#define DFL_THROTL_SLICE_HD (HZ / 10)
#define DFL_THROTL_SLICE_SSD (HZ / 50)
#define MAX_THROTL_SLICE (HZ)
+#define DFL_IDLE_THRESHOLD_SSD (1000L) /* 1 ms */
+#define DFL_IDLE_THRESHOLD_HD (100L * 1000) /* 100 ms */
+#define MAX_IDLE_TIME (5L * 1000 * 1000) /* 5 s */
+/* default latency target is 0, eg, guarantee IO latency by default */
+#define DFL_LATENCY_TARGET (0)
+
+#define SKIP_LATENCY (((u64)1) << BLK_STAT_RES_SHIFT)
static struct blkcg_policy blkcg_policy_throtl;
unsigned long last_check_time;
- unsigned long last_dispatch_time[2];
-
+ unsigned long latency_target; /* us */
/* When did we start a new slice */
unsigned long slice_start[2];
unsigned long slice_end[2];
+
+ unsigned long last_finish_time; /* ns / 1024 */
+ unsigned long checked_last_finish_time; /* ns / 1024 */
+ unsigned long avg_idletime; /* ns / 1024 */
+ unsigned long idletime_threshold; /* us */
+
+ unsigned int bio_cnt; /* total bios */
+ unsigned int bad_bio_cnt; /* bios exceeding latency threshold */
+ unsigned long bio_cnt_reset_time;
+};
+
+/* We measure latency for request size from <= 4k to >= 1M */
+#define LATENCY_BUCKET_SIZE 9
+
+struct latency_bucket {
+ unsigned long total_latency; /* ns / 1024 */
+ int samples;
+};
+
+struct avg_latency_bucket {
+ unsigned long latency; /* ns / 1024 */
+ bool valid;
};
struct throtl_data
unsigned int limit_index;
bool limit_valid[LIMIT_CNT];
+ unsigned long dft_idletime_threshold; /* us */
+
unsigned long low_upgrade_time;
unsigned long low_downgrade_time;
unsigned int scale;
+
+ struct latency_bucket tmp_buckets[LATENCY_BUCKET_SIZE];
+ struct avg_latency_bucket avg_buckets[LATENCY_BUCKET_SIZE];
+ struct latency_bucket __percpu *latency_buckets;
+ unsigned long last_calculate_time;
+
+ bool track_bio_latency;
};
static void throtl_pending_timer_fn(unsigned long arg);
return ret;
}
+#define request_bucket_index(sectors) \
+ clamp_t(int, order_base_2(sectors) - 3, 0, LATENCY_BUCKET_SIZE - 1)
+
/**
* throtl_log - log debug message via blktrace
* @sq: the service_queue being reported
tg->iops_conf[WRITE][LIMIT_MAX] = UINT_MAX;
/* LIMIT_LOW will have default value 0 */
+ tg->latency_target = DFL_LATENCY_TARGET;
+
return &tg->pd;
}
if (cgroup_subsys_on_dfl(io_cgrp_subsys) && blkg->parent)
sq->parent_sq = &blkg_to_tg(blkg->parent)->service_queue;
tg->td = td;
+
+ tg->idletime_threshold = td->dft_idletime_threshold;
}
/*
* Update has_rules[] after a new group is brought online.
