[<prev] [next>] [day] [month] [year] [list]
Message-ID: <164914777661.389.2436402911593584852.tip-bot2@tip-bot2>
Date: Tue, 05 Apr 2022 08:36:16 -0000
From: "tip-bot2 for Peter Zijlstra" <tip-bot2@...utronix.de>
To: linux-tip-commits@...r.kernel.org
Cc: "Peter Zijlstra (Intel)" <peterz@...radead.org>, x86@...nel.org,
linux-kernel@...r.kernel.org
Subject: [tip: locking/core] locking/mutex: Make contention tracepoints more
consistent wrt adaptive spinning
The following commit has been merged into the locking/core branch of tip:
Commit-ID: dc1f7893a70fe403983bd8492f177bf993940e2c
Gitweb: https://git.kernel.org/tip/dc1f7893a70fe403983bd8492f177bf993940e2c
Author: Peter Zijlstra <peterz@...radead.org>
AuthorDate: Wed, 30 Mar 2022 13:06:54 +02:00
Committer: Peter Zijlstra <peterz@...radead.org>
CommitterDate: Tue, 05 Apr 2022 10:24:36 +02:00
locking/mutex: Make contention tracepoints more consistent wrt adaptive spinning
Have the trace_contention_*() tracepoints consistently include
adaptive spinning. In order to differentiate between the spinning and
non-spinning states add LCB_F_MUTEX and combine with LCB_F_SPIN.
The consequence is that a mutex contention can now triggler multiple
_begin() tracepoints before triggering an _end().
Additionally, this fixes one path where mutex would trigger _end()
without ever seeing a _begin().
Signed-off-by: Peter Zijlstra (Intel) <peterz@...radead.org>
---
include/trace/events/lock.h | 4 +++-
kernel/locking/mutex.c | 16 ++++++++++++----
2 files changed, 15 insertions(+), 5 deletions(-)
diff --git a/include/trace/events/lock.h b/include/trace/events/lock.h
index b9b6e3e..9ebd081 100644
--- a/include/trace/events/lock.h
+++ b/include/trace/events/lock.h
@@ -14,6 +14,7 @@
#define LCB_F_WRITE (1U << 2)
#define LCB_F_RT (1U << 3)
#define LCB_F_PERCPU (1U << 4)
+#define LCB_F_MUTEX (1U << 5)
#ifdef CONFIG_LOCKDEP
@@ -113,7 +114,8 @@ TRACE_EVENT(contention_begin,
{ LCB_F_READ, "READ" },
{ LCB_F_WRITE, "WRITE" },
{ LCB_F_RT, "RT" },
- { LCB_F_PERCPU, "PERCPU" }
+ { LCB_F_PERCPU, "PERCPU" },
+ { LCB_F_MUTEX, "MUTEX" }
))
);
diff --git a/kernel/locking/mutex.c b/kernel/locking/mutex.c
index c88deda..d973fe6 100644
--- a/kernel/locking/mutex.c
+++ b/kernel/locking/mutex.c
@@ -602,12 +602,14 @@ __mutex_lock_common(struct mutex *lock, unsigned int state, unsigned int subclas
preempt_disable();
mutex_acquire_nest(&lock->dep_map, subclass, 0, nest_lock, ip);
+ trace_contention_begin(lock, LCB_F_MUTEX | LCB_F_SPIN);
if (__mutex_trylock(lock) ||
mutex_optimistic_spin(lock, ww_ctx, NULL)) {
/* got the lock, yay! */
lock_acquired(&lock->dep_map, ip);
if (ww_ctx)
ww_mutex_set_context_fastpath(ww, ww_ctx);
+ trace_contention_end(lock, 0);
preempt_enable();
return 0;
}
@@ -644,7 +646,7 @@ __mutex_lock_common(struct mutex *lock, unsigned int state, unsigned int subclas
}
set_current_state(state);
- trace_contention_begin(lock, 0);
+ trace_contention_begin(lock, LCB_F_MUTEX);
for (;;) {
bool first;
@@ -684,10 +686,16 @@ __mutex_lock_common(struct mutex *lock, unsigned int state, unsigned int subclas
* state back to RUNNING and fall through the next schedule(),
* or we must see its unlock and acquire.
*/
- if (__mutex_trylock_or_handoff(lock, first) ||
- (first && mutex_optimistic_spin(lock, ww_ctx, &waiter)))
+ if (__mutex_trylock_or_handoff(lock, first))
break;
+ if (first) {
+ trace_contention_begin(lock, LCB_F_MUTEX | LCB_F_SPIN);
+ if (mutex_optimistic_spin(lock, ww_ctx, &waiter))
+ break;
+ trace_contention_begin(lock, LCB_F_MUTEX);
+ }
+
raw_spin_lock(&lock->wait_lock);
}
raw_spin_lock(&lock->wait_lock);
@@ -723,8 +731,8 @@ skip_wait:
err:
__set_current_state(TASK_RUNNING);
__mutex_remove_waiter(lock, &waiter);
- trace_contention_end(lock, ret);
err_early_kill:
+ trace_contention_end(lock, ret);
raw_spin_unlock(&lock->wait_lock);
debug_mutex_free_waiter(&waiter);
mutex_release(&lock->dep_map, ip);
Powered by blists - more mailing lists