[PATCH 4/4] sched: Remove set_task_state()

From: Davidlohr Bueso
Date: Tue Jan 03 2017 - 16:44:29 EST


This is a nasty interface and setting the state of a foreign task
must not be done. As of be628be0956 (bcache: Make gc wakeup sane,
remove set_task_state()) everyone in the kernel calls
set_task_state() with current, allowing the helper to be removed.
However, as the comment indicates, it is still around for those
archs where computing current is more expensive than using a pointer,
at least in theory. An important arch that is affected is arm64[1],
however this has been addressed now[2] and performance is up to par
making no difference with either calls.

Of all the callers, if any, it's the locking bits that would care
most about this -- ie: we end up passing a tsk pointer to a lot of
the lock slowpath, and setting ->state on that. The following numbers
are based on two tests: a custom ad-hoc microbenchmark that just
measures latencies (for ~65 million calls) between get_task_state()
vs get_current_state().

Secondly for a higher overview, an unlink microbenchmark was used,
which pounds on a single file with open, close,unlink combos with
increasing thread counts (up to 4x ncpus). While the workload is
quite unrealistic, it does contend a lot on the inode mutex or now
rwsem.

[1] https://lkml.org/lkml/2016/12/30/230
[2] http://lists.infradead.org/pipermail/linux-arm-kernel/2017-January/476461.html

== 1. x86-64 ==

Avg runtime set_task_state(): 601 msecs
Avg runtime set_current_state(): 552 msecs

vanilla dirty
Hmean unlink1-processes-2 36089.26 ( 0.00%) 38977.33 ( 8.00%)
Hmean unlink1-processes-5 28555.01 ( 0.00%) 29832.55 ( 4.28%)
Hmean unlink1-processes-8 37323.75 ( 0.00%) 44974.57 ( 20.50%)
Hmean unlink1-processes-12 43571.88 ( 0.00%) 44283.01 ( 1.63%)
Hmean unlink1-processes-21 34431.52 ( 0.00%) 38284.45 ( 11.19%)
Hmean unlink1-processes-30 34813.26 ( 0.00%) 37975.17 ( 9.08%)
Hmean unlink1-processes-48 37048.90 ( 0.00%) 39862.78 ( 7.59%)
Hmean unlink1-processes-79 35630.01 ( 0.00%) 36855.30 ( 3.44%)
Hmean unlink1-processes-110 36115.85 ( 0.00%) 39843.91 ( 10.32%)
Hmean unlink1-processes-141 32546.96 ( 0.00%) 35418.52 ( 8.82%)
Hmean unlink1-processes-172 34674.79 ( 0.00%) 36899.21 ( 6.42%)
Hmean unlink1-processes-203 37303.11 ( 0.00%) 36393.04 ( -2.44%)
Hmean unlink1-processes-224 35712.13 ( 0.00%) 36685.96 ( 2.73%)

== 2. ppc64le ==

Avg runtime set_task_state(): 938 msecs
Avg runtime set_current_state: 940 msecs

vanilla dirty
Hmean unlink1-processes-2 19269.19 ( 0.00%) 30704.50 ( 59.35%)
Hmean unlink1-processes-5 20106.15 ( 0.00%) 21804.15 ( 8.45%)
Hmean unlink1-processes-8 17496.97 ( 0.00%) 17243.28 ( -1.45%)
Hmean unlink1-processes-12 14224.15 ( 0.00%) 17240.21 ( 21.20%)
Hmean unlink1-processes-21 14155.66 ( 0.00%) 15681.23 ( 10.78%)
Hmean unlink1-processes-30 14450.70 ( 0.00%) 15995.83 ( 10.69%)
Hmean unlink1-processes-48 16945.57 ( 0.00%) 16370.42 ( -3.39%)
Hmean unlink1-processes-79 15788.39 ( 0.00%) 14639.27 ( -7.28%)
Hmean unlink1-processes-110 14268.48 ( 0.00%) 14377.40 ( 0.76%)
Hmean unlink1-processes-141 14023.65 ( 0.00%) 16271.69 ( 16.03%)
Hmean unlink1-processes-172 13417.62 ( 0.00%) 16067.55 ( 19.75%)
Hmean unlink1-processes-203 15293.08 ( 0.00%) 15440.40 ( 0.96%)
Hmean unlink1-processes-234 13719.32 ( 0.00%) 16190.74 ( 18.01%)
Hmean unlink1-processes-265 16400.97 ( 0.00%) 16115.22 ( -1.74%)
Hmean unlink1-processes-296 14388.60 ( 0.00%) 16216.13 ( 12.70%)
Hmean unlink1-processes-320 15771.85 ( 0.00%) 15905.96 ( 0.85%)

x86-64 (known to be fast for get_current()/this_cpu_read_stable() caching)
and ppc64 (with paca) show similar improvements in the unlink microbenches.
The small delta for ppc64 (2ms), does not represent the gains on the unlink
runs. In the case of x86, there was a decent amount of variation in the
latency runs, but always within a 20 to 50ms increase), ppc was more constant.

