block: add support for IO CPU affinity
This patch adds support for controlling the IO completion CPU of
either all requests on a queue, or on a per-request basis. We export
a sysfs variable (rq_affinity) which, if set, migrates completions
of requests to the CPU that originally submitted it. A bio helper
(bio_set_completion_cpu()) is also added, so that queuers can ask
for completion on that specific CPU.
In testing, this has been show to cut the system time by as much
as 20-40% on synthetic workloads where CPU affinity is desired.
This requires a little help from the architecture, so it'll only
work as designed for archs that are using the new generic smp
helper infrastructure.
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
diff --git a/block/blk-softirq.c b/block/blk-softirq.c
index 9e1c43b..3a1af55 100644
--- a/block/blk-softirq.c
+++ b/block/blk-softirq.c
@@ -13,6 +13,70 @@
static DEFINE_PER_CPU(struct list_head, blk_cpu_done);
+/*
+ * Softirq action handler - move entries to local list and loop over them
+ * while passing them to the queue registered handler.
+ */
+static void blk_done_softirq(struct softirq_action *h)
+{
+ struct list_head *cpu_list, local_list;
+
+ local_irq_disable();
+ cpu_list = &__get_cpu_var(blk_cpu_done);
+ list_replace_init(cpu_list, &local_list);
+ local_irq_enable();
+
+ while (!list_empty(&local_list)) {
+ struct request *rq;
+
+ rq = list_entry(local_list.next, struct request, csd.list);
+ list_del_init(&rq->csd.list);
+ rq->q->softirq_done_fn(rq);
+ }
+}
+
+#if defined(CONFIG_SMP) && defined(CONFIG_USE_GENERIC_SMP_HELPERS)
+static void trigger_softirq(void *data)
+{
+ struct request *rq = data;
+ unsigned long flags;
+ struct list_head *list;
+
+ local_irq_save(flags);
+ list = &__get_cpu_var(blk_cpu_done);
+ list_add_tail(&rq->csd.list, list);
+
+ if (list->next == &rq->csd.list)
+ raise_softirq_irqoff(BLOCK_SOFTIRQ);
+
+ local_irq_restore(flags);
+}
+
+/*
+ * Setup and invoke a run of 'trigger_softirq' on the given cpu.
+ */
+static int raise_blk_irq(int cpu, struct request *rq)
+{
+ if (cpu_online(cpu)) {
+ struct call_single_data *data = &rq->csd;
+
+ data->func = trigger_softirq;
+ data->info = rq;
+ data->flags = 0;
+
+ __smp_call_function_single(cpu, data);
+ return 0;
+ }
+
+ return 1;
+}
+#else /* CONFIG_SMP && CONFIG_USE_GENERIC_SMP_HELPERS */
+static int raise_blk_irq(int cpu, struct request *rq)
+{
+ return 1;
+}
+#endif
+
static int __cpuinit blk_cpu_notify(struct notifier_block *self,
unsigned long action, void *hcpu)
{
@@ -33,33 +97,10 @@
return NOTIFY_OK;
}
-
-static struct notifier_block blk_cpu_notifier __cpuinitdata = {
+static struct notifier_block __cpuinitdata blk_cpu_notifier = {
.notifier_call = blk_cpu_notify,
};
-/*
- * splice the completion data to a local structure and hand off to
- * process_completion_queue() to complete the requests
- */
-static void blk_done_softirq(struct softirq_action *h)
-{
- struct list_head *cpu_list, local_list;
-
- local_irq_disable();
- cpu_list = &__get_cpu_var(blk_cpu_done);
- list_replace_init(cpu_list, &local_list);
- local_irq_enable();
-
- while (!list_empty(&local_list)) {
- struct request *rq;
-
- rq = list_entry(local_list.next, struct request, donelist);
- list_del_init(&rq->donelist);
- rq->q->softirq_done_fn(rq);
- }
-}
-
/**
* blk_complete_request - end I/O on a request
* @req: the request being processed
@@ -71,25 +112,48 @@
* through a softirq handler. The user must have registered a completion
* callback through blk_queue_softirq_done().
**/
-
void blk_complete_request(struct request *req)
{
- struct list_head *cpu_list;
+ struct request_queue *q = req->q;
unsigned long flags;
+ int ccpu, cpu, group_cpu;
- BUG_ON(!req->q->softirq_done_fn);
+ BUG_ON(!q->softirq_done_fn);
local_irq_save(flags);
+ cpu = smp_processor_id();
+ group_cpu = blk_cpu_to_group(cpu);
- cpu_list = &__get_cpu_var(blk_cpu_done);
- list_add_tail(&req->donelist, cpu_list);
- raise_softirq_irqoff(BLOCK_SOFTIRQ);
+ /*
+ * Select completion CPU
+ */
+ if (test_bit(QUEUE_FLAG_SAME_COMP, &q->queue_flags) && req->cpu != -1)
+ ccpu = req->cpu;
+ else
+ ccpu = cpu;
+
+ if (ccpu == cpu || ccpu == group_cpu) {
+ struct list_head *list;
+do_local:
+ list = &__get_cpu_var(blk_cpu_done);
+ list_add_tail(&req->csd.list, list);
+
+ /*
+ * if the list only contains our just added request,
+ * signal a raise of the softirq. If there are already
+ * entries there, someone already raised the irq but it
+ * hasn't run yet.
+ */
+ if (list->next == &req->csd.list)
+ raise_softirq_irqoff(BLOCK_SOFTIRQ);
+ } else if (raise_blk_irq(ccpu, req))
+ goto do_local;
local_irq_restore(flags);
}
EXPORT_SYMBOL(blk_complete_request);
-int __init blk_softirq_init(void)
+__init int blk_softirq_init(void)
{
int i;