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


Groups > linux.kernel > #1172553

[RFC PATCH 3/6] stop_machine: introduce stop_work_alloc() and stop_work_free()

From Oleg Nesterov <oleg@redhat.com>
Newsgroups linux.kernel
Subject [RFC PATCH 3/6] stop_machine: introduce stop_work_alloc() and stop_work_free()
Date 2015-06-26 04:20 +0200
Message-ID <pFrpg-1YQ-13@gated-at.bofh.it> (permalink)
References <pFrpf-1YQ-3@gated-at.bofh.it>
Organization linux.* mail to news gateway

Show all headers | View raw


A separate and intentionally suboptimal patch to simplify the review of
this and the next changes.

And the new helpers, stop_work_alloc(cpumask) and stop_work_free(cpumask),
should be called if the caller is going to use cpu_stopper->stop_work's.
Note that 2 callers can never deadlock even if their cpumask's overlap,
they always "lock" cpu_stopper->stop_owner's in the same order as if we
had another per-cpu mutex.

This is obviously greatly inefficient, this will be fixed later.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
---
 kernel/stop_machine.c |   50 +++++++++++++++++++++++++++++++++++++++++++++++++
 1 files changed, 50 insertions(+), 0 deletions(-)

diff --git a/kernel/stop_machine.c b/kernel/stop_machine.c
index 6212208..3d5d810 100644
--- a/kernel/stop_machine.c
+++ b/kernel/stop_machine.c
@@ -42,11 +42,61 @@ struct cpu_stopper {
 	struct list_head	works;		/* list of pending works */
 
 	struct cpu_stop_work	stop_work;	/* for stop_cpus */
+	struct task_struct	*stop_owner;
 };
 
 static DEFINE_PER_CPU(struct cpu_stopper, cpu_stopper);
 static bool stop_machine_initialized = false;
 
+static DECLARE_WAIT_QUEUE_HEAD(stop_work_wq);
+
+static void stop_work_free_one(int cpu)
+{
+	struct cpu_stopper *stopper = &per_cpu(cpu_stopper, cpu);
+	/* Can be NULL if stop_work_alloc(wait => false) fails */
+	if (likely(stopper->stop_owner == current))
+		stopper->stop_owner = NULL;
+}
+
+static void stop_work_free(const struct cpumask *cpumask)
+{
+	int cpu;
+
+	for_each_cpu(cpu, cpumask)
+		stop_work_free_one(cpu);
+	wake_up_all(&stop_work_wq);
+}
+
+static struct cpu_stop_work *stop_work_alloc_one(int cpu, bool wait)
+{
+	struct cpu_stopper *stopper = &per_cpu(cpu_stopper, cpu);
+
+	if (cmpxchg(&stopper->stop_owner, NULL, current) == NULL)
+		goto done;
+
+	if (!wait)
+		return NULL;
+
+	__wait_event(stop_work_wq,
+		cmpxchg(&stopper->stop_owner, NULL, current) == NULL);
+done:
+	return &stopper->stop_work;
+}
+
+static bool stop_work_alloc(const struct cpumask *cpumask, bool wait)
+{
+	int cpu;
+
+	for_each_cpu(cpu, cpumask) {
+		if (stop_work_alloc_one(cpu, wait))
+			continue;
+		stop_work_free(cpumask);
+		return false;
+	}
+
+	return true;
+}
+
 /*
  * Avoids a race between stop_two_cpus and global stop_cpus, where
  * the stoppers could get queued up in reverse order, leading to
-- 
1.5.5.1

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

Back to linux.kernel | Previous | Next | Find similar | Unroll thread


Thread

[RFC PATCH 3/6] stop_machine: introduce stop_work_alloc() and  stop_work_free() Oleg Nesterov <oleg@redhat.com> - 2015-06-26 04:20 +0200

csiph-web