From 6b41fd1785f4effe2f3bd40da864415812f5b8c9 Mon Sep 17 00:00:00 2001
From: Linus Torvalds <torvalds@g5.osdl.org>
Date: Tue, 20 Jun 2006 19:44:03 -0700
Subject: Fix up CFQ scheduler for recent rbtree node shrinkage

The color is now in the low bits of the parent pointer, and initializing
it to 0 happens as part of the whole memset above, so just remove the
unnecessary RB_CLEAR_COLOR.

Signed-off-by: Linus Torvalds <torvalds@osdl.org>
---
 block/cfq-iosched.c | 1 -
 1 file changed, 1 deletion(-)

(limited to 'block')

diff --git a/block/cfq-iosched.c b/block/cfq-iosched.c
index 6200d9b9af28..e2e6ad0a158e 100644
--- a/block/cfq-iosched.c
+++ b/block/cfq-iosched.c
@@ -1318,7 +1318,6 @@ cfq_alloc_io_context(struct cfq_data *cfqd, gfp_t gfp_mask)
 
 	if (cic) {
 		memset(cic, 0, sizeof(*cic));
-		RB_CLEAR_COLOR(&cic->rb_node);
 		cic->last_end_request = jiffies;
 		INIT_LIST_HEAD(&cic->queue_list);
 		cic->dtor = cfq_free_io_context;
-- 
cgit v1.2.3


From b9d9c82b4d081feb464f62dfc786c8621d09ecd2 Mon Sep 17 00:00:00 2001
From: Kay Sievers <kay.sievers@suse.de>
Date: Thu, 15 Jun 2006 15:31:56 +0200
Subject: [PATCH] Driver core: add generic "subsystem" link to all devices

Like the SUBSYTEM= key we find in the environment of the uevent, this
creates a generic "subsystem" link in sysfs for every device. Userspace
usually doesn't care at all if its a "class" or a "bus" device. This
provides an unified way to determine the subsytem of a device, regardless
of the way the driver core has created it.

Signed-off-by: Kay Sievers <kay.sievers@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
---
 block/genhd.c         | 7 ++-----
 drivers/base/bus.c    | 2 ++
 drivers/base/class.c  | 2 ++
 drivers/base/core.c   | 9 +++++++--
 fs/partitions/check.c | 4 ++++
 5 files changed, 17 insertions(+), 7 deletions(-)

(limited to 'block')

diff --git a/block/genhd.c b/block/genhd.c
index 5a8d3bf02f17..8d7339511e5e 100644
--- a/block/genhd.c
+++ b/block/genhd.c
@@ -17,8 +17,7 @@
 #include <linux/buffer_head.h>
 #include <linux/mutex.h>
 
-static struct subsystem block_subsys;
-
+struct subsystem block_subsys;
 static DEFINE_MUTEX(block_subsys_lock);
 
 /*
@@ -511,9 +510,7 @@ static struct kset_uevent_ops block_uevent_ops = {
 	.uevent		= block_uevent,
 };
 
-/* declare block_subsys. */
-static decl_subsys(block, &ktype_block, &block_uevent_ops);
-
+decl_subsys(block, &ktype_block, &block_uevent_ops);
 
 /*
  * aggregate disk stat collector.  Uses the same stats that the sysfs
diff --git a/drivers/base/bus.c b/drivers/base/bus.c
index 64ba9011d1a8..050d86d0b872 100644
--- a/drivers/base/bus.c
+++ b/drivers/base/bus.c
@@ -374,6 +374,7 @@ int bus_add_device(struct device * dev)
 		error = device_add_attrs(bus, dev);
 		if (!error) {
 			sysfs_create_link(&bus->devices.kobj, &dev->kobj, dev->bus_id);
+			sysfs_create_link(&dev->kobj, &dev->bus->subsys.kset.kobj, "subsystem");
 			sysfs_create_link(&dev->kobj, &dev->bus->subsys.kset.kobj, "bus");
 		}
 	}
@@ -408,6 +409,7 @@ void bus_attach_device(struct device * dev)
 void bus_remove_device(struct device * dev)
 {
 	if (dev->bus) {
+		sysfs_remove_link(&dev->kobj, "subsystem");
 		sysfs_remove_link(&dev->kobj, "bus");
 		sysfs_remove_link(&dev->bus->devices.kobj, dev->bus_id);
 		device_remove_attrs(dev->bus, dev);
diff --git a/drivers/base/class.c b/drivers/base/class.c
index 50e841a33af0..9aa127460262 100644
--- a/drivers/base/class.c
+++ b/drivers/base/class.c
@@ -561,6 +561,7 @@ int class_device_add(struct class_device *class_dev)
 		goto out2;
 
