0/7 IO scheduler abstractions

index 5da56d4..1c44ce3 100644
@@ -14,7 +14,6 @@ #include <linux/module.h>
 #include <linux/slab.h>
 #include <linux/init.h>
 #include <linux/compiler.h>
-#include <linux/hash.h>
 #include <linux/rbtree.h>
 #include <linux/interrupt.h>
 

@@ -95,7 +94,6 @@ struct as_data {
 
 	struct as_rq *next_arq[2];	/* next in sort order */
 	sector_t last_sector[2];	/* last REQ_SYNC & REQ_ASYNC sectors */
-	struct hlist_head *hash;	/* request hash */
 
 	unsigned long exit_prob;	/* probability a task will exit while
 					   being waited on */

@@ -162,11 +160,6 @@ struct as_rq {
 	struct io_context *io_context;	/* The submitting task */
 
 	/*
-	 * request hash, key is the ending offset (for back merge lookup)
-	 */
-	struct hlist_node hash;
-
-	/*
 	 * expire fifo
 	 */
 	struct list_head fifo;

@@ -273,77 +266,6 @@ static void as_put_io_context(struct as_
 }
 
 /*
- * the back merge hash support functions
- */
-static const int as_hash_shift = 6;
-#define AS_HASH_BLOCK(sec)	((sec) >> 3)
-#define AS_HASH_FN(sec)		(hash_long(AS_HASH_BLOCK((sec)), as_hash_shift))
-#define AS_HASH_ENTRIES		(1 << as_hash_shift)
-#define rq_hash_key(rq)		((rq)->sector + (rq)->nr_sectors)
-
-static inline void __as_del_arq_hash(struct as_rq *arq)
-{
-	hlist_del_init(&arq->hash);
-}
-
-static inline void as_del_arq_hash(struct as_rq *arq)
-{
-	if (!hlist_unhashed(&arq->hash))
-		__as_del_arq_hash(arq);
-}
-
-static void as_add_arq_hash(struct as_data *ad, struct as_rq *arq)
-{
-	struct request *rq = arq->request;
-
-	BUG_ON(!hlist_unhashed(&arq->hash));
-
-	hlist_add_head(&arq->hash, &ad->hash[AS_HASH_FN(rq_hash_key(rq))]);
-}
-
-/*
- * move hot entry to front of chain
- */
-static inline void as_hot_arq_hash(struct as_data *ad, struct as_rq *arq)
-{
-	struct request *rq = arq->request;
-	struct hlist_head *head = &ad->hash[AS_HASH_FN(rq_hash_key(rq))];
-
-	if (hlist_unhashed(&arq->hash)) {
-		WARN_ON(1);
-		return;
-	}
-
-	if (&arq->hash != head->first) {
-		hlist_del(&arq->hash);
-		hlist_add_head(&arq->hash, head);
-	}
-}
-
-static struct request *as_find_arq_hash(struct as_data *ad, sector_t offset)
-{
-	struct hlist_head *hash_list = &ad->hash[AS_HASH_FN(offset)];
-	struct hlist_node *entry, *next;
-	struct as_rq *arq;
-
-	hlist_for_each_entry_safe(arq, entry, next, hash_list, hash) {
-		struct request *__rq = arq->request;
-
-		BUG_ON(hlist_unhashed(&arq->hash));
-
-		if (!rq_mergeable(__rq)) {
-			as_del_arq_hash(arq);
-			continue;
-		}
-
-		if (rq_hash_key(__rq) == offset)
-			return __rq;
-	}
-
-	return NULL;
-}
-
-/*
  * rb tree support functions
  */
 #define rb_entry_arq(node)	rb_entry((node), struct as_rq, rb_node)

@@ -1060,7 +982,6 @@ static void as_remove_queued_request(req
 		ad->next_arq[data_dir] = as_find_next_arq(ad, arq);
 
 	list_del_init(&arq->fifo);
-	as_del_arq_hash(arq);
 	as_del_arq_rb(ad, arq);
 }
 

@@ -1349,8 +1270,6 @@ static void as_add_request(request_queue
 	}
 
 	as_add_arq_rb(ad, arq);
-	if (rq_mergeable(arq->request))
-		as_add_arq_hash(ad, arq);
 
 	/*
 	 * set expire time (only used for reads) and add to fifo list

@@ -1428,42 +1347,17 @@ as_merge(request_queue_t *q, struct requ
 	struct as_data *ad = q->elevator->elevator_data;
 	sector_t rb_key = bio->bi_sector + bio_sectors(bio);
 	struct request *__rq;
-	int ret;
-
-	/*
-	 * see if the merge hash can satisfy a back merge
-	 */
-	__rq = as_find_arq_hash(ad, bio->bi_sector);
-	if (__rq) {
-		BUG_ON(__rq->sector + __rq->nr_sectors != bio->bi_sector);
-
-		if (elv_rq_merge_ok(__rq, bio)) {
-			ret = ELEVATOR_BACK_MERGE;
-			goto out;
-		}
-	}
 
 	/*
 	 * check for front merge
 	 */
 	__rq = as_find_arq_rb(ad, rb_key, bio_data_dir(bio));
-	if (__rq) {
-		BUG_ON(rb_key != rq_rb_key(__rq));
-
-		if (elv_rq_merge_ok(__rq, bio)) {
-			ret = ELEVATOR_FRONT_MERGE;
-			goto out;
-		}
+	if (__rq && elv_rq_merge_ok(__rq, bio)) {
+		*req = __rq;
+		return ELEVATOR_FRONT_MERGE;
 	}
 
 	return ELEVATOR_NO_MERGE;
-out:
-	if (ret) {
-		if (rq_mergeable(__rq))
-			as_hot_arq_hash(ad, RQ_DATA(__rq));
-	}
-	*req = __rq;
-	return ret;
 }
 
 static void as_merged_request(request_queue_t *q, struct request *req)

@@ -1472,12 +1366,6 @@ static void as_merged_request(request_qu
 	struct as_rq *arq = RQ_DATA(req);
 
 	/*
-	 * hash always needs to be repositioned, key is end sector
-	 */
-	as_del_arq_hash(arq);
-	as_add_arq_hash(ad, arq);
-
-	/*
 	 * if the merge was a front merge, we need to reposition request
 	 */
 	if (rq_rb_key(req) != arq->rb_key) {

