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

Groups > linux.kernel > #1538168 > unrolled thread

[PATCHSET/RFC] blk-mq scheduling framework

Back to article view | Back to linux.kernel

Contents

  [PATCHSET/RFC] blk-mq scheduling framework Jens Axboe <axboe@fb.com> - 2016-12-08 00:20 +0100
    [PATCH 5/7] blk-mq-sched: add framework for MQ capable IO schedulers Jens Axboe <axboe@fb.com> - 2016-12-08 00:20 +0100

#1538168 — [PATCHSET/RFC] blk-mq scheduling framework

I hacked up some support for registering blk-mq capable IO schedulers,
and when that was done, I adapted deadline to work with it as a new
mq-deadline scheduler.

Basically this is similar to the legacy scheduling patches I posted
recently, in that we setup a list of fake requests (I called them
shadows) that the IO scheduler works on. We then transform those
into a real blk-mq request, when we have to dispatch to hardware.

There's a hack in place to make this work with the flush/fua
requests, as those bypass the regular software queue insertion.
For now we simply ensure that we allocate a real request for
those. Howeer, I'd prefer if we simply inserted the request as
we usually would, and then start the flush state machinery when
we pull the request out of the queue. That would both be cleaner
from a flush perspective, and from the scheduling side as well.

I'm reusing the existing elevator interface, just augmenting
that with mq_ops and a ->uses_mq so we can tell which is which.
They show up automatically, for instance on a scsi-mq device:

$ cat /sys/block/sda/queue/scheduler 
[mq-deadline] none

vs just a legacy device:

$ cat /sys/block/nullb0/queue/scheduler 
noop deadline [cfq]

Changing schedulers is done in the same way as it always has,
by echoing into the 'scheduler' file. For MQ, there's a 'none'
setting as well that isn't a real scheduler, it simply turns off
the scheduler. Handy for comparison.

Obviously a direct deadline adaptation has performance implications,
so it can be improved. A _real_ MQ scheduler is forth coming, which
will sit on top of this interface. Paolo, for BFQ, this is the
interface you should target. Let me know if you have any questions
about how it works.


 block/Kconfig.iosched    |    6 
 block/Makefile           |    3 
 block/blk-core.c         |    3 
 block/blk-exec.c         |    3 
 block/blk-flush.c        |    7 
 block/blk-mq-sched.c     |  243 ++++++++++++++++++
 block/blk-mq-sched.h     |  168 ++++++++++++
 block/blk-mq-tag.c       |    1 
 block/blk-mq.c           |  241 ++++++++++--------
 block/blk-mq.h           |   33 --
 block/elevator.c         |  140 +++++++---
 block/mq-deadline.c      |  622 +++++++++++++++++++++++++++++++++++++++++++++++
 include/linux/blk-mq.h   |    3 
 include/linux/elevator.h |   30 ++
 14 files changed, 1331 insertions(+), 172 deletions(-)


-- 
Jens Axboe

[toc] | [next] | [standalone]

#1538169 — [PATCH 5/7] blk-mq-sched: add framework for MQ capable IO schedulers

Signed-off-by: Jens Axboe <axboe@fb.com>
---
 block/blk-mq-sched.c | 243 +++++++++++++++++++++++++++++++++++++++++++++++++++
 block/blk-mq-sched.h | 168 +++++++++++++++++++++++++++++++++++
 2 files changed, 411 insertions(+)
 create mode 100644 block/blk-mq-sched.c
 create mode 100644 block/blk-mq-sched.h

