This commit was manufactured by cvs2svn to create tag
[linux-2.6.git] / drivers / block / cfq-iosched.c
index 7b45a80..068f4ea 100644 (file)
@@ -6,18 +6,6 @@
  *  Based on ideas from a previously unfinished io
  *  scheduler (round robin per-process disk scheduling) and Andrea Arcangeli.
  *
- *  IO priorities are supported, from 0% to 100% in 5% increments. Both of
- *  those values have special meaning - 0% class is allowed to do io if
- *  noone else wants to use the disk. 100% is considered real-time io, and
- *  always get priority. Default process io rate is 95%. In absence of other
- *  io, a class may consume 100% disk bandwidth regardless. Withing a class,
- *  bandwidth is distributed equally among the citizens.
- *
- * TODO:
- *     - cfq_select_requests() needs some work for 5-95% io
- *     - barriers not supported
- *     - export grace periods in ms, not jiffies
- *
  *  Copyright (C) 2003 Jens Axboe <axboe@suse.de>
  */
 #include <linux/kernel.h>
 #include <linux/hash.h>
 #include <linux/rbtree.h>
 #include <linux/mempool.h>
-#include <asm/div64.h>
-
-#if IOPRIO_NR > BITS_PER_LONG
-#error Cannot support this many io priority levels
-#endif
-
-#define LIMIT_DEBUG   1
 
 /*
  * tunables
  */
-static int cfq_quantum = 6;
-static int cfq_quantum_io = 256;
-static int cfq_idle_quantum = 1;
-static int cfq_idle_quantum_io = 64;
-static int cfq_queued = 4;
-static int cfq_grace_rt = HZ / 100 ?: 1;
-static int cfq_grace_idle = HZ / 10;
+static int cfq_quantum = 4;
+static int cfq_queued = 8;
 
 #define CFQ_QHASH_SHIFT                6
 #define CFQ_QHASH_ENTRIES      (1 << CFQ_QHASH_SHIFT)
-#define list_entry_qhash(entry)        hlist_entry((entry), struct cfq_queue, cfq_hash)
+#define list_entry_qhash(entry)        list_entry((entry), struct cfq_queue, cfq_hash)
 
 #define CFQ_MHASH_SHIFT                8
 #define CFQ_MHASH_BLOCK(sec)   ((sec) >> 3)
 #define CFQ_MHASH_ENTRIES      (1 << CFQ_MHASH_SHIFT)
 #define CFQ_MHASH_FN(sec)      (hash_long(CFQ_MHASH_BLOCK((sec)),CFQ_MHASH_SHIFT))
+#define ON_MHASH(crq)          !list_empty(&(crq)->hash)
 #define rq_hash_key(rq)                ((rq)->sector + (rq)->nr_sectors)
-#define list_entry_hash(ptr)   hlist_entry((ptr), struct cfq_rq, hash)
+#define list_entry_hash(ptr)   list_entry((ptr), struct cfq_rq, hash)
 
 #define list_entry_cfqq(ptr)   list_entry((ptr), struct cfq_queue, cfq_list)
-#define list_entry_prio(ptr)   list_entry((ptr), struct cfq_rq, prio_list)
-
-#define cfq_account_io(crq)    \
-       ((crq)->ioprio != IOPRIO_IDLE && (crq)->ioprio != IOPRIO_RT)
-
-/* define to be 50 ms for now; make tunable later */
-#define CFQ_EPOCH              50000
-/* Needs to be made tunable right away, in MiB/s */
-#define CFQ_DISKBW             10       
-/* Temporary global limit, as percent of available b/w, for each "class" */
-#define CFQ_TEMPLIM            10
-
-/*
- * defines how we distribute bandwidth (can be tgid, uid, etc)
- */
-
-/* FIXME: change hash_key to be sizeof(void *) rather than sizeof(int) 
- * otherwise the cast of cki_tsk_icls will not work reliably on 64-bit arches.
- * OR, change cki_tsk_icls to return ints (will need another id space to be 
- * managed)
- */
-
-#if defined(CONFIG_CKRM_RES_BLKIO) || defined(CONFIG_CKRM_RES_BLKIO_MODULE)
-extern inline void *cki_hash_key(struct task_struct *tsk);
-extern inline int cki_ioprio(struct task_struct *tsk);
-#define cfq_hash_key(current)   ((int)cki_hash_key((current)))
-#define cfq_ioprio(current)    (cki_ioprio((current)))
-
-#else
-#define cfq_hash_key(current)  ((current)->tgid)
-
-/*
- * move to io_context
- */
-#define cfq_ioprio(current)    ((current)->ioprio)
-#endif
 
-#define CFQ_WAIT_RT    0
-#define CFQ_WAIT_NORM  1
+#define RQ_DATA(rq)            ((struct cfq_rq *) (rq)->elevator_private)
 
 static kmem_cache_t *crq_pool;
 static kmem_cache_t *cfq_pool;
 static mempool_t *cfq_mpool;
 
-/*
- * defines an io priority level
- */
-struct io_prio_data {
-       struct list_head rr_list;
-       int busy_queues;
-       int busy_rq;
-       unsigned long busy_sectors;
-       
-       /* requests, sectors and queues 
-         * added(in),dispatched/deleted(out) 
-        * at this priority level. 
-        */
-       atomic_t cum_rq_in,cum_rq_out;              
-       atomic_t cum_sectors_in,cum_sectors_out;    
-       atomic_t cum_queues_in,cum_queues_out;
-
-#ifdef LIMIT_DEBUG
-       int nskip;
-       unsigned long navsec;
-       unsigned long csectorate;
-       unsigned long lsectorate;
-#endif
-
-       struct list_head prio_list;
-       int last_rq;
-       int last_sectors;
-};
-
-/*
- * per-request queue structure
- */
 struct cfq_data {
        struct list_head rr_list;
        struct list_head *dispatch;
-       struct hlist_head *cfq_hash;
-       struct hlist_head *crq_hash;
-       mempool_t *crq_pool;
+       struct list_head *cfq_hash;
 
-       struct io_prio_data cid[IOPRIO_NR];
+       struct list_head *crq_hash;
 
-       /*
-        * total number of busy queues and requests
-        */
-       int busy_rq;
-       int busy_queues;
-       unsigned long busy_sectors;
+       unsigned int busy_queues;
+       unsigned int max_queued;
 
+       mempool_t *crq_pool;
 
        request_queue_t *queue;
-       unsigned long rq_starved_mask;
-
-       /*
-        * grace period handling
-        */
-       struct timer_list timer;
-       unsigned long wait_end;
-       unsigned long flags;
-       struct work_struct work;
 
        /*
         * tunables
         */
        unsigned int cfq_quantum;
-       unsigned int cfq_quantum_io;
-       unsigned int cfq_idle_quantum;
-       unsigned int cfq_idle_quantum_io;
        unsigned int cfq_queued;
-       unsigned int cfq_grace_rt;
-       unsigned int cfq_grace_idle;
-
-       unsigned long cfq_epoch;        /* duration for limit enforcement */
-       unsigned long cfq_epochsectors; /* max sectors dispatchable/epoch */
 };
 
-/*
- * per-class structure
- */
 struct cfq_queue {
+       struct list_head cfq_hash;
        struct list_head cfq_list;
-       struct hlist_node cfq_hash;
-       int hash_key;
        struct rb_root sort_list;
+       int pid;
        int queued[2];
-       int ioprio;
-
-       unsigned long avsec;            /* avg sectors dispatched/epoch */
-       unsigned long long lastime;     /* timestamp of last request served */
-       unsigned long sectorate;        /* limit for sectors served/epoch */
-       int skipped;                    /* queue skipped at last dispatch ? */
+#if 0
+       /*
+        * with a simple addition like this, we can do io priorities. almost.
+        * does need a split request free list, too.
+        */
+       int io_prio
+#endif
 };
 
