[PATCH 4/4 V5] workqueue: Allow modifying low level unbound workqueue cpumask

[Lai Jiangshan]

Allow to modify the low-level unbound workqueues cpumask through
sysfs. This is performed by traversing the entire workqueue list
and calling wq_unbound_install_ctx_prepare() on the unbound workqueues
with the low level mask passed in. Only after all the preparation are done,
we commit them all together.

The oreder-workquue is ignore from the low level unbound workqueue cpumask,
it will be handled in near future.

The per-nodes' pwqs are mandatorily controlled by the low level cpumask, while
the default pwq ignores the low level cpumask when (and ONLY when) the cpumask set
by the user doesn't overlap with the low level cpumask. In this case, we can't
apply the empty cpumask to the default pwq, so we use the user-set cpumask
directly.

The default wq_unbound_cpumask is still cpu_possible_mask due to the workqueue
subsystem doesn't know what is the best default value for the runtime, the
system manager or other subsystem which knows the sufficient information should set
it when needed.

Cc: Christoph Lameter <cl@xxxxxxxxx>
Cc: Kevin Hilman <khilman@xxxxxxxxxx>
Cc: Lai Jiangshan <laijs@xxxxxxxxxxxxxx>
Cc: Mike Galbraith <bitbucket@xxxxxxxxx>
Cc: Paul E. McKenney <paulmck@xxxxxxxxxxxxxxxxxx>
Cc: Tejun Heo <tj@xxxxxxxxxx>
Cc: Viresh Kumar <viresh.kumar@xxxxxxxxxx>
Cc: Frederic Weisbecker <fweisbec@xxxxxxxxx>
Original-patch-by: Frederic Weisbecker <fweisbec@xxxxxxxxx>
Signed-off-by: Lai Jiangshan <laijs@xxxxxxxxxxxxxx>
---
 include/linux/workqueue.h |   1 +
 kernel/workqueue.c        | 121 +++++++++++++++++++++++++++++++++++++++++++---
 2 files changed, 114 insertions(+), 8 deletions(-)

diff --git a/include/linux/workqueue.h b/include/linux/workqueue.h
index deee212..052d1bf 100644
--- a/include/linux/workqueue.h
+++ b/include/linux/workqueue.h
@@ -424,6 +424,7 @@ struct workqueue_attrs *alloc_workqueue_attrs(gfp_t gfp_mask);
 void free_workqueue_attrs(struct workqueue_attrs *attrs);
 int apply_workqueue_attrs(struct workqueue_struct *wq,
 			  const struct workqueue_attrs *attrs);
+int workqueue_unbounds_cpumask_set(cpumask_var_t cpumask);
 
 extern bool queue_work_on(int cpu, struct workqueue_struct *wq,
 			struct work_struct *work);
diff --git a/kernel/workqueue.c b/kernel/workqueue.c
index d1197f0..47f1a9d 100644
--- a/kernel/workqueue.c
+++ b/kernel/workqueue.c
@@ -299,7 +299,7 @@ static DEFINE_SPINLOCK(wq_mayday_lock);	/* protects wq->maydays list */
 static LIST_HEAD(workqueues);		/* PR: list of all workqueues */
 static bool workqueue_freezing;		/* PL: have wqs started freezing? */
 
-static cpumask_var_t wq_unbound_cpumask;
+static cpumask_var_t wq_unbound_cpumask; /* PL: low level cpumask for all unbound wqs */
 
 /* the per-cpu worker pools */
 static DEFINE_PER_CPU_SHARED_ALIGNED(struct worker_pool [NR_STD_WORKER_POOLS],
@@ -3492,6 +3492,7 @@ static struct pool_workqueue *numa_pwq_tbl_install(struct workqueue_struct *wq,
 struct wq_unbound_install_ctx {
 	struct workqueue_struct	*wq;	/* target to be installed */
 	struct workqueue_attrs	*attrs;	/* attrs for installing */
+	struct list_head	list;	/* queued for batching commit */
 	struct pool_workqueue	*dfl_pwq;
 	struct pool_workqueue	*pwq_tbl[];
 };
@@ -3516,10 +3517,11 @@ static void wq_unbound_install_ctx_free(struct wq_unbound_install_ctx *ctx)
 /* Allocates the attrs and pwqs for later installment */
 static struct wq_unbound_install_ctx *
 wq_unbound_install_ctx_prepare(struct workqueue_struct *wq,
-			       const struct workqueue_attrs *attrs)
+			       const struct workqueue_attrs *attrs,
+			       cpumask_var_t unbound_cpumask)
 {
 	struct wq_unbound_install_ctx *ctx;
-	struct workqueue_attrs *new_attrs, *tmp_attrs;
+	struct workqueue_attrs *new_attrs, *pwq_attrs, *tmp_attrs;
 	int node;
 
