From dd879364df83542bd16aa36ce373006a9a4a1f35 Mon Sep 17 00:00:00 2001
From: Kent Overstreet <kmo@daterainc.com>
Date: Tue, 10 Sep 2013 19:02:45 -0700
Subject: [PATCH] bcache: Break up struct search

With all the recent refactoring around struct btree op struct search has
gotten rather large.

But we can now easily break it up in a different way - we break out
struct btree_insert_op which is for inserting data into the cache, and
that's now what the copying gc code uses - struct search is now specific
to request.c

Signed-off-by: Kent Overstreet <kmo@daterainc.com>
---
 drivers/md/bcache/debug.c     |  22 +-
 drivers/md/bcache/debug.h     |   4 +-
 drivers/md/bcache/movinggc.c  |  56 ++--
 drivers/md/bcache/request.c   | 586 ++++++++++++++++++----------------
 drivers/md/bcache/request.h   |  37 +--
 drivers/md/bcache/stats.c     |  26 +-
 drivers/md/bcache/stats.h     |  13 +-
 drivers/md/bcache/trace.c     |   1 -
 include/trace/events/bcache.h |  18 +-
 9 files changed, 381 insertions(+), 382 deletions(-)

diff --git a/drivers/md/bcache/debug.c b/drivers/md/bcache/debug.c
index 88e6411eab4f..542b451d7f77 100644
--- a/drivers/md/bcache/debug.c
+++ b/drivers/md/bcache/debug.c
@@ -8,7 +8,6 @@
 #include "bcache.h"
 #include "btree.h"
 #include "debug.h"
-#include "request.h"
 
 #include <linux/console.h>
 #include <linux/debugfs.h>
@@ -182,20 +181,17 @@ static void data_verify_endio(struct bio *bio, int error)
 	closure_put(cl);
 }
 
-void bch_data_verify(struct search *s)
+void bch_data_verify(struct cached_dev *dc, struct bio *bio)
 {
 	char name[BDEVNAME_SIZE];
-	struct cached_dev *dc = container_of(s->d, struct cached_dev, disk);
-	struct closure *cl = &s->cl;
+	struct closure cl;
 	struct bio *check;
 	struct bio_vec *bv;
 	int i;
 
-	if (!s->unaligned_bvec)
-		bio_for_each_segment(bv, s->orig_bio, i)
-			bv->bv_offset = 0, bv->bv_len = PAGE_SIZE;
+	closure_init_stack(&cl);
 
-	check = bio_clone(s->orig_bio, GFP_NOIO);
+	check = bio_clone(bio, GFP_NOIO);
 	if (!check)
 		return;
 
@@ -203,13 +199,13 @@ void bch_data_verify(struct search *s)
 		goto out_put;
 
 	check->bi_rw		= READ_SYNC;
-	check->bi_private	= cl;
+	check->bi_private	= &cl;
 	check->bi_end_io	= data_verify_endio;
 
-	closure_bio_submit(check, cl, &dc->disk);
-	closure_sync(cl);
+	closure_bio_submit(check, &cl, &dc->disk);
+	closure_sync(&cl);
 
-	bio_for_each_segment(bv, s->orig_bio, i) {
+	bio_for_each_segment(bv, bio, i) {
 		void *p1 = kmap(bv->bv_page);
 		void *p2 = kmap(check->bi_io_vec[i].bv_page);
 
@@ -219,7 +215,7 @@ void bch_data_verify(struct search *s)
 			printk(KERN_ERR
 			       "bcache (%s): verify failed at sector %llu\n",
 			       bdevname(dc->bdev, name),
-			       (uint64_t) s->orig_bio->bi_sector);
+			       (uint64_t) bio->bi_sector);
 
 		kunmap(bv->bv_page);
 		kunmap(check->bi_io_vec[i].bv_page);
diff --git a/drivers/md/bcache/debug.h b/drivers/md/bcache/debug.h
index 1c39b5a2489b..0f4b3440512c 100644
--- a/drivers/md/bcache/debug.h
+++ b/drivers/md/bcache/debug.h
@@ -29,12 +29,12 @@ void bch_check_keys(struct btree *, const char *, ...);
 #ifdef CONFIG_BCACHE_DEBUG
 
 void bch_btree_verify(struct btree *, struct bset *);
-void bch_data_verify(struct search *);
+void bch_data_verify(struct cached_dev *, struct bio *);
 
 #else /* DEBUG */
 
 static inline void bch_btree_verify(struct btree *b, struct bset *i) {}
-static inline void bch_data_verify(struct search *s) {};
+static inline void bch_data_verify(struct cached_dev *dc, struct bio *bio) {};
 
 #endif
 
diff --git a/drivers/md/bcache/movinggc.c b/drivers/md/bcache/movinggc.c
index debcf4c11a01..db72d659efb1 100644
--- a/drivers/md/bcache/movinggc.c
+++ b/drivers/md/bcache/movinggc.c
@@ -12,8 +12,9 @@
 #include <trace/events/bcache.h>
 
 struct moving_io {
+	struct closure		cl;
 	struct keybuf_key	*w;
-	struct search		s;
+	struct data_insert_op	op;
 	struct bbio		bio;
 };
 
@@ -38,13 +39,13 @@ static bool moving_pred(struct keybuf *buf, struct bkey *k)
 
 static void moving_io_destructor(struct closure *cl)
 {
-	struct moving_io *io = container_of(cl, struct moving_io, s.cl);
+	struct moving_io *io = container_of(cl, struct moving_io, cl);
 	kfree(io);
 }
 
 static void write_moving_finish(struct closure *cl)
 {
-	struct moving_io *io = container_of(cl, struct moving_io, s.cl);
+	struct moving_io *io = container_of(cl, struct moving_io, cl);
 	struct bio *bio = &io->bio.bio;
 	struct bio_vec *bv;
 	int i;
@@ -52,12 +53,12 @@ static void write_moving_finish(struct closure *cl)
 	bio_for_each_segment_all(bv, bio, i)
 		__free_page(bv->bv_page);
 
-	if (io->s.insert_collision)
+	if (io->op.replace_collision)
 		trace_bcache_gc_copy_collision(&io->w->key);
 
-	bch_keybuf_del(&io->s.c->moving_gc_keys, io->w);
+	bch_keybuf_del(&io->op.c->moving_gc_keys, io->w);
 
-	up(&io->s.c->moving_in_flight);
+	up(&io->op.c->moving_in_flight);
 
 	closure_return_with_destructor(cl, moving_io_destructor);
 }
@@ -65,12 +66,12 @@ static void write_moving_finish(struct closure *cl)
 static void read_moving_endio(struct bio *bio, int error)
 {
 	struct moving_io *io = container_of(bio->bi_private,
-					    struct moving_io, s.cl);
+					    struct moving_io, cl);
 
 	if (error)
-		io->s.error = error;
+		io->op.error = error;
 
-	bch_bbio_endio(io->s.c, bio, error, "reading data to move");
+	bch_bbio_endio(io->op.c, bio, error, "reading data to move");
 }
 
 static void moving_init(struct moving_io *io)
@@ -84,32 +85,30 @@ static void moving_init(struct moving_io *io)
 	bio->bi_size		= KEY_SIZE(&io->w->key) << 9;
 	bio->bi_max_vecs	= DIV_ROUND_UP(KEY_SIZE(&io->w->key),
 					       PAGE_SECTORS);
-	bio->bi_private		= &io->s.cl;
+	bio->bi_private		= &io->cl;
 	bio->bi_io_vec		= bio->bi_inline_vecs;
 	bch_bio_map(bio, NULL);
 }
 
 static void write_moving(struct closure *cl)
 {
-	struct search *s = container_of(cl, struct search, cl);
-	struct moving_io *io = container_of(s, struct moving_io, s);
+	struct moving_io *io = container_of(cl, struct moving_io, cl);
+	struct data_insert_op *op = &io->op;
 