-/*
- * per-request structure
- */
 struct cfq_rq {
-       struct cfq_queue *cfq_queue;
        struct rb_node rb_node;
-       struct hlist_node hash;
        sector_t rb_key;
 
        struct request *request;
-       struct list_head prio_list;
-       unsigned long nr_sectors;
-       int ioprio;
+
+       struct cfq_queue *cfq_queue;
+
+       struct list_head hash;
 };
 
 static void cfq_put_queue(struct cfq_data *cfqd, struct cfq_queue *cfqq);
@@ -223,13 +103,18 @@ static void cfq_dispatch_sort(struct cfq_data *cfqd, struct cfq_queue *cfqq,
 /*
  * lots of deadline iosched dupes, can be abstracted later...
  */
+static inline void __cfq_del_crq_hash(struct cfq_rq *crq)
+{
+       list_del_init(&crq->hash);
+}
+
 static inline void cfq_del_crq_hash(struct cfq_rq *crq)
 {
-       hlist_del_init(&crq->hash);
+       if (ON_MHASH(crq))
+               __cfq_del_crq_hash(crq);
 }
 
-static inline void
-cfq_remove_merge_hints(request_queue_t *q, struct cfq_rq *crq)
+static void cfq_remove_merge_hints(request_queue_t *q, struct cfq_rq *crq)
 {
        cfq_del_crq_hash(crq);
 
@@ -240,26 +125,27 @@ cfq_remove_merge_hints(request_queue_t *q, struct cfq_rq *crq)
 static inline void cfq_add_crq_hash(struct cfq_data *cfqd, struct cfq_rq *crq)
 {
        struct request *rq = crq->request;
-       const int hash_idx = CFQ_MHASH_FN(rq_hash_key(rq));
 
-       BUG_ON(!hlist_unhashed(&crq->hash));
-       hlist_add_head(&crq->hash, &cfqd->crq_hash[hash_idx]);
+       BUG_ON(ON_MHASH(crq));
+
+       list_add(&crq->hash, &cfqd->crq_hash[CFQ_MHASH_FN(rq_hash_key(rq))]);
 }
 
 static struct request *cfq_find_rq_hash(struct cfq_data *cfqd, sector_t offset)
 {
-       struct hlist_head *hash_list = &cfqd->crq_hash[CFQ_MHASH_FN(offset)];
-       struct hlist_node *entry, *next;
+       struct list_head *hash_list = &cfqd->crq_hash[CFQ_MHASH_FN(offset)];
+       struct list_head *entry, *next = hash_list->next;
 
-       hlist_for_each_safe(entry, next, hash_list) {
+       while ((entry = next) != hash_list) {
                struct cfq_rq *crq = list_entry_hash(entry);
                struct request *__rq = crq->request;
 
-               BUG_ON(hlist_unhashed(&crq->hash));
+               next = entry->next;
+
+               BUG_ON(!ON_MHASH(crq));
 
                if (!rq_mergeable(__rq)) {
-                       cfq_del_crq_hash(crq);
+                       __cfq_del_crq_hash(crq);
                        continue;
                }
 
@@ -273,27 +159,20 @@ static struct request *cfq_find_rq_hash(struct cfq_data *cfqd, sector_t offset)
 /*
  * rb tree support functions
  */
-#define RB_EMPTY(node)         ((node)->rb_node == NULL)
+#define RB_NONE                (2)
+#define RB_EMPTY(node) ((node)->rb_node == NULL)
+#define RB_CLEAR(node) ((node)->rb_color = RB_NONE)
+#define RB_CLEAR_ROOT(root)    ((root)->rb_node = NULL)
+#define ON_RB(node)    ((node)->rb_color != RB_NONE)
 #define rb_entry_crq(node)     rb_entry((node), struct cfq_rq, rb_node)
 #define rq_rb_key(rq)          (rq)->sector
 
-static void
-cfq_del_crq_rb(struct cfq_data *cfqd, struct cfq_queue *cfqq,struct cfq_rq *crq)
+static inline void cfq_del_crq_rb(struct cfq_queue *cfqq, struct cfq_rq *crq)
 {
-       if (crq->cfq_queue) {
-               crq->cfq_queue = NULL;
-
-               if (cfq_account_io(crq)) {
-                       cfqd->busy_rq--;
-                       cfqd->busy_sectors -= crq->nr_sectors;
-                       cfqd->cid[crq->ioprio].busy_rq--;
-                       cfqd->cid[crq->ioprio].busy_sectors -= crq->nr_sectors;
-               }
-               atomic_inc(&(cfqd->cid[crq->ioprio].cum_rq_out));
-               atomic_add(crq->nr_sectors,
-                          &(cfqd->cid[crq->ioprio].cum_sectors_out));
+       if (ON_RB(&crq->rb_node)) {
                cfqq->queued[rq_data_dir(crq->request)]--;
                rb_erase(&crq->rb_node, &cfqq->sort_list);
+               crq->cfq_queue = NULL;
        }
 }
 
@@ -326,22 +205,12 @@ cfq_add_crq_rb(struct cfq_data *cfqd, struct cfq_queue *cfqq,struct cfq_rq *crq)
        struct request *rq = crq->request;
        struct cfq_rq *__alias;
 
-
+       crq->rb_key = rq_rb_key(rq);
        cfqq->queued[rq_data_dir(rq)]++;
-       if (cfq_account_io(crq)) {
-               cfqd->busy_rq++;
-               cfqd->busy_sectors += crq->nr_sectors;
-               cfqd->cid[crq->ioprio].busy_rq++;
-               cfqd->cid[crq->ioprio].busy_sectors += crq->nr_sectors;
-       }
-       atomic_inc(&(cfqd->cid[crq->ioprio].cum_rq_in));
-       atomic_add(crq->nr_sectors,
-                  &(cfqd->cid[crq->ioprio].cum_sectors_in));
 retry:
        __alias = __cfq_add_crq_rb(cfqq, crq);
        if (!__alias) {
                rb_insert_color(&crq->rb_node, &cfqq->sort_list);
-               crq->rb_key = rq_rb_key(rq);
                crq->cfq_queue = cfqq;
                return;
        }
@@ -353,7 +222,7 @@ retry:
 static struct request *
 cfq_find_rq_rb(struct cfq_data *cfqd, sector_t sector)
 {
-       struct cfq_queue *cfqq = cfq_find_cfq_hash(cfqd, cfq_hash_key(current));
+       struct cfq_queue *cfqq = cfq_find_cfq_hash(cfqd, current->tgid);
        struct rb_node *n;
 
        if (!cfqq)
@@ -378,31 +247,16 @@ out:
 static void cfq_remove_request(request_queue_t *q, struct request *rq)
 {
        struct cfq_data *cfqd = q->elevator.elevator_data;
-       struct cfq_rq *crq = RQ_ELV_DATA(rq);
+       struct cfq_rq *crq = RQ_DATA(rq);
 
        if (crq) {
+               struct cfq_queue *cfqq = crq->cfq_queue;
 
                cfq_remove_merge_hints(q, crq);
-               list_del_init(&crq->prio_list);
                list_del_init(&rq->queuelist);
 
-               /*
-                * set a grace period timer to allow realtime io to make real
-                * progress, if we release an rt request. for normal request,
-                * set timer so idle io doesn't interfere with other io
-                */
-               if (crq->ioprio == IOPRIO_RT) {
-                       set_bit(CFQ_WAIT_RT, &cfqd->flags);
-                       cfqd->wait_end = jiffies + cfqd->cfq_grace_rt;
-               } else if (crq->ioprio != IOPRIO_IDLE) {
-                       set_bit(CFQ_WAIT_NORM, &cfqd->flags);
-                       cfqd->wait_end = jiffies + cfqd->cfq_grace_idle;
-               }
-
-               if (crq->cfq_queue) {
-                       struct cfq_queue *cfqq = crq->cfq_queue;
-
-                       cfq_del_crq_rb(cfqd, cfqq, crq);
+               if (cfqq) {
+                       cfq_del_crq_rb(cfqq, crq);
 