@@ -1501,13 +1389,6 @@ static void as_merged_requests(request_q
 	BUG_ON(!arq);
 	BUG_ON(!anext);
 
-	/*
-	 * reposition arq (this is the merged request) in hash, and in rbtree
-	 * in case of a front merge
-	 */
-	as_del_arq_hash(arq);
-	as_add_arq_hash(ad, arq);
-
 	if (rq_rb_key(req) != arq->rb_key) {
 		as_del_arq_rb(ad, arq);
 		as_add_arq_rb(ad, arq);

@@ -1591,7 +1472,6 @@ static int as_set_request(request_queue_
 		arq->request = rq;
 		arq->state = AS_RQ_PRESCHED;
 		arq->io_context = NULL;
-		INIT_HLIST_NODE(&arq->hash);
 		INIT_LIST_HEAD(&arq->fifo);
 		rq->elevator_private = arq;
 		return 0;

@@ -1628,7 +1508,6 @@ static void as_exit_queue(elevator_t *e)
 
 	mempool_destroy(ad->arq_pool);
 	put_io_context(ad->io_context);
-	kfree(ad->hash);
 	kfree(ad);
 }
 

@@ -1639,7 +1518,6 @@ static void as_exit_queue(elevator_t *e)
 static void *as_init_queue(request_queue_t *q, elevator_t *e)
 {
 	struct as_data *ad;
-	int i;
 
 	if (!arq_pool)
 		return NULL;

@@ -1651,17 +1529,9 @@ static void *as_init_queue(request_queue
 
 	ad->q = q; /* Identify what queue the data belongs to */
 
-	ad->hash = kmalloc_node(sizeof(struct hlist_head)*AS_HASH_ENTRIES,
-				GFP_KERNEL, q->node);
-	if (!ad->hash) {
-		kfree(ad);
-		return NULL;
-	}
-
 	ad->arq_pool = mempool_create_node(BLKDEV_MIN_RQ, mempool_alloc_slab,
 				mempool_free_slab, arq_pool, q->node);
 	if (!ad->arq_pool) {
-		kfree(ad->hash);
 		kfree(ad);
 		return NULL;
 	}

@@ -1672,9 +1542,6 @@ static void *as_init_queue(request_queue
 	init_timer(&ad->antic_timer);
 	INIT_WORK(&ad->antic_work, as_work_handler, q);
 
-	for (i = 0; i < AS_HASH_ENTRIES; i++)
-		INIT_HLIST_HEAD(&ad->hash[i]);
-
 	INIT_LIST_HEAD(&ad->fifo_list[REQ_SYNC]);
 	INIT_LIST_HEAD(&ad->fifo_list[REQ_ASYNC]);
 	ad->sort_list[REQ_SYNC] = RB_ROOT;

171 lines not on this page · raw message

236 lines not on this page · raw message

226 lines not on this page · raw message

index 61d6b3c..84c7b1c 100644
@@ -281,6 +281,7 @@ static inline void rq_init(request_queue
 {
 	INIT_LIST_HEAD(&rq->queuelist);
 	INIT_LIST_HEAD(&rq->donelist);
+	INIT_HLIST_NODE(&rq->hash);
 
 	rq->errors = 0;
 	rq->rq_status = RQ_ACTIVE;

@@ -2665,6 +2666,7 @@ void __blk_put_request(request_queue_t *
 		int priv = req->flags & REQ_ELVPRIV;
 
 		BUG_ON(!list_empty(&req->queuelist));
+		BUG_ON(!hlist_unhashed(&req->hash));
 
 		blk_free_request(q, req);
 		freed_request(q, rw, priv);

index aafe827..9b23cbe 100644
@@ -148,6 +148,8 @@ struct request {
 	struct bio *bio;
 	struct bio *biotail;
 
+	struct hlist_node hash;	/* merge hash */
+
 	void *elevator_private;
 	void *completion_data;
 

@@ -679,21 +681,6 @@ static inline void blkdev_dequeue_reques
 }
 
 /*
- * This should be in elevator.h, but that requires pulling in rq and q
- */
-static inline void elv_dispatch_add_tail(struct request_queue *q,
-					 struct request *rq)
-{
-	if (q->last_merge == rq)
-		q->last_merge = NULL;
-	q->nr_sorted--;
-
-	q->end_sector = rq_end_sector(rq);
-	q->boundary_rq = rq;
-	list_add_tail(&rq->queuelist, &q->queue_head);
-}
-
-/*
  * Access functions for manipulating queue properties
  */
 extern request_queue_t *blk_init_queue_node(request_fn_proc *rfn,

index 1713ace..2c270e9 100644
@@ -82,12 +82,14 @@ struct elevator_queue
 	struct kobject kobj;
 	struct elevator_type *elevator_type;
 	struct mutex sysfs_lock;
+	struct hlist_head *hash;
 };
 
 /*
  * block elevator interface
  */
 extern void elv_dispatch_sort(request_queue_t *, struct request *);
+extern void elv_dispatch_add_tail(request_queue_t *, struct request *);
 extern void elv_add_request(request_queue_t *, struct request *, int, int);
 extern void __elv_add_request(request_queue_t *, struct request *, int, int);
 extern void elv_insert(request_queue_t *, struct request *, int);

index 1c44ce3..d677029 100644
@@ -336,7 +336,7 @@ static void as_add_arq_rb(struct as_data
 
 static inline void as_del_arq_rb(struct as_data *ad, struct as_rq *arq)
 {
-	if (!RB_EMPTY_NODE(&arq->rb_node)) {
+	if (RB_EMPTY_NODE(&arq->rb_node)) {
 		WARN_ON(1);
 		return;
 	}

@@ -1039,7 +1039,7 @@ static void as_move_to_dispatch(struct a
 	struct request *rq = arq->request;
 	const int data_dir = arq->is_sync;
 
-	BUG_ON(!RB_EMPTY_NODE(&arq->rb_node));
+	BUG_ON(RB_EMPTY_NODE(&arq->rb_node));
 
 	as_antic_stop(ad);
 	ad->antic_status = ANTIC_OFF;

index 8d5382e..344bc34 100644
@@ -133,7 +133,7 @@ #define RB_ROOT	(struct rb_root) { NULL,
 #define	rb_entry(ptr, type, member) container_of(ptr, type, member)
 