 	/* add the needed attributes to this device */
+	sysfs_create_link(&class_dev->kobj, &parent_class->subsys.kset.kobj, "subsystem");
 	class_dev->uevent_attr.attr.name = "uevent";
 	class_dev->uevent_attr.attr.mode = S_IWUSR;
 	class_dev->uevent_attr.attr.owner = parent_class->owner;
@@ -737,6 +738,7 @@ void class_device_del(struct class_device *class_dev)
 		sysfs_remove_link(&class_dev->kobj, "device");
 		sysfs_remove_link(&class_dev->dev->kobj, class_name);
 	}
+	sysfs_remove_link(&class_dev->kobj, "subsystem");
 	class_device_remove_file(class_dev, &class_dev->uevent_attr);
 	if (class_dev->devt_attr)
 		class_device_remove_file(class_dev, class_dev->devt_attr);
diff --git a/drivers/base/core.c b/drivers/base/core.c
index 252cf403f891..cc8bb97427d0 100644
--- a/drivers/base/core.c
+++ b/drivers/base/core.c
@@ -319,9 +319,12 @@ int device_add(struct device *dev)
 		dev->devt_attr = attr;
 	}
 
-	if (dev->class)
+	if (dev->class) {
+		sysfs_create_link(&dev->kobj, &dev->class->subsys.kset.kobj,
+				  "subsystem");
 		sysfs_create_link(&dev->class->subsys.kset.kobj, &dev->kobj,
 				  dev->bus_id);
+	}
 
 	if ((error = device_pm_add(dev)))
 		goto PMError;
@@ -422,8 +425,10 @@ void device_del(struct device * dev)
 		klist_del(&dev->knode_parent);
 	if (dev->devt_attr)
 		device_remove_file(dev, dev->devt_attr);
-	if (dev->class)
+	if (dev->class) {
+		sysfs_remove_link(&dev->kobj, "subsystem");
 		sysfs_remove_link(&dev->class->subsys.kset.kobj, dev->bus_id);
+	}
 	device_remove_file(dev, &dev->uevent_attr);
 
 	/* Notify the platform of the removal, in case they
diff --git a/fs/partitions/check.c b/fs/partitions/check.c
index 7ef1f094de91..8851b81e7c5a 100644
--- a/fs/partitions/check.c
+++ b/fs/partitions/check.c
@@ -329,6 +329,7 @@ void delete_partition(struct gendisk *disk, int part)
 	p->ios[0] = p->ios[1] = 0;
 	p->sectors[0] = p->sectors[1] = 0;
 	devfs_remove("%s/part%d", disk->devfs_name, part);
+	sysfs_remove_link(&p->kobj, "subsystem");
 	if (p->holder_dir)
 		kobject_unregister(p->holder_dir);
 	kobject_uevent(&p->kobj, KOBJ_REMOVE);
@@ -363,6 +364,7 @@ void add_partition(struct gendisk *disk, int part, sector_t start, sector_t len)
 	kobject_add(&p->kobj);
 	if (!disk->part_uevent_suppress)
 		kobject_uevent(&p->kobj, KOBJ_ADD);
+	sysfs_create_link(&p->kobj, &block_subsys.kset.kobj, "subsystem");
 	partition_sysfs_add_subdir(p);
 	disk->part[part-1] = p;
 }
@@ -398,6 +400,7 @@ static void disk_sysfs_symlinks(struct gendisk *disk)
 			kfree(disk_name);
 		}
 	}
+	sysfs_create_link(&disk->kobj, &block_subsys.kset.kobj, "subsystem");
 }
 
 /* Not exported, helper to add_disk(). */
@@ -548,5 +551,6 @@ void del_gendisk(struct gendisk *disk)
 		put_device(disk->driverfs_dev);
 		disk->driverfs_dev = NULL;
 	}
+	sysfs_remove_link(&disk->kobj, "subsystem");
 	kobject_del(&disk->kobj);
 }
-- 
cgit v1.2.3


From 626ab0e69d376fa07599af669af8ba92d58e87c1 Mon Sep 17 00:00:00 2001
From: Oleg Nesterov <oleg@tv-sign.ru>
Date: Fri, 23 Jun 2006 02:05:55 -0700
Subject: [PATCH] list: use list_replace_init() instead of list_splice_init()

list_splice_init(list, head) does unneeded job if it is known that
list_empty(head) == 1.  We can use list_replace_init() instead.

