest.c
@@ -1433,6 +1433,7 @@ static void dotest(void (*testcase_fn)(void), int expected, int lockclass_mask)
 #ifdef CONFIG_PREEMPT_RT
 	int saved_mgd_count = current->migration_disabled;
 	int saved_rcu_count = current->rcu_read_lock_nesting;
+	int saved_sched_rt_mutex = current->sched_rt_mutex;
 #endif
 
 	WARN_ON(irqs_disabled());
@@ -1469,6 +1470,8 @@ static void dotest(void (*testcase_fn)(void), int expected, int lockclass_mask)
 	preempt_count_set(saved_preempt_count);
 
 #ifdef CONFIG_PREEMPT_RT
+	current->sched_rt_mutex = saved_sched_rt_mutex;
+
 	while (current->migration_disabled > saved_mgd_count)
 		migrate_enable();
 

Re: [PATCH 0/2] lockdep/selftests: Restore PREEMPT_RT task state

[Sebastian Andrzej Siewior]

On 2026-05-23 20:51:21 [+0200], Karl Mehltretter wrote:
> Hi,
> 
> the lockdep selftests deliberately run broken locking patterns. On
> PREEMPT_RT, some of those paths also leave task-local RT state behind, and
> dotest() has to restore that state before the next testcase.
…

Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>

Sebastian