-	if (!s->error) {
+	if (!op->error) {
 		moving_init(io);
 
-		io->bio.bio.bi_sector	= KEY_START(&io->w->key);
-		s->op.lock		= -1;
-		s->write_prio		= 1;
-		s->cache_bio		= &io->bio.bio;
+		io->bio.bio.bi_sector = KEY_START(&io->w->key);
+		op->write_prio		= 1;
+		op->bio			= &io->bio.bio;
 
-		s->writeback		= KEY_DIRTY(&io->w->key);
-		s->csum			= KEY_CSUM(&io->w->key);
+		op->writeback		= KEY_DIRTY(&io->w->key);
+		op->csum		= KEY_CSUM(&io->w->key);
 
-		bkey_copy(&s->replace_key, &io->w->key);
-		s->replace = true;
+		bkey_copy(&op->replace_key, &io->w->key);
+		op->replace		= true;
 
-		closure_init(&s->btree, cl);
-		bch_data_insert(&s->btree);
+		closure_call(&op->cl, bch_data_insert, NULL, cl);
 	}
 
 	continue_at(cl, write_moving_finish, NULL);
@@ -117,11 +116,10 @@ static void write_moving(struct closure *cl)
 
 static void read_moving_submit(struct closure *cl)
 {
-	struct search *s = container_of(cl, struct search, cl);
-	struct moving_io *io = container_of(s, struct moving_io, s);
+	struct moving_io *io = container_of(cl, struct moving_io, cl);
 	struct bio *bio = &io->bio.bio;
 
-	bch_submit_bbio(bio, s->c, &io->w->key, 0);
+	bch_submit_bbio(bio, io->op.c, &io->w->key, 0);
 
 	continue_at(cl, write_moving, system_wq);
 }
@@ -151,8 +149,8 @@ static void read_moving(struct cache_set *c)
 
 		w->private	= io;
 		io->w		= w;
-		io->s.inode	= KEY_INODE(&w->key);
-		io->s.c		= c;
+		io->op.inode	= KEY_INODE(&w->key);
+		io->op.c	= c;
 
 		moving_init(io);
 		bio = &io->bio.bio;
@@ -166,7 +164,7 @@ static void read_moving(struct cache_set *c)
 		trace_bcache_gc_copy(&w->key);
 
 		down(&c->moving_in_flight);
-		closure_call(&io->s.cl, read_moving_submit, NULL, &cl);
+		closure_call(&io->cl, read_moving_submit, NULL, &cl);
 	}
 
 	if (0) {
diff --git a/drivers/md/bcache/request.c b/drivers/md/bcache/request.c
index 0dcb6e3bc38d..d0fafa83ef1b 100644
--- a/drivers/md/bcache/request.c
+++ b/drivers/md/bcache/request.c
@@ -215,9 +215,9 @@ static void bio_csum(struct bio *bio, struct bkey *k)
 
 static void bch_data_insert_keys(struct closure *cl)
 {
-	struct search *s = container_of(cl, struct search, btree);
+	struct data_insert_op *op = container_of(cl, struct data_insert_op, cl);
 	atomic_t *journal_ref = NULL;
-	struct bkey *replace_key = s->replace ? &s->replace_key : NULL;
+	struct bkey *replace_key = op->replace ? &op->replace_key : NULL;
 	int ret;
 
 	/*
@@ -232,27 +232,26 @@ static void bch_data_insert_keys(struct closure *cl)
 		closure_sync(&s->cl);
 #endif
 
-	if (s->write)
-		journal_ref = bch_journal(s->c, &s->insert_keys,
-					  s->flush_journal
-					  ? &s->cl : NULL);
+	if (!op->replace)
+		journal_ref = bch_journal(op->c, &op->insert_keys,
+					  op->flush_journal ? cl : NULL);
 
-	ret = bch_btree_insert(s->c, &s->insert_keys,
+	ret = bch_btree_insert(op->c, &op->insert_keys,
 			       journal_ref, replace_key);
 	if (ret == -ESRCH) {
-		s->insert_collision = true;
+		op->replace_collision = true;
 	} else if (ret) {
-		s->error		= -ENOMEM;
-		s->insert_data_done	= true;
+		op->error		= -ENOMEM;
+		op->insert_data_done	= true;
 	}
 
 	if (journal_ref)
 		atomic_dec_bug(journal_ref);
 
-	if (!s->insert_data_done)
+	if (!op->insert_data_done)
 		continue_at(cl, bch_data_insert_start, bcache_wq);
 
-	bch_keylist_free(&s->insert_keys);
+	bch_keylist_free(&op->insert_keys);
 	closure_return(cl);
 }
 
@@ -349,10 +348,10 @@ found:
  *
  * If s->writeback is true, will not fail.
  */
-static bool bch_alloc_sectors(struct bkey *k, unsigned sectors,
-			      struct search *s)
+static bool bch_alloc_sectors(struct data_insert_op *op,
+			      struct bkey *k, unsigned sectors)
 {
-	struct cache_set *c = s->c;
+	struct cache_set *c = op->c;
 	struct open_bucket *b;
 	BKEY_PADDED(key) alloc;
 	unsigned i;
@@ -367,15 +366,15 @@ static bool bch_alloc_sectors(struct bkey *k, unsigned sectors,
 	bkey_init(&alloc.key);
 	spin_lock(&c->data_bucket_lock);
 
-	while (!(b = pick_data_bucket(c, k, s->task, &alloc.key))) {
-		unsigned watermark = s->write_prio
+	while (!(b = pick_data_bucket(c, k, op->task, &alloc.key))) {
+		unsigned watermark = op->write_prio
 			? WATERMARK_MOVINGGC
 			: WATERMARK_NONE;
 
 		spin_unlock(&c->data_bucket_lock);
 
 		if (bch_bucket_alloc_set(c, watermark, &alloc.key,
-					 1, s->writeback))
+					 1, op->writeback))
 			return false;
 
 		spin_lock(&c->data_bucket_lock);
@@ -409,7 +408,7 @@ static bool bch_alloc_sectors(struct bkey *k, unsigned sectors,
 	 */
 	list_move_tail(&b->list, &c->data_buckets);
 	bkey_copy_key(&b->key, k);
-	b->last = s->task;
+	b->last = op->task;
 
 	b->sectors_free	-= sectors;
 
@@ -438,8 +437,8 @@ static bool bch_alloc_sectors(struct bkey *k, unsigned sectors,
 
 static void bch_data_invalidate(struct closure *cl)
 {
-	struct search *s = container_of(cl, struct search, btree);
-	struct bio *bio = s->cache_bio;
+	struct data_insert_op *op = container_of(cl, struct data_insert_op, cl);
+	struct bio *bio = op->bio;
 
 	pr_debug("invalidating %i sectors from %llu",
 		 bio_sectors(bio), (uint64_t) bio->bi_sector);
@@ -447,17 +446,17 @@ static void bch_data_invalidate(struct closure *cl)
 	while (bio_sectors(bio)) {
 		unsigned len = min(bio_sectors(bio), 1U << 14);
 
-		if (bch_keylist_realloc(&s->insert_keys, 0, s->c))
+		if (bch_keylist_realloc(&op->insert_keys, 0, op->c))
 			goto out;
 
 		bio->bi_sector	+= len;
 		bio->bi_size	-= len << 9;
 