Signed-off-by: Oleg Nesterov <oleg@tv-sign.ru>
Acked-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
---
 arch/i386/mm/pageattr.c | 8 ++++----
 block/ll_rw_blk.c       | 5 ++---
 fs/aio.c                | 4 ++--
 kernel/timer.c          | 8 ++++----
 kernel/workqueue.c      | 4 ++--
 net/core/dev.c          | 6 +++---
 net/core/link_watch.c   | 5 ++---
 7 files changed, 19 insertions(+), 21 deletions(-)

(limited to 'block')

diff --git a/arch/i386/mm/pageattr.c b/arch/i386/mm/pageattr.c
index 92c3d9f0e731..0887b34bc59b 100644
--- a/arch/i386/mm/pageattr.c
+++ b/arch/i386/mm/pageattr.c
@@ -209,19 +209,19 @@ int change_page_attr(struct page *page, int numpages, pgprot_t prot)
 }
 
 void global_flush_tlb(void)
-{ 
-	LIST_HEAD(l);
+{
+	struct list_head l;
 	struct page *pg, *next;
 
 	BUG_ON(irqs_disabled());
 
 	spin_lock_irq(&cpa_lock);
-	list_splice_init(&df_list, &l);
+	list_replace_init(&df_list, &l);
 	spin_unlock_irq(&cpa_lock);
 	flush_map();
 	list_for_each_entry_safe(pg, next, &l, lru)
 		__free_page(pg);
-} 
+}
 
 #ifdef CONFIG_DEBUG_PAGEALLOC
 void kernel_map_pages(struct page *page, int numpages, int enable)
diff --git a/block/ll_rw_blk.c b/block/ll_rw_blk.c
index 7eb36c53f4b7..465b54312c59 100644
--- a/block/ll_rw_blk.c
+++ b/block/ll_rw_blk.c
@@ -3359,12 +3359,11 @@ EXPORT_SYMBOL(end_that_request_chunk);
  */
 static void blk_done_softirq(struct softirq_action *h)
 {
-	struct list_head *cpu_list;
-	LIST_HEAD(local_list);
+	struct list_head *cpu_list, local_list;
 
 	local_irq_disable();
 	cpu_list = &__get_cpu_var(blk_cpu_done);
-	list_splice_init(cpu_list, &local_list);
+	list_replace_init(cpu_list, &local_list);
 	local_irq_enable();
 
 	while (!list_empty(&local_list)) {
diff --git a/fs/aio.c b/fs/aio.c
index e41e932ba489..8c34a62df7d7 100644
--- a/fs/aio.c
+++ b/fs/aio.c
@@ -777,11 +777,11 @@ out:
 static int __aio_run_iocbs(struct kioctx *ctx)
 {
 	struct kiocb *iocb;
-	LIST_HEAD(run_list);
+	struct list_head run_list;
 
 	assert_spin_locked(&ctx->ctx_lock);
 
-	list_splice_init(&ctx->run_list, &run_list);
+	list_replace_init(&ctx->run_list, &run_list);
 	while (!list_empty(&run_list)) {
 		iocb = list_entry(run_list.next, struct kiocb,
 			ki_run_list);
diff --git a/kernel/timer.c b/kernel/timer.c
index 9e49deed468c..3bf0e9ed2dbe 100644
--- a/kernel/timer.c
+++ b/kernel/timer.c
@@ -419,10 +419,10 @@ static inline void __run_timers(tvec_base_t *base)
 
 	spin_lock_irq(&base->lock);
 	while (time_after_eq(jiffies, base->timer_jiffies)) {
-		struct list_head work_list = LIST_HEAD_INIT(work_list);
+		struct list_head work_list;
 		struct list_head *head = &work_list;
  		int index = base->timer_jiffies & TVR_MASK;
- 
+
 		/*
 		 * Cascade timers:
 		 */
@@ -431,8 +431,8 @@ static inline void __run_timers(tvec_base_t *base)
 				(!cascade(base, &base->tv3, INDEX(1))) &&
 					!cascade(base, &base->tv4, INDEX(2)))
 			cascade(base, &base->tv5, INDEX(3));
-		++base->timer_jiffies; 
-		list_splice_init(base->tv1.vec + index, &work_list);
+		++base->timer_jiffies;
+		list_replace_init(base->tv1.vec + index, &work_list);
 		while (!list_empty(head)) {
 			void (*fn)(unsigned long);
 			unsigned long data;
diff --git a/kernel/workqueue.c b/kernel/workqueue.c
index 880fb415a8f6..740c5abceb07 100644
--- a/kernel/workqueue.c
+++ b/kernel/workqueue.c
@@ -531,11 +531,11 @@ int current_is_keventd(void)
 static void take_over_work(struct workqueue_struct *wq, unsigned int cpu)
 {
 	struct cpu_workqueue_struct *cwq = per_cpu_ptr(wq->cpu_wq, cpu);
-	LIST_HEAD(list);
+	struct list_head list;
 	struct work_struct *work;
 