 	lockdep_assert_held(&wq_pool_mutex);
@@ -3528,13 +3530,16 @@ wq_unbound_install_ctx_prepare(struct workqueue_struct *wq,
 		      GFP_KERNEL);
 
 	new_attrs = alloc_workqueue_attrs(GFP_KERNEL);
+	pwq_attrs = alloc_workqueue_attrs(GFP_KERNEL);
 	tmp_attrs = alloc_workqueue_attrs(GFP_KERNEL);
 	if (!ctx || !new_attrs || !tmp_attrs)
 		goto out_free;
 
 	/* make a copy of @attrs and sanitize it */
 	copy_workqueue_attrs(new_attrs, attrs);
-	cpumask_and(new_attrs->cpumask, new_attrs->cpumask, wq_unbound_cpumask);
+	copy_workqueue_attrs(pwq_attrs, attrs);
+	cpumask_and(new_attrs->cpumask, new_attrs->cpumask, cpu_possible_mask);
+	cpumask_and(pwq_attrs->cpumask, pwq_attrs->cpumask, unbound_cpumask);
 
 	/*
 	 * We may create multiple pwqs with differing cpumasks.  Make a
@@ -3547,13 +3552,21 @@ wq_unbound_install_ctx_prepare(struct workqueue_struct *wq,
 	 * If something goes wrong during CPU up/down, we'll fall back to
 	 * the default pwq covering whole @attrs->cpumask.  Always create
 	 * it even if we don't use it immediately.
+	 *
+	 * If the cpumask set by the user doesn't overlap with the global
+	 * wq_unbound_cpumask, we ignore the wq_unbound_cpumask for this wq
+	 * which means all its nodes' pwqs are its default pwq and its default
+	 * pwq's workers' cpumask is totally equals to the user setting.
 	 */
-	ctx->dfl_pwq = alloc_unbound_pwq(wq, new_attrs);
+	if (cpumask_empty(pwq_attrs->cpumask))
+		ctx->dfl_pwq = alloc_unbound_pwq(wq, new_attrs);
+	else
+		ctx->dfl_pwq = alloc_unbound_pwq(wq, pwq_attrs);
 	if (!ctx->dfl_pwq)
 		goto out_free;
 
 	for_each_node(node) {
-		if (wq_calc_node_cpumask(attrs, node, -1, tmp_attrs->cpumask)) {
+		if (wq_calc_node_cpumask(pwq_attrs, node, -1, tmp_attrs->cpumask)) {
 			ctx->pwq_tbl[node] = alloc_unbound_pwq(wq, tmp_attrs);
 			if (!ctx->pwq_tbl[node])
 				goto out_free;
@@ -3567,6 +3580,7 @@ wq_unbound_install_ctx_prepare(struct workqueue_struct *wq,
 	ctx->attrs = new_attrs;
 
 out_free:
+	free_workqueue_attrs(pwq_attrs);
 	free_workqueue_attrs(tmp_attrs);
 
 	if (!ctx || !ctx->wq) {
@@ -3638,7 +3652,7 @@ int apply_workqueue_attrs(struct workqueue_struct *wq,
 	get_online_cpus();
 
 	mutex_lock(&wq_pool_mutex);
-	ctx = wq_unbound_install_ctx_prepare(wq, attrs);
+	ctx = wq_unbound_install_ctx_prepare(wq, attrs, wq_unbound_cpumask);
 	mutex_unlock(&wq_pool_mutex);
 
 	put_online_cpus();
@@ -3712,6 +3726,9 @@ static void wq_update_unbound_numa(struct workqueue_struct *wq, int cpu,
 	 * wq's, the default pwq should be used.
 	 */
 	if (wq_calc_node_cpumask(wq->unbound_attrs, node, cpu_off, cpumask)) {
+		cpumask_and(cpumask, cpumask, wq_unbound_cpumask);
+		if (cpumask_empty(cpumask))
+			goto use_dfl_pwq;
 		if (cpumask_equal(cpumask, pwq->pool->attrs->cpumask))
 			goto out_unlock;
 	} else {
@@ -3749,6 +3766,74 @@ out_unlock:
 	put_pwq_unlocked(old_pwq);
 }
 