*/
tg_update_has_rules(tg);
- tg->last_dispatch_time[READ] = jiffies;
- tg->last_dispatch_time[WRITE] = jiffies;
}
static void blk_throtl_update_limit_valid(struct throtl_data *td)
char bufs[4][21] = { "max", "max", "max", "max" };
u64 bps_dft;
unsigned int iops_dft;
+ char idle_time[26] = "";
+ char latency_time[26] = "";
if (!dname)
return 0;
if (tg->bps_conf[READ][off] == bps_dft &&
tg->bps_conf[WRITE][off] == bps_dft &&
tg->iops_conf[READ][off] == iops_dft &&
- tg->iops_conf[WRITE][off] == iops_dft)
+ tg->iops_conf[WRITE][off] == iops_dft &&
+ (off != LIMIT_LOW ||
+ (tg->idletime_threshold == tg->td->dft_idletime_threshold &&
+ tg->latency_target == DFL_LATENCY_TARGET)))
return 0;
if (tg->bps_conf[READ][off] != bps_dft)
if (tg->iops_conf[WRITE][off] != iops_dft)
snprintf(bufs[3], sizeof(bufs[3]), "%u",
tg->iops_conf[WRITE][off]);
+ if (off == LIMIT_LOW) {
+ if (tg->idletime_threshold == ULONG_MAX)
+ strcpy(idle_time, " idle=max");
+ else
+ snprintf(idle_time, sizeof(idle_time), " idle=%lu",
+ tg->idletime_threshold);
+
+ if (tg->latency_target == ULONG_MAX)
+ strcpy(latency_time, " latency=max");
+ else
+ snprintf(latency_time, sizeof(latency_time),
+ " latency=%lu", tg->latency_target);
+ }
- seq_printf(sf, "%s rbps=%s wbps=%s riops=%s wiops=%s\n",
- dname, bufs[0], bufs[1], bufs[2], bufs[3]);
+ seq_printf(sf, "%s rbps=%s wbps=%s riops=%s wiops=%s%s%s\n",
+ dname, bufs[0], bufs[1], bufs[2], bufs[3], idle_time,
+ latency_time);
return 0;
}
struct blkg_conf_ctx ctx;
struct throtl_grp *tg;
u64 v[4];
+ unsigned long idle_time;
+ unsigned long latency_time;
int ret;
int index = of_cft(of)->private;
v[2] = tg->iops_conf[READ][index];
v[3] = tg->iops_conf[WRITE][index];
+ idle_time = tg->idletime_threshold;
+ latency_time = tg->latency_target;
while (true) {
char tok[27]; /* wiops=18446744073709551616 */
char *p;
v[2] = min_t(u64, val, UINT_MAX);
else if (!strcmp(tok, "wiops"))
v[3] = min_t(u64, val, UINT_MAX);
+ else if (off == LIMIT_LOW && !strcmp(tok, "idle"))
+ idle_time = val;
+ else if (off == LIMIT_LOW && !strcmp(tok, "latency"))
+ latency_time = val;
else
goto out_finish;
}
blk_throtl_update_limit_valid(tg->td);
if (tg->td->limit_valid[LIMIT_LOW])
tg->td->limit_index = LIMIT_LOW;
+ tg->idletime_threshold = (idle_time == ULONG_MAX) ?
+ ULONG_MAX : idle_time;
+ tg->latency_target = (latency_time == ULONG_MAX) ?
+ ULONG_MAX : latency_time;
}
tg_conf_updated(tg);
ret = 0;
return ret;
}
+static bool throtl_tg_is_idle(struct throtl_grp *tg)
+{
+ /*
+ * cgroup is idle if:
+ * - single idle is too long, longer than a fixed value (in case user
+ * configure a too big threshold) or 4 times of slice
+ * - average think time is more than threshold
+ * - IO latency is largely below threshold
+ */
+ unsigned long time = jiffies_to_usecs(4 * tg->td->throtl_slice);
+
+ time = min_t(unsigned long, MAX_IDLE_TIME, time);
+ return (ktime_get_ns() >> 10) - tg->last_finish_time > time ||
+ tg->avg_idletime > tg->idletime_threshold ||
+ (tg->latency_target && tg->bio_cnt &&
+ tg->bad_bio_cnt * 5 < tg->bio_cnt);
+}
+
static bool throtl_tg_can_upgrade(struct throtl_grp *tg)
{