 	spin_lock_irq(&cwq->lock);
-	list_splice_init(&cwq->worklist, &list);
+	list_replace_init(&cwq->worklist, &list);
 
 	while (!list_empty(&list)) {
 		printk("Taking work for %s\n", wq->name);
diff --git a/net/core/dev.c b/net/core/dev.c
index ab39fe17cb58..195a5e96b2d1 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -2980,7 +2980,7 @@ static void netdev_wait_allrefs(struct net_device *dev)
 static DEFINE_MUTEX(net_todo_run_mutex);
 void netdev_run_todo(void)
 {
-	struct list_head list = LIST_HEAD_INIT(list);
+	struct list_head list;
 
 	/* Need to guard against multiple cpu's getting out of order. */
 	mutex_lock(&net_todo_run_mutex);
@@ -2995,9 +2995,9 @@ void netdev_run_todo(void)
 
 	/* Snapshot list, allow later requests */
 	spin_lock(&net_todo_list_lock);
-	list_splice_init(&net_todo_list, &list);
+	list_replace_init(&net_todo_list, &list);
 	spin_unlock(&net_todo_list_lock);
-		
+
 	while (!list_empty(&list)) {
 		struct net_device *dev
 			= list_entry(list.next, struct net_device, todo_list);
diff --git a/net/core/link_watch.c b/net/core/link_watch.c
index 646937cc2d84..0f37266411b5 100644
--- a/net/core/link_watch.c
+++ b/net/core/link_watch.c
@@ -91,11 +91,10 @@ static void rfc2863_policy(struct net_device *dev)
 /* Must be called with the rtnl semaphore held */
 void linkwatch_run_queue(void)
 {
-	LIST_HEAD(head);
-	struct list_head *n, *next;
+	struct list_head head, *n, *next;
 
 	spin_lock_irq(&lweventlist_lock);
-	list_splice_init(&lweventlist, &head);
+	list_replace_init(&lweventlist, &head);
 	spin_unlock_irq(&lweventlist_lock);
 
 	list_for_each_safe(n, next, &head) {
-- 
cgit v1.2.3


From bae386f7884aa3720cc7880b36a41a1d2b9c327b Mon Sep 17 00:00:00 2001
From: Akinobu Mita <mita@miraclelinux.com>
Date: Mon, 24 Apr 2006 21:12:59 +0200
Subject: [PATCH] iosched: use hlist for request hashtable

Use hlist instead of list_head for request hashtable in deadline-iosched
and as-iosched. It also can remove the flag to know hashed or unhashed.

Signed-off-by: Akinobu Mita <mita@miraclelinux.com>
Signed-off-by: Jens Axboe <axboe@suse.de>

 block/as-iosched.c       |   45 +++++++++++++++++++--------------------------
 block/deadline-iosched.c |   39 ++++++++++++++++-----------------------
 2 files changed, 35 insertions(+), 49 deletions(-)
---
 block/as-iosched.c       | 45 +++++++++++++++++++--------------------------
 block/deadline-iosched.c | 39 ++++++++++++++++-----------------------
 2 files changed, 35 insertions(+), 49 deletions(-)

(limited to 'block')

diff --git a/block/as-iosched.c b/block/as-iosched.c
index 0c750393be4a..9b13d72ffefa 100644
--- a/block/as-iosched.c
+++ b/block/as-iosched.c
@@ -96,7 +96,7 @@ 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 list_head *hash;		/* request hash */
+	struct hlist_head *hash;	/* request hash */
 
 	unsigned long exit_prob;	/* probability a task will exit while
 					   being waited on */
@@ -165,8 +165,7 @@ struct as_rq {
 	/*
 	 * request hash, key is the ending offset (for back merge lookup)
 	 */
-	struct list_head hash;
-	unsigned int on_hash;
+	struct hlist_node hash;
 