diff --git a/block/blk-mq-sched.c b/block/blk-mq-sched.c
new file mode 100644
index 000000000000..8317b26990f8
--- /dev/null
+++ b/block/blk-mq-sched.c
@@ -0,0 +1,243 @@
+#include <linux/kernel.h>
+#include <linux/module.h>
+
+#include <linux/blk-mq.h>
+#include "blk.h"
+#include "blk-mq.h"
+#include "blk-mq-sched.h"
+#include "blk-mq-tag.h"
+#include "blk-wbt.h"
+
+/*
+ * Empty set
+ */
+static struct blk_mq_ops mq_sched_tag_ops = {
+	.queue_rq	= NULL,
+};
+
+void blk_mq_sched_free_requests(struct blk_mq_tags *tags)
+{
+	blk_mq_free_rq_map(NULL, tags, 0);
+}
+EXPORT_SYMBOL_GPL(blk_mq_sched_free_requests);
+
+struct blk_mq_tags *blk_mq_sched_alloc_requests(unsigned int depth,
+						unsigned int numa_node)
+{
+	struct blk_mq_tag_set set = {
+		.ops		= &mq_sched_tag_ops,
+		.nr_hw_queues	= 1,
+		.queue_depth	= depth,
+		.numa_node	= numa_node,
+	};
+
+	return blk_mq_init_rq_map(&set, 0);
+}
+EXPORT_SYMBOL_GPL(blk_mq_sched_alloc_requests);
+
+void blk_mq_sched_free_hctx_data(struct request_queue *q,
+				 void (*exit)(struct blk_mq_hw_ctx *))
+{
+	struct blk_mq_hw_ctx *hctx;
+	int i;
+
+	queue_for_each_hw_ctx(q, hctx, i) {
+		if (exit)
+			exit(hctx);
+		kfree(hctx->sched_data);
+		hctx->sched_data = NULL;
+	}
+}
+EXPORT_SYMBOL_GPL(blk_mq_sched_free_hctx_data);
+
+int blk_mq_sched_init_hctx_data(struct request_queue *q, size_t size,
+				void (*init)(struct blk_mq_hw_ctx *))
+{
+	struct blk_mq_hw_ctx *hctx;
+	int i;
+
+	queue_for_each_hw_ctx(q, hctx, i) {
+		hctx->sched_data = kmalloc_node(size, GFP_KERNEL, hctx->numa_node);
+		if (!hctx->sched_data)
+			goto error;
+
+		if (init)
+			init(hctx);
+	}
+
+	return 0;
+error:
+	blk_mq_sched_free_hctx_data(q, NULL);
+	return -ENOMEM;
+}
+EXPORT_SYMBOL_GPL(blk_mq_sched_init_hctx_data);
+
+struct request *blk_mq_sched_alloc_shadow_request(struct request_queue *q,
+						  struct blk_mq_alloc_data *data,
+						  struct blk_mq_tags *tags,
+						  atomic_t *wait_index)
+{
+	struct sbq_wait_state *ws;
+	DEFINE_WAIT(wait);
+	struct request *rq;
+	int tag;
+
+	tag = __sbitmap_queue_get(&tags->bitmap_tags);
+	if (tag != -1)
+		goto done;
+
+	if (data->flags & BLK_MQ_REQ_NOWAIT)
+		return NULL;
+
+	ws = sbq_wait_ptr(&tags->bitmap_tags, wait_index);
+	do {
+		prepare_to_wait(&ws->wait, &wait, TASK_UNINTERRUPTIBLE);
+
+		tag = __sbitmap_queue_get(&tags->bitmap_tags);
+		if (tag != -1)
+			break;
+
+		blk_mq_run_hw_queue(data->hctx, false);
+
+		tag = __sbitmap_queue_get(&tags->bitmap_tags);
+		if (tag != -1)
+			break;
+
+		blk_mq_put_ctx(data->ctx);
+		io_schedule();
+
+		data->ctx = blk_mq_get_ctx(data->q);
+		data->hctx = blk_mq_map_queue(data->q, data->ctx->cpu);
+		finish_wait(&ws->wait, &wait);
+		ws = sbq_wait_ptr(&tags->bitmap_tags, wait_index);
+	} while (1);
+
+	finish_wait(&ws->wait, &wait);
+done:
+	rq = tags->rqs[tag];
+	rq->tag = tag;
+	return rq;
+}
+EXPORT_SYMBOL_GPL(blk_mq_sched_alloc_shadow_request);
+
+void blk_mq_sched_free_shadow_request(struct blk_mq_tags *tags,
+				      struct request *rq)
+{
+	sbitmap_queue_clear(&tags->bitmap_tags, rq->tag, rq->mq_ctx->cpu);
+}
+EXPORT_SYMBOL_GPL(blk_mq_sched_free_shadow_request);
+
+static void rq_copy(struct request *rq, struct request *src)
+{
+#define FIELD_COPY(dst, src, name)	((dst)->name = (src)->name)
+	FIELD_COPY(rq, src, cpu);
+	FIELD_COPY(rq, src, cmd_type);
+	FIELD_COPY(rq, src, cmd_flags);
+	rq->rq_flags |= (src->rq_flags & (RQF_PREEMPT | RQF_QUIET | RQF_PM | RQF_DONTPREP));
+	rq->rq_flags &= ~RQF_IO_STAT;
+	FIELD_COPY(rq, src, __data_len);
+	FIELD_COPY(rq, src, __sector);