struct throtl_service_queue *sq = &tg->service_queue;
return true;
if (time_after_eq(jiffies,
- tg->last_dispatch_time[READ] + tg->td->throtl_slice) &&
- time_after_eq(jiffies,
- tg->last_dispatch_time[WRITE] + tg->td->throtl_slice))
+ tg_last_low_overflow_time(tg) + tg->td->throtl_slice) &&
+ throtl_tg_is_idle(tg))
return true;
return false;
}
return true;
}
+static void throtl_upgrade_check(struct throtl_grp *tg)
+{
+ unsigned long now = jiffies;
+
+ if (tg->td->limit_index != LIMIT_LOW)
+ return;
+
+ if (time_after(tg->last_check_time + tg->td->throtl_slice, now))
+ return;
+
+ tg->last_check_time = now;
+
+ if (!time_after_eq(now,
+ __tg_last_low_overflow_time(tg) + tg->td->throtl_slice))
+ return;
+
+ if (throtl_can_upgrade(tg->td, NULL))
+ throtl_upgrade_state(tg->td);
+}
+
static void throtl_upgrade_state(struct throtl_data *td)
{
struct cgroup_subsys_state *pos_css;
struct throtl_data *td = tg->td;
unsigned long now = jiffies;
- if (time_after_eq(now, tg->last_dispatch_time[READ] +
- td->throtl_slice) &&
- time_after_eq(now, tg->last_dispatch_time[WRITE] +
- td->throtl_slice))
- return false;
/*
* If cgroup is below low limit, consider downgrade and throttle other
* cgroups
*/
if (time_after_eq(now, td->low_upgrade_time + td->throtl_slice) &&
time_after_eq(now, tg_last_low_overflow_time(tg) +
- td->throtl_slice))
+ td->throtl_slice) &&
+ (!throtl_tg_is_idle(tg) ||
+ !list_empty(&tg_to_blkg(tg)->blkcg->css.children)))
return true;
return false;
}
tg->last_io_disp[WRITE] = 0;
}
+static void blk_throtl_update_idletime(struct throtl_grp *tg)
+{
+ unsigned long now = ktime_get_ns() >> 10;
+ unsigned long last_finish_time = tg->last_finish_time;
+
+ if (now <= last_finish_time || last_finish_time == 0 ||
+ last_finish_time == tg->checked_last_finish_time)
+ return;
+
+ tg->avg_idletime = (tg->avg_idletime * 7 + now - last_finish_time) >> 3;
+ tg->checked_last_finish_time = last_finish_time;
+}
+
+#ifdef CONFIG_BLK_DEV_THROTTLING_LOW
+static void throtl_update_latency_buckets(struct throtl_data *td)
+{
+ struct avg_latency_bucket avg_latency[LATENCY_BUCKET_SIZE];
+ int i, cpu;
+ unsigned long last_latency = 0;
+ unsigned long latency;
+
+ if (!blk_queue_nonrot(td->queue))
+ return;
+ if (time_before(jiffies, td->last_calculate_time + HZ))
+ return;
+ td->last_calculate_time = jiffies;
+
+ memset(avg_latency, 0, sizeof(avg_latency));
+ for (i = 0; i < LATENCY_BUCKET_SIZE; i++) {
+ struct latency_bucket *tmp = &td->tmp_buckets[i];
+
+ for_each_possible_cpu(cpu) {
+ struct latency_bucket *bucket;
+
+ /* this isn't race free, but ok in practice */
+ bucket = per_cpu_ptr(td->latency_buckets, cpu);
+ tmp->total_latency += bucket[i].total_latency;
+ tmp->samples += bucket[i].samples;
+ bucket[i].total_latency = 0;
+ bucket[i].samples = 0;
+ }
+
+ if (tmp->samples >= 32) {
+ int samples = tmp->samples;
+
+ latency = tmp->total_latency;
+
+ tmp->total_latency = 0;
+ tmp->samples = 0;
+ latency /= samples;
+ if (latency == 0)
+ continue;
+ avg_latency[i].latency = latency;
+ }
+ }
+
+ for (i = 0; i < LATENCY_BUCKET_SIZE; i++) {
+ if (!avg_latency[i].latency) {
