Unbound workqueues are going to be NUMA-affine. Add wq_numa_tbl_len
and wq_numa_possible_cpumask[] in preparation. The former is the
highest NUMA node ID + 1 and the latter is masks of possibles CPUs for
each NUMA node.
This patch only introduces these. Future patches will make use of
them.
Signed-off-by: Tejun Heo <t...@kernel.org>
---
kernel/workqueue.c | 35 ++++++++++++++++++++++++++++++++++-
1 file changed, 34 insertions(+), 1 deletion(-)
diff --git a/kernel/workqueue.c b/kernel/workqueue.c
index 775c2f4..9b096e3 100644
--- a/kernel/workqueue.c
+++ b/kernel/workqueue.c
@@ -44,6 +44,7 @@
#include <linux/jhash.h>
#include <linux/hashtable.h>
#include <linux/rculist.h>
+#include <linux/nodemask.h>
#include "workqueue_internal.h"
@@ -256,6 +257,11 @@ struct workqueue_struct {
static struct kmem_cache *pwq_cache;
+static int wq_numa_tbl_len; /* highest possible NUMA node id + 1 */
+static cpumask_var_t *wq_numa_possible_cpumask;
+ /* possible CPUs of each node, may
+ be NULL if init failed */
+
static DEFINE_MUTEX(wq_mutex); /* protects workqueues and pools */
static DEFINE_SPINLOCK(pwq_lock); /* protects pool_workqueues */
static DEFINE_SPINLOCK(wq_mayday_lock); /* protects wq->maydays list */
@@ -4416,7 +4422,7 @@ out_unlock:
static int __init init_workqueues(void)
{
int std_nice[NR_STD_WORKER_POOLS] = { 0, HIGHPRI_NICE_LEVEL };
- int i, cpu;
+ int i, node, cpu;
/* make sure we have enough bits for OFFQ pool ID */
BUILD_BUG_ON((1LU << (BITS_PER_LONG - WORK_OFFQ_POOL_SHIFT)) <
@@ -4429,6 +4435,33 @@ static int __init init_workqueues(void)
cpu_notifier(workqueue_cpu_up_callback, CPU_PRI_WORKQUEUE_UP);
hotcpu_notifier(workqueue_cpu_down_callback, CPU_PRI_WORKQUEUE_DOWN);
+ /* determine NUMA pwq table len - highest node id + 1 */
+ for_each_node(node)
+ wq_numa_tbl_len = max(wq_numa_tbl_len, node + 1);
+
+ /*
+ * We want masks of possible CPUs of each node which isn't readily
+ * available. Build one from cpu_to_node() which should have been
+ * fully initialized by now.
+ */
+ wq_numa_possible_cpumask = kzalloc(wq_numa_tbl_len *
+ sizeof(wq_numa_possible_cpumask[0]),
+ GFP_KERNEL);
+ BUG_ON(!wq_numa_possible_cpumask);
+
+ for_each_node(node)
+ BUG_ON(!alloc_cpumask_var_node(&wq_numa_possible_cpumask[node],
+ GFP_KERNEL, node));
+ for_each_possible_cpu(cpu) {
+ node = cpu_to_node(cpu);
+ if (WARN_ON(node == NUMA_NO_NODE)) {
+ pr_err("workqueue: NUMA node mapping not available for
cpu%d, disabling NUMA support\n", cpu);
+ wq_numa_possible_cpumask = NULL;
+ break;
+ }
+ cpumask_set_cpu(cpu, wq_numa_possible_cpumask[node]);
+ }
+
/* initialize CPU pools */
for_each_possible_cpu(cpu) {
struct worker_pool *pool;
--
1.8.1.4
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