 #define RB_EMPTY_ROOT(root)	((root)->rb_node == NULL)
-#define RB_EMPTY_NODE(node)	(rb_parent(node) != node)
+#define RB_EMPTY_NODE(node)	(rb_parent(node) == node)
 #define RB_CLEAR_NODE(node)	(rb_set_parent(node, node))
 
 extern void rb_insert_color(struct rb_node *, struct rb_root *);

index 1e55ba1..48499c2 100644
@@ -322,6 +322,9 @@ struct rb_node *rb_next(struct rb_node *
 {
 	struct rb_node *parent;
 
+	if (rb_parent(node) == node)
+		return NULL;
+
 	/* If we have a right-hand child, go down and then left as far
 	   as we can. */
 	if (node->rb_right) {

@@ -348,6 +351,9 @@ struct rb_node *rb_prev(struct rb_node *
 {
 	struct rb_node *parent;
 
+	if (rb_parent(node) == node)
+		return NULL;
+
 	/* If we have a left-hand child, go down and then right as far
 	   as we can. */
 	if (node->rb_left) {

index d677029..000e776 100644
@@ -149,12 +149,6 @@ enum arq_state {
 };
 
 struct as_rq {
-	/*
-	 * rbtree index, key is the starting offset
-	 */
-	struct rb_node rb_node;
-	sector_t rb_key;
-
 	struct request *request;
 
 	struct io_context *io_context;	/* The submitting task */

@@ -268,101 +262,22 @@ static void as_put_io_context(struct as_
 /*
  * rb tree support functions
  */
-#define rb_entry_arq(node)	rb_entry((node), struct as_rq, rb_node)
 #define ARQ_RB_ROOT(ad, arq)	(&(ad)->sort_list[(arq)->is_sync])
-#define rq_rb_key(rq)		(rq)->sector
-
-/*
- * as_find_first_arq finds the first (lowest sector numbered) request
- * for the specified data_dir. Used to sweep back to the start of the disk
- * (1-way elevator) after we process the last (highest sector) request.
- */
-static struct as_rq *as_find_first_arq(struct as_data *ad, int data_dir)
-{
-	struct rb_node *n = ad->sort_list[data_dir].rb_node;
-
-	if (n == NULL)
-		return NULL;
-
-	for (;;) {
-		if (n->rb_left == NULL)
-			return rb_entry_arq(n);
-
-		n = n->rb_left;
-	}
-}
-
-/*
- * Add the request to the rb tree if it is unique.  If there is an alias (an
- * existing request against the same sector), which can happen when using
- * direct IO, then return the alias.
- */
-static struct as_rq *__as_add_arq_rb(struct as_data *ad, struct as_rq *arq)
-{
-	struct rb_node **p = &ARQ_RB_ROOT(ad, arq)->rb_node;
-	struct rb_node *parent = NULL;
-	struct as_rq *__arq;
-	struct request *rq = arq->request;
-
-	arq->rb_key = rq_rb_key(rq);
-
-	while (*p) {
-		parent = *p;
-		__arq = rb_entry_arq(parent);
-
-		if (arq->rb_key < __arq->rb_key)
-			p = &(*p)->rb_left;
-		else if (arq->rb_key > __arq->rb_key)
-			p = &(*p)->rb_right;
-		else
-			return __arq;
-	}
-
-	rb_link_node(&arq->rb_node, parent, p);
-	rb_insert_color(&arq->rb_node, ARQ_RB_ROOT(ad, arq));
 
-	return NULL;
-}
-
-static void as_add_arq_rb(struct as_data *ad, struct as_rq *arq)
+static void as_add_arq_rb(struct as_data *ad, struct request *rq)
 {
-	struct as_rq *alias;
+	struct as_rq *arq = RQ_DATA(rq);
+	struct request *alias;
 
-	while ((unlikely(alias = __as_add_arq_rb(ad, arq)))) {
-		as_move_to_dispatch(ad, alias);
+	while ((unlikely(alias = elv_rb_add(ARQ_RB_ROOT(ad, arq), rq)))) {
+		as_move_to_dispatch(ad, RQ_DATA(alias));
 		as_antic_stop(ad);
 	}
 }
 
-static inline void as_del_arq_rb(struct as_data *ad, struct as_rq *arq)
-{
-	if (RB_EMPTY_NODE(&arq->rb_node)) {
-		WARN_ON(1);
-		return;
-	}
-
-	rb_erase(&arq->rb_node, ARQ_RB_ROOT(ad, arq));
-	RB_CLEAR_NODE(&arq->rb_node);
-}
-
-static struct request *
-as_find_arq_rb(struct as_data *ad, sector_t sector, int data_dir)
+static inline void as_del_arq_rb(struct as_data *ad, struct request *rq)
 {
-	struct rb_node *n = ad->sort_list[data_dir].rb_node;
-	struct as_rq *arq;
-
-	while (n) {
-		arq = rb_entry_arq(n);
-
-		if (sector < arq->rb_key)
-			n = n->rb_left;
-		else if (sector > arq->rb_key)
-			n = n->rb_right;
-		else
-			return arq->request;
-	}
-
-	return NULL;
+	elv_rb_del(ARQ_RB_ROOT(ad, RQ_DATA(rq)), rq);
 }
 
 /*

@@ -455,32 +370,29 @@ as_choose_req(struct as_data *ad, struct
  * this with as_choose_req form the basis for how the scheduler chooses
  * what request to process next. Anticipation works on top of this.
  */
-static struct as_rq *as_find_next_arq(struct as_data *ad, struct as_rq *last)
+static struct as_rq *as_find_next_arq(struct as_data *ad, struct as_rq *arq)
 {
-	const int data_dir = last->is_sync;
-	struct as_rq *ret;
+	struct request *last = arq->request;
 	struct rb_node *rbnext = rb_next(&last->rb_node);
 	struct rb_node *rbprev = rb_prev(&last->rb_node);
-	struct as_rq *arq_next, *arq_prev;
+	struct as_rq *next = NULL, *prev = NULL;
 