-		bch_keylist_add(&s->insert_keys,
-				&KEY(s->inode, bio->bi_sector, len));
+		bch_keylist_add(&op->insert_keys,
+				&KEY(op->inode, bio->bi_sector, len));
 	}
 
-	s->insert_data_done = true;
+	op->insert_data_done = true;
 	bio_put(bio);
 out:
 	continue_at(cl, bch_data_insert_keys, bcache_wq);
@@ -465,7 +464,7 @@ out:
 
 static void bch_data_insert_error(struct closure *cl)
 {
-	struct search *s = container_of(cl, struct search, btree);
+	struct data_insert_op *op = container_of(cl, struct data_insert_op, cl);
 
 	/*
 	 * Our data write just errored, which means we've got a bunch of keys to
@@ -476,9 +475,9 @@ static void bch_data_insert_error(struct closure *cl)
 	 * from the keys we'll accomplish just that.
 	 */
 
-	struct bkey *src = s->insert_keys.keys, *dst = s->insert_keys.keys;
+	struct bkey *src = op->insert_keys.keys, *dst = op->insert_keys.keys;
 
-	while (src != s->insert_keys.top) {
+	while (src != op->insert_keys.top) {
 		struct bkey *n = bkey_next(src);
 
 		SET_KEY_PTRS(src, 0);
@@ -488,7 +487,7 @@ static void bch_data_insert_error(struct closure *cl)
 		src = n;
 	}
 
-	s->insert_keys.top = dst;
+	op->insert_keys.top = dst;
 
 	bch_data_insert_keys(cl);
 }
@@ -496,32 +495,32 @@ static void bch_data_insert_error(struct closure *cl)
 static void bch_data_insert_endio(struct bio *bio, int error)
 {
 	struct closure *cl = bio->bi_private;
-	struct search *s = container_of(cl, struct search, btree);
+	struct data_insert_op *op = container_of(cl, struct data_insert_op, cl);
 
 	if (error) {
 		/* TODO: We could try to recover from this. */
-		if (s->writeback)
-			s->error = error;
-		else if (s->write)
+		if (op->writeback)
+			op->error = error;
+		else if (!op->replace)
 			set_closure_fn(cl, bch_data_insert_error, bcache_wq);
 		else
 			set_closure_fn(cl, NULL, NULL);
 	}
 
-	bch_bbio_endio(s->c, bio, error, "writing data to cache");
+	bch_bbio_endio(op->c, bio, error, "writing data to cache");
 }
 
 static void bch_data_insert_start(struct closure *cl)
 {
-	struct search *s = container_of(cl, struct search, btree);
-	struct bio *bio = s->cache_bio, *n;
+	struct data_insert_op *op = container_of(cl, struct data_insert_op, cl);
+	struct bio *bio = op->bio, *n;
 
-	if (s->bypass)
+	if (op->bypass)
 		return bch_data_invalidate(cl);
 
-	if (atomic_sub_return(bio_sectors(bio), &s->c->sectors_to_gc) < 0) {
-		set_gc_sectors(s->c);
-		wake_up_gc(s->c);
+	if (atomic_sub_return(bio_sectors(bio), &op->c->sectors_to_gc) < 0) {
+		set_gc_sectors(op->c);
+		wake_up_gc(op->c);
 	}
 
 	/*
@@ -533,21 +532,20 @@ static void bch_data_insert_start(struct closure *cl)
 	do {
 		unsigned i;
 		struct bkey *k;
-		struct bio_set *split = s->d
-			? s->d->bio_split : s->c->bio_split;
+		struct bio_set *split = op->c->bio_split;
 
 		/* 1 for the device pointer and 1 for the chksum */
-		if (bch_keylist_realloc(&s->insert_keys,
-					1 + (s->csum ? 1 : 0),
-					s->c))
+		if (bch_keylist_realloc(&op->insert_keys,
+					1 + (op->csum ? 1 : 0),
+					op->c))
 			continue_at(cl, bch_data_insert_keys, bcache_wq);
 
-		k = s->insert_keys.top;
+		k = op->insert_keys.top;
 		bkey_init(k);
-		SET_KEY_INODE(k, s->inode);
+		SET_KEY_INODE(k, op->inode);
 		SET_KEY_OFFSET(k, bio->bi_sector);
 
-		if (!bch_alloc_sectors(k, bio_sectors(bio), s))
+		if (!bch_alloc_sectors(op, k, bio_sectors(bio)))
 			goto err;
 
 		n = bch_bio_split(bio, KEY_SIZE(k), GFP_NOIO, split);
@@ -555,30 +553,30 @@ static void bch_data_insert_start(struct closure *cl)
 		n->bi_end_io	= bch_data_insert_endio;
 		n->bi_private	= cl;
 
-		if (s->writeback) {
+		if (op->writeback) {
 			SET_KEY_DIRTY(k, true);
 
 			for (i = 0; i < KEY_PTRS(k); i++)
-				SET_GC_MARK(PTR_BUCKET(s->c, k, i),
+				SET_GC_MARK(PTR_BUCKET(op->c, k, i),
 					    GC_MARK_DIRTY);
 		}
 
-		SET_KEY_CSUM(k, s->csum);
+		SET_KEY_CSUM(k, op->csum);
 		if (KEY_CSUM(k))
 			bio_csum(n, k);
 
 		trace_bcache_cache_insert(k);
-		bch_keylist_push(&s->insert_keys);
+		bch_keylist_push(&op->insert_keys);
 
 		n->bi_rw |= REQ_WRITE;
-		bch_submit_bbio(n, s->c, k, 0);
+		bch_submit_bbio(n, op->c, k, 0);
 	} while (n != bio);
 
-	s->insert_data_done = true;
+	op->insert_data_done = true;
 	continue_at(cl, bch_data_insert_keys, bcache_wq);
 err:
 	/* bch_alloc_sectors() blocks if s->writeback = true */
-	BUG_ON(s->writeback);
+	BUG_ON(op->writeback);
 
 	/*
 	 * But if it's not a writeback write we'd rather just bail out if
@@ -586,24 +584,24 @@ err:
 	 * we might be starving btree writes for gc or something.
 	 */
 
-	if (s->write) {
+	if (!op->replace) {
 		/*
 		 * Writethrough write: We can't complete the write until we've
 		 * updated the index. But we don't want to delay the write while
 		 * we wait for buckets to be freed up, so just invalidate the
 		 * rest of the write.
 		 */
-		s->bypass = true;
+		op->bypass = true;
 		return bch_data_invalidate(cl);
 	} else {
 		/*
 		 * From a cache miss, we can just insert the keys for the data
 		 * we have written or bail out if we didn't do anything.
 		 */
-		s->insert_data_done = true;
+		op->insert_data_done = true;
 		bio_put(bio);
 
-		if (!bch_keylist_empty(&s->insert_keys))
+		if (!bch_keylist_empty(&op->insert_keys))
 			continue_at(cl, bch_data_insert_keys, bcache_wq);
 		else
 			closure_return(cl);
@@ -631,15 +629,169 @@ err:
  */
 void bch_data_insert(struct closure *cl)
 {
-	struct search *s = container_of(cl, struct search, btree);
+	struct data_insert_op *op = container_of(cl, struct data_insert_op, cl);
 
-	bch_keylist_init(&s->insert_keys);
-	bio_get(s->cache_bio);
+	trace_bcache_write(op->bio, op->writeback, op->bypass);
+
+	bch_keylist_init(&op->insert_keys);
+	bio_get(op->bio);
 	bch_data_insert_start(cl);
 }
 