                        if (RB_EMPTY(&cfqq->sort_list))
                                cfq_put_queue(cfqd, cfqq);
@@ -452,26 +306,18 @@ out_insert:
 static void cfq_merged_request(request_queue_t *q, struct request *req)
 {
        struct cfq_data *cfqd = q->elevator.elevator_data;
-       struct cfq_rq *crq = RQ_ELV_DATA(req);
-       int tmp;
+       struct cfq_rq *crq = RQ_DATA(req);
 
        cfq_del_crq_hash(crq);
        cfq_add_crq_hash(cfqd, crq);
 
-       if (crq->cfq_queue && (rq_rb_key(req) != crq->rb_key)) {
+       if (ON_RB(&crq->rb_node) && (rq_rb_key(req) != crq->rb_key)) {
                struct cfq_queue *cfqq = crq->cfq_queue;
 
-               cfq_del_crq_rb(cfqd, cfqq, crq);
+               cfq_del_crq_rb(cfqq, crq);
                cfq_add_crq_rb(cfqd, cfqq, crq);
        }
 
-       tmp = req->hard_nr_sectors - crq->nr_sectors;
-       cfqd->busy_sectors += tmp;
-       cfqd->cid[crq->ioprio].busy_sectors += tmp;
-       atomic_add(tmp,&(cfqd->cid[crq->ioprio].cum_sectors_in));
-
-       crq->nr_sectors = req->hard_nr_sectors;
-
        q->last_merge = req;
 }
 
@@ -483,9 +329,6 @@ cfq_merged_requests(request_queue_t *q, struct request *req,
        cfq_remove_request(q, next);
 }
 
-/*
- * sort into dispatch list, in optimal ascending order
- */
 static void
 cfq_dispatch_sort(struct cfq_data *cfqd, struct cfq_queue *cfqq,
                  struct cfq_rq *crq)
@@ -493,7 +336,7 @@ cfq_dispatch_sort(struct cfq_data *cfqd, struct cfq_queue *cfqq,
        struct list_head *head = cfqd->dispatch, *entry = head;
        struct request *__rq;
 
-       cfq_del_crq_rb(cfqd, cfqq, crq);
+       cfq_del_crq_rb(cfqq, crq);
        cfq_remove_merge_hints(cfqd->queue, crq);
 
        if (!list_empty(head)) {
@@ -516,219 +359,47 @@ link:
        list_add_tail(&crq->request->queuelist, entry);
 }
 
-/*
- * remove from io scheduler core and put on dispatch list for service
- */
-static inline int
+static inline void
 __cfq_dispatch_requests(request_queue_t *q, struct cfq_data *cfqd,
                        struct cfq_queue *cfqq)
 {
-       struct cfq_rq *crq;
-       unsigned long long ts, gap;
-       unsigned long newavsec;
-
-       crq = rb_entry_crq(rb_first(&cfqq->sort_list));
-
-#if 1
-       /* Determine if queue should be skipped for being overshare */
-       ts = sched_clock();
-       gap = ts - cfqq->lastime;
-#ifdef LIMIT_DEBUG
-       cfqq->sectorate = (cfqd->cfq_epochsectors 
-                          * CFQ_TEMPLIM)/100;
-       
-#endif
-       if ((gap >= cfqd->cfq_epoch) || (gap < 0)) {
-               cfqq->avsec = crq->nr_sectors ; 
-               cfqq->lastime = ts;
-       } else {
-               u64 tmp;
-               /* Age old average and accumalate request to be served */
-
-//             tmp = (u64) (cfqq->avsec * gap) ;
-//             do_div(tmp, cfqd->cfq_epoch);
-               newavsec = (unsigned long)(cfqq->avsec >> 1) + crq->nr_sectors;
-//             if (crq->ioprio >= 0 && crq->ioprio <= 20)
-//                     cfqd->cid[crq->ioprio].lsectorate = newavsec; 
-//             atomic_set(&(cfqd->cid[crq->ioprio].lsectorate),
-//                        newavsec);
-
-               if ((newavsec < cfqq->sectorate) || cfqq->skipped) {
-                       cfqq->avsec = newavsec ;
-                       cfqq->lastime = ts;
-                       cfqq->skipped = 0;
-               } else {
-                       /* queue over share ; skip once */
-                       cfqq->skipped = 1;
-#ifdef LIMIT_DEBUG     
-//                     atomic_inc(&(cfqd->cid[crq->ioprio].nskip));
-//                     if (crq->ioprio >= 0 && crq->ioprio <= 20)
-//                             cfqd->cid[crq->ioprio].nskip++;
-#endif
-                       return 0;
-               }
-       }
-#endif
-
-#ifdef LIMIT_DEBUG
-//     if (crq->ioprio >= 0 && crq->ioprio <= 20) {
-//             cfqd->cid[crq->ioprio].navsec = cfqq->avsec;
-//             cfqd->cid[crq->ioprio].csectorate = cfqq->sectorate;
-//     }
+       struct cfq_rq *crq = rb_entry_crq(rb_first(&cfqq->sort_list));
 
-//     atomic_set(&(cfqd->cid[crq->ioprio].navsec),cfqq->avsec);
-//     atomic_set(&(cfqd->cid[crq->ioprio].csectorate),cfqq->sectorate);
-#endif
        cfq_dispatch_sort(cfqd, cfqq, crq);
-
-       /*
-        * technically, for IOPRIO_RT we don't need to add it to the list.
-        */
-       list_add_tail(&crq->prio_list, &cfqd->cid[cfqq->ioprio].prio_list);
-       return crq->nr_sectors;
 }
 
-static int
-cfq_dispatch_requests(request_queue_t *q, int prio, int max_rq, int max_sectors)
+static int cfq_dispatch_requests(request_queue_t *q, struct cfq_data *cfqd)
 {
-       struct cfq_data *cfqd = q->elevator.elevator_data;
-       struct list_head *plist = &cfqd->cid[prio].rr_list;
-       struct list_head *entry, *nxt;
-       int q_rq, q_io;
-       int ret ;
+       struct cfq_queue *cfqq;
+       struct list_head *entry, *tmp;
+       int ret, queued, good_queues;
 
-       /*
-        * for each queue at this prio level, dispatch a request
-        */
-       q_rq = q_io = 0;
-       list_for_each_safe(entry, nxt, plist) {
-               struct cfq_queue *cfqq = list_entry_cfqq(entry);
+       if (list_empty(&cfqd->rr_list))
+               return 0;
+
+       queued = ret = 0;
+restart:
+       good_queues = 0;
+       list_for_each_safe(entry, tmp, &cfqd->rr_list) {
+               cfqq = list_entry_cfqq(cfqd->rr_list.next);
 
                BUG_ON(RB_EMPTY(&cfqq->sort_list));
 
-               ret = __cfq_dispatch_requests(q, cfqd, cfqq);
-               if (ret <= 0) {
-                       continue; /* skip queue */
-                       /* can optimize more by moving q to end of plist ? */
-               }
-               q_io += ret ;
-               q_rq++ ;
+               __cfq_dispatch_requests(q, cfqd, cfqq);
 