-	BUG_ON(!RB_EMPTY_NODE(&last->rb_node));
+	BUG_ON(RB_EMPTY_NODE(&last->rb_node));
 
 	if (rbprev)
-		arq_prev = rb_entry_arq(rbprev);
-	else
-		arq_prev = NULL;
+		prev = RQ_DATA(rb_entry_rq(rbprev));
 
 	if (rbnext)
-		arq_next = rb_entry_arq(rbnext);
+		next = RQ_DATA(rb_entry_rq(rbnext));
 	else {
-		arq_next = as_find_first_arq(ad, data_dir);
-		if (arq_next == last)
-			arq_next = NULL;
-	}
+		const int data_dir = arq->is_sync;
 
-	ret = as_choose_req(ad,	arq_next, arq_prev);
+		rbnext = rb_first(&ad->sort_list[data_dir]);
+		if (rbnext && rbnext != &last->rb_node)
+			next = RQ_DATA(rb_entry_rq(rbnext));
+	}
 
-	return ret;
+	return as_choose_req(ad, next, prev);
 }
 
 /*

@@ -982,7 +894,7 @@ static void as_remove_queued_request(req
 		ad->next_arq[data_dir] = as_find_next_arq(ad, arq);
 
 	list_del_init(&arq->fifo);
-	as_del_arq_rb(ad, arq);
+	as_del_arq_rb(ad, rq);
 }
 
 /*

@@ -1039,7 +951,7 @@ static void as_move_to_dispatch(struct a
 	struct request *rq = arq->request;
 	const int data_dir = arq->is_sync;
 
-	BUG_ON(RB_EMPTY_NODE(&arq->rb_node));
+	BUG_ON(RB_EMPTY_NODE(&rq->rb_node));
 
 	as_antic_stop(ad);
 	ad->antic_status = ANTIC_OFF;

@@ -1064,8 +976,6 @@ static void as_move_to_dispatch(struct a
 	}
 	ad->ioc_finished = 0;
 
-	ad->next_arq[data_dir] = as_find_next_arq(ad, arq);
-
 	/*
 	 * take it off the sort and fifo list, add to dispatch queue
 	 */

@@ -1269,7 +1179,7 @@ static void as_add_request(request_queue
 		atomic_inc(&arq->io_context->aic->nr_queued);
 	}
 
-	as_add_arq_rb(ad, arq);
+	as_add_arq_rb(ad, rq);
 
 	/*
 	 * set expire time (only used for reads) and add to fifo list

@@ -1315,32 +1225,6 @@ static int as_queue_empty(request_queue_
 		&& list_empty(&ad->fifo_list[REQ_SYNC]);
 }
 
-static struct request *as_former_request(request_queue_t *q,
-					struct request *rq)
-{
-	struct as_rq *arq = RQ_DATA(rq);
-	struct rb_node *rbprev = rb_prev(&arq->rb_node);
-	struct request *ret = NULL;
-
-	if (rbprev)
-		ret = rb_entry_arq(rbprev)->request;
-
-	return ret;
-}
-
-static struct request *as_latter_request(request_queue_t *q,
-					struct request *rq)
-{
-	struct as_rq *arq = RQ_DATA(rq);
-	struct rb_node *rbnext = rb_next(&arq->rb_node);
-	struct request *ret = NULL;
-
-	if (rbnext)
-		ret = rb_entry_arq(rbnext)->request;
-
-	return ret;
-}
-
 static int
 as_merge(request_queue_t *q, struct request **req, struct bio *bio)
 {

@@ -1351,7 +1235,7 @@ as_merge(request_queue_t *q, struct requ
 	/*
 	 * check for front merge
 	 */
-	__rq = as_find_arq_rb(ad, rb_key, bio_data_dir(bio));
+	__rq = elv_rb_find(&ad->sort_list[bio_data_dir(bio)], rb_key);
 	if (__rq && elv_rq_merge_ok(__rq, bio)) {
 		*req = __rq;
 		return ELEVATOR_FRONT_MERGE;

@@ -1360,7 +1244,7 @@ as_merge(request_queue_t *q, struct requ
 	return ELEVATOR_NO_MERGE;
 }
 
-static void as_merged_request(request_queue_t *q, struct request *req)
+static void as_merged_request(request_queue_t *q, struct request *req, int type)
 {
 	struct as_data *ad = q->elevator->elevator_data;
 	struct as_rq *arq = RQ_DATA(req);

@@ -1368,9 +1252,8 @@ static void as_merged_request(request_qu
 	/*
 	 * if the merge was a front merge, we need to reposition request
 	 */
-	if (rq_rb_key(req) != arq->rb_key) {
-		as_del_arq_rb(ad, arq);
-		as_add_arq_rb(ad, arq);
+	if (type == ELEVATOR_FRONT_MERGE) {
+		elv_rb_reposition(ARQ_RB_ROOT(ad, arq), req);
 		/*
 		 * Note! At this stage of this and the next function, our next
 		 * request may not be optimal - eg the request may have "grown"

@@ -1382,18 +1265,12 @@ static void as_merged_request(request_qu
 static void as_merged_requests(request_queue_t *q, struct request *req,
 			 	struct request *next)
 {
-	struct as_data *ad = q->elevator->elevator_data;
 	struct as_rq *arq = RQ_DATA(req);
 	struct as_rq *anext = RQ_DATA(next);
 
 	BUG_ON(!arq);
 	BUG_ON(!anext);
 
-	if (rq_rb_key(req) != arq->rb_key) {
-		as_del_arq_rb(ad, arq);
-		as_add_arq_rb(ad, arq);
-	}
-
 	/*
 	 * if anext expires before arq, assign its expire time to arq
 	 * and move into anext position (anext will be deleted) in fifo

@@ -1468,7 +1345,6 @@ static int as_set_request(request_queue_
 
 	if (arq) {
 		memset(arq, 0, sizeof(*arq));
-		RB_CLEAR_NODE(&arq->rb_node);
 		arq->request = rq;
 		arq->state = AS_RQ_PRESCHED;
 		arq->io_context = NULL;

@@ -1654,8 +1530,8 @@ static struct elevator_type iosched_as =
 		.elevator_deactivate_req_fn = 	as_deactivate_request,
 		.elevator_queue_empty_fn =	as_queue_empty,
 		.elevator_completed_req_fn =	as_completed_request,
-		.elevator_former_req_fn =	as_former_request,
-		.elevator_latter_req_fn =	as_latter_request,
+		.elevator_former_req_fn =	elv_rb_former_request,
+		.elevator_latter_req_fn =	elv_rb_latter_request,
 		.elevator_set_req_fn =		as_set_request,
 		.elevator_put_req_fn =		as_put_request,
 		.elevator_may_queue_fn =	as_may_queue,

