diff --git a/kernel/src/process/signal/pause.rs b/kernel/src/process/signal/pause.rs
index 85744dfba..56625ac73 100644
--- a/kernel/src/process/signal/pause.rs
+++ b/kernel/src/process/signal/pause.rs
@@ -32,6 +32,7 @@ pub trait Pause: WaitTimeout {
     /// condition is met.
     ///
     /// [`EINTR`]: crate::error::Errno::EINTR
+    #[track_caller]
     fn pause_until<F, R>(&self, cond: F) -> Result<R>
     where
         F: FnMut() -> Option<R>,
@@ -49,6 +50,7 @@ pub trait Pause: WaitTimeout {
     ///
     /// [`ETIME`]: crate::error::Errno::ETIME
     /// [`EINTR`]: crate::error::Errno::EINTR
+    #[track_caller]
     fn pause_until_or_timeout<'a, F, T, R>(&self, mut cond: F, timeout: T) -> Result<R>
     where
         F: FnMut() -> Option<R>,
@@ -74,6 +76,7 @@ pub trait Pause: WaitTimeout {
     /// [`ETIME`]: crate::error::Errno::ETIME
     /// [`EINTR`]: crate::error::Errno::EINTR
     #[doc(hidden)]
+    #[track_caller]
     fn pause_until_or_timeout_impl<F, R>(
         &self,
         cond: F,
@@ -94,6 +97,7 @@ pub trait Pause: WaitTimeout {
     ///
     /// [`ETIME`]: crate::error::Errno::ETIME
     /// [`EINTR`]: crate::error::Errno::EINTR
+    #[track_caller]
     fn pause_timeout<'a>(&self, timeout: impl Into<TimeoutExt<'a>>) -> Result<()>;
 }
 
diff --git a/kernel/src/process/signal/poll.rs b/kernel/src/process/signal/poll.rs
index 1f97a6f93..bf3c63665 100644
--- a/kernel/src/process/signal/poll.rs
+++ b/kernel/src/process/signal/poll.rs
@@ -368,6 +368,7 @@ pub trait Pollable {
     /// The user must ensure that a call to `try_op()` does not fail with `EAGAIN` when the
     /// interesting events occur. However, it is allowed to have spurious `EAGAIN` failures due to
     /// race opitions where the events are consumed by another thread.
+    #[track_caller]
     fn wait_events<F, R>(
         &self,
         mask: IoEvents,
diff --git a/kernel/src/thread/kernel_thread.rs b/kernel/src/thread/kernel_thread.rs
index 04ee8cf0b..9bfb67b0b 100644
--- a/kernel/src/thread/kernel_thread.rs
+++ b/kernel/src/thread/kernel_thread.rs
@@ -73,6 +73,7 @@ impl ThreadOptions {
     }
 
     /// Builds a new kernel thread and runs it immediately.
+    #[track_caller]
     pub fn spawn(self) -> Arc<Thread> {
         let task = self.build();
         let thread = task.as_thread().unwrap().clone();
diff --git a/kernel/src/thread/mod.rs b/kernel/src/thread/mod.rs
index 84b937200..a7d2f0b9c 100644
--- a/kernel/src/thread/mod.rs
+++ b/kernel/src/thread/mod.rs
@@ -78,6 +78,7 @@ impl Thread {
     }
 