Signed-off-by: Davidlohr Bueso <dbueso@xxxxxxx>
---
arch/um/drivers/random.c | 2 +-
drivers/md/dm-bufio.c | 2 +-
drivers/md/dm-crypt.c | 4 ++--
drivers/md/persistent-data/dm-block-manager.c | 4 ++--
.../staging/lustre/lnet/libcfs/linux/linux-debug.c | 2 +-
drivers/tty/tty_ldsem.c | 10 ++++----
include/linux/sched.h | 27 +---------------------
kernel/exit.c | 4 ++--
kernel/locking/mutex.c | 8 +++----
kernel/locking/rwsem-spinlock.c | 8 +++----
kernel/locking/rwsem-xadd.c | 4 ++--
kernel/locking/semaphore.c | 2 +-
12 files changed, 26 insertions(+), 51 deletions(-)

diff --git a/arch/um/drivers/random.c b/arch/um/drivers/random.c
index 05523f14d7b2..57f03050c850 100644
--- a/arch/um/drivers/random.c
+++ b/arch/um/drivers/random.c
@@ -76,7 +76,7 @@ static ssize_t rng_dev_read (struct file *filp, char __user *buf, size_t size,
add_sigio_fd(random_fd);

add_wait_queue(&host_read_wait, &wait);
- set_task_state(current, TASK_INTERRUPTIBLE);
+ set_current_state(TASK_INTERRUPTIBLE);

schedule();
remove_wait_queue(&host_read_wait, &wait);
diff --git a/drivers/md/dm-bufio.c b/drivers/md/dm-bufio.c
index 84d2f0e4c754..d36d427a9efb 100644
--- a/drivers/md/dm-bufio.c
+++ b/drivers/md/dm-bufio.c
@@ -794,7 +794,7 @@ static void __wait_for_free_buffer(struct dm_bufio_client *c)
DECLARE_WAITQUEUE(wait, current);

add_wait_queue(&c->free_buffer_wait, &wait);
- set_task_state(current, TASK_UNINTERRUPTIBLE);
+ set_current_state(TASK_UNINTERRUPTIBLE);
dm_bufio_unlock(c);

io_schedule();
diff --git a/drivers/md/dm-crypt.c b/drivers/md/dm-crypt.c
index 7c6c57216bf2..96692d13a6e4 100644
--- a/drivers/md/dm-crypt.c
+++ b/drivers/md/dm-crypt.c
@@ -1210,14 +1210,14 @@ static int dmcrypt_write(void *data)
spin_unlock_irq(&cc->write_thread_wait.lock);

if (unlikely(kthread_should_stop())) {
- set_task_state(current, TASK_RUNNING);
+ set_current_state(TASK_RUNNING);
remove_wait_queue(&cc->write_thread_wait, &wait);
break;
}

schedule();

- set_task_state(current, TASK_RUNNING);
+ set_current_state(TASK_RUNNING);
spin_lock_irq(&cc->write_thread_wait.lock);
__remove_wait_queue(&cc->write_thread_wait, &wait);
goto continue_locked;
diff --git a/drivers/md/persistent-data/dm-block-manager.c b/drivers/md/persistent-data/dm-block-manager.c
index a6dde7cab458..758d90cc2733 100644
--- a/drivers/md/persistent-data/dm-block-manager.c
+++ b/drivers/md/persistent-data/dm-block-manager.c
@@ -120,7 +120,7 @@ static int __check_holder(struct block_lock *lock)
static void __wait(struct waiter *w)
{
for (;;) {
- set_task_state(current, TASK_UNINTERRUPTIBLE);
+ set_current_state(TASK_UNINTERRUPTIBLE);

if (!w->task)
break;
@@ -128,7 +128,7 @@ static void __wait(struct waiter *w)
schedule();
}