+ if (td->avg_buckets[i].latency < last_latency)
+ td->avg_buckets[i].latency = last_latency;
+ continue;
+ }
+
+ if (!td->avg_buckets[i].valid)
+ latency = avg_latency[i].latency;
+ else
+ latency = (td->avg_buckets[i].latency * 7 +
+ avg_latency[i].latency) >> 3;
+
+ td->avg_buckets[i].latency = max(latency, last_latency);
+ td->avg_buckets[i].valid = true;
+ last_latency = td->avg_buckets[i].latency;
+ }
+}
+#else
+static inline void throtl_update_latency_buckets(struct throtl_data *td)
+{
+}
+#endif
+
+static void blk_throtl_assoc_bio(struct throtl_grp *tg, struct bio *bio)
+{
+#ifdef CONFIG_BLK_DEV_THROTTLING_LOW
+ int ret;
+
+ ret = bio_associate_current(bio);
+ if (ret == 0 || ret == -EBUSY)
+ bio->bi_cg_private = tg;
+ blk_stat_set_issue(&bio->bi_issue_stat, bio_sectors(bio));
+#else
+ bio_associate_current(bio);
+#endif
+}
+
bool blk_throtl_bio(struct request_queue *q, struct blkcg_gq *blkg,
struct bio *bio)
{
struct throtl_service_queue *sq;
bool rw = bio_data_dir(bio);
bool throttled = false;
+ struct throtl_data *td = tg->td;
WARN_ON_ONCE(!rcu_read_lock_held());
spin_lock_irq(q->queue_lock);
+ throtl_update_latency_buckets(td);
+
if (unlikely(blk_queue_bypass(q)))
goto out_unlock;
+ blk_throtl_assoc_bio(tg, bio);
+ blk_throtl_update_idletime(tg);
+
sq = &tg->service_queue;
again:
while (true) {
- tg->last_dispatch_time[rw] = jiffies;
if (tg->last_low_overflow_time[rw] == 0)
tg->last_low_overflow_time[rw] = jiffies;
throtl_downgrade_check(tg);
+ throtl_upgrade_check(tg);
/* throtl is FIFO - if bios are already queued, should queue */
if (sq->nr_queued[rw])
break;
/* if above limits, break to queue */
if (!tg_may_dispatch(tg, bio, NULL)) {
tg->last_low_overflow_time[rw] = jiffies;
- if (throtl_can_upgrade(tg->td, tg)) {
- throtl_upgrade_state(tg->td);
+ if (throtl_can_upgrade(td, tg)) {
+ throtl_upgrade_state(td);
goto again;
}
break;
tg->last_low_overflow_time[rw] = jiffies;
- bio_associate_current(bio);
- tg->td->nr_queued[rw]++;
+ td->nr_queued[rw]++;
throtl_add_bio_tg(bio, qn, tg);
throttled = true;
*/
if (!throttled)
bio_clear_flag(bio, BIO_THROTTLED);
+
+#ifdef CONFIG_BLK_DEV_THROTTLING_LOW
+ if (throttled || !td->track_bio_latency)
+ bio->bi_issue_stat.stat |= SKIP_LATENCY;
+#endif
return throttled;
}
+#ifdef CONFIG_BLK_DEV_THROTTLING_LOW
+static void throtl_track_latency(struct throtl_data *td, sector_t size,
+ int op, unsigned long time)
+{
+ struct latency_bucket *latency;
+ int index;
+
+ if (!td || td->limit_index != LIMIT_LOW || op != REQ_OP_READ ||
+ !blk_queue_nonrot(td->queue))
+ return;
+
+ index = request_bucket_index(size);
+
+ latency = get_cpu_ptr(td->latency_buckets);
+ latency[index].total_latency += time;
+ latency[index].samples++;
+ put_cpu_ptr(td->latency_buckets);
+}
+
+void blk_throtl_stat_add(struct request *rq, u64 time_ns)
+{
+ struct request_queue *q = rq->q;
+ struct throtl_data *td = q->td;
+
+ throtl_track_latency(td, blk_stat_size(&rq->issue_stat),
+ req_op(rq), time_ns >> 10);
+}
+
+void blk_throtl_bio_endio(struct bio *bio)
+{
+ struct throtl_grp *tg;
+ u64 finish_time_ns;
+ unsigned long finish_time;