     /// Runs this thread at once.
+    #[track_caller]
     pub fn run(&self) {
         self.status.store(ThreadStatus::Running, Ordering::Release);
         self.task.upgrade().unwrap().run();
@@ -151,6 +152,7 @@ impl Thread {
     /// Yields the execution to another thread.
     ///
     /// This method will return once the current thread is scheduled again.
+    #[track_caller]
     pub fn yield_now() {
         Task::yield_now()
     }
@@ -158,6 +160,7 @@ impl Thread {
     /// Joins the execution of the thread.
     ///
     /// This method will return after the thread exits.
+    #[track_caller]
     pub fn join(&self) {
         while !self.is_exited() {
             Self::yield_now();
diff --git a/kernel/src/time/wait.rs b/kernel/src/time/wait.rs
index ce308249b..a246211fa 100644
--- a/kernel/src/time/wait.rs
+++ b/kernel/src/time/wait.rs
@@ -17,6 +17,7 @@ pub trait WaitTimeout {
     /// will return an error with [`ETIME`].
     ///
     /// [`ETIME`]: crate::error::Errno::ETIME
+    #[track_caller]
     fn wait_until_or_timeout<'a, F, T, R>(&self, mut cond: F, timeout: T) -> Result<R>
     where
         F: FnMut() -> Option<R>,
@@ -41,6 +42,7 @@ pub trait WaitTimeout {
     ///  - an error with [`ETIME`] if the timeout is reached;
     ///  - the error returned by the cancel condition if the cancel condition returns `Err(_)`.
     #[doc(hidden)]
+    #[track_caller]
     fn wait_until_or_timeout_cancelled<F, R, FCancel>(
         &self,
         cond: F,
diff --git a/ostd/src/sync/mutex.rs b/ostd/src/sync/mutex.rs
index 75036e181..34bc4f4ed 100644
--- a/ostd/src/sync/mutex.rs
+++ b/ostd/src/sync/mutex.rs
@@ -32,6 +32,7 @@ impl<T: ?Sized> Mutex<T> {
     /// Acquires the mutex.
     ///
     /// This method runs in a block way until the mutex can be acquired.
+    #[track_caller]
     pub fn lock(&self) -> MutexGuard<T> {
         self.queue.wait_until(|| self.try_lock())
     }
diff --git a/ostd/src/sync/rwmutex.rs b/ostd/src/sync/rwmutex.rs
index d5fc9a843..2ba98744a 100644
--- a/ostd/src/sync/rwmutex.rs
+++ b/ostd/src/sync/rwmutex.rs
@@ -121,6 +121,7 @@ impl<T: ?Sized> RwMutex<T> {
     /// upreaders present. The implementation of [`WaitQueue`] guarantees the
     /// order in which other concurrent readers or writers waiting simultaneously
     /// will acquire the mutex.
+    #[track_caller]
     pub fn read(&self) -> RwMutexReadGuard<T> {
         self.queue.wait_until(|| self.try_read())
     }
@@ -131,6 +132,7 @@ impl<T: ?Sized> RwMutex<T> {
     /// or readers present. The implementation of [`WaitQueue`] guarantees the
     /// order in which other concurrent readers or writers waiting simultaneously
     /// will acquire the mutex.
+    #[track_caller]
     pub fn write(&self) -> RwMutexWriteGuard<T> {
         self.queue.wait_until(|| self.try_write())
     }
@@ -145,6 +147,7 @@ impl<T: ?Sized> RwMutex<T> {
     /// and reader do not differ before invoking the upgread method. However,
     /// only one upreader can exist at any time to avoid deadlock in the
     /// upgread method.
+    #[track_caller]
     pub fn upread(&self) -> RwMutexUpgradeableGuard<T> {
         self.queue.wait_until(|| self.try_upread())
     }
diff --git a/ostd/src/sync/wait.rs b/ostd/src/sync/wait.rs
index dd4d6ce8f..40301b670 100644
--- a/ostd/src/sync/wait.rs
+++ b/ostd/src/sync/wait.rs
@@ -65,6 +65,7 @@ impl WaitQueue {
     ///
     /// By taking a condition closure, this wait-wakeup mechanism becomes
     /// more efficient and robust.
+    #[track_caller]
     pub fn wait_until<F, R>(&self, mut cond: F) -> R
     where
         F: FnMut() -> Option<R>,
@@ -194,6 +195,7 @@ impl Waiter {
     /// This method returns immediately if the waiter has been woken since the end of the last call
     /// to this method (or since the waiter was created, if this method has not been called
     /// before). Otherwise, it puts the current thread to sleep until the waiter is woken up.
+    #[track_caller]
     pub fn wait(&self) {
         self.waker.do_wait();
     }
@@ -203,6 +205,7 @@ impl Waiter {
     /// This method will return `Ok(_)` if the condition returns `Some(_)`, and will stop waiting
     /// if the cancel condition returns `Err(_)`. In this situation, this method will return the `Err(_)`
     /// generated by the cancel condition.
+    #[track_caller]
     pub fn wait_until_or_cancelled<F, R, FCancel, E>(
         &self,
         mut cond: F,
@@ -265,6 +268,7 @@ impl Waker {
         true
     }
 