 	/*
 	 * expire fifo
@@ -282,17 +281,15 @@ static const int as_hash_shift = 6;
 #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)
-#define list_entry_hash(ptr)	list_entry((ptr), struct as_rq, hash)
 
 static inline void __as_del_arq_hash(struct as_rq *arq)
 {
-	arq->on_hash = 0;
-	list_del_init(&arq->hash);
+	hlist_del_init(&arq->hash);
 }
 
 static inline void as_del_arq_hash(struct as_rq *arq)
 {
-	if (arq->on_hash)
+	if (!hlist_unhashed(&arq->hash))
 		__as_del_arq_hash(arq);
 }
 
@@ -300,10 +297,9 @@ static void as_add_arq_hash(struct as_data *ad, struct as_rq *arq)
 {
 	struct request *rq = arq->request;
 
-	BUG_ON(arq->on_hash);
+	BUG_ON(!hlist_unhashed(&arq->hash));
 
-	arq->on_hash = 1;
-	list_add(&arq->hash, &ad->hash[AS_HASH_FN(rq_hash_key(rq))]);
+	hlist_add_head(&arq->hash, &ad->hash[AS_HASH_FN(rq_hash_key(rq))]);
 }
 
 /*
@@ -312,31 +308,29 @@ static void as_add_arq_hash(struct as_data *ad, struct as_rq *arq)
 static inline void as_hot_arq_hash(struct as_data *ad, struct as_rq *arq)
 {
 	struct request *rq = arq->request;
-	struct list_head *head = &ad->hash[AS_HASH_FN(rq_hash_key(rq))];
+	struct hlist_head *head = &ad->hash[AS_HASH_FN(rq_hash_key(rq))];
 
-	if (!arq->on_hash) {
+	if (hlist_unhashed(&arq->hash)) {
 		WARN_ON(1);
 		return;
 	}
 
-	if (arq->hash.prev != head) {
-		list_del(&arq->hash);
-		list_add(&arq->hash, head);
+	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 list_head *hash_list = &ad->hash[AS_HASH_FN(offset)];
-	struct list_head *entry, *next = hash_list->next;
+	struct hlist_head *hash_list = &ad->hash[AS_HASH_FN(offset)];
+	struct hlist_node *entry, *next;
+	struct as_rq *arq;
 
-	while ((entry = next) != hash_list) {
-		struct as_rq *arq = list_entry_hash(entry);
+	hlist_for_each_entry_safe(arq, entry, next, hash_list, hash) {
 		struct request *__rq = arq->request;
 
-		next = entry->next;
-
-		BUG_ON(!arq->on_hash);
+		BUG_ON(hlist_unhashed(&arq->hash));
 
 		if (!rq_mergeable(__rq)) {
 			as_del_arq_hash(arq);
@@ -1601,8 +1595,7 @@ static int as_set_request(request_queue_t *q, struct request *rq,
 		arq->request = rq;
 		arq->state = AS_RQ_PRESCHED;
 		arq->io_context = NULL;
-		INIT_LIST_HEAD(&arq->hash);
-		arq->on_hash = 0;
+		INIT_HLIST_NODE(&arq->hash);
 		INIT_LIST_HEAD(&arq->fifo);
 		rq->elevator_private = arq;
 		return 0;
@@ -1662,7 +1655,7 @@ static void *as_init_queue(request_queue_t *q, elevator_t *e)
 
 	ad->q = q; /* Identify what queue the data belongs to */
 
-	ad->hash = kmalloc_node(sizeof(struct list_head)*AS_HASH_ENTRIES,
+	ad->hash = kmalloc_node(sizeof(struct hlist_head)*AS_HASH_ENTRIES,
 				GFP_KERNEL, q->node);
 	if (!ad->hash) {
 		kfree(ad);
@@ -1684,7 +1677,7 @@ static void *as_init_queue(request_queue_t *q, elevator_t *e)
 	INIT_WORK(&ad->antic_work, as_work_handler, q);
 
 	for (i = 0; i < AS_HASH_ENTRIES; i++)
-		INIT_LIST_HEAD(&ad->hash[i]);
+		INIT_HLIST_HEAD(&ad->hash[i]);
 
