linux-kernel - [PATCH v5 20/23] locking/rtmutex: Add block_on_rt

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-Id: <2f7168be0b92cffe1ddc762914344136ef471220.1745926331.git.namcao@linutronix.de>
Date: Tue, 29 Apr 2025 14:01:05 +0200
From: Nam Cao <namcao@...utronix.de>
To: Steven Rostedt <rostedt@...dmis.org>,
	Gabriele Monaco <gmonaco@...hat.com>,
	linux-trace-kernel@...r.kernel.org,
	linux-kernel@...r.kernel.org
Cc: john.ogness@...utronix.de,
	Nam Cao <namcao@...utronix.de>,
	Peter Zijlstra <peterz@...radead.org>,
	Ingo Molnar <mingo@...hat.com>,
	Will Deacon <will@...nel.org>,
	Boqun Feng <boqun.feng@...il.com>,
	Waiman Long <longman@...hat.com>
Subject: [PATCH v5 20/23] locking/rtmutex: Add block_on_rt_mutex tracepoints

Add block_on_rt_mutex_begin_tp and block_on_rt_mutex_end_tp tracepoints.
They are useful to implement runtime verification monitor which detects
priority inversion.

trace_contention_begin and trace_contention_end are similar to these new
tracepoints, but unfortunately they cannot be used without breaking
userspace:

  - userspace tool (perf-lock) assumes "contention_begin" tracepoint means
    "current" is contending the lock.
  - The runtime verification monitor needs the tracepoint in
    rt_mutex_start_proxy_lock(). In this case, it is not "current" who is
    contending.

Signed-off-by: Nam Cao <namcao@...utronix.de>
---
Cc: Peter Zijlstra <peterz@...radead.org>
Cc: Ingo Molnar <mingo@...hat.com>
Cc: Will Deacon <will@...nel.org>
Cc: Boqun Feng <boqun.feng@...il.com>
Cc: Waiman Long <longman@...hat.com>
---
 include/trace/events/lock.h  | 8 ++++++++
 kernel/locking/rtmutex.c     | 2 ++
 kernel/locking/rtmutex_api.c | 4 ++++
 3 files changed, 14 insertions(+)

diff --git a/include/trace/events/lock.h b/include/trace/events/lock.h
index 8e89baa3775f..d83ec2eaab22 100644
--- a/include/trace/events/lock.h
+++ b/include/trace/events/lock.h
@@ -138,6 +138,14 @@ TRACE_EVENT(contention_end,
 	TP_printk("%p (ret=%d)", __entry->lock_addr, __entry->ret)
 );
 
+DECLARE_TRACE(block_on_rt_mutex_begin_tp,
+	TP_PROTO(struct task_struct *task),
+	TP_ARGS(task));
+
+DECLARE_TRACE(block_on_rt_mutex_end_tp,
+	TP_PROTO(struct task_struct *task),
+	TP_ARGS(task));
+
 #endif /* _TRACE_LOCK_H */
 
 /* This part must be outside protection */
diff --git a/kernel/locking/rtmutex.c b/kernel/locking/rtmutex.c
index 4a8df1800cbb..08d33b74be13 100644
--- a/kernel/locking/rtmutex.c
+++ b/kernel/locking/rtmutex.c
@@ -1707,6 +1707,7 @@ static int __sched __rt_mutex_slowlock(struct rt_mutex_base *lock,
 	set_current_state(state);
 
 	trace_contention_begin(lock, LCB_F_RT);
+	trace_block_on_rt_mutex_begin_tp(current);
 
 	ret = task_blocks_on_rt_mutex(lock, waiter, current, ww_ctx, chwalk, wake_q);
 	if (likely(!ret))
@@ -1732,6 +1733,7 @@ static int __sched __rt_mutex_slowlock(struct rt_mutex_base *lock,
 	fixup_rt_mutex_waiters(lock, true);
 
 	trace_contention_end(lock, ret);
+	trace_block_on_rt_mutex_end_tp(current);
 
 	return ret;
 }
diff --git a/kernel/locking/rtmutex_api.c b/kernel/locking/rtmutex_api.c
index 191e4720e546..35f9bd7cbd54 100644
--- a/kernel/locking/rtmutex_api.c
+++ b/kernel/locking/rtmutex_api.c
@@ -302,6 +302,8 @@ int __sched __rt_mutex_start_proxy_lock(struct rt_mutex_base *lock,
 	if (try_to_take_rt_mutex(lock, task, NULL))
 		return 1;
 
+	trace_block_on_rt_mutex_begin_tp(task);
+
 	/* We enforce deadlock detection for futexes */
 	ret = task_blocks_on_rt_mutex(lock, waiter, task, NULL,
 				      RT_MUTEX_FULL_CHAINWALK, wake_q);
@@ -391,6 +393,8 @@ int __sched rt_mutex_wait_proxy_lock(struct rt_mutex_base *lock,
 	fixup_rt_mutex_waiters(lock, true);
 	raw_spin_unlock_irq(&lock->wait_lock);
 
+	trace_block_on_rt_mutex_end_tp(current);
+
 	return ret;
 }
 
-- 
2.39.5