                if (RB_EMPTY(&cfqq->sort_list))
                        cfq_put_queue(cfqd, cfqq);
-               /*
-                * if we hit the queue limit, put the string of serviced
-                * queues at the back of the pending list
-                */
-               if (q_io >= max_sectors || q_rq >= max_rq) {
-                       struct list_head *prv = nxt->prev;
-
-                       if (prv != plist) {
-                               list_del(plist);
-                               list_add(plist, prv);
-                       }
-                       break;
-               }
-       }
-
-       cfqd->cid[prio].last_rq = q_rq;
-       cfqd->cid[prio].last_sectors = q_io;
-       return q_rq;
-}
-
-/*
- * try to move some requests to the dispatch list. return 0 on success
- */
-static int cfq_select_requests(request_queue_t *q, struct cfq_data *cfqd)
-{
-       int queued, busy_rq, busy_sectors, i;
-
-       /*
-        * if there's any realtime io, only schedule that
-        */
-       if (cfq_dispatch_requests(q, IOPRIO_RT, cfqd->cfq_quantum, cfqd->cfq_quantum_io))
-               return 1;
-
-       /*
-        * if RT io was last serviced and grace time hasn't expired,
-        * arm the timer to restart queueing if no other RT io has been
-        * submitted in the mean time
-        */
-       if (test_bit(CFQ_WAIT_RT, &cfqd->flags)) {
-               if (time_before(jiffies, cfqd->wait_end)) {
-                       mod_timer(&cfqd->timer, cfqd->wait_end);
-                       return 0;
-               }
-               clear_bit(CFQ_WAIT_RT, &cfqd->flags);
-       }
-
-       /*
-        * for each priority level, calculate number of requests we
-        * are allowed to put into service.
-        */
-       queued = 0;
-       busy_rq = cfqd->busy_rq;
-       busy_sectors = cfqd->busy_sectors;
-       for (i = IOPRIO_RT - 1; i > IOPRIO_IDLE; i--) {
-               const int o_rq = busy_rq - cfqd->cid[i].busy_rq;
-               const int o_sectors = busy_sectors - cfqd->cid[i].busy_sectors;
-               int q_rq = cfqd->cfq_quantum * (i + 1) / IOPRIO_NR;
-               int q_io = cfqd->cfq_quantum_io * (i + 1) / IOPRIO_NR;
-
-               /*
-                * no need to keep iterating the list, if there are no
-                * requests pending anymore
-                */
-               if (!cfqd->busy_rq)
-                       break;
-
-               /*
-                * find out how many requests and sectors we are allowed to
-                * service
-                */
-               if (o_rq)
-                       q_rq = o_sectors * (i + 1) / IOPRIO_NR;
-               if (q_rq > cfqd->cfq_quantum)
-                       q_rq = cfqd->cfq_quantum;
-
-               if (o_sectors)
-                       q_io = o_sectors * (i + 1) / IOPRIO_NR;
-               if (q_io > cfqd->cfq_quantum_io)
-                       q_io = cfqd->cfq_quantum_io;
-
-               /*
-                * average with last dispatched for fairness
-                */
-               if (cfqd->cid[i].last_rq != -1)
-                       q_rq = (cfqd->cid[i].last_rq + q_rq) / 2;
-               if (cfqd->cid[i].last_sectors != -1)
-                       q_io = (cfqd->cid[i].last_sectors + q_io) / 2;
-
-               queued += cfq_dispatch_requests(q, i, q_rq, q_io);
-       }
-
-       if (queued)
-               return 1;
+               else
+                       good_queues++;
 
-       /*
-        * only allow dispatch of idle io, if the queue has been idle from
-        * servicing RT or normal io for the grace period
-        */
-       if (test_bit(CFQ_WAIT_NORM, &cfqd->flags)) {
-               if (time_before(jiffies, cfqd->wait_end)) {
-                       mod_timer(&cfqd->timer, cfqd->wait_end);
-                       return 0;
-               }
-               clear_bit(CFQ_WAIT_NORM, &cfqd->flags);
+               queued++;
+               ret = 1;
        }
 
-       /*
-        * if we found nothing to do, allow idle io to be serviced
-        */
-       if (cfq_dispatch_requests(q, IOPRIO_IDLE, cfqd->cfq_idle_quantum, cfqd->cfq_idle_quantum_io))
-               return 1;
+       if ((queued < cfqd->cfq_quantum) && good_queues)
+               goto restart;
 
-       return 0;
+       return ret;
 }
 
 static struct request *cfq_next_request(request_queue_t *q)
@@ -739,82 +410,61 @@ static struct request *cfq_next_request(request_queue_t *q)
        if (!list_empty(cfqd->dispatch)) {
                struct cfq_rq *crq;
 dispatch:
-               /*
-                * end grace period, we are servicing a request
-                */
-               del_timer(&cfqd->timer);
-               clear_bit(CFQ_WAIT_RT, &cfqd->flags);
-               clear_bit(CFQ_WAIT_NORM, &cfqd->flags);
-
-               BUG_ON(list_empty(cfqd->dispatch));
                rq = list_entry_rq(cfqd->dispatch->next);
 
-               BUG_ON(q->last_merge == rq);
-               crq = RQ_ELV_DATA(rq);
-               if (crq) {
-                       BUG_ON(!hlist_unhashed(&crq->hash));
-                       list_del_init(&crq->prio_list);
-               }
+               crq = RQ_DATA(rq);
+               if (crq)
+                       cfq_remove_merge_hints(q, crq);
 
                return rq;
        }
 
-       /*
-        * we moved requests to dispatch list, go back end serve one
-        */
-       if (cfq_select_requests(q, cfqd))
+       if (cfq_dispatch_requests(q, cfqd))
                goto dispatch;
 
        return NULL;
 }
 
 static inline struct cfq_queue *
-__cfq_find_cfq_hash(struct cfq_data *cfqd, int hashkey, const int hashval)
+__cfq_find_cfq_hash(struct cfq_data *cfqd, int pid, const int hashval)
 {
-       struct hlist_head *hash_list = &cfqd->cfq_hash[hashval];
-       struct hlist_node *entry;
+       struct list_head *hash_list = &cfqd->cfq_hash[hashval];
+       struct list_head *entry;
 
-       hlist_for_each(entry, hash_list) {
+       list_for_each(entry, hash_list) {
                struct cfq_queue *__cfqq = list_entry_qhash(entry);
 
-               if (__cfqq->hash_key == hashkey)
+               if (__cfqq->pid == pid)
                        return __cfqq;
        }
 
        return NULL;
 }
 
-
-static struct cfq_queue *cfq_find_cfq_hash(struct cfq_data *cfqd, int hashkey)
+static struct cfq_queue *cfq_find_cfq_hash(struct cfq_data *cfqd, int pid)
 {
-       const int hashval = hash_long(hashkey, CFQ_QHASH_SHIFT);
+       const int hashval = hash_long(current->tgid, CFQ_QHASH_SHIFT);
 
-       return __cfq_find_cfq_hash(cfqd, hashkey, hashval);
+       return __cfq_find_cfq_hash(cfqd, pid, hashval);
 }
 
 static void cfq_put_queue(struct cfq_data *cfqd, struct cfq_queue *cfqq)
 {
        cfqd->busy_queues--;
-       WARN_ON(cfqd->busy_queues < 0);
-
-       cfqd->cid[cfqq->ioprio].busy_queues--;
-       WARN_ON(cfqd->cid[cfqq->ioprio].busy_queues < 0);
-       atomic_inc(&(cfqd->cid[cfqq->ioprio].cum_queues_out));
-
        list_del(&cfqq->cfq_list);
-       hlist_del(&cfqq->cfq_hash);
+       list_del(&cfqq->cfq_hash);
        mempool_free(cfqq, cfq_mpool);
 }
 
-static struct cfq_queue *__cfq_get_queue(struct cfq_data *cfqd, int hashkey,
+static struct cfq_queue *__cfq_get_queue(struct cfq_data *cfqd, int pid,
                                         int gfp_mask)
 {
-       const int hashval = hash_long(hashkey, CFQ_QHASH_SHIFT);
+       const int hashval = hash_long(current->tgid, CFQ_QHASH_SHIFT);
        struct cfq_queue *cfqq, *new_cfqq = NULL;
        request_queue_t *q = cfqd->queue;
 
 retry:
-       cfqq = __cfq_find_cfq_hash(cfqd, hashkey, hashval);
+       cfqq = __cfq_find_cfq_hash(cfqd, pid, hashval);
 
        if (!cfqq) {
                if (new_cfqq) {
@@ -828,15 +478,13 @@ retry:
                } else
                        return NULL;
 