 	INIT_LIST_HEAD(&ad->fifo_list[REQ_SYNC]);
 	INIT_LIST_HEAD(&ad->fifo_list[REQ_ASYNC]);
diff --git a/block/deadline-iosched.c b/block/deadline-iosched.c
index c94de8e12fbf..e5bccaaed563 100644
--- a/block/deadline-iosched.c
+++ b/block/deadline-iosched.c
@@ -30,8 +30,7 @@ static const int deadline_hash_shift = 5;
 #define DL_HASH_FN(sec)		(hash_long(DL_HASH_BLOCK((sec)), deadline_hash_shift))
 #define DL_HASH_ENTRIES		(1 << deadline_hash_shift)
 #define rq_hash_key(rq)		((rq)->sector + (rq)->nr_sectors)
-#define list_entry_hash(ptr)	list_entry((ptr), struct deadline_rq, hash)
-#define ON_HASH(drq)		(drq)->on_hash
+#define ON_HASH(drq)		(!hlist_unhashed(&(drq)->hash))
 
 struct deadline_data {
 	/*
@@ -48,7 +47,7 @@ struct deadline_data {
 	 * next in sort order. read, write or both are NULL
 	 */
 	struct deadline_rq *next_drq[2];
-	struct list_head *hash;		/* request hash */
+	struct hlist_head *hash;	/* request hash */
 	unsigned int batching;		/* number of sequential requests made */
 	sector_t last_sector;		/* head position */
 	unsigned int starved;		/* times reads have starved writes */
@@ -79,8 +78,7 @@ struct deadline_rq {
 	/*
 	 * request hash, key is the ending offset (for back merge lookup)
 	 */
-	struct list_head hash;
-	char on_hash;
+	struct hlist_node hash;
 
 	/*
 	 * expire fifo
@@ -100,8 +98,7 @@ static kmem_cache_t *drq_pool;
  */
 static inline void __deadline_del_drq_hash(struct deadline_rq *drq)
 {
-	drq->on_hash = 0;
-	list_del_init(&drq->hash);
+	hlist_del_init(&drq->hash);
 }
 
 static inline void deadline_del_drq_hash(struct deadline_rq *drq)
@@ -117,8 +114,7 @@ deadline_add_drq_hash(struct deadline_data *dd, struct deadline_rq *drq)
 
 	BUG_ON(ON_HASH(drq));
 
-	drq->on_hash = 1;
-	list_add(&drq->hash, &dd->hash[DL_HASH_FN(rq_hash_key(rq))]);
+	hlist_add_head(&drq->hash, &dd->hash[DL_HASH_FN(rq_hash_key(rq))]);
 }
 
 /*
@@ -128,26 +124,24 @@ static inline void
 deadline_hot_drq_hash(struct deadline_data *dd, struct deadline_rq *drq)
 {
 	struct request *rq = drq->request;
-	struct list_head *head = &dd->hash[DL_HASH_FN(rq_hash_key(rq))];
+	struct hlist_head *head = &dd->hash[DL_HASH_FN(rq_hash_key(rq))];
 
-	if (ON_HASH(drq) && drq->hash.prev != head) {
-		list_del(&drq->hash);
-		list_add(&drq->hash, head);
+	if (ON_HASH(drq) && &drq->hash != head->first) {
+		hlist_del(&drq->hash);
+		hlist_add_head(&drq->hash, head);
 	}
 }
 
 static struct request *
 deadline_find_drq_hash(struct deadline_data *dd, sector_t offset)
 {
-	struct list_head *hash_list = &dd->hash[DL_HASH_FN(offset)];
-	struct list_head *entry, *next = hash_list->next;
+	struct hlist_head *hash_list = &dd->hash[DL_HASH_FN(offset)];
+	struct hlist_node *entry, *next;
+	struct deadline_rq *drq;
 
-	while ((entry = next) != hash_list) {
-		struct deadline_rq *drq = list_entry_hash(entry);
+	hlist_for_each_entry_safe(drq, entry, next, hash_list, hash) {
 		struct request *__rq = drq->request;
 
-		next = entry->next;
-		
 		BUG_ON(!ON_HASH(drq));
 
 		if (!rq_mergeable(__rq)) {
@@ -625,7 +619,7 @@ static void *deadline_init_queue(request_queue_t *q, elevator_t *e)
 		return NULL;
 	memset(dd, 0, sizeof(*dd));
 