+/* Congested? */
+
+unsigned bch_get_congested(struct cache_set *c)
+{
+	int i;
+	long rand;
+
+	if (!c->congested_read_threshold_us &&
+	    !c->congested_write_threshold_us)
+		return 0;
+
+	i = (local_clock_us() - c->congested_last_us) / 1024;
+	if (i < 0)
+		return 0;
+
+	i += atomic_read(&c->congested);
+	if (i >= 0)
+		return 0;
+
+	i += CONGESTED_MAX;
+
+	if (i > 0)
+		i = fract_exp_two(i, 6);
+
+	rand = get_random_int();
+	i -= bitmap_weight(&rand, BITS_PER_LONG);
+
+	return i > 0 ? i : 1;
+}
+
+static void add_sequential(struct task_struct *t)
+{
+	ewma_add(t->sequential_io_avg,
+		 t->sequential_io, 8, 0);
+
+	t->sequential_io = 0;
+}
+
+static struct hlist_head *iohash(struct cached_dev *dc, uint64_t k)
+{
+	return &dc->io_hash[hash_64(k, RECENT_IO_BITS)];
+}
+
+static bool check_should_bypass(struct cached_dev *dc, struct bio *bio)
+{
+	struct cache_set *c = dc->disk.c;
+	unsigned mode = cache_mode(dc, bio);
+	unsigned sectors, congested = bch_get_congested(c);
+	struct task_struct *task = current;
+
+	if (atomic_read(&dc->disk.detaching) ||
+	    c->gc_stats.in_use > CUTOFF_CACHE_ADD ||
+	    (bio->bi_rw & REQ_DISCARD))
+		goto skip;
+
+	if (mode == CACHE_MODE_NONE ||
+	    (mode == CACHE_MODE_WRITEAROUND &&
+	     (bio->bi_rw & REQ_WRITE)))
+		goto skip;
+
+	if (bio->bi_sector & (c->sb.block_size - 1) ||
+	    bio_sectors(bio) & (c->sb.block_size - 1)) {
+		pr_debug("skipping unaligned io");
+		goto skip;
+	}
+
+	if (!congested && !dc->sequential_cutoff)
+		goto rescale;
+
+	if (!congested &&
+	    mode == CACHE_MODE_WRITEBACK &&
+	    (bio->bi_rw & REQ_WRITE) &&
+	    (bio->bi_rw & REQ_SYNC))
+		goto rescale;
+
+	if (dc->sequential_merge) {
+		struct io *i;
+
+		spin_lock(&dc->io_lock);
+
+		hlist_for_each_entry(i, iohash(dc, bio->bi_sector), hash)
+			if (i->last == bio->bi_sector &&
+			    time_before(jiffies, i->jiffies))
+				goto found;
+
+		i = list_first_entry(&dc->io_lru, struct io, lru);
+
+		add_sequential(task);
+		i->sequential = 0;
+found:
+		if (i->sequential + bio->bi_size > i->sequential)
+			i->sequential	+= bio->bi_size;
+
+		i->last			 = bio_end_sector(bio);
+		i->jiffies		 = jiffies + msecs_to_jiffies(5000);
+		task->sequential_io	 = i->sequential;
+
+		hlist_del(&i->hash);
+		hlist_add_head(&i->hash, iohash(dc, i->last));
+		list_move_tail(&i->lru, &dc->io_lru);
+
+		spin_unlock(&dc->io_lock);
+	} else {
+		task->sequential_io = bio->bi_size;
+
+		add_sequential(task);
+	}
+
+	sectors = max(task->sequential_io,
+		      task->sequential_io_avg) >> 9;
+
+	if (dc->sequential_cutoff &&
+	    sectors >= dc->sequential_cutoff >> 9) {
+		trace_bcache_bypass_sequential(bio);
+		goto skip;
+	}
+
+	if (congested && sectors >= congested) {
+		trace_bcache_bypass_congested(bio);
+		goto skip;
+	}
+
+rescale:
+	bch_rescale_priorities(c, bio_sectors(bio));
+	return false;
+skip:
+	bch_mark_sectors_bypassed(c, dc, bio_sectors(bio));
+	return true;
+}
+
 /* Cache lookup */
 
+struct search {
+	/* Stack frame for bio_complete */
+	struct closure		cl;
+
+	struct bcache_device	*d;
+
+	struct bbio		bio;
+	struct bio		*orig_bio;
+	struct bio		*cache_miss;
+
+	unsigned		insert_bio_sectors;
+
+	unsigned		recoverable:1;
+	unsigned		unaligned_bvec:1;
+	unsigned		write:1;
+
+	unsigned long		start_time;
+
+	struct btree_op		op;
+	struct data_insert_op	iop;
+};
+
 static void bch_cache_read_endio(struct bio *bio, int error)
 {
 	struct bbio *b = container_of(bio, struct bbio, bio);
@@ -654,13 +806,13 @@ static void bch_cache_read_endio(struct bio *bio, int error)
 	 */
 
 	if (error)
-		s->error = error;
-	else if (ptr_stale(s->c, &b->key, 0)) {
-		atomic_long_inc(&s->c->cache_read_races);
-		s->error = -EINTR;
+		s->iop.error = error;
+	else if (ptr_stale(s->iop.c, &b->key, 0)) {
+		atomic_long_inc(&s->iop.c->cache_read_races);
+		s->iop.error = -EINTR;
 	}
 
-	bch_bbio_endio(s->c, bio, error, "reading from cache");
+	bch_bbio_endio(s->iop.c, bio, error, "reading from cache");
 }
 
 /*
@@ -674,13 +826,13 @@ static int cache_lookup_fn(struct btree_op *op, struct btree *b, struct bkey *k)
 	struct bkey *bio_key;
 	unsigned ptr;
 
-	if (bkey_cmp(k, &KEY(s->inode, bio->bi_sector, 0)) <= 0)
+	if (bkey_cmp(k, &KEY(s->iop.inode, bio->bi_sector, 0)) <= 0)
 		return MAP_CONTINUE;
 
-	if (KEY_INODE(k) != s->inode ||
+	if (KEY_INODE(k) != s->iop.inode ||
 	    KEY_START(k) > bio->bi_sector) {
 		unsigned bio_sectors = bio_sectors(bio);
-		unsigned sectors = KEY_INODE(k) == s->inode
+		unsigned sectors = KEY_INODE(k) == s->iop.inode
 			? min_t(uint64_t, INT_MAX,
 				KEY_START(k) - bio->bi_sector)
 			: INT_MAX;
@@ -708,8 +860,8 @@ static int cache_lookup_fn(struct btree_op *op, struct btree *b, struct bkey *k)
 	bio_key = &container_of(n, struct bbio, bio)->key;
 	bch_bkey_copy_single_ptr(bio_key, k, ptr);
 
-	bch_cut_front(&KEY(s->inode, n->bi_sector, 0), bio_key);
-	bch_cut_back(&KEY(s->inode, bio_end_sector(n), 0), bio_key);
+	bch_cut_front(&KEY(s->iop.inode, n->bi_sector, 0), bio_key);
+	bch_cut_back(&KEY(s->iop.inode, bio_end_sector(n), 0), bio_key);
 
 	n->bi_end_io	= bch_cache_read_endio;
 	n->bi_private	= &s->cl;
@@ -731,11 +883,11 @@ static int cache_lookup_fn(struct btree_op *op, struct btree *b, struct bkey *k)
 
 static void cache_lookup(struct closure *cl)
 {
-	struct search *s = container_of(cl, struct search, btree);
+	struct search *s = container_of(cl, struct search, iop.cl);
 	struct bio *bio = &s->bio.bio;
 