-               memset(cfqq, 0, sizeof(*cfqq));
-               INIT_HLIST_NODE(&cfqq->cfq_hash);
+               INIT_LIST_HEAD(&cfqq->cfq_hash);
                INIT_LIST_HEAD(&cfqq->cfq_list);
-               cfqq->hash_key = cfq_hash_key(current);
-               cfqq->ioprio = cfq_ioprio(current);
-               cfqq->avsec = 0 ;
-               cfqq->lastime = sched_clock();
-               cfqq->sectorate = (cfqd->cfq_epochsectors * CFQ_TEMPLIM)/100;
-               hlist_add_head(&cfqq->cfq_hash, &cfqd->cfq_hash[hashval]);
+               RB_CLEAR_ROOT(&cfqq->sort_list);
+
+               cfqq->pid = pid;
+               cfqq->queued[0] = cfqq->queued[1] = 0;
+               list_add(&cfqq->cfq_hash, &cfqd->cfq_hash[hashval]);
        }
 
        if (new_cfqq)
@@ -845,63 +493,31 @@ retry:
        return cfqq;
 }
 
-static struct cfq_queue *cfq_get_queue(struct cfq_data *cfqd, int hashkey,
+static struct cfq_queue *cfq_get_queue(struct cfq_data *cfqd, int pid,
                                       int gfp_mask)
 {
        request_queue_t *q = cfqd->queue;
        struct cfq_queue *cfqq;
 
        spin_lock_irq(q->queue_lock);
-       cfqq = __cfq_get_queue(cfqd, hashkey, gfp_mask);
+       cfqq = __cfq_get_queue(cfqd, pid, gfp_mask);
        spin_unlock_irq(q->queue_lock);
 
        return cfqq;
 }
 
-static void
-__cfq_enqueue(request_queue_t *q, struct cfq_data *cfqd, struct cfq_rq *crq)
+static void cfq_enqueue(struct cfq_data *cfqd, struct cfq_rq *crq)
 {
-       const int prio = crq->ioprio;
        struct cfq_queue *cfqq;
 
-       cfqq = __cfq_get_queue(cfqd, cfq_hash_key(current), GFP_ATOMIC);
+       cfqq = __cfq_get_queue(cfqd, current->tgid, GFP_ATOMIC);
        if (cfqq) {
-
-               /*
-                * not too good...
-                */
-               if (prio > cfqq->ioprio) {
-                       printk("prio hash collision %d %d\n", 
-                              prio, cfqq->ioprio);
-                       if (!list_empty(&cfqq->cfq_list)) {
-                               cfqd->cid[cfqq->ioprio].busy_queues--;
-                               WARN_ON(cfqd->cid[cfqq->ioprio].busy_queues<0);
-                               atomic_inc(&(cfqd->cid[cfqq->ioprio].cum_queues_out));
-                               cfqd->cid[prio].busy_queues++;
-                               atomic_inc(&(cfqd->cid[prio].cum_queues_in));
-                               list_move_tail(&cfqq->cfq_list, 
-                                              &cfqd->cid[prio].rr_list);
-                       }
-                       cfqq->ioprio = prio;
-               }
-
                cfq_add_crq_rb(cfqd, cfqq, crq);
 
                if (list_empty(&cfqq->cfq_list)) {
-                       list_add_tail(&cfqq->cfq_list, 
-                                     &cfqd->cid[prio].rr_list);
-                       cfqd->cid[prio].busy_queues++;
-                       atomic_inc(&(cfqd->cid[prio].cum_queues_in));
+                       list_add(&cfqq->cfq_list, &cfqd->rr_list);
                        cfqd->busy_queues++;
                }
-
-               if (rq_mergeable(crq->request)) {
-                       cfq_add_crq_hash(cfqd, crq);
-                       
-                       if (!q->last_merge)
-                               q->last_merge = crq->request;
-               }
-
        } else {
                /*
                 * should can only happen if the request wasn't allocated
@@ -912,57 +528,16 @@ __cfq_enqueue(request_queue_t *q, struct cfq_data *cfqd, struct cfq_rq *crq)
        }
 }
 
-static void cfq_reenqueue(request_queue_t *q, struct cfq_data *cfqd, int prio)
-{
-       struct list_head *prio_list = &cfqd->cid[prio].prio_list;
-       struct list_head *entry, *tmp;
-
-       list_for_each_safe(entry, tmp, prio_list) {
-               struct cfq_rq *crq = list_entry_prio(entry);
-
-               list_del_init(entry);
-               list_del_init(&crq->request->queuelist);
-               __cfq_enqueue(q, cfqd, crq);
-       }
-}
-
-static void
-cfq_enqueue(request_queue_t *q, struct cfq_data *cfqd, struct cfq_rq *crq)
-{
-       const int prio = cfq_ioprio(current);
-
-       crq->ioprio = prio;
-       crq->nr_sectors = crq->request->hard_nr_sectors;
-       __cfq_enqueue(q, cfqd, crq);
-
-       if (prio == IOPRIO_RT) {
-               int i;
-
-               /*
-                * realtime io gets priority, move all other io back
-                */
-               for (i = IOPRIO_IDLE; i < IOPRIO_RT; i++)
-                       cfq_reenqueue(q, cfqd, i);
-       } else if (prio != IOPRIO_IDLE) {
-               /*
-                * check if we need to move idle io back into queue
-                */
-               cfq_reenqueue(q, cfqd, IOPRIO_IDLE);
-       }
-}
-
 static void
 cfq_insert_request(request_queue_t *q, struct request *rq, int where)
 {
        struct cfq_data *cfqd = q->elevator.elevator_data;
-       struct cfq_rq *crq = RQ_ELV_DATA(rq);
+       struct cfq_rq *crq = RQ_DATA(rq);
 
        switch (where) {
                case ELEVATOR_INSERT_BACK:
-#if 0
                        while (cfq_dispatch_requests(q, cfqd))
                                ;
-#endif
                        list_add_tail(&rq->queuelist, cfqd->dispatch);
                        break;
                case ELEVATOR_INSERT_FRONT:
@@ -970,20 +545,26 @@ cfq_insert_request(request_queue_t *q, struct request *rq, int where)
                        break;
                case ELEVATOR_INSERT_SORT:
                        BUG_ON(!blk_fs_request(rq));
-                       cfq_enqueue(q, cfqd, crq);
+                       cfq_enqueue(cfqd, crq);
                        break;
                default:
-                       printk("%s: bad insert point %d\n", 
-                              __FUNCTION__,where);
+                       printk("%s: bad insert point %d\n", __FUNCTION__,where);
                        return;
        }
+
+       if (rq_mergeable(rq)) {
+               cfq_add_crq_hash(cfqd, crq);
+
+               if (!q->last_merge)
+                       q->last_merge = rq;
+       }
 }
 
 static int cfq_queue_empty(request_queue_t *q)
 {
        struct cfq_data *cfqd = q->elevator.elevator_data;
 
-       if (list_empty(cfqd->dispatch) && !cfqd->busy_queues)
+       if (list_empty(cfqd->dispatch) && list_empty(&cfqd->rr_list))
                return 1;
 
        return 0;
@@ -992,7 +573,7 @@ static int cfq_queue_empty(request_queue_t *q)
 static struct request *
 cfq_former_request(request_queue_t *q, struct request *rq)
 {
-       struct cfq_rq *crq = RQ_ELV_DATA(rq);
+       struct cfq_rq *crq = RQ_DATA(rq);
        struct rb_node *rbprev = rb_prev(&crq->rb_node);
 
        if (rbprev)
@@ -1004,7 +585,7 @@ cfq_former_request(request_queue_t *q, struct request *rq)
 static struct request *
 cfq_latter_request(request_queue_t *q, struct request *rq)
 {
-       struct cfq_rq *crq = RQ_ELV_DATA(rq);
+       struct cfq_rq *crq = RQ_DATA(rq);
        struct rb_node *rbnext = rb_next(&crq->rb_node);
 
        if (rbnext)
@@ -1013,46 +594,27 @@ cfq_latter_request(request_queue_t *q, struct request *rq)
        return NULL;
 }
 
