Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1474223 > unrolled thread

[PATCH 0/3] sched/cputime: sum_exec_runtime fixes for 32-bit cpus

Back to article view | Back to linux.kernel

Contents

  [PATCH 0/3] sched/cputime: sum_exec_runtime fixes for 32-bit cpus Stanislaw Gruszka <sgruszka@redhat.com> - 2016-09-01 11:40 +0200
    [PATCH 3/3] sched/cputime: Protect other sum_exec_runtime reads on 32 bit cpus Stanislaw Gruszka <sgruszka@redhat.com> - 2016-09-01 11:40 +0200

#1474223 — [PATCH 0/3] sched/cputime: sum_exec_runtime fixes for 32-bit cpus

We already improved scalability of SYS_times() and
SYS_clock_gettimes(CLOCK_PROCESS_CPUTIME_ID) syscalls on 64-bit cpus.
However performance on 32-bit cpus is still poor. Patches improved that
and fix sum_exec_runtime 64 bit variable possible (but very very unlike)
inconsistency problem on 32-bit architectures.

Second patch could be merged with first one, but I posted it separately
for eventual discussion.

Stanislaw Gruszka (3):
  sched/cputime: Improve scalability of times()/clock_gettime() on 32
    bit cpus
  sched/cputime: Make nr_migrations u32 on 32 bit cpus
  sched/cputime: Protect other sum_exec_runtime reads on 32 bit cpus

 fs/proc/base.c                 |  2 +-
 include/linux/sched.h          | 40 ++++++++++++++++++++++++++++++++++++++--
 kernel/delayacct.c             |  2 +-
 kernel/exit.c                  |  2 +-
 kernel/sched/core.c            | 17 +++++------------
 kernel/sched/cputime.c         | 24 ++----------------------
 kernel/sched/deadline.c        |  2 +-
 kernel/sched/debug.c           |  4 ++--
 kernel/sched/fair.c            |  2 +-
 kernel/sched/rt.c              |  2 +-
 kernel/time/posix-cpu-timers.c |  7 ++++---
 11 files changed, 57 insertions(+), 47 deletions(-)

-- 
1.8.3.1

[toc] | [next] | [standalone]

#1474231 — [PATCH 3/3] sched/cputime: Protect other sum_exec_runtime reads on 32 bit cpus

We protect sum_exec_runtime variable against inconsistency when reading
it on thread_group_cputime(), but we also read it on other places not
called with task_rq_lock() taken.

Patch changes that, except file kernel/sched/debug.c , where we also
read some other u64 variables (vrunime, exec_start) without protection.
But since this is for debug purpose and inconsistency is very very
improbable, we can ignore it.

Signed-off-by: Stanislaw Gruszka <sgruszka@redhat.com>
---
 fs/proc/base.c                 | 2 +-
 kernel/delayacct.c             | 2 +-
 kernel/exit.c                  | 2 +-
 kernel/sched/cputime.c         | 2 +-
 kernel/time/posix-cpu-timers.c | 4 ++--
 5 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/fs/proc/base.c b/fs/proc/base.c
index e9ff186..8118d97 100644
--- a/fs/proc/base.c
+++ b/fs/proc/base.c
@@ -505,7 +505,7 @@ static int proc_pid_schedstat(struct seq_file *m, struct pid_namespace *ns,
 		seq_printf(m, "0 0 0\n");
 	else
 		seq_printf(m, "%llu %llu %lu\n",
-		   (unsigned long long)task->se.sum_exec_runtime,
+		   (unsigned long long)read_sum_exec_runtime(task),
 		   (unsigned long long)task->sched_info.run_delay,
 		   task->sched_info.pcount);
 
diff --git a/kernel/delayacct.c b/kernel/delayacct.c
index 435c14a..023b2ca 100644
--- a/kernel/delayacct.c
+++ b/kernel/delayacct.c
@@ -104,7 +104,7 @@ int __delayacct_add_tsk(struct taskstats *d, struct task_struct *tsk)
 	 */
 	t1 = tsk->sched_info.pcount;
 	t2 = tsk->sched_info.run_delay;
-	t3 = tsk->se.sum_exec_runtime;
+	t3 = read_sum_exec_runtime(tsk);
 
 	d->cpu_count += t1;
 
diff --git a/kernel/exit.c b/kernel/exit.c
index 2f974ae..a46f96f 100644
--- a/kernel/exit.c
+++ b/kernel/exit.c
@@ -134,7 +134,7 @@ static void __exit_signal(struct task_struct *tsk)
 	sig->inblock += task_io_get_inblock(tsk);
 	sig->oublock += task_io_get_oublock(tsk);
 	task_io_accounting_add(&sig->ioac, &tsk->ioac);
-	sig->sum_sched_runtime += tsk->se.sum_exec_runtime;
+	sig->sum_sched_runtime += read_sum_exec_runtime(tsk);
 	sig->nr_threads--;
 	__unhash_process(tsk, group_dead);
 	write_sequnlock(&sig->stats_lock);
diff --git a/kernel/sched/cputime.c b/kernel/sched/cputime.c
index 026c1c4..cd1dd43 100644
--- a/kernel/sched/cputime.c
+++ b/kernel/sched/cputime.c
@@ -682,7 +682,7 @@ out:
 void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st)
 {
 	struct task_cputime cputime = {
-		.sum_exec_runtime = p->se.sum_exec_runtime,
+		.sum_exec_runtime = read_sum_exec_runtime(p),
 	};
 
 	task_cputime(p, &cputime.utime, &cputime.stime);
diff --git a/kernel/time/posix-cpu-timers.c b/kernel/time/posix-cpu-timers.c
index 4d8466b..b1e7275 100644
--- a/kernel/time/posix-cpu-timers.c
+++ b/kernel/time/posix-cpu-timers.c
@@ -849,7 +849,7 @@ static void check_thread_timers(struct task_struct *tsk,
 	tsk_expires->virt_exp = expires_to_cputime(expires);
 
 	tsk_expires->sched_exp = check_timers_list(++timers, firing,
-						   tsk->se.sum_exec_runtime);
+						   read_sum_exec_runtime(tsk));
 
 	/*
 	 * Check for the special case thread timers.
@@ -1116,7 +1116,7 @@ static inline int fastpath_timer_check(struct task_struct *tsk)
 		struct task_cputime task_sample;
 
 		task_cputime(tsk, &task_sample.utime, &task_sample.stime);
-		task_sample.sum_exec_runtime = tsk->se.sum_exec_runtime;
+		task_sample.sum_exec_runtime = read_sum_exec_runtime(tsk);
 		if (task_cputime_expired(&task_sample, &tsk->cputime_expires))
 			return 1;
 	}
-- 
1.8.3.1

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web