298 lines not on this page · raw message

279 lines not on this page · raw message

109 lines not on this page · raw message

index 84c7b1c..08c1615 100644
@@ -281,11 +281,12 @@ static inline void rq_init(request_queue
 {
 	INIT_LIST_HEAD(&rq->queuelist);
 	INIT_LIST_HEAD(&rq->donelist);
-	INIT_HLIST_NODE(&rq->hash);
 
 	rq->errors = 0;
 	rq->rq_status = RQ_ACTIVE;
 	rq->bio = rq->biotail = NULL;
+	INIT_HLIST_NODE(&rq->hash);
+	RB_CLEAR_NODE(&rq->rb_node);
 	rq->ioprio = 0;
 	rq->buffer = NULL;
 	rq->ref_count = 1;

@@ -2896,7 +2897,7 @@ static int __make_request(request_queue_
 			req->ioprio = ioprio_best(req->ioprio, prio);
 			drive_stat_acct(req, nr_sectors, 0);
 			if (!attempt_back_merge(q, req))
-				elv_merged_request(q, req);
+				elv_merged_request(q, req, el_ret);
 			goto out;
 
 		case ELEVATOR_FRONT_MERGE:

@@ -2923,7 +2924,7 @@ static int __make_request(request_queue_
 			req->ioprio = ioprio_best(req->ioprio, prio);
 			drive_stat_acct(req, nr_sectors, 0);
 			if (!attempt_front_merge(q, req))
-				elv_merged_request(q, req);
+				elv_merged_request(q, req, el_ret);
 			goto out;
 
 		/* ELV_NO_MERGE: elevator says don't/can't merge. */

index 9b23cbe..e296719 100644
@@ -149,6 +149,7 @@ struct request {
 	struct bio *biotail;
 
 	struct hlist_node hash;	/* merge hash */
+	struct rb_node rb_node;	/* sort/lookup */
 
 	void *elevator_private;
 	void *completion_data;

index 2c270e9..38f0f0d 100644
@@ -6,7 +6,7 @@ typedef int (elevator_merge_fn) (request
 
 typedef void (elevator_merge_req_fn) (request_queue_t *, struct request *, struct request *);
 
-typedef void (elevator_merged_fn) (request_queue_t *, struct request *);
+typedef void (elevator_merged_fn) (request_queue_t *, struct request *, int);
 
 typedef int (elevator_dispatch_fn) (request_queue_t *, int);
 

@@ -96,7 +96,7 @@ extern void elv_insert(request_queue_t *
 extern int elv_merge(request_queue_t *, struct request **, struct bio *);
 extern void elv_merge_requests(request_queue_t *, struct request *,
 			       struct request *);
-extern void elv_merged_request(request_queue_t *, struct request *);
+extern void elv_merged_request(request_queue_t *, struct request *, int);
 extern void elv_dequeue_request(request_queue_t *, struct request *);
 extern void elv_requeue_request(request_queue_t *, struct request *);
 extern int elv_queue_empty(request_queue_t *);

@@ -127,6 +127,20 @@ extern void elevator_exit(elevator_t *);
 extern int elv_rq_merge_ok(struct request *, struct bio *);
 
 /*
+ * Helper functions.
+ */
+extern struct request *elv_rb_former_request(request_queue_t *, struct request *);
+extern struct request *elv_rb_latter_request(request_queue_t *, struct request *);
+
+/*
+ * rb support functions.
+ */
+extern struct request *elv_rb_add(struct rb_root *, struct request *);
+extern void elv_rb_del(struct rb_root *, struct request *);
+extern struct request *elv_rb_find(struct rb_root *, sector_t);
+extern void elv_rb_reposition(struct rb_root *, struct request *);
+
+/*
  * Return values from elevator merger
  */
 #define ELEVATOR_NO_MERGE	0

@@ -151,5 +165,6 @@ enum {
 };
 
 #define rq_end_sector(rq)	((rq)->sector + (rq)->nr_sectors)
+#define rb_entry_rq(node)	rb_entry((node), struct request, rb_node)
 
 #endif

index 95bc2e8..54cc8e3 100644
@@ -42,7 +42,6 @@ #define CFQ_QHASH_ENTRIES	(1 << CFQ_QHAS
 #define list_entry_qhash(entry)	hlist_entry((entry), struct cfq_queue, cfq_hash)
 
 #define list_entry_cfqq(ptr)	list_entry((ptr), struct cfq_queue, cfq_list)
-#define list_entry_fifo(ptr)	list_entry((ptr), struct request, queuelist)
 
 #define RQ_DATA(rq)		(rq)->elevator_private
 

@@ -839,7 +838,7 @@ static inline struct cfq_rq *cfq_check_f
 	if (!list_empty(&cfqq->fifo)) {
 		int fifo = cfq_cfqq_class_sync(cfqq);
 
-		crq = RQ_DATA(list_entry_fifo(cfqq->fifo.next));
+		crq = RQ_DATA(rq_entry_fifo(cfqq->fifo.next));
 		rq = crq->request;
 		if (time_after(jiffies, rq->start_time + cfqd->cfq_fifo_expire[fifo])) {
 			cfq_mark_cfqq_fifo_expire(cfqq);

index ce86b1f..15da9b2 100644
@@ -37,7 +37,7 @@ struct deadline_data {
 	/*
 	 * next in sort order. read, write or both are NULL
 	 */
-	struct deadline_rq *next_drq[2];
+	struct request *next_rq[2];
 	unsigned int batching;		/* number of sequential requests made */
 	sector_t last_sector;		/* head position */
 	unsigned int starved;		/* times reads have starved writes */

@@ -49,34 +49,14 @@ struct deadline_data {
 	int fifo_batch;
 	int writes_starved;
 	int front_merges;
-
-	mempool_t *drq_pool;
 };
 