-	int ret = bch_btree_map_keys(&s->op, s->c,
-				     &KEY(s->inode, bio->bi_sector, 0),
+	int ret = bch_btree_map_keys(&s->op, s->iop.c,
+				     &KEY(s->iop.inode, bio->bi_sector, 0),
 				     cache_lookup_fn, MAP_END_KEY);
 	if (ret == -EAGAIN)
 		continue_at(cl, cache_lookup, bcache_wq);
@@ -751,7 +903,7 @@ static void request_endio(struct bio *bio, int error)
 
 	if (error) {
 		struct search *s = container_of(cl, struct search, cl);
-		s->error = error;
+		s->iop.error = error;
 		/* Only cache read errors are recoverable */
 		s->recoverable = false;
 	}
@@ -771,8 +923,8 @@ static void bio_complete(struct search *s)
 		part_stat_add(cpu, &s->d->disk->part0, ticks[rw], duration);
 		part_stat_unlock();
 
-		trace_bcache_request_end(s, s->orig_bio);
-		bio_endio(s->orig_bio, s->error);
+		trace_bcache_request_end(s->d, s->orig_bio);
+		bio_endio(s->orig_bio, s->iop.error);
 		s->orig_bio = NULL;
 	}
 }
@@ -792,8 +944,8 @@ static void search_free(struct closure *cl)
 	struct search *s = container_of(cl, struct search, cl);
 	bio_complete(s);
 
-	if (s->cache_bio)
-		bio_put(s->cache_bio);
+	if (s->iop.bio)
+		bio_put(s->iop.bio);
 
 	if (s->unaligned_bvec)
 		mempool_free(s->bio.bio.bi_io_vec, s->d->unaligned_bvec);
@@ -808,18 +960,18 @@ static struct search *search_alloc(struct bio *bio, struct bcache_device *d)
 	struct search *s;
 
 	s = mempool_alloc(d->c->search, GFP_NOIO);
-	memset(s, 0, offsetof(struct search, insert_keys));
+	memset(s, 0, offsetof(struct search, iop.insert_keys));
 
 	__closure_init(&s->cl, NULL);
 
-	s->inode		= d->id;
-	s->c			= d->c;
+	s->iop.inode		= d->id;
+	s->iop.c		= d->c;
 	s->d			= d;
 	s->op.lock		= -1;
-	s->task			= current;
+	s->iop.task		= current;
 	s->orig_bio		= bio;
 	s->write		= (bio->bi_rw & REQ_WRITE) != 0;
-	s->flush_journal	= (bio->bi_rw & (REQ_FLUSH|REQ_FUA)) != 0;
+	s->iop.flush_journal	= (bio->bi_rw & (REQ_FLUSH|REQ_FUA)) != 0;
 	s->recoverable		= 1;
 	s->start_time		= jiffies;
 	do_bio_hook(s);
@@ -847,148 +999,20 @@ static void cached_dev_bio_complete(struct closure *cl)
 	cached_dev_put(dc);
 }
 
-unsigned bch_get_congested(struct cache_set *c)
-{
-	int i;
-	long rand;
-
-	if (!c->congested_read_threshold_us &&
-	    !c->congested_write_threshold_us)
-		return 0;
-
-	i = (local_clock_us() - c->congested_last_us) / 1024;
-	if (i < 0)
-		return 0;
-
-	i += atomic_read(&c->congested);
-	if (i >= 0)
-		return 0;
-
-	i += CONGESTED_MAX;
-
-	if (i > 0)
-		i = fract_exp_two(i, 6);
-
-	rand = get_random_int();
-	i -= bitmap_weight(&rand, BITS_PER_LONG);
-
-	return i > 0 ? i : 1;
-}
-
-static void add_sequential(struct task_struct *t)
-{
-	ewma_add(t->sequential_io_avg,
-		 t->sequential_io, 8, 0);
-
-	t->sequential_io = 0;
-}
-
-static struct hlist_head *iohash(struct cached_dev *dc, uint64_t k)
-{
-	return &dc->io_hash[hash_64(k, RECENT_IO_BITS)];
-}
-
-static bool check_should_bypass(struct cached_dev *dc, struct search *s)
-{
-	struct cache_set *c = s->c;
-	struct bio *bio = &s->bio.bio;
-	unsigned mode = cache_mode(dc, bio);
-	unsigned sectors, congested = bch_get_congested(c);
-
-	if (atomic_read(&dc->disk.detaching) ||
-	    c->gc_stats.in_use > CUTOFF_CACHE_ADD ||
-	    (bio->bi_rw & REQ_DISCARD))
-		goto skip;
-
-	if (mode == CACHE_MODE_NONE ||
-	    (mode == CACHE_MODE_WRITEAROUND &&
-	     (bio->bi_rw & REQ_WRITE)))
-		goto skip;
-
-	if (bio->bi_sector & (c->sb.block_size - 1) ||
-	    bio_sectors(bio) & (c->sb.block_size - 1)) {
-		pr_debug("skipping unaligned io");
-		goto skip;
-	}
-
-	if (!congested && !dc->sequential_cutoff)
-		goto rescale;
-
-	if (!congested &&
-	    mode == CACHE_MODE_WRITEBACK &&
-	    (bio->bi_rw & REQ_WRITE) &&
-	    (bio->bi_rw & REQ_SYNC))
-		goto rescale;
-
-	if (dc->sequential_merge) {
-		struct io *i;
-
-		spin_lock(&dc->io_lock);
-
-		hlist_for_each_entry(i, iohash(dc, bio->bi_sector), hash)
-			if (i->last == bio->bi_sector &&
-			    time_before(jiffies, i->jiffies))
-				goto found;
-
-		i = list_first_entry(&dc->io_lru, struct io, lru);
-
-		add_sequential(s->task);
-		i->sequential = 0;
-found:
-		if (i->sequential + bio->bi_size > i->sequential)
-			i->sequential	+= bio->bi_size;
-
-		i->last			 = bio_end_sector(bio);
-		i->jiffies		 = jiffies + msecs_to_jiffies(5000);
-		s->task->sequential_io	 = i->sequential;
-
-		hlist_del(&i->hash);
-		hlist_add_head(&i->hash, iohash(dc, i->last));
-		list_move_tail(&i->lru, &dc->io_lru);
-
-		spin_unlock(&dc->io_lock);
-	} else {
-		s->task->sequential_io = bio->bi_size;
-
-		add_sequential(s->task);
-	}
-
-	sectors = max(s->task->sequential_io,
-		      s->task->sequential_io_avg) >> 9;
-
-	if (dc->sequential_cutoff &&
-	    sectors >= dc->sequential_cutoff >> 9) {
-		trace_bcache_bypass_sequential(s->orig_bio);
-		goto skip;
-	}
-
-	if (congested && sectors >= congested) {
-		trace_bcache_bypass_congested(s->orig_bio);
-		goto skip;
-	}
-
-rescale:
-	bch_rescale_priorities(c, bio_sectors(bio));
-	return false;
-skip:
-	bch_mark_sectors_bypassed(s, bio_sectors(bio));
-	return true;
-}
-
 /* Process reads */
 
 static void cached_dev_cache_miss_done(struct closure *cl)
 {
 	struct search *s = container_of(cl, struct search, cl);
 
-	if (s->insert_collision)
-		bch_mark_cache_miss_collision(s);
+	if (s->iop.replace_collision)
+		bch_mark_cache_miss_collision(s->iop.c, s->d);
 
-	if (s->cache_bio) {
+	if (s->iop.bio) {
 		int i;
 		struct bio_vec *bv;
 
-		bio_for_each_segment_all(bv, s->cache_bio, i)
+		bio_for_each_segment_all(bv, s->iop.bio, i)
 			__free_page(bv->bv_page);
 	}
 