+    #[track_caller]
     fn do_wait(&self) {
         while !self.has_woken.swap(false, Ordering::Acquire) {
             scheduler::park_current(|| self.has_woken.load(Ordering::Acquire));
diff --git a/ostd/src/task/atomic_mode.rs b/ostd/src/task/atomic_mode.rs
index 799c8b288..eb959a7ac 100644
--- a/ostd/src/task/atomic_mode.rs
+++ b/ostd/src/task/atomic_mode.rs
@@ -32,6 +32,7 @@ use core::sync::atomic::Ordering;
 /// Marks a function as one that might sleep.
 ///
 /// This function will panic if it is executed in atomic mode.
+#[track_caller]
 pub fn might_sleep() {
     let preempt_count = super::preempt::cpu_local::get_guard_count();
     let is_local_irq_enabled = crate::arch::irq::is_local_enabled();
diff --git a/ostd/src/task/mod.rs b/ostd/src/task/mod.rs
index fcac605ec..ae3a9984d 100644
--- a/ostd/src/task/mod.rs
+++ b/ostd/src/task/mod.rs
@@ -83,6 +83,7 @@ impl Task {
     ///
     /// Note that this method cannot be simply named "yield" as the name is
     /// a Rust keyword.
+    #[track_caller]
     pub fn yield_now() {
         scheduler::yield_now()
     }
@@ -90,6 +91,7 @@ impl Task {
     /// Kicks the task scheduler to run the task.
     ///
     /// BUG: This method highly depends on the current scheduling policy.
+    #[track_caller]
     pub fn run(self: &Arc<Self>) {
         scheduler::run_new_task(self.clone());
     }
@@ -239,6 +241,7 @@ impl TaskOptions {
     }
 
     /// Builds a new task and runs it immediately.
+    #[track_caller]
     pub fn spawn(self) -> Result<Arc<Task>> {
         let task = Arc::new(self.build()?);
         task.run();
diff --git a/ostd/src/task/processor.rs b/ostd/src/task/processor.rs
index 51527cc5d..1eadb4be3 100644
--- a/ostd/src/task/processor.rs
+++ b/ostd/src/task/processor.rs
@@ -33,6 +33,7 @@ pub(super) fn current_task() -> Option<NonNull<Task>> {
 ///
 /// This function will panic if called while holding preemption locks or with
 /// local IRQ disabled.
+#[track_caller]
 pub(super) fn switch_to_task(next_task: Arc<Task>) {
     super::atomic_mode::might_sleep();
 
diff --git a/ostd/src/task/scheduler/mod.rs b/ostd/src/task/scheduler/mod.rs
index e714ee280..ae5baa3c3 100644
--- a/ostd/src/task/scheduler/mod.rs
+++ b/ostd/src/task/scheduler/mod.rs
@@ -105,6 +105,7 @@ pub enum UpdateFlags {
 }
 
 /// Preempts the current task.
+#[track_caller]
 pub(crate) fn might_preempt() {
     if !cpu_local::should_preempt() {
         return;
@@ -116,6 +117,7 @@ pub(crate) fn might_preempt() {
 ///
 /// Note that this method may return due to spurious wake events. It's the caller's responsibility
 /// to detect them (if necessary).
+#[track_caller]
 pub(crate) fn park_current<F>(has_woken: F)
 where
     F: Fn() -> bool,
@@ -168,6 +170,7 @@ pub(crate) fn unpark_target(runnable: Arc<Task>) {
 /// Enqueues a newly built task.
 ///
 /// Note that the new task is not guaranteed to run at once.
+#[track_caller]
 pub(super) fn run_new_task(runnable: Arc<Task>) {
     // FIXME: remove this check for `SCHEDULER`.
     // Currently OSTD cannot know whether its user has injected a scheduler.
@@ -199,6 +202,7 @@ fn set_need_preempt(cpu_id: CpuId) {
 /// Dequeues the current task from its runqueue.
 ///
 /// This should only be called if the current is to exit.
+#[track_caller]
 pub(super) fn exit_current() -> ! {
     reschedule(|local_rq: &mut dyn LocalRunQueue| {
         let _ = local_rq.dequeue_current();
@@ -213,6 +217,7 @@ pub(super) fn exit_current() -> ! {
 }
 
 /// Yields execution.
+#[track_caller]
 pub(super) fn yield_now() {
     reschedule(|local_rq| {
         local_rq.update_current(UpdateFlags::Yield);
@@ -228,6 +233,7 @@ pub(super) fn yield_now() {
 /// user-given closure.
 ///
 /// The closure makes the scheduling decision by taking the local runqueue has its input.
+#[track_caller]
 fn reschedule<F>(mut f: F)
 where
     F: FnMut(&mut dyn LocalRunQueue) -> ReschedAction,
diff --git a/ostd/src/user.rs b/ostd/src/user.rs
index 16fe7fe9d..070b3cfaa 100644
--- a/ostd/src/user.rs
+++ b/ostd/src/user.rs
@@ -151,6 +151,7 @@ impl<'a> UserMode<'a> {
     /// cause the method to return
     /// and updating the user-mode CPU context,
     /// this method can be invoked again to go back to the user space.
+    #[track_caller]
     pub fn execute<F>(&mut self, has_kernel_event: F) -> ReturnReason
     where
         F: FnMut() -> bool,