-/*
- * pre-request data.
- */
-struct deadline_rq {
-	struct request *request;
-
-	/*
-	 * expire fifo
-	 */
-	struct list_head fifo;
-	unsigned long expires;
-};
-
-static void deadline_move_request(struct deadline_data *dd, struct deadline_rq *drq);
-
-static kmem_cache_t *drq_pool;
-
-#define RQ_DATA(rq)	((struct deadline_rq *) (rq)->elevator_private)
+static void deadline_move_request(struct deadline_data *, struct request *);
 
 #define RQ_RB_ROOT(dd, rq)	(&(dd)->sort_list[rq_data_dir((rq))])
-#define DRQ_RB_ROOT(dd, drq)	RQ_RB_ROOT((drq)->request)
 
 static void
-deadline_add_drq_rb(struct deadline_data *dd, struct request *rq)
+deadline_add_rq_rb(struct deadline_data *dd, struct request *rq)
 {
 	struct rb_root *root = RQ_RB_ROOT(dd, rq);
 	struct request *__alias;

@@ -84,45 +64,43 @@ deadline_add_drq_rb(struct deadline_data
 retry:
 	__alias = elv_rb_add(root, rq);
 	if (unlikely(__alias)) {
-		deadline_move_request(dd, RQ_DATA(__alias));
+		deadline_move_request(dd, __alias);
 		goto retry;
 	}
 }
 
 static inline void
-deadline_del_drq_rb(struct deadline_data *dd, struct deadline_rq *drq)
+deadline_del_rq_rb(struct deadline_data *dd, struct request *rq)
 {
-	struct request *rq = drq->request;
 	const int data_dir = rq_data_dir(rq);
 
-	if (dd->next_drq[data_dir] == drq) {
+	if (dd->next_rq[data_dir] == rq) {
 		struct rb_node *rbnext = rb_next(&rq->rb_node);
 
-		dd->next_drq[data_dir] = NULL;
+		dd->next_rq[data_dir] = NULL;
 		if (rbnext)
-			dd->next_drq[data_dir] = RQ_DATA(rb_entry_rq(rbnext));
+			dd->next_rq[data_dir] = rb_entry_rq(rbnext);
 	}
 
 	elv_rb_del(RQ_RB_ROOT(dd, rq), rq);
 }
 
 /*
- * add drq to rbtree and fifo
+ * add rq to rbtree and fifo
  */
 static void
 deadline_add_request(struct request_queue *q, struct request *rq)
 {
 	struct deadline_data *dd = q->elevator->elevator_data;
-	struct deadline_rq *drq = RQ_DATA(rq);
-	const int data_dir = rq_data_dir(drq->request);
+	const int data_dir = rq_data_dir(rq);
 
-	deadline_add_drq_rb(dd, rq);
+	deadline_add_rq_rb(dd, rq);
 
 	/*
 	 * set expire time (only used for reads) and add to fifo list
 	 */
-	drq->expires = jiffies + dd->fifo_expire[data_dir];
-	list_add_tail(&drq->fifo, &dd->fifo_list[data_dir]);
+	rq_set_fifo_time(rq, jiffies + dd->fifo_expire[data_dir]);
+	list_add_tail(&rq->queuelist, &dd->fifo_list[data_dir]);
 }
 
 /*

@@ -130,11 +108,10 @@ deadline_add_request(struct request_queu
  */
 static void deadline_remove_request(request_queue_t *q, struct request *rq)
 {
-	struct deadline_rq *drq = RQ_DATA(rq);
 	struct deadline_data *dd = q->elevator->elevator_data;
 
-	list_del_init(&drq->fifo);
-	deadline_del_drq_rb(dd, drq);
+	rq_fifo_clear(rq);
+	deadline_del_rq_rb(dd, rq);
 }
 
 static int

@@ -183,20 +160,14 @@ static void
 deadline_merged_requests(request_queue_t *q, struct request *req,
 			 struct request *next)
 {
-	struct deadline_rq *drq = RQ_DATA(req);
-	struct deadline_rq *dnext = RQ_DATA(next);
-
-	BUG_ON(!drq);
-	BUG_ON(!dnext);
-
 	/*
-	 * if dnext expires before drq, assign its expire time to drq
-	 * and move into dnext position (dnext will be deleted) in fifo
+	 * if next expires before rq, assign its expire time to rq
+	 * and move into next position (next will be deleted) in fifo
 	 */
-	if (!list_empty(&drq->fifo) && !list_empty(&dnext->fifo)) {
-		if (time_before(dnext->expires, drq->expires)) {
-			list_move(&drq->fifo, &dnext->fifo);
-			drq->expires = dnext->expires;
+	if (!list_empty(&req->queuelist) && !list_empty(&next->queuelist)) {
+		if (time_before(rq_fifo_time(next), rq_fifo_time(req))) {
+			list_move(&req->queuelist, &next->queuelist);
+			rq_set_fifo_time(req, rq_fifo_time(next));
 		}
 	}
 

@@ -210,53 +181,50 @@ deadline_merged_requests(request_queue_t
  * move request from sort list to dispatch queue.
  */
 static inline void
-deadline_move_to_dispatch(struct deadline_data *dd, struct deadline_rq *drq)
+deadline_move_to_dispatch(struct deadline_data *dd, struct request *rq)
 {
-	request_queue_t *q = drq->request->q;
+	request_queue_t *q = rq->q;
 
-	deadline_remove_request(q, drq->request);
-	elv_dispatch_add_tail(q, drq->request);
+	deadline_remove_request(q, rq);
+	elv_dispatch_add_tail(q, rq);
 }
 
 /*
  * move an entry to dispatch queue
  */
 static void
-deadline_move_request(struct deadline_data *dd, struct deadline_rq *drq)
+deadline_move_request(struct deadline_data *dd, struct request *rq)
 {
-	struct request *rq = drq->request;
 	const int data_dir = rq_data_dir(rq);
 	struct rb_node *rbnext = rb_next(&rq->rb_node);
 
-	dd->next_drq[READ] = NULL;
-	dd->next_drq[WRITE] = NULL;
+	dd->next_rq[READ] = NULL;
+	dd->next_rq[WRITE] = NULL;
 
 	if (rbnext)
-		dd->next_drq[data_dir] = RQ_DATA(rb_entry_rq(rbnext));
+		dd->next_rq[data_dir] = rb_entry_rq(rbnext);
 	