@@ -1006,7 +1030,7 @@ static void cached_dev_read_error(struct closure *cl)
 		/* Retry from the backing device: */
 		trace_bcache_read_retry(s->orig_bio);
 
-		s->error = 0;
+		s->iop.error = 0;
 		bv = s->bio.bio.bi_io_vec;
 		do_bio_hook(s);
 		s->bio.bio.bi_io_vec = bv;
@@ -1041,28 +1065,28 @@ static void cached_dev_read_done(struct closure *cl)
 	 * to the buffers the original bio pointed to:
 	 */
 
-	if (s->cache_bio) {
-		bio_reset(s->cache_bio);
-		s->cache_bio->bi_sector	= s->cache_miss->bi_sector;
-		s->cache_bio->bi_bdev	= s->cache_miss->bi_bdev;
-		s->cache_bio->bi_size	= s->cache_bio_sectors << 9;
-		bch_bio_map(s->cache_bio, NULL);
+	if (s->iop.bio) {
+		bio_reset(s->iop.bio);
+		s->iop.bio->bi_sector	= s->cache_miss->bi_sector;
+		s->iop.bio->bi_bdev	= s->cache_miss->bi_bdev;
+		s->iop.bio->bi_size	= s->insert_bio_sectors << 9;
+		bch_bio_map(s->iop.bio, NULL);
 
-		bio_copy_data(s->cache_miss, s->cache_bio);
+		bio_copy_data(s->cache_miss, s->iop.bio);
 
 		bio_put(s->cache_miss);
 		s->cache_miss = NULL;
 	}
 
-	if (verify(dc, &s->bio.bio) && s->recoverable)
-		bch_data_verify(s);
+	if (verify(dc, &s->bio.bio) && s->recoverable && !s->unaligned_bvec)
+		bch_data_verify(dc, s->orig_bio);
 
 	bio_complete(s);
 
-	if (s->cache_bio &&
-	    !test_bit(CACHE_SET_STOPPING, &s->c->flags)) {
-		BUG_ON(!s->replace);
-		closure_call(&s->btree, bch_data_insert, NULL, cl);
+	if (s->iop.bio &&
+	    !test_bit(CACHE_SET_STOPPING, &s->iop.c->flags)) {
+		BUG_ON(!s->iop.replace);
+		closure_call(&s->iop.cl, bch_data_insert, NULL, cl);
 	}
 
 	continue_at(cl, cached_dev_cache_miss_done, NULL);
@@ -1073,12 +1097,13 @@ static void cached_dev_read_done_bh(struct closure *cl)
 	struct search *s = container_of(cl, struct search, cl);
 	struct cached_dev *dc = container_of(s->d, struct cached_dev, disk);
 
-	bch_mark_cache_accounting(s, !s->cache_miss, s->bypass);
-	trace_bcache_read(s->orig_bio, !s->cache_miss, s->bypass);
+	bch_mark_cache_accounting(s->iop.c, s->d,
+				  !s->cache_miss, s->iop.bypass);
+	trace_bcache_read(s->orig_bio, !s->cache_miss, s->iop.bypass);
 
-	if (s->error)
+	if (s->iop.error)
 		continue_at_nobarrier(cl, cached_dev_read_error, bcache_wq);
-	else if (s->cache_bio || verify(dc, &s->bio.bio))
+	else if (s->iop.bio || verify(dc, &s->bio.bio))
 		continue_at_nobarrier(cl, cached_dev_read_done, bcache_wq);
 	else
 		continue_at_nobarrier(cl, cached_dev_bio_complete, NULL);
@@ -1092,7 +1117,7 @@ static int cached_dev_cache_miss(struct btree *b, struct search *s,
 	struct cached_dev *dc = container_of(s->d, struct cached_dev, disk);
 	struct bio *miss, *cache_bio;
 
-	if (s->cache_miss || s->bypass) {
+	if (s->cache_miss || s->iop.bypass) {
 		miss = bch_bio_split(bio, sectors, GFP_NOIO, s->d->bio_split);
 		ret = miss == bio ? MAP_DONE : MAP_CONTINUE;
 		goto out_submit;
@@ -1100,21 +1125,21 @@ static int cached_dev_cache_miss(struct btree *b, struct search *s,
 
 	if (!(bio->bi_rw & REQ_RAHEAD) &&
 	    !(bio->bi_rw & REQ_META) &&
-	    s->c->gc_stats.in_use < CUTOFF_CACHE_READA)
+	    s->iop.c->gc_stats.in_use < CUTOFF_CACHE_READA)
 		reada = min_t(sector_t, dc->readahead >> 9,
 			      bdev_sectors(bio->bi_bdev) - bio_end_sector(bio));
 
-	s->cache_bio_sectors = min(sectors, bio_sectors(bio) + reada);
+	s->insert_bio_sectors = min(sectors, bio_sectors(bio) + reada);
 
-	s->replace_key = KEY(s->inode, bio->bi_sector +
-			     s->cache_bio_sectors,
-			     s->cache_bio_sectors);
+	s->iop.replace_key = KEY(s->iop.inode,
+				 bio->bi_sector + s->insert_bio_sectors,
+				 s->insert_bio_sectors);
 
-	ret = bch_btree_insert_check_key(b, &s->op, &s->replace_key);
+	ret = bch_btree_insert_check_key(b, &s->op, &s->iop.replace_key);
 	if (ret)
 		return ret;
 
-	s->replace = true;
+	s->iop.replace = true;
 
 	miss = bch_bio_split(bio, sectors, GFP_NOIO, s->d->bio_split);
 
@@ -1122,14 +1147,14 @@ static int cached_dev_cache_miss(struct btree *b, struct search *s,
 	ret = miss == bio ? MAP_DONE : -EINTR;
 
 	cache_bio = bio_alloc_bioset(GFP_NOWAIT,
-			DIV_ROUND_UP(s->cache_bio_sectors, PAGE_SECTORS),
+			DIV_ROUND_UP(s->insert_bio_sectors, PAGE_SECTORS),
 			dc->disk.bio_split);
 	if (!cache_bio)
 		goto out_submit;
 
 	cache_bio->bi_sector	= miss->bi_sector;
 	cache_bio->bi_bdev	= miss->bi_bdev;
-	cache_bio->bi_size	= s->cache_bio_sectors << 9;
+	cache_bio->bi_size	= s->insert_bio_sectors << 9;
 
 	cache_bio->bi_end_io	= request_endio;
 	cache_bio->bi_private	= &s->cl;
@@ -1138,8 +1163,11 @@ static int cached_dev_cache_miss(struct btree *b, struct search *s,
 	if (bio_alloc_pages(cache_bio, __GFP_NOWARN|GFP_NOIO))
 		goto out_put;
 
+	if (reada)
+		bch_mark_cache_readahead(s->iop.c, s->d);
+
 	s->cache_miss	= miss;
-	s->cache_bio = cache_bio;
+	s->iop.bio	= cache_bio;
 	bio_get(cache_bio);
 	closure_bio_submit(cache_bio, &s->cl, s->d);
 
@@ -1157,7 +1185,7 @@ static void cached_dev_read(struct cached_dev *dc, struct search *s)
 {
 	struct closure *cl = &s->cl;
 
-	closure_call(&s->btree, cache_lookup, NULL, cl);
+	closure_call(&s->iop.cl, cache_lookup, NULL, cl);
 	continue_at(cl, cached_dev_read_done_bh, NULL);
 }
 
@@ -1179,7 +1207,7 @@ static void cached_dev_write(struct cached_dev *dc, struct search *s)
 	struct bkey start = KEY(dc->disk.id, bio->bi_sector, 0);
 	struct bkey end = KEY(dc->disk.id, bio_end_sector(bio), 0);
 