-static void cfq_queue_congested(request_queue_t *q)
-{
-       struct cfq_data *cfqd = q->elevator.elevator_data;
-
-       set_bit(cfq_ioprio(current), &cfqd->rq_starved_mask);
-}
-
 static int cfq_may_queue(request_queue_t *q, int rw)
 {
        struct cfq_data *cfqd = q->elevator.elevator_data;
        struct cfq_queue *cfqq;
-       const int prio = cfq_ioprio(current);
-       int limit, ret = 1;
+       int ret = 1;
 
        if (!cfqd->busy_queues)
                goto out;
 
-       cfqq = cfq_find_cfq_hash(cfqd, cfq_hash_key(current));
-       if (!cfqq)
-               goto out;
-
-       cfqq = cfq_find_cfq_hash(cfqd, cfq_hash_key(current));
-       if (!cfqq)
-               goto out;
-
-       /*
-        * if higher or equal prio io is sleeping waiting for a request, don't
-        * allow this one to allocate one. as long as ll_rw_blk does fifo
-        * waitqueue wakeups this should work...
-        */
-       if (cfqd->rq_starved_mask & ~((1 << prio) - 1))
-               goto out;
+       cfqq = cfq_find_cfq_hash(cfqd, current->tgid);
+       if (cfqq) {
+               int limit = (q->nr_requests - cfqd->cfq_queued) / cfqd->busy_queues;
 
-       if (cfqq->queued[rw] < cfqd->cfq_queued || !cfqd->cid[prio].busy_queues)
-               goto out;
+               if (limit < 3)
+                       limit = 3;
+               else if (limit > cfqd->max_queued)
+                       limit = cfqd->max_queued;
 
-       limit = q->nr_requests * (prio + 1) / IOPRIO_NR;
-       limit /= cfqd->cid[prio].busy_queues;
-       if (cfqq->queued[rw] > limit)
-               ret = 0;
+               if (cfqq->queued[rw] > limit)
+                       ret = 0;
+       }
 out:
        return ret;
 }
@@ -1060,13 +622,13 @@ out:
 static void cfq_put_request(request_queue_t *q, struct request *rq)
 {
        struct cfq_data *cfqd = q->elevator.elevator_data;
-       struct cfq_rq *crq = RQ_ELV_DATA(rq);
+       struct cfq_rq *crq = RQ_DATA(rq);
        struct request_list *rl;
        int other_rw;
 
        if (crq) {
                BUG_ON(q->last_merge == rq);
-               BUG_ON(!hlist_unhashed(&crq->hash));
+               BUG_ON(ON_MHASH(crq));
 
                mempool_free(crq, cfqd->crq_pool);
                rq->elevator_private = NULL;
@@ -1099,21 +661,17 @@ static int cfq_set_request(request_queue_t *q, struct request *rq, int gfp_mask)
        /*
         * prepare a queue up front, so cfq_enqueue() doesn't have to
         */
-       cfqq = cfq_get_queue(cfqd, cfq_hash_key(current), gfp_mask);
+       cfqq = cfq_get_queue(cfqd, current->tgid, gfp_mask);
        if (!cfqq)
                return 1;
 
        crq = mempool_alloc(cfqd->crq_pool, gfp_mask);
        if (crq) {
-               /*
-                * process now has one request
-                */
-               clear_bit(cfq_ioprio(current), &cfqd->rq_starved_mask);
-
                memset(crq, 0, sizeof(*crq));
+               RB_CLEAR(&crq->rb_node);
                crq->request = rq;
-               INIT_HLIST_NODE(&crq->hash);
-               INIT_LIST_HEAD(&crq->prio_list);
+               crq->cfq_queue = NULL;
+               INIT_LIST_HEAD(&crq->hash);
                rq->elevator_private = crq;
                return 0;
        }
@@ -1132,26 +690,6 @@ static void cfq_exit(request_queue_t *q, elevator_t *e)
        kfree(cfqd);
 }
 
-static void cfq_timer(unsigned long data)
-{
-       struct cfq_data *cfqd = (struct cfq_data *) data;
-
-       clear_bit(CFQ_WAIT_RT, &cfqd->flags);
-       clear_bit(CFQ_WAIT_NORM, &cfqd->flags);
-       kblockd_schedule_work(&cfqd->work);
-}
-
-static void cfq_work(void *data)
-{
-       request_queue_t *q = data;
-       unsigned long flags;
-
-       spin_lock_irqsave(q->queue_lock, flags);
-       if (cfq_next_request(q))
-               q->request_fn(q);
-       spin_unlock_irqrestore(q->queue_lock, flags);
-}
-
 static int cfq_init(request_queue_t *q, elevator_t *e)
 {
        struct cfq_data *cfqd;
@@ -1162,75 +700,38 @@ static int cfq_init(request_queue_t *q, elevator_t *e)
                return -ENOMEM;
 
        memset(cfqd, 0, sizeof(*cfqd));
-       init_timer(&cfqd->timer);
-       cfqd->timer.function = cfq_timer;
-       cfqd->timer.data = (unsigned long) cfqd;
-
-       INIT_WORK(&cfqd->work, cfq_work, q);
-
-       for (i = 0; i < IOPRIO_NR; i++) {
-               struct io_prio_data *cid = &cfqd->cid[i];
-
-               INIT_LIST_HEAD(&cid->rr_list);
-               INIT_LIST_HEAD(&cid->prio_list);
-               cid->last_rq = -1;
-               cid->last_sectors = -1;
-
-               atomic_set(&cid->cum_rq_in,0);          
-               atomic_set(&cid->cum_rq_out,0);
-               atomic_set(&cid->cum_sectors_in,0);
-               atomic_set(&cid->cum_sectors_out,0);            
-               atomic_set(&cid->cum_queues_in,0);
-               atomic_set(&cid->cum_queues_out,0);
-#if 0
-               atomic_set(&cid->nskip,0);
-               atomic_set(&cid->navsec,0);
-               atomic_set(&cid->csectorate,0);
-               atomic_set(&cid->lsectorate,0);
-#endif
-       }
+       INIT_LIST_HEAD(&cfqd->rr_list);
 
-       cfqd->crq_hash = kmalloc(sizeof(struct hlist_head) * CFQ_MHASH_ENTRIES,
-                                GFP_KERNEL);
+       cfqd->crq_hash = kmalloc(sizeof(struct list_head) * CFQ_MHASH_ENTRIES, GFP_KERNEL);
        if (!cfqd->crq_hash)
                goto out_crqhash;
 
-       cfqd->cfq_hash = kmalloc(sizeof(struct hlist_head) * CFQ_QHASH_ENTRIES,
-                                GFP_KERNEL);
+       cfqd->cfq_hash = kmalloc(sizeof(struct list_head) * CFQ_QHASH_ENTRIES, GFP_KERNEL);
        if (!cfqd->cfq_hash)
                goto out_cfqhash;
 
-       cfqd->crq_pool = mempool_create(BLKDEV_MIN_RQ, mempool_alloc_slab, 
-                                       mempool_free_slab, crq_pool);
+       cfqd->crq_pool = mempool_create(BLKDEV_MIN_RQ, mempool_alloc_slab, mempool_free_slab, crq_pool);
        if (!cfqd->crq_pool)
                goto out_crqpool;
 
        for (i = 0; i < CFQ_MHASH_ENTRIES; i++)
-               INIT_HLIST_HEAD(&cfqd->crq_hash[i]);
+               INIT_LIST_HEAD(&cfqd->crq_hash[i]);
        for (i = 0; i < CFQ_QHASH_ENTRIES; i++)
-               INIT_HLIST_HEAD(&cfqd->cfq_hash[i]);
-
-       cfqd->cfq_queued = cfq_queued;
-       cfqd->cfq_quantum = cfq_quantum;
-       cfqd->cfq_quantum_io = cfq_quantum_io;
-       cfqd->cfq_idle_quantum = cfq_idle_quantum;
-       cfqd->cfq_idle_quantum_io = cfq_idle_quantum_io;
-       cfqd->cfq_grace_rt = cfq_grace_rt;
-       cfqd->cfq_grace_idle = cfq_grace_idle;
-
-       q->nr_requests <<= 2;
+               INIT_LIST_HEAD(&cfqd->cfq_hash[i]);
 
        cfqd->dispatch = &q->queue_head;
        e->elevator_data = cfqd;
        cfqd->queue = q;
 