-	dd->last_sector = drq->request->sector + drq->request->nr_sectors;
+	dd->last_sector = rq->sector + rq->nr_sectors;
 
 	/*
 	 * take it off the sort and fifo list, move
 	 * to dispatch queue
 	 */
-	deadline_move_to_dispatch(dd, drq);
+	deadline_move_to_dispatch(dd, rq);
 }
 
-#define list_entry_fifo(ptr)	list_entry((ptr), struct deadline_rq, fifo)
-
 /*
  * deadline_check_fifo returns 0 if there are no expired reads on the fifo,
  * 1 otherwise. Requires !list_empty(&dd->fifo_list[data_dir])
  */
 static inline int deadline_check_fifo(struct deadline_data *dd, int ddir)
 {
-	struct deadline_rq *drq = list_entry_fifo(dd->fifo_list[ddir].next);
+	struct request *rq = rq_entry_fifo(dd->fifo_list[ddir].next);
 
 	/*
-	 * drq is expired!
+	 * rq is expired!
 	 */
-	if (time_after(jiffies, drq->expires))
+	if (time_after(jiffies, rq_fifo_time(rq)))
 		return 1;
 
 	return 0;

@@ -271,21 +239,21 @@ static int deadline_dispatch_requests(re
 	struct deadline_data *dd = q->elevator->elevator_data;
 	const int reads = !list_empty(&dd->fifo_list[READ]);
 	const int writes = !list_empty(&dd->fifo_list[WRITE]);
-	struct deadline_rq *drq;
+	struct request *rq;
 	int data_dir;
 
 	/*
 	 * batches are currently reads XOR writes
 	 */
-	if (dd->next_drq[WRITE])
-		drq = dd->next_drq[WRITE];
+	if (dd->next_rq[WRITE])
+		rq = dd->next_rq[WRITE];
 	else
-		drq = dd->next_drq[READ];
+		rq = dd->next_rq[READ];
 
-	if (drq) {
+	if (rq) {
 		/* we have a "next request" */
 		
-		if (dd->last_sector != drq->request->sector)
+		if (dd->last_sector != rq->sector)
 			/* end the batch on a non sequential request */
 			dd->batching += dd->fifo_batch;
 		

@@ -334,34 +302,33 @@ dispatch_find_request:
 	if (deadline_check_fifo(dd, data_dir)) {
 		/* An expired request exists - satisfy it */
 		dd->batching = 0;
-		drq = list_entry_fifo(dd->fifo_list[data_dir].next);
+		rq = rq_entry_fifo(dd->fifo_list[data_dir].next);
 		
-	} else if (dd->next_drq[data_dir]) {
+	} else if (dd->next_rq[data_dir]) {
 		/*
 		 * The last req was the same dir and we have a next request in
 		 * sort order. No expired requests so continue on from here.
 		 */
-		drq = dd->next_drq[data_dir];
+		rq = dd->next_rq[data_dir];
 	} else {
-		struct rb_node *n;
-
+		struct rb_node *node;
 		/*
 		 * The last req was the other direction or we have run out of
 		 * higher-sectored requests. Go back to the lowest sectored
 		 * request (1 way elevator) and start a new batch.
 		 */
 		dd->batching = 0;
-		n = rb_first(&dd->sort_list[data_dir]);
-		if (n)
-			drq = RQ_DATA(rb_entry_rq(n));
+		node = rb_first(&dd->sort_list[data_dir]);
+		if (node)
+			rq = rb_entry_rq(node);
 	}
 
 dispatch_request:
 	/*
-	 * drq is the selected appropriate request.
+	 * rq is the selected appropriate request.
 	 */
 	dd->batching++;
-	deadline_move_request(dd, drq);
+	deadline_move_request(dd, rq);
 
 	return 1;
 }

@@ -381,33 +348,21 @@ static void deadline_exit_queue(elevator
 	BUG_ON(!list_empty(&dd->fifo_list[READ]));
 	BUG_ON(!list_empty(&dd->fifo_list[WRITE]));
 
-	mempool_destroy(dd->drq_pool);
 	kfree(dd);
 }
 
 /*
- * initialize elevator private data (deadline_data), and alloc a drq for
- * each request on the free lists
+ * initialize elevator private data (deadline_data).
  */
 static void *deadline_init_queue(request_queue_t *q, elevator_t *e)
 {
 	struct deadline_data *dd;
 
-	if (!drq_pool)
-		return NULL;
-
 	dd = kmalloc_node(sizeof(*dd), GFP_KERNEL, q->node);
 	if (!dd)
 		return NULL;
 	memset(dd, 0, sizeof(*dd));
 
-	dd->drq_pool = mempool_create_node(BLKDEV_MIN_RQ, mempool_alloc_slab,
-					mempool_free_slab, drq_pool, q->node);
-	if (!dd->drq_pool) {
-		kfree(dd);
-		return NULL;
-	}
-
 	INIT_LIST_HEAD(&dd->fifo_list[READ]);
 	INIT_LIST_HEAD(&dd->fifo_list[WRITE]);
 	dd->sort_list[READ] = RB_ROOT;

@@ -420,36 +375,6 @@ static void *deadline_init_queue(request
 	return dd;
 }
 
-static void deadline_put_request(request_queue_t *q, struct request *rq)
-{
-	struct deadline_data *dd = q->elevator->elevator_data;
-	struct deadline_rq *drq = RQ_DATA(rq);
-
-	mempool_free(drq, dd->drq_pool);
-	rq->elevator_private = NULL;
-}
-
-static int
-deadline_set_request(request_queue_t *q, struct request *rq, struct bio *bio,
-		     gfp_t gfp_mask)
-{
-	struct deadline_data *dd = q->elevator->elevator_data;
-	struct deadline_rq *drq;
-
-	drq = mempool_alloc(dd->drq_pool, gfp_mask);
-	if (drq) {
-		memset(drq, 0, sizeof(*drq));
-		drq->request = rq;
-
-		INIT_LIST_HEAD(&drq->fifo);
-
-		rq->elevator_private = drq;
-		return 0;
-	}
-
-	return 1;
-}
-
 /*
  * sysfs parts below
  */