-	dd->hash = kmalloc_node(sizeof(struct list_head)*DL_HASH_ENTRIES,
+	dd->hash = kmalloc_node(sizeof(struct hlist_head)*DL_HASH_ENTRIES,
 				GFP_KERNEL, q->node);
 	if (!dd->hash) {
 		kfree(dd);
@@ -641,7 +635,7 @@ static void *deadline_init_queue(request_queue_t *q, elevator_t *e)
 	}
 
 	for (i = 0; i < DL_HASH_ENTRIES; i++)
-		INIT_LIST_HEAD(&dd->hash[i]);
+		INIT_HLIST_HEAD(&dd->hash[i]);
 
 	INIT_LIST_HEAD(&dd->fifo_list[READ]);
 	INIT_LIST_HEAD(&dd->fifo_list[WRITE]);
@@ -677,8 +671,7 @@ deadline_set_request(request_queue_t *q, struct request *rq, struct bio *bio,
 		RB_CLEAR(&drq->rb_node);
 		drq->request = rq;
 
-		INIT_LIST_HEAD(&drq->hash);
-		drq->on_hash = 0;
+		INIT_HLIST_NODE(&drq->hash);
 
 		INIT_LIST_HEAD(&drq->fifo);
 
-- 
cgit v1.2.3


From a038e2536472b4dd932399b5277e65f188811de5 Mon Sep 17 00:00:00 2001
From: Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it>
Date: Mon, 5 Jun 2006 12:09:01 +0200
Subject: [PATCH] blk_start_queue() must be called with irq disabled - add
 warning

The queue lock can be taken from interrupts so it must always be taken with
irq disabling primitives.  Some primitives already verify this.
blk_start_queue() is called under this lock, so interrupts must be
disabled.

Also document this requirement clearly in blk_init_queue(), where the queue
spinlock is set.

Signed-off-by: Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Jens Axboe <axboe@suse.de>
---
 block/ll_rw_blk.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

(limited to 'block')

diff --git a/block/ll_rw_blk.c b/block/ll_rw_blk.c
index 465b54312c59..17c42ddd31db 100644
--- a/block/ll_rw_blk.c
+++ b/block/ll_rw_blk.c
@@ -1663,6 +1663,8 @@ static void blk_unplug_timeout(unsigned long data)
  **/
 void blk_start_queue(request_queue_t *q)
 {
+	WARN_ON(!irqs_disabled());
+
 	clear_bit(QUEUE_FLAG_STOPPED, &q->queue_flags);
 
 	/*
@@ -1878,7 +1880,8 @@ EXPORT_SYMBOL(blk_alloc_queue_node);
  *    get dealt with eventually.
  *
  *    The queue spin lock must be held while manipulating the requests on the
- *    request queue.
+ *    request queue; this lock will be taken also from interrupt context, so irq
+ *    disabling is needed for it.
  *
  *    Function returns a pointer to the initialized request queue, or NULL if
  *    it didn't succeed.
-- 
cgit v1.2.3


From acf421755593f7d7bd9352d57eda796c6eb4fa43 Mon Sep 17 00:00:00 2001
From: Dave Jones <davej@redhat.com>
Date: Mon, 12 Jun 2006 14:20:58 +0200
Subject: [PATCH] remove dead code from elevator switching

We already drop the refcount in elevator_exit(), and as
we're setting 'e' to NULL, we'll never take that branch anyway.
Finally, as 'e' is a local var that isn't referenced afterwards,
setting it to NULL is pointless.

Signed-off-by: Dave Jones <davej@redhat.com>
Signed-off-by: Jens Axboe <axboe@suse.de>
---
 block/elevator.c | 3 ---
 1 file changed, 3 deletions(-)

(limited to 'block')

diff --git a/block/elevator.c b/block/elevator.c
index a0afdd317cef..d00b283f31d2 100644
--- a/block/elevator.c
+++ b/block/elevator.c
@@ -850,12 +850,9 @@ fail_register:
 	 * one again (along with re-adding the sysfs dir)
 	 */
 	elevator_exit(e);
-	e = NULL;
 	q->elevator = old_elevator;
 	elv_register_queue(q);
 	clear_bit(QUEUE_FLAG_ELVSWITCH, &q->queue_flags);
-	if (e)
-		kobject_put(&e->kobj);
 	return 0;
 }
 
-- 
cgit v1.2.3


From 271f18f102c789f59644bb6c53a69da1df72b2f4 Mon Sep 17 00:00:00 2001
From: Jens Axboe <axboe@suse.de>
Date: Tue, 13 Jun 2006 08:08:38 +0200
Subject: [PATCH] cfq-iosched: Don't set the queue batching limits

We cannot update them if the user changes nr_requests, so don't
set it in the first place. The gains are pretty questionable as
well. The batching loss has been shown to decrease throughput.