+	FIELD_COPY(rq, src, bio);
+	FIELD_COPY(rq, src, biotail);
+	FIELD_COPY(rq, src, rq_disk);
+	FIELD_COPY(rq, src, part);
+	FIELD_COPY(rq, src, nr_phys_segments);
+#if defined(CONFIG_BLK_DEV_INTEGRITY)
+	FIELD_COPY(rq, src, nr_integrity_segments);
+#endif
+	FIELD_COPY(rq, src, ioprio);
+	FIELD_COPY(rq, src, timeout);
+
+	if (src->cmd_type == REQ_TYPE_BLOCK_PC) {
+		FIELD_COPY(rq, src, cmd);
+		FIELD_COPY(rq, src, cmd_len);
+		FIELD_COPY(rq, src, extra_len);
+		FIELD_COPY(rq, src, sense_len);
+		FIELD_COPY(rq, src, resid_len);
+		FIELD_COPY(rq, src, sense);
+		FIELD_COPY(rq, src, retries);
+	}
+
+	src->bio = src->biotail = NULL;
+}
+
+static void sched_rq_end_io(struct request *rq, int error)
+{
+	struct request *sched_rq = rq->end_io_data;
+
+	FIELD_COPY(sched_rq, rq, resid_len);
+	FIELD_COPY(sched_rq, rq, extra_len);
+	FIELD_COPY(sched_rq, rq, sense_len);
+	FIELD_COPY(sched_rq, rq, errors);
+	FIELD_COPY(sched_rq, rq, retries);
+
+	blk_account_io_completion(sched_rq, blk_rq_bytes(sched_rq));
+	blk_account_io_done(sched_rq);
+
+	wbt_done(sched_rq->q->rq_wb, &sched_rq->issue_stat);
+
+	if (sched_rq->end_io)
+		sched_rq->end_io(sched_rq, error);
+
+	blk_mq_free_request(rq);
+}
+
+struct request *
+blk_mq_sched_request_from_shadow(struct blk_mq_hw_ctx *hctx,
+				 struct request *(*get_sched_rq)(struct blk_mq_hw_ctx *))
+{
+	struct blk_mq_alloc_data data;
+	struct request *sched_rq, *rq;
+
+	data.q = hctx->queue;
+	data.flags = BLK_MQ_REQ_NOWAIT;
+	data.ctx = blk_mq_get_ctx(hctx->queue);
+	data.hctx = hctx;
+
+	rq = __blk_mq_alloc_request(&data, 0);
+	blk_mq_put_ctx(data.ctx);
+
+	if (!rq) {
+		blk_mq_stop_hw_queue(hctx);
+		return NULL;
+	}
+
+	sched_rq = get_sched_rq(hctx);
+
+	if (!sched_rq) {
+		blk_queue_enter_live(hctx->queue);
+		__blk_mq_free_request(hctx, data.ctx, rq);
+		return NULL;
+	}
+
+	rq_copy(rq, sched_rq);
+	rq->end_io = sched_rq_end_io;
+	rq->end_io_data = sched_rq;
+
+	return rq;
+}
+EXPORT_SYMBOL_GPL(blk_mq_sched_request_from_shadow);
+
+void blk_mq_sched_dispatch_requests(struct blk_mq_hw_ctx *hctx)
+{
+	struct elevator_queue *e = hctx->queue->elevator;
+	struct request *rq;
+	LIST_HEAD(rq_list);
+
+	if (unlikely(blk_mq_hctx_stopped(hctx)))
+		return;
+
+	hctx->run++;
+
+	if (!list_empty(&hctx->dispatch)) {
+		spin_lock(&hctx->lock);
+		if (!list_empty(&hctx->dispatch))
+			list_splice_init(&hctx->dispatch, &rq_list);
+		spin_unlock(&hctx->lock);
+	}
+
+	while ((rq = e->type->mq_ops.dispatch_request(hctx)) != NULL)
+		list_add_tail(&rq->queuelist, &rq_list);
+
+	blk_mq_dispatch_rq_list(hctx, &rq_list);
+}
diff --git a/block/blk-mq-sched.h b/block/blk-mq-sched.h
new file mode 100644
index 000000000000..125e14e5274a
--- /dev/null
+++ b/block/blk-mq-sched.h
@@ -0,0 +1,168 @@
+#ifndef BLK_MQ_SCHED_H
+#define BLK_MQ_SCHED_H
+
+#include "blk-mq.h"
+
+struct blk_mq_hw_ctx;
+struct blk_mq_ctx;
+struct request_queue;
+
+struct blk_mq_tags *blk_mq_sched_alloc_requests(unsigned int depth, unsigned int numa_node);
+void blk_mq_sched_free_requests(struct blk_mq_tags *tags);
+
+int blk_mq_sched_init_hctx_data(struct request_queue *q, size_t size,
+				void (*init)(struct blk_mq_hw_ctx *));
+void blk_mq_sched_free_hctx_data(struct request_queue *q,
+				 void (*exit)(struct blk_mq_hw_ctx *));
+
+void blk_mq_sched_free_shadow_request(struct blk_mq_tags *tags,
+				      struct request *rq);
+struct request *blk_mq_sched_alloc_shadow_request(struct request_queue *q,
+						  struct blk_mq_alloc_data *data,
+						  struct blk_mq_tags *tags,
+						  atomic_t *wait_index);