-       cfqd->cfq_epoch = CFQ_EPOCH;
-       if (q->hardsect_size)
-               cfqd->cfq_epochsectors = ((CFQ_DISKBW * 1000000)/
-                                     q->hardsect_size)* (1000000 / CFQ_EPOCH);
-       else
-               cfqd->cfq_epochsectors = ((CFQ_DISKBW * 1000000)/512)
-                       * (1000000 / CFQ_EPOCH) ;
+       /*
+        * just set it to some high value, we want anyone to be able to queue
+        * some requests. fairness is handled differently
+        */
+       cfqd->max_queued = q->nr_requests;
+       q->nr_requests = 8192;
+
+       cfqd->cfq_queued = cfq_queued;
+       cfqd->cfq_quantum = cfq_quantum;
 
        return 0;
 out_crqpool:
@@ -1296,12 +797,7 @@ static ssize_t __FUNC(struct cfq_data *cfqd, char *page)          \
        return cfq_var_show(__VAR, (page));                             \
 }
 SHOW_FUNCTION(cfq_quantum_show, cfqd->cfq_quantum);
-SHOW_FUNCTION(cfq_quantum_io_show, cfqd->cfq_quantum_io);
-SHOW_FUNCTION(cfq_idle_quantum_show, cfqd->cfq_idle_quantum);
-SHOW_FUNCTION(cfq_idle_quantum_io_show, cfqd->cfq_idle_quantum_io);
 SHOW_FUNCTION(cfq_queued_show, cfqd->cfq_queued);
-SHOW_FUNCTION(cfq_grace_rt_show, cfqd->cfq_grace_rt);
-SHOW_FUNCTION(cfq_grace_idle_show, cfqd->cfq_grace_idle);
 #undef SHOW_FUNCTION
 
 #define STORE_FUNCTION(__FUNC, __PTR, MIN, MAX)                                \
@@ -1315,271 +811,23 @@ static ssize_t __FUNC(struct cfq_data *cfqd, const char *page, size_t count)    \
        return ret;                                                     \
 }
 STORE_FUNCTION(cfq_quantum_store, &cfqd->cfq_quantum, 1, INT_MAX);
-STORE_FUNCTION(cfq_quantum_io_store, &cfqd->cfq_quantum_io, 4, INT_MAX);
-STORE_FUNCTION(cfq_idle_quantum_store, &cfqd->cfq_idle_quantum, 1, INT_MAX);
-STORE_FUNCTION(cfq_idle_quantum_io_store, &cfqd->cfq_idle_quantum_io, 4, INT_MAX);
 STORE_FUNCTION(cfq_queued_store, &cfqd->cfq_queued, 1, INT_MAX);
-STORE_FUNCTION(cfq_grace_rt_store, &cfqd->cfq_grace_rt, 0, INT_MAX);
-STORE_FUNCTION(cfq_grace_idle_store, &cfqd->cfq_grace_idle, 0, INT_MAX);
 #undef STORE_FUNCTION
 