+static int unbounds_cpumask_apply(cpumask_var_t cpumask)
+{
+	LIST_HEAD(ctxs);
+	int ret = 0;
+	struct workqueue_struct *wq;
+	struct wq_unbound_install_ctx *ctx, *n;
+
+	lockdep_assert_held(&wq_pool_mutex);
+
+	list_for_each_entry(wq, &workqueues, list) {
+		if (!(wq->flags & WQ_UNBOUND))
+			continue;
+		/* creating multiple pwqs breaks ordering guarantee */
+		if (wq->flags & __WQ_ORDERED)
+			continue;
+
+		ctx = wq_unbound_install_ctx_prepare(wq, wq->unbound_attrs,
+						     cpumask);
+		if (!ctx) {
+			ret = -ENOMEM;
+			break;
+		}
+
+		list_add_tail(&ctx->list, &ctxs);
+	}
+
+	list_for_each_entry_safe(ctx, n, &ctxs, list) {
+		if (ret >= 0)
+			wq_unbound_install_ctx_commit(ctx);
+		wq_unbound_install_ctx_free(ctx);
+	}
+
+	return ret;
+}
+
+/**
+ *  workqueue_unbounds_cpumask_set - Set the low-level unbound cpumask
+ *  @cpumask: the cpumask to set
+ *
+ *  The low-level workqueues cpumask is a global cpumask that limits
+ *  the affinity of all unbound workqueues.  This function check the @cpumask
+ *  and apply it to all unbound workqueues and updates all pwqs of them.
+ *  When all succeed, it saves @cpumask to the global low-level unbound
+ *  cpumask.
+ *
+ *  Retun:	0	- Success
+ *  		-EINVAL	- No online cpu in the @cpumask
+ *  		-ENOMEM	- Failed to allocate memory for attrs or pwqs.
+ */
+int workqueue_unbounds_cpumask_set(cpumask_var_t cpumask)
+{
+	int ret = -EINVAL;
+
+	get_online_cpus();
+	cpumask_and(cpumask, cpumask, cpu_possible_mask);
+	if (cpumask_intersects(cpumask, cpu_online_mask)) {
+		mutex_lock(&wq_pool_mutex);
+		ret = unbounds_cpumask_apply(cpumask);
+		if (ret >= 0)
+			cpumask_copy(wq_unbound_cpumask, cpumask);
+		mutex_unlock(&wq_pool_mutex);
+	}
+	put_online_cpus();
+
+	return ret;
+}
+EXPORT_SYMBOL_GPL(workqueue_unbounds_cpumask_set);
+
 #ifdef CONFIG_SYSFS
 /*
  * Workqueues with WQ_SYSFS flag set is visible to userland via
@@ -3965,19 +4050,39 @@ static struct bus_type wq_subsys = {
 	.dev_groups			= wq_sysfs_groups,
 };
 
+static ssize_t unbounds_cpumask_store(struct device *dev,
+				      struct device_attribute *attr,
+				      const char *buf, size_t count)
+{
+	cpumask_var_t cpumask;
+	int ret;
+
+	if (!zalloc_cpumask_var(&cpumask, GFP_KERNEL))
+		return -ENOMEM;
+
+	ret = cpumask_parse(buf, cpumask);
+	if (!ret)
+		ret = workqueue_unbounds_cpumask_set(cpumask);
+
+	free_cpumask_var(cpumask);
+	return ret ? ret : count;
+}
+
 static ssize_t unbounds_cpumask_show(struct device *dev,
 				     struct device_attribute *attr, char *buf)
 {
 	int written;
 
+	mutex_lock(&wq_pool_mutex);
 	written = scnprintf(buf, PAGE_SIZE, "%*pb\n",
 			    cpumask_pr_args(wq_unbound_cpumask));
+	mutex_unlock(&wq_pool_mutex);
 
 	return written;
 }
 
 static struct device_attribute wq_sysfs_cpumask_attr =
-	__ATTR(cpumask, 0444, unbounds_cpumask_show, NULL);
+	__ATTR(cpumask, 0644, unbounds_cpumask_show, unbounds_cpumask_store);
 
 static int __init wq_sysfs_init(void)
 {
-- 
2.1.0

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