- set_task_state(current, TASK_RUNNING);
+ set_current_state(TASK_RUNNING);
}

static void __wake_waiter(struct waiter *w)
diff --git a/drivers/staging/lustre/lnet/libcfs/linux/linux-debug.c b/drivers/staging/lustre/lnet/libcfs/linux/linux-debug.c
index 39a72e3f0c18..7035356e56b3 100644
--- a/drivers/staging/lustre/lnet/libcfs/linux/linux-debug.c
+++ b/drivers/staging/lustre/lnet/libcfs/linux/linux-debug.c
@@ -107,7 +107,7 @@ void __noreturn lbug_with_loc(struct libcfs_debug_msg_data *msgdata)
libcfs_debug_dumplog();
if (libcfs_panic_on_lbug)
panic("LBUG");
- set_task_state(current, TASK_UNINTERRUPTIBLE);
+ set_current_state(TASK_UNINTERRUPTIBLE);
while (1)
schedule();
}
diff --git a/drivers/tty/tty_ldsem.c b/drivers/tty/tty_ldsem.c
index 3e6722954f29..9229de43e19d 100644
--- a/drivers/tty/tty_ldsem.c
+++ b/drivers/tty/tty_ldsem.c
@@ -231,7 +231,7 @@ down_read_failed(struct ld_semaphore *sem, long count, long timeout)

/* wait to be given the lock */
for (;;) {
- set_task_state(current, TASK_UNINTERRUPTIBLE);
+ set_current_state(TASK_UNINTERRUPTIBLE);

if (!waiter.task)
break;
@@ -240,7 +240,7 @@ down_read_failed(struct ld_semaphore *sem, long count, long timeout)
timeout = schedule_timeout(timeout);
}

- __set_task_state(current, TASK_RUNNING);
+ __set_current_state(TASK_RUNNING);

if (!timeout) {
/* lock timed out but check if this task was just
@@ -289,14 +289,14 @@ down_write_failed(struct ld_semaphore *sem, long count, long timeout)

waiter.task = current;

- set_task_state(current, TASK_UNINTERRUPTIBLE);
+ set_current_state(TASK_UNINTERRUPTIBLE);
for (;;) {
if (!timeout)
break;
raw_spin_unlock_irq(&sem->wait_lock);
timeout = schedule_timeout(timeout);
raw_spin_lock_irq(&sem->wait_lock);
- set_task_state(current, TASK_UNINTERRUPTIBLE);
+ set_current_state(TASK_UNINTERRUPTIBLE);
locked = writer_trylock(sem);
if (locked)
break;
@@ -307,7 +307,7 @@ down_write_failed(struct ld_semaphore *sem, long count, long timeout)
list_del(&waiter.list);
raw_spin_unlock_irq(&sem->wait_lock);

- __set_task_state(current, TASK_RUNNING);
+ __set_current_state(TASK_RUNNING);

/* lock wait may have timed out */
if (!locked)
diff --git a/include/linux/sched.h b/include/linux/sched.h
index 4d1905245c7a..8edf16d82f8c 100644
--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@ -227,7 +227,7 @@ extern void proc_sched_set_task(struct task_struct *p);
extern char ___assert_task_state[1 - 2*!!(
sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];

-/* Convenience macros for the sake of set_task_state */
+/* Convenience macros for the sake of set_current_state */
#define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
#define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
#define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
@@ -254,17 +254,6 @@ extern char ___assert_task_state[1 - 2*!!(

#ifdef CONFIG_DEBUG_ATOMIC_SLEEP

-#define __set_task_state(tsk, state_value) \
- do { \
- (tsk)->task_state_change = _THIS_IP_; \
- (tsk)->state = (state_value); \
- } while (0)
-#define set_task_state(tsk, state_value) \
- do { \
- (tsk)->task_state_change = _THIS_IP_; \
- smp_store_mb((tsk)->state, (state_value)); \
- } while (0)
-
#define __set_current_state(state_value) \
do { \
current->task_state_change = _THIS_IP_; \
@@ -277,20 +266,6 @@ extern char ___assert_task_state[1 - 2*!!(
} while (0)

#else
-
-/*
- * @tsk had better be current, or you get to keep the pieces.
- *
- * The only reason is that computing current can be more expensive than
- * using a pointer that's already available.
- *
- * Therefore, see set_current_state().
- */
-#define __set_task_state(tsk, state_value) \
- do { (tsk)->state = (state_value); } while (0)
-#define set_task_state(tsk, state_value) \
- smp_store_mb((tsk)->state, (state_value))
-
/*
* set_current_state() includes a barrier so that the write of current->state
* is correctly serialised wrt the caller's subsequent test of whether to
diff --git a/kernel/exit.c b/kernel/exit.c
index 2385d434a46e..27c68653e2fc 100644
--- a/kernel/exit.c
+++ b/kernel/exit.c
@@ -501,12 +501,12 @@ static void exit_mm(void)
complete(&core_state->startup);

for (;;) {
- set_task_state(current, TASK_UNINTERRUPTIBLE);
+ set_current_state(TASK_UNINTERRUPTIBLE);
if (!self.task) /* see coredump_finish() */
break;
freezable_schedule();
}
- __set_task_state(current, TASK_RUNNING);
+ __set_current_state(TASK_RUNNING);
down_read(&mm->mmap_sem);
}
atomic_inc(&mm->mm_count);
diff --git a/kernel/locking/mutex.c b/kernel/locking/mutex.c
index 4c7d04362c95..97d142486f93 100644
--- a/kernel/locking/mutex.c
+++ b/kernel/locking/mutex.c
@@ -666,7 +666,7 @@ __mutex_lock_common(struct mutex *lock, long state, unsigned int subclass,

lock_contended(&lock->dep_map, ip);