-
-static ssize_t cfq_epoch_show(struct cfq_data *cfqd, char *page)
-{
-       return sprintf(page, "%lu\n", cfqd->cfq_epoch);
-}
-
-static ssize_t cfq_epoch_store(struct cfq_data *cfqd, const char *page, size_t count)
-{
-       char *p = (char *) page;
-       cfqd->cfq_epoch = simple_strtoul(p, &p, 10);
-       return count;
-}
-
-static ssize_t cfq_epochsectors_show(struct cfq_data *cfqd, char *page)
-{
-       return sprintf(page, "%lu\n", cfqd->cfq_epochsectors);
-}
-
-static ssize_t 
-cfq_epochsectors_store(struct cfq_data *cfqd, const char *page, size_t count)
-{
-       char *p = (char *) page;
-       cfqd->cfq_epochsectors = simple_strtoul(p, &p, 10);
-       return count;
-}
-
-/* Additional entries to get priority level data */
-static ssize_t
-cfq_prio_show(struct cfq_data *cfqd, char *page, unsigned int priolvl)
-{
-       int r1,r2,s1,s2,q1,q2;
-
-       if (!(priolvl >= IOPRIO_IDLE && priolvl <= IOPRIO_RT)) 
-               return 0;
-       
-       r1 = (int)atomic_read(&(cfqd->cid[priolvl].cum_rq_in));
-       r2 = (int)atomic_read(&(cfqd->cid[priolvl].cum_rq_out));
-       s1 = (int)atomic_read(&(cfqd->cid[priolvl].cum_sectors_in));
-       s2 = (int)atomic_read(&(cfqd->cid[priolvl].cum_sectors_out));
-       q1 = (int)atomic_read(&(cfqd->cid[priolvl].cum_queues_in)); 
-       q2 = (int)atomic_read(&(cfqd->cid[priolvl].cum_queues_out));
-       
-       return sprintf(page,"skip %d avsec %lu rate %lu new %lu"
-                      "rq (%d,%d) sec (%d,%d) q (%d,%d)\n",
-                      cfqd->cid[priolvl].nskip,
-                      cfqd->cid[priolvl].navsec,
-                      cfqd->cid[priolvl].csectorate,
-                      cfqd->cid[priolvl].lsectorate,
-//                    atomic_read(&cfqd->cid[priolvl].nskip),
-//                    atomic_read(&cfqd->cid[priolvl].navsec),
-//                    atomic_read(&cfqd->cid[priolvl].csectorate),
-//                    atomic_read(&cfqd->cid[priolvl].lsectorate),
-                      r1,r2,
-                      s1,s2,
-                      q1,q2);
-}
-
-#define SHOW_PRIO_DATA(__PRIOLVL)                                               \
-static ssize_t cfq_prio_##__PRIOLVL##_show(struct cfq_data *cfqd, char *page)  \
-{                                                                              \
-       return cfq_prio_show(cfqd,page,__PRIOLVL);                              \
-}
-SHOW_PRIO_DATA(0);
-SHOW_PRIO_DATA(1);
-SHOW_PRIO_DATA(2);
-SHOW_PRIO_DATA(3);
-SHOW_PRIO_DATA(4);
-SHOW_PRIO_DATA(5);
-SHOW_PRIO_DATA(6);
-SHOW_PRIO_DATA(7);
-SHOW_PRIO_DATA(8);
-SHOW_PRIO_DATA(9);
-SHOW_PRIO_DATA(10);
-SHOW_PRIO_DATA(11);
-SHOW_PRIO_DATA(12);
-SHOW_PRIO_DATA(13);
-SHOW_PRIO_DATA(14);
-SHOW_PRIO_DATA(15);
-SHOW_PRIO_DATA(16);
-SHOW_PRIO_DATA(17);
-SHOW_PRIO_DATA(18);
-SHOW_PRIO_DATA(19);
-SHOW_PRIO_DATA(20);
-#undef SHOW_PRIO_DATA
-
-
-static ssize_t cfq_prio_store(struct cfq_data *cfqd, const char *page, size_t count, int priolvl)
-{      
-       atomic_set(&(cfqd->cid[priolvl].cum_rq_in),0);
-       atomic_set(&(cfqd->cid[priolvl].cum_rq_out),0);
-       atomic_set(&(cfqd->cid[priolvl].cum_sectors_in),0);
-       atomic_set(&(cfqd->cid[priolvl].cum_sectors_out),0);
-       atomic_set(&(cfqd->cid[priolvl].cum_queues_in),0);
-       atomic_set(&(cfqd->cid[priolvl].cum_queues_out),0);
-
-       return count;
-}
-
-
-#define STORE_PRIO_DATA(__PRIOLVL)                                                                \
-static ssize_t cfq_prio_##__PRIOLVL##_store(struct cfq_data *cfqd, const char *page, size_t count) \
-{                                                                                                 \
-        return cfq_prio_store(cfqd,page,count,__PRIOLVL);                                          \
-}                  
-STORE_PRIO_DATA(0);     
-STORE_PRIO_DATA(1);
-STORE_PRIO_DATA(2);
-STORE_PRIO_DATA(3);
-STORE_PRIO_DATA(4);
-STORE_PRIO_DATA(5);
-STORE_PRIO_DATA(6);
-STORE_PRIO_DATA(7);
-STORE_PRIO_DATA(8);
-STORE_PRIO_DATA(9);
-STORE_PRIO_DATA(10);
-STORE_PRIO_DATA(11);
-STORE_PRIO_DATA(12);
-STORE_PRIO_DATA(13);
-STORE_PRIO_DATA(14);
-STORE_PRIO_DATA(15);
-STORE_PRIO_DATA(16);
-STORE_PRIO_DATA(17);
-STORE_PRIO_DATA(18);
-STORE_PRIO_DATA(19);
-STORE_PRIO_DATA(20);
-#undef STORE_PRIO_DATA
-
-
 static struct cfq_fs_entry cfq_quantum_entry = {
        .attr = {.name = "quantum", .mode = S_IRUGO | S_IWUSR },
        .show = cfq_quantum_show,
        .store = cfq_quantum_store,
 };
-static struct cfq_fs_entry cfq_quantum_io_entry = {
-       .attr = {.name = "quantum_io", .mode = S_IRUGO | S_IWUSR },
-       .show = cfq_quantum_io_show,
-       .store = cfq_quantum_io_store,
-};
-static struct cfq_fs_entry cfq_idle_quantum_entry = {
-       .attr = {.name = "idle_quantum", .mode = S_IRUGO | S_IWUSR },
-       .show = cfq_idle_quantum_show,
-       .store = cfq_idle_quantum_store,
-};
-static struct cfq_fs_entry cfq_idle_quantum_io_entry = {
-       .attr = {.name = "idle_quantum_io", .mode = S_IRUGO | S_IWUSR },
-       .show = cfq_idle_quantum_io_show,
-       .store = cfq_idle_quantum_io_store,
-};
 static struct cfq_fs_entry cfq_queued_entry = {
        .attr = {.name = "queued", .mode = S_IRUGO | S_IWUSR },
        .show = cfq_queued_show,
        .store = cfq_queued_store,
 };
-static struct cfq_fs_entry cfq_grace_rt_entry = {
-       .attr = {.name = "grace_rt", .mode = S_IRUGO | S_IWUSR },
-       .show = cfq_grace_rt_show,
-       .store = cfq_grace_rt_store,
-};
-static struct cfq_fs_entry cfq_grace_idle_entry = {
-       .attr = {.name = "grace_idle", .mode = S_IRUGO | S_IWUSR },
-       .show = cfq_grace_idle_show,
-       .store = cfq_grace_idle_store,
-};
-static struct cfq_fs_entry cfq_epoch_entry = {
-       .attr = {.name = "epoch", .mode = S_IRUGO | S_IWUSR },
-       .show = cfq_epoch_show,
-       .store = cfq_epoch_store,
-};
-static struct cfq_fs_entry cfq_epochsectors_entry = {
-       .attr = {.name = "epochsectors", .mode = S_IRUGO | S_IWUSR },
-       .show = cfq_epochsectors_show,
-       .store = cfq_epochsectors_store,
-};
-
-#define P_0_STR   "p0"
-#define P_1_STR   "p1"
-#define P_2_STR   "p2"
-#define P_3_STR   "p3"
-#define P_4_STR   "p4"
-#define P_5_STR   "p5"
-#define P_6_STR   "p6"
-#define P_7_STR   "p7"
-#define P_8_STR   "p8"
-#define P_9_STR   "p9"
-#define P_10_STR  "p10"
-#define P_11_STR  "p11"
-#define P_12_STR  "p12"
-#define P_13_STR  "p13"
-#define P_14_STR  "p14"
-#define P_15_STR  "p15"
-#define P_16_STR  "p16"
-#define P_17_STR  "p17"
-#define P_18_STR  "p18"
-#define P_19_STR  "p19"
-#define P_20_STR  "p20"
-
-
-#define CFQ_PRIO_SYSFS_ENTRY(__PRIOLVL)                                           \
-static struct cfq_fs_entry cfq_prio_##__PRIOLVL##_entry = {                \
-       .attr = {.name = P_##__PRIOLVL##_STR, .mode = S_IRUGO | S_IWUSR }, \
-       .show = cfq_prio_##__PRIOLVL##_show,                               \
-       .store = cfq_prio_##__PRIOLVL##_store,                             \
-};
-CFQ_PRIO_SYSFS_ENTRY(0);
-CFQ_PRIO_SYSFS_ENTRY(1);
-CFQ_PRIO_SYSFS_ENTRY(2);
-CFQ_PRIO_SYSFS_ENTRY(3);
-CFQ_PRIO_SYSFS_ENTRY(4);
-CFQ_PRIO_SYSFS_ENTRY(5);
-CFQ_PRIO_SYSFS_ENTRY(6);
-CFQ_PRIO_SYSFS_ENTRY(7);
-CFQ_PRIO_SYSFS_ENTRY(8);
-CFQ_PRIO_SYSFS_ENTRY(9);
-CFQ_PRIO_SYSFS_ENTRY(10);
-CFQ_PRIO_SYSFS_ENTRY(11);
-CFQ_PRIO_SYSFS_ENTRY(12);
-CFQ_PRIO_SYSFS_ENTRY(13);
-CFQ_PRIO_SYSFS_ENTRY(14);
-CFQ_PRIO_SYSFS_ENTRY(15);
-CFQ_PRIO_SYSFS_ENTRY(16);
-CFQ_PRIO_SYSFS_ENTRY(17);
-CFQ_PRIO_SYSFS_ENTRY(18);
-CFQ_PRIO_SYSFS_ENTRY(19);
-CFQ_PRIO_SYSFS_ENTRY(20);
-#undef CFQ_PRIO_SYSFS_ENTRY
 
 static struct attribute *default_attrs[] = {
        &cfq_quantum_entry.attr,
-       &cfq_quantum_io_entry.attr,
-       &cfq_idle_quantum_entry.attr,
-       &cfq_idle_quantum_io_entry.attr,
        &cfq_queued_entry.attr,
-       &cfq_grace_rt_entry.attr,
-       &cfq_grace_idle_entry.attr,
-       &cfq_epoch_entry.attr,
-       &cfq_epochsectors_entry.attr,
-       &cfq_prio_0_entry.attr,
-       &cfq_prio_1_entry.attr,
-       &cfq_prio_2_entry.attr,
-       &cfq_prio_3_entry.attr,
-       &cfq_prio_4_entry.attr,
-       &cfq_prio_5_entry.attr,
-       &cfq_prio_6_entry.attr,
-       &cfq_prio_7_entry.attr,
-       &cfq_prio_8_entry.attr,
-       &cfq_prio_9_entry.attr,
-       &cfq_prio_10_entry.attr,
-       &cfq_prio_11_entry.attr,
-       &cfq_prio_12_entry.attr,
-       &cfq_prio_13_entry.attr,
-       &cfq_prio_14_entry.attr,
-       &cfq_prio_15_entry.attr,
-       &cfq_prio_16_entry.attr,
-       &cfq_prio_17_entry.attr,
-       &cfq_prio_18_entry.attr,
-       &cfq_prio_19_entry.attr,
-       &cfq_prio_20_entry.attr,
        NULL,
 };
 
@@ -1635,7 +883,6 @@ elevator_t iosched_cfq = {
        .elevator_set_req_fn =          cfq_set_request,
        .elevator_put_req_fn =          cfq_put_request,
        .elevator_may_queue_fn =        cfq_may_queue,
-       .elevator_set_congested_fn =    cfq_queue_congested,
        .elevator_init_fn =             cfq_init,
        .elevator_exit_fn =             cfq_exit,
 };