+struct request *
+blk_mq_sched_request_from_shadow(struct blk_mq_hw_ctx *hctx,
+				 struct request *(*get_sched_rq)(struct blk_mq_hw_ctx *));
+
+
+struct blk_mq_alloc_data {
+	/* input parameter */
+	struct request_queue *q;
+	unsigned int flags;
+
+	/* input & output parameter */
+	struct blk_mq_ctx *ctx;
+	struct blk_mq_hw_ctx *hctx;
+};
+
+static inline void blk_mq_set_alloc_data(struct blk_mq_alloc_data *data,
+		struct request_queue *q, unsigned int flags,
+		struct blk_mq_ctx *ctx, struct blk_mq_hw_ctx *hctx)
+{
+	data->q = q;
+	data->flags = flags;
+	data->ctx = ctx;
+	data->hctx = hctx;
+}
+
+void blk_mq_sched_dispatch_requests(struct blk_mq_hw_ctx *hctx);
+
+static inline bool
+blk_mq_sched_bio_merge(struct request_queue *q, struct bio *bio)
+{
+	struct elevator_queue *e = q->elevator;
+
+	if (blk_queue_nomerges(q) || !bio_mergeable(bio))
+		return false;
+
+	if (e) {
+		struct blk_mq_ctx *ctx = blk_mq_get_ctx(q);
+		struct blk_mq_hw_ctx *hctx = blk_mq_map_queue(q, ctx->cpu);
+
+		blk_mq_put_ctx(ctx);
+		return e->type->mq_ops.bio_merge(hctx, bio);
+	}
+
+	return false;
+}
+
+static inline struct request *
+blk_mq_sched_get_request(struct request_queue *q, struct bio *bio,
+			 struct blk_mq_alloc_data *data)
+{
+	struct elevator_queue *e = q->elevator;
+	struct blk_mq_hw_ctx *hctx;
+	struct blk_mq_ctx *ctx;
+	struct request *rq;
+
+	blk_queue_enter_live(q);
+	ctx = blk_mq_get_ctx(q);
+	hctx = blk_mq_map_queue(q, ctx->cpu);
+
+	blk_mq_set_alloc_data(data, q, 0, ctx, hctx);
+
+	if (e)
+		rq = e->type->mq_ops.get_request(q, bio, data);
+	else
+		rq = __blk_mq_alloc_request(data, bio->bi_opf);
+
+	if (rq)
+		data->hctx->queued++;
+
+	return rq;
+
+}
+
+static inline void
+blk_mq_sched_insert_request(struct request *rq, bool at_head, bool run_queue,
+			    bool async)
+{
+	struct request_queue *q = rq->q;
+	struct elevator_queue *e = q->elevator;
+	struct blk_mq_ctx *ctx = rq->mq_ctx;
+	struct blk_mq_hw_ctx *hctx = blk_mq_map_queue(q, ctx->cpu);
+
+	if (e)
+		e->type->mq_ops.insert_request(hctx, rq, at_head);
+	else {
+		spin_lock(&ctx->lock);
+		__blk_mq_insert_request(hctx, rq, at_head);
+		spin_unlock(&ctx->lock);
+	}
+
+	if (run_queue)
+		blk_mq_run_hw_queue(hctx, async);
+}
+
+static inline bool
+blk_mq_sched_allow_merge(struct request_queue *q, struct request *rq,
+			 struct bio *bio)
+{
+	struct elevator_queue *e = q->elevator;
+
+	if (e && e->type->mq_ops.allow_merge)
+		return e->type->mq_ops.allow_merge(q, rq, bio);
+
+	return true;
+}
+
+static inline void
+blk_mq_sched_completed_request(struct blk_mq_hw_ctx *hctx, struct request *rq)
+{
+	struct elevator_queue *e = hctx->queue->elevator;
+
+	if (e && e->type->mq_ops.completed_request)
+		e->type->mq_ops.completed_request(hctx, rq);
+}
+
+static inline void blk_mq_sched_started_request(struct request *rq)
+{
+	struct request_queue *q = rq->q;
+	struct elevator_queue *e = q->elevator;
+
+	if (e && e->type->mq_ops.started_request)
+		e->type->mq_ops.started_request(rq);
+}
+
+static inline void blk_mq_sched_requeue_request(struct request *rq)
+{
+	struct request_queue *q = rq->q;
+	struct elevator_queue *e = q->elevator;
+
+	if (e && e->type->mq_ops.requeue_request)
+		e->type->mq_ops.requeue_request(rq);
+}
+
+static inline bool blk_mq_sched_has_work(struct blk_mq_hw_ctx *hctx)
+{
+	struct elevator_queue *e = hctx->queue->elevator;
+
+	if (e && e->type->mq_ops.has_work)
+		return e->type->mq_ops.has_work(hctx);
+
+	return false;
+}
+
+
+#endif
-- 
2.7.4

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web