-	bch_keybuf_check_overlapping(&s->c->moving_gc_keys, &start, &end);
+	bch_keybuf_check_overlapping(&s->iop.c->moving_gc_keys, &start, &end);
 
 	down_read_non_owner(&dc->writeback_lock);
 	if (bch_keybuf_check_overlapping(&dc->writeback_keys, &start, &end)) {
@@ -1187,8 +1215,8 @@ static void cached_dev_write(struct cached_dev *dc, struct search *s)
 		 * We overlap with some dirty data undergoing background
 		 * writeback, force this write to writeback
 		 */
-		s->bypass	= false;
-		s->writeback	= true;
+		s->iop.bypass = false;
+		s->iop.writeback = true;
 	}
 
 	/*
@@ -1199,46 +1227,44 @@ static void cached_dev_write(struct cached_dev *dc, struct search *s)
 	 * so we still want to call it.
 	 */
 	if (bio->bi_rw & REQ_DISCARD)
-		s->bypass = true;
+		s->iop.bypass = true;
 
 	if (should_writeback(dc, s->orig_bio,
 			     cache_mode(dc, bio),
-			     s->bypass)) {
-		s->bypass = false;
-		s->writeback = true;
+			     s->iop.bypass)) {
+		s->iop.bypass = false;
+		s->iop.writeback = true;
 	}
 
-	trace_bcache_write(s->orig_bio, s->writeback, s->bypass);
-
-	if (s->bypass) {
-		s->cache_bio = s->orig_bio;
-		bio_get(s->cache_bio);
+	if (s->iop.bypass) {
+		s->iop.bio = s->orig_bio;
+		bio_get(s->iop.bio);
 
 		if (!(bio->bi_rw & REQ_DISCARD) ||
 		    blk_queue_discard(bdev_get_queue(dc->bdev)))
 			closure_bio_submit(bio, cl, s->d);
-	} else if (s->writeback) {
+	} else if (s->iop.writeback) {
 		bch_writeback_add(dc);
 
-		if (s->flush_journal) {
+		if (s->iop.flush_journal) {
 			/* Also need to send a flush to the backing device */
-			s->cache_bio = bio_clone_bioset(bio, GFP_NOIO,
-							dc->disk.bio_split);
+			s->iop.bio = bio_clone_bioset(bio, GFP_NOIO,
+						      dc->disk.bio_split);
 
 			bio->bi_size = 0;
 			bio->bi_vcnt = 0;
 			closure_bio_submit(bio, cl, s->d);
 		} else {
-			s->cache_bio = bio;
+			s->iop.bio = bio;
 		}
 	} else {
-		s->cache_bio = bio_clone_bioset(bio, GFP_NOIO,
-						dc->disk.bio_split);
+		s->iop.bio = bio_clone_bioset(bio, GFP_NOIO,
+					      dc->disk.bio_split);
 
 		closure_bio_submit(bio, cl, s->d);
 	}
 
-	closure_call(&s->btree, bch_data_insert, NULL, cl);
+	closure_call(&s->iop.cl, bch_data_insert, NULL, cl);
 	continue_at(cl, cached_dev_write_complete, NULL);
 }
 
@@ -1247,8 +1273,8 @@ static void cached_dev_nodata(struct closure *cl)
 	struct search *s = container_of(cl, struct search, cl);
 	struct bio *bio = &s->bio.bio;
 
-	if (s->flush_journal)
-		bch_journal_meta(s->c, cl);
+	if (s->iop.flush_journal)
+		bch_journal_meta(s->iop.c, cl);
 
 	/* If it's a flush, we send the flush to the backing device too */
 	closure_bio_submit(bio, cl, s->d);
@@ -1275,7 +1301,7 @@ static void cached_dev_make_request(struct request_queue *q, struct bio *bio)
 
 	if (cached_dev_get(dc)) {
 		s = search_alloc(bio, d);
-		trace_bcache_request_start(s, bio);
+		trace_bcache_request_start(s->d, bio);
 
 		if (!bio->bi_size) {
 			/*
@@ -1286,7 +1312,7 @@ static void cached_dev_make_request(struct request_queue *q, struct bio *bio)
 					      cached_dev_nodata,
 					      bcache_wq);
 		} else {
-			s->bypass = check_should_bypass(dc, s);
+			s->iop.bypass = check_should_bypass(dc, bio);
 
 			if (rw)
 				cached_dev_write(dc, s);
@@ -1376,8 +1402,8 @@ static void flash_dev_nodata(struct closure *cl)
 {
 	struct search *s = container_of(cl, struct search, cl);
 
-	if (s->flush_journal)
-		bch_journal_meta(s->c, cl);
+	if (s->iop.flush_journal)
+		bch_journal_meta(s->iop.c, cl);
 
 	continue_at(cl, search_free, NULL);
 }
@@ -1398,7 +1424,7 @@ static void flash_dev_make_request(struct request_queue *q, struct bio *bio)
 	cl = &s->cl;
 	bio = &s->bio.bio;
 
-	trace_bcache_request_start(s, bio);
+	trace_bcache_request_start(s->d, bio);
 
 	if (!bio->bi_size) {
 		/*
@@ -1409,17 +1435,17 @@ static void flash_dev_make_request(struct request_queue *q, struct bio *bio)
 				      flash_dev_nodata,
 				      bcache_wq);
 	} else if (rw) {
-		bch_keybuf_check_overlapping(&s->c->moving_gc_keys,
+		bch_keybuf_check_overlapping(&s->iop.c->moving_gc_keys,
 					&KEY(d->id, bio->bi_sector, 0),
 					&KEY(d->id, bio_end_sector(bio), 0));
 
-		s->bypass	= (bio->bi_rw & REQ_DISCARD) != 0;
-		s->writeback	= true;
-		s->cache_bio	= bio;
+		s->iop.bypass		= (bio->bi_rw & REQ_DISCARD) != 0;
+		s->iop.writeback	= true;
+		s->iop.bio		= bio;
 
-		closure_call(&s->btree, bch_data_insert, NULL, cl);
+		closure_call(&s->iop.cl, bch_data_insert, NULL, cl);
 	} else {
-		closure_call(&s->btree, cache_lookup, NULL, cl);
+		closure_call(&s->iop.cl, cache_lookup, NULL, cl);
 	}
 
 	continue_at(cl, search_free, NULL);
diff --git a/drivers/md/bcache/request.h b/drivers/md/bcache/request.h
index f0e930b4ca89..54d7de27356f 100644
--- a/drivers/md/bcache/request.h
+++ b/drivers/md/bcache/request.h
@@ -3,46 +3,25 @@
 
 #include <linux/cgroup.h>
 
-struct search {
-	/* Stack frame for bio_complete */
+struct data_insert_op {
 	struct closure		cl;
-	struct closure		btree;
-
-	struct bcache_device	*d;
 	struct cache_set	*c;
 	struct task_struct	*task;
-
-	struct bbio		bio;
-	struct bio		*orig_bio;
-	struct bio		*cache_miss;
-
-	/* Bio to be inserted into the cache */
-	struct bio		*cache_bio;
-	unsigned		cache_bio_sectors;
+	struct bio		*bio;
 
 	unsigned		inode;
+	uint16_t		write_prio;
+	short			error;
 
-	unsigned		recoverable:1;
-	unsigned		unaligned_bvec:1;
-
-	unsigned		write:1;
-	unsigned		writeback:1;
-
-	unsigned		csum:1;
 	unsigned		bypass:1;
+	unsigned		writeback:1;
 	unsigned		flush_journal:1;
+	unsigned		csum:1;
 