- set_task_state(current, state);
+ set_current_state(state);
for (;;) {
/*
* Once we hold wait_lock, we're serialized against
@@ -701,7 +701,7 @@ __mutex_lock_common(struct mutex *lock, long state, unsigned int subclass,
__mutex_set_flag(lock, MUTEX_FLAG_HANDOFF);
}

- set_task_state(current, state);
+ set_current_state(state);
/*
* Here we order against unlock; we must either see it change
* state back to RUNNING and fall through the next schedule(),
@@ -715,7 +715,7 @@ __mutex_lock_common(struct mutex *lock, long state, unsigned int subclass,
}
spin_lock_mutex(&lock->wait_lock, flags);
acquired:
- __set_task_state(current, TASK_RUNNING);
+ __set_current_state(TASK_RUNNING);

mutex_remove_waiter(lock, &waiter, current);
if (likely(list_empty(&lock->wait_list)))
@@ -735,7 +735,7 @@ __mutex_lock_common(struct mutex *lock, long state, unsigned int subclass,
return 0;

err:
- __set_task_state(current, TASK_RUNNING);
+ __set_current_state(TASK_RUNNING);
mutex_remove_waiter(lock, &waiter, current);
spin_unlock_mutex(&lock->wait_lock, flags);
debug_mutex_free_waiter(&waiter);
diff --git a/kernel/locking/rwsem-spinlock.c b/kernel/locking/rwsem-spinlock.c
index 60d15d3bb2a8..5eacab880f67 100644
--- a/kernel/locking/rwsem-spinlock.c
+++ b/kernel/locking/rwsem-spinlock.c
@@ -139,7 +139,7 @@ void __sched __down_read(struct rw_semaphore *sem)
goto out;
}

- set_task_state(current, TASK_UNINTERRUPTIBLE);
+ set_current_state(TASK_UNINTERRUPTIBLE);

/* set up my own style of waitqueue */
waiter.task = current;
@@ -156,10 +156,10 @@ void __sched __down_read(struct rw_semaphore *sem)
if (!waiter.task)
break;
schedule();
- set_task_state(current, TASK_UNINTERRUPTIBLE);
+ set_current_state(TASK_UNINTERRUPTIBLE);
}

- __set_task_state(current, TASK_RUNNING);
+ __set_current_state(TASK_RUNNING);
out:
;
}
@@ -216,7 +216,7 @@ int __sched __down_write_common(struct rw_semaphore *sem, int state)
ret = -EINTR;
goto out;
}
- set_task_state(current, state);
+ set_current_state(state);
raw_spin_unlock_irqrestore(&sem->wait_lock, flags);
schedule();
raw_spin_lock_irqsave(&sem->wait_lock, flags);
diff --git a/kernel/locking/rwsem-xadd.c b/kernel/locking/rwsem-xadd.c
index d3b819c7f07f..a3a381aee24b 100644
--- a/kernel/locking/rwsem-xadd.c
+++ b/kernel/locking/rwsem-xadd.c
@@ -253,13 +253,13 @@ struct rw_semaphore __sched *rwsem_down_read_failed(struct rw_semaphore *sem)

/* wait to be given the lock */
while (true) {
- set_task_state(current, TASK_UNINTERRUPTIBLE);
+ set_current_state(TASK_UNINTERRUPTIBLE);
if (!waiter.task)
break;
schedule();
}

- __set_task_state(current, TASK_RUNNING);
+ __set_current_state(TASK_RUNNING);
return sem;
}
EXPORT_SYMBOL(rwsem_down_read_failed);
diff --git a/kernel/locking/semaphore.c b/kernel/locking/semaphore.c
index 29aac9f9e5e4..9512e37637dc 100644
--- a/kernel/locking/semaphore.c
+++ b/kernel/locking/semaphore.c
@@ -215,7 +215,7 @@ static inline int __sched __down_common(struct semaphore *sem, long state,
goto interrupted;
if (unlikely(timeout <= 0))
goto timed_out;
- __set_task_state(current, state);
+ __set_current_state(state);
raw_spin_unlock_irq(&sem->lock);
timeout = schedule_timeout(timeout);
raw_spin_lock_irq(&sem->lock);
--
2.6.6