+ unsigned long start_time;
+ unsigned long lat;
+
+ tg = bio->bi_cg_private;
+ if (!tg)
+ return;
+ bio->bi_cg_private = NULL;
+
+ finish_time_ns = ktime_get_ns();
+ tg->last_finish_time = finish_time_ns >> 10;
+
+ start_time = blk_stat_time(&bio->bi_issue_stat) >> 10;
+ finish_time = __blk_stat_time(finish_time_ns) >> 10;
+ if (!start_time || finish_time <= start_time)
+ return;
+
+ lat = finish_time - start_time;
+ /* this is only for bio based driver */
+ if (!(bio->bi_issue_stat.stat & SKIP_LATENCY))
+ throtl_track_latency(tg->td, blk_stat_size(&bio->bi_issue_stat),
+ bio_op(bio), lat);
+
+ if (tg->latency_target) {
+ int bucket;
+ unsigned int threshold;
+
+ bucket = request_bucket_index(
+ blk_stat_size(&bio->bi_issue_stat));
+ threshold = tg->td->avg_buckets[bucket].latency +
+ tg->latency_target;
+ if (lat > threshold)
+ tg->bad_bio_cnt++;
+ /*
+ * Not race free, could get wrong count, which means cgroups
+ * will be throttled
+ */
+ tg->bio_cnt++;
+ }
+
+ if (time_after(jiffies, tg->bio_cnt_reset_time) || tg->bio_cnt > 1024) {
+ tg->bio_cnt_reset_time = tg->td->throtl_slice + jiffies;
+ tg->bio_cnt /= 2;
+ tg->bad_bio_cnt /= 2;
+ }
+}
+#endif
+
/*
* Dispatch all bios from all children tg's queued on @parent_sq. On
* return, @parent_sq is guaranteed to not have any active children tg's
td = kzalloc_node(sizeof(*td), GFP_KERNEL, q->node);
if (!td)
return -ENOMEM;
+ td->latency_buckets = __alloc_percpu(sizeof(struct latency_bucket) *
+ LATENCY_BUCKET_SIZE, __alignof__(u64));
+ if (!td->latency_buckets) {
+ kfree(td);
+ return -ENOMEM;
+ }
INIT_WORK(&td->dispatch_work, blk_throtl_dispatch_work_fn);
throtl_service_queue_init(&td->service_queue);
td->limit_index = LIMIT_MAX;
td->low_upgrade_time = jiffies;
td->low_downgrade_time = jiffies;
+
/* activate policy */
ret = blkcg_activate_policy(q, &blkcg_policy_throtl);
- if (ret)
+ if (ret) {
+ free_percpu(td->latency_buckets);
kfree(td);
+ }
return ret;
}
BUG_ON(!q->td);
throtl_shutdown_wq(q);
blkcg_deactivate_policy(q, &blkcg_policy_throtl);
+ free_percpu(q->td->latency_buckets);
kfree(q->td);
}
void blk_throtl_register_queue(struct request_queue *q)
{
struct throtl_data *td;
+ struct cgroup_subsys_state *pos_css;
+ struct blkcg_gq *blkg;
td = q->td;
BUG_ON(!td);
- if (blk_queue_nonrot(q))
+ if (blk_queue_nonrot(q)) {
td->throtl_slice = DFL_THROTL_SLICE_SSD;
- else
+ td->dft_idletime_threshold = DFL_IDLE_THRESHOLD_SSD;
+ } else {
td->throtl_slice = DFL_THROTL_SLICE_HD;
+ td->dft_idletime_threshold = DFL_IDLE_THRESHOLD_HD;
+ }
#ifndef CONFIG_BLK_DEV_THROTTLING_LOW
/* if no low limit, use previous default */
td->throtl_slice = DFL_THROTL_SLICE_HD;
#endif
+
+ td->track_bio_latency = !q->mq_ops && !q->request_fn;
+ if (!td->track_bio_latency)
+ blk_stat_enable_accounting(q);
+
+ /*
+ * some tg are created before queue is fully initialized, eg, nonrot
+ * isn't initialized yet
+ */
+ rcu_read_lock();
+ blkg_for_each_descendant_post(blkg, pos_css, q->root_blkg) {
+ struct throtl_grp *tg = blkg_to_tg(blkg);
+
+ tg->idletime_threshold = td->dft_idletime_threshold;
+ }
+ rcu_read_unlock();
}
#ifdef CONFIG_BLK_DEV_THROTTLING_LOW