-	unsigned		insert_data_done:1;
 	unsigned		replace:1;
-	unsigned		insert_collision:1;
-
-	uint16_t		write_prio;
-
-	/* IO error returned to s->bio */
-	short			error;
-	unsigned long		start_time;
+	unsigned		replace_collision:1;
 
-	struct btree_op		op;
+	unsigned		insert_data_done:1;
 
 	/* Anything past this point won't get zeroed in search_alloc() */
 	struct keylist		insert_keys;
diff --git a/drivers/md/bcache/stats.c b/drivers/md/bcache/stats.c
index ea77263cf7ef..84d0782f702e 100644
--- a/drivers/md/bcache/stats.c
+++ b/drivers/md/bcache/stats.c
@@ -7,7 +7,6 @@
 #include "bcache.h"
 #include "stats.h"
 #include "btree.h"
-#include "request.h"
 #include "sysfs.h"
 
 /*
@@ -196,35 +195,36 @@ static void mark_cache_stats(struct cache_stat_collector *stats,
 			atomic_inc(&stats->cache_bypass_misses);
 }
 
-void bch_mark_cache_accounting(struct search *s, bool hit, bool bypass)
+void bch_mark_cache_accounting(struct cache_set *c, struct bcache_device *d,
+			       bool hit, bool bypass)
 {
-	struct cached_dev *dc = container_of(s->d, struct cached_dev, disk);
+	struct cached_dev *dc = container_of(d, struct cached_dev, disk);
 	mark_cache_stats(&dc->accounting.collector, hit, bypass);
-	mark_cache_stats(&s->c->accounting.collector, hit, bypass);
+	mark_cache_stats(&c->accounting.collector, hit, bypass);
 #ifdef CONFIG_CGROUP_BCACHE
 	mark_cache_stats(&(bch_bio_to_cgroup(s->orig_bio)->stats), hit, bypass);
 #endif
 }
 
-void bch_mark_cache_readahead(struct search *s)
+void bch_mark_cache_readahead(struct cache_set *c, struct bcache_device *d)
 {
-	struct cached_dev *dc = container_of(s->d, struct cached_dev, disk);
+	struct cached_dev *dc = container_of(d, struct cached_dev, disk);
 	atomic_inc(&dc->accounting.collector.cache_readaheads);
-	atomic_inc(&s->c->accounting.collector.cache_readaheads);
+	atomic_inc(&c->accounting.collector.cache_readaheads);
 }
 
-void bch_mark_cache_miss_collision(struct search *s)
+void bch_mark_cache_miss_collision(struct cache_set *c, struct bcache_device *d)
 {
-	struct cached_dev *dc = container_of(s->d, struct cached_dev, disk);
+	struct cached_dev *dc = container_of(d, struct cached_dev, disk);
 	atomic_inc(&dc->accounting.collector.cache_miss_collisions);
-	atomic_inc(&s->c->accounting.collector.cache_miss_collisions);
+	atomic_inc(&c->accounting.collector.cache_miss_collisions);
 }
 
-void bch_mark_sectors_bypassed(struct search *s, int sectors)
+void bch_mark_sectors_bypassed(struct cache_set *c, struct cached_dev *dc,
+			       int sectors)
 {
-	struct cached_dev *dc = container_of(s->d, struct cached_dev, disk);
 	atomic_add(sectors, &dc->accounting.collector.sectors_bypassed);
-	atomic_add(sectors, &s->c->accounting.collector.sectors_bypassed);
+	atomic_add(sectors, &c->accounting.collector.sectors_bypassed);
 }
 
 void bch_cache_accounting_init(struct cache_accounting *acc,
diff --git a/drivers/md/bcache/stats.h b/drivers/md/bcache/stats.h
index c7c7a8fd29fe..adbff141c887 100644
--- a/drivers/md/bcache/stats.h
+++ b/drivers/md/bcache/stats.h
@@ -38,7 +38,9 @@ struct cache_accounting {
 	struct cache_stats day;
 };
 
-struct search;
+struct cache_set;
+struct cached_dev;
+struct bcache_device;
 
 void bch_cache_accounting_init(struct cache_accounting *acc,
 			       struct closure *parent);
@@ -50,9 +52,10 @@ void bch_cache_accounting_clear(struct cache_accounting *acc);
 
 void bch_cache_accounting_destroy(struct cache_accounting *acc);
 
-void bch_mark_cache_accounting(struct search *s, bool hit, bool bypass);
-void bch_mark_cache_readahead(struct search *s);
-void bch_mark_cache_miss_collision(struct search *s);
-void bch_mark_sectors_bypassed(struct search *s, int sectors);
+void bch_mark_cache_accounting(struct cache_set *, struct bcache_device *,
+			       bool, bool);
+void bch_mark_cache_readahead(struct cache_set *, struct bcache_device *);
+void bch_mark_cache_miss_collision(struct cache_set *, struct bcache_device *);
+void bch_mark_sectors_bypassed(struct cache_set *, struct cached_dev *, int);
 
 #endif /* _BCACHE_STATS_H_ */
diff --git a/drivers/md/bcache/trace.c b/drivers/md/bcache/trace.c
index f7b6c197f90f..adbc3df17a80 100644
--- a/drivers/md/bcache/trace.c
+++ b/drivers/md/bcache/trace.c
@@ -1,6 +1,5 @@
 #include "bcache.h"
 #include "btree.h"
-#include "request.h"
 
 #include <linux/blktrace_api.h>
 #include <linux/module.h>
diff --git a/include/trace/events/bcache.h b/include/trace/events/bcache.h
index 5ebda976ea93..32c89b33c391 100644
--- a/include/trace/events/bcache.h
+++ b/include/trace/events/bcache.h
@@ -6,11 +6,9 @@
 
 #include <linux/tracepoint.h>
 
-struct search;
-
 DECLARE_EVENT_CLASS(bcache_request,
-	TP_PROTO(struct search *s, struct bio *bio),
-	TP_ARGS(s, bio),
+	TP_PROTO(struct bcache_device *d, struct bio *bio),
+	TP_ARGS(d, bio),
 
 	TP_STRUCT__entry(
 		__field(dev_t,		dev			)
@@ -24,8 +22,8 @@ DECLARE_EVENT_CLASS(bcache_request,
 
 	TP_fast_assign(
 		__entry->dev		= bio->bi_bdev->bd_dev;
-		__entry->orig_major	= s->d->disk->major;
-		__entry->orig_minor	= s->d->disk->first_minor;
+		__entry->orig_major	= d->disk->major;
+		__entry->orig_minor	= d->disk->first_minor;
 		__entry->sector		= bio->bi_sector;
 		__entry->orig_sector	= bio->bi_sector - 16;
 		__entry->nr_sector	= bio->bi_size >> 9;
@@ -79,13 +77,13 @@ DECLARE_EVENT_CLASS(btree_node,
 /* request.c */
 
 DEFINE_EVENT(bcache_request, bcache_request_start,
-	TP_PROTO(struct search *s, struct bio *bio),
-	TP_ARGS(s, bio)
+	TP_PROTO(struct bcache_device *d, struct bio *bio),
+	TP_ARGS(d, bio)
 );
 
 DEFINE_EVENT(bcache_request, bcache_request_end,
-	TP_PROTO(struct search *s, struct bio *bio),
-	TP_ARGS(s, bio)
+	TP_PROTO(struct bcache_device *d, struct bio *bio),
+	TP_ARGS(d, bio)
 );
 
 DECLARE_EVENT_CLASS(bcache_bio,
-- 
2.39.5