@@ -531,8 +456,6 @@ static struct elevator_type iosched_dead
 		.elevator_queue_empty_fn =	deadline_queue_empty,
 		.elevator_former_req_fn =	elv_rb_former_request,
 		.elevator_latter_req_fn =	elv_rb_latter_request,
-		.elevator_set_req_fn =		deadline_set_request,
-		.elevator_put_req_fn = 		deadline_put_request,
 		.elevator_init_fn =		deadline_init_queue,
 		.elevator_exit_fn =		deadline_exit_queue,
 	},

@@ -544,24 +467,11 @@ static struct elevator_type iosched_dead
 
 static int __init deadline_init(void)
 {
-	int ret;
-
-	drq_pool = kmem_cache_create("deadline_drq", sizeof(struct deadline_rq),
-				     0, 0, NULL, NULL);
-
-	if (!drq_pool)
-		return -ENOMEM;
-
-	ret = elv_register(&iosched_deadline);
-	if (ret)
-		kmem_cache_destroy(drq_pool);
-
-	return ret;
+	return elv_register(&iosched_deadline);
 }
 
 static void __exit deadline_exit(void)
 {
-	kmem_cache_destroy(drq_pool);
 	elv_unregister(&iosched_deadline);
 }
 

index 38f0f0d..6e1c903 100644
@@ -167,4 +167,16 @@ enum {
 #define rq_end_sector(rq)	((rq)->sector + (rq)->nr_sectors)
 #define rb_entry_rq(node)	rb_entry((node), struct request, rb_node)
 
+/*
+ * Hack to reuse the donelist list_head as the fifo time holder while
+ * the request is in the io scheduler. Saves an unsigned long in rq.
+ */
+#define rq_fifo_time(rq)	((unsigned long) (rq)->donelist.next)
+#define rq_set_fifo_time(rq,exp)	((rq)->donelist.next = (void *) (exp))
+#define rq_entry_fifo(ptr)	list_entry((ptr), struct request, queuelist)
+#define rq_fifo_clear(rq)	do {		\
+	list_del_init(&(rq)->queuelist);	\
+	INIT_LIST_HEAD(&(rq)->donelist);	\
+	} while (0)
+ 
 #endif

index 000e776..66bd0bc 100644
@@ -153,12 +153,6 @@ struct as_rq {
 
 	struct io_context *io_context;	/* The submitting task */
 
-	/*
-	 * expire fifo
-	 */
-	struct list_head fifo;
-	unsigned long expires;
-
 	unsigned int is_sync;
 	enum arq_state state;
 };

@@ -893,7 +887,7 @@ static void as_remove_queued_request(req
 	if (ad->next_arq[data_dir] == arq)
 		ad->next_arq[data_dir] = as_find_next_arq(ad, arq);
 
-	list_del_init(&arq->fifo);
+	rq_fifo_clear(rq);
 	as_del_arq_rb(ad, rq);
 }
 

@@ -907,7 +901,7 @@ static void as_remove_queued_request(req
  */
 static int as_fifo_expired(struct as_data *ad, int adir)
 {
-	struct as_rq *arq;
+	struct request *rq;
 	long delta_jif;
 
 	delta_jif = jiffies - ad->last_check_fifo[adir];

@@ -921,9 +915,9 @@ static int as_fifo_expired(struct as_dat
 	if (list_empty(&ad->fifo_list[adir]))
 		return 0;
 
-	arq = list_entry_fifo(ad->fifo_list[adir].next);
+	rq = rq_entry_fifo(ad->fifo_list[adir].next);
 
-	return time_after(jiffies, arq->expires);
+	return time_after(jiffies, rq_fifo_time(rq));
 }
 
 /*

@@ -1087,7 +1081,7 @@ static int as_dispatch_request(request_q
 			ad->changed_batch = 1;
 		}
 		ad->batch_data_dir = REQ_SYNC;
-		arq = list_entry_fifo(ad->fifo_list[ad->batch_data_dir].next);
+		arq = RQ_DATA(rq_entry_fifo(ad->fifo_list[REQ_SYNC].next));
 		ad->last_check_fifo[ad->batch_data_dir] = jiffies;
 		goto dispatch_request;
 	}

@@ -1127,8 +1121,7 @@ dispatch_request:
 
 	if (as_fifo_expired(ad, ad->batch_data_dir)) {
 fifo_expired:
-		arq = list_entry_fifo(ad->fifo_list[ad->batch_data_dir].next);
-		BUG_ON(arq == NULL);
+		arq = RQ_DATA(rq_entry_fifo(ad->fifo_list[ad->batch_data_dir].next));
 	}
 
 	if (ad->changed_batch) {

@@ -1184,8 +1177,8 @@ static void as_add_request(request_queue
 	/*
 	 * set expire time (only used for reads) and add to fifo list
 	 */
-	arq->expires = jiffies + ad->fifo_expire[data_dir];
-	list_add_tail(&arq->fifo, &ad->fifo_list[data_dir]);
+	rq_set_fifo_time(rq, jiffies + ad->fifo_expire[data_dir]);
+	list_add_tail(&rq->queuelist, &ad->fifo_list[data_dir]);
 
 	as_update_arq(ad, arq); /* keep state machine up to date */
 	arq->state = AS_RQ_QUEUED;

@@ -1275,10 +1268,10 @@ static void as_merged_requests(request_q
 	 * if anext expires before arq, assign its expire time to arq
 	 * and move into anext position (anext will be deleted) in fifo
 	 */
-	if (!list_empty(&arq->fifo) && !list_empty(&anext->fifo)) {
-		if (time_before(anext->expires, arq->expires)) {
-			list_move(&arq->fifo, &anext->fifo);
-			arq->expires = anext->expires;
+	if (!list_empty(&req->queuelist) && !list_empty(&next->queuelist)) {
+		if (time_before(rq_fifo_time(next), rq_fifo_time(req))) {
+			list_move(&req->queuelist, &next->queuelist);
+			rq_set_fifo_time(req, rq_fifo_time(next));
 			/*
 			 * Don't copy here but swap, because when anext is
 			 * removed below, it must contain the unused context

@@ -1348,7 +1341,6 @@ static int as_set_request(request_queue_
 		arq->request = rq;
 		arq->state = AS_RQ_PRESCHED;
 		arq->io_context = NULL;
-		INIT_LIST_HEAD(&arq->fifo);
 		rq->elevator_private = arq;
 		return 0;
 	}