Signed-off-by: Jens Axboe <axboe@suse.de>
---
 block/cfq-iosched.c | 48 +++---------------------------------------------
 1 file changed, 3 insertions(+), 45 deletions(-)

(limited to 'block')

diff --git a/block/cfq-iosched.c b/block/cfq-iosched.c
index e2e6ad0a158e..c88f161d3fb3 100644
--- a/block/cfq-iosched.c
+++ b/block/cfq-iosched.c
@@ -123,8 +123,6 @@ struct cfq_data {
 	 */
 	struct hlist_head *crq_hash;
 
-	unsigned int max_queued;
-
 	mempool_t *crq_pool;
 
 	int rq_in_driver;
@@ -1910,7 +1908,6 @@ static inline int
 __cfq_may_queue(struct cfq_data *cfqd, struct cfq_queue *cfqq,
 		struct task_struct *task, int rw)
 {
-#if 1
 	if ((cfq_cfqq_wait_request(cfqq) || cfq_cfqq_must_alloc(cfqq)) &&
 	    !cfq_cfqq_must_alloc_slice(cfqq)) {
 		cfq_mark_cfqq_must_alloc_slice(cfqq);
@@ -1918,39 +1915,6 @@ __cfq_may_queue(struct cfq_data *cfqd, struct cfq_queue *cfqq,
 	}
 
 	return ELV_MQUEUE_MAY;
-#else
-	if (!cfqq || task->flags & PF_MEMALLOC)
-		return ELV_MQUEUE_MAY;
-	if (!cfqq->allocated[rw] || cfq_cfqq_must_alloc(cfqq)) {
-		if (cfq_cfqq_wait_request(cfqq))
-			return ELV_MQUEUE_MUST;
-
-		/*
-		 * only allow 1 ELV_MQUEUE_MUST per slice, otherwise we
-		 * can quickly flood the queue with writes from a single task
-		 */
-		if (rw == READ || !cfq_cfqq_must_alloc_slice(cfqq)) {
-			cfq_mark_cfqq_must_alloc_slice(cfqq);
-			return ELV_MQUEUE_MUST;
-		}
-
-		return ELV_MQUEUE_MAY;
-	}
-	if (cfq_class_idle(cfqq))
-		return ELV_MQUEUE_NO;
-	if (cfqq->allocated[rw] >= cfqd->max_queued) {
-		struct io_context *ioc = get_io_context(GFP_ATOMIC);
-		int ret = ELV_MQUEUE_NO;
-
-		if (ioc && ioc->nr_batch_requests)
-			ret = ELV_MQUEUE_MAY;
-
-		put_io_context(ioc);
-		return ret;
-	}
-
-	return ELV_MQUEUE_MAY;
-#endif
 }
 
 static int cfq_may_queue(request_queue_t *q, int rw, struct bio *bio)
@@ -1979,16 +1943,13 @@ static int cfq_may_queue(request_queue_t *q, int rw, struct bio *bio)
 static void cfq_check_waiters(request_queue_t *q, struct cfq_queue *cfqq)
 {
 	struct cfq_data *cfqd = q->elevator->elevator_data;
-	struct request_list *rl = &q->rq;
 
-	if (cfqq->allocated[READ] <= cfqd->max_queued || cfqd->rq_starved) {
+	if (unlikely(cfqd->rq_starved)) {
+		struct request_list *rl = &q->rq;
+
 		smp_mb();
 		if (waitqueue_active(&rl->wait[READ]))
 			wake_up(&rl->wait[READ]);
-	}
-
-	if (cfqq->allocated[WRITE] <= cfqd->max_queued || cfqd->rq_starved) {
-		smp_mb();
 		if (waitqueue_active(&rl->wait[WRITE]))
 			wake_up(&rl->wait[WRITE]);
 	}
@@ -2278,9 +2239,6 @@ static void *cfq_init_queue(request_queue_t *q, elevator_t *e)
 
 	cfqd->queue = q;
 
-	cfqd->max_queued = q->nr_requests / 4;
-	q->nr_batching = cfq_queued;
-
 	init_timer(&cfqd->idle_slice_timer);
 	cfqd->idle_slice_timer.function = cfq_idle_slice_timer;
 	cfqd->idle_slice_timer.data = (unsigned long) cfqd;
-- 
cgit v1.2.3


From b31dc66a54ad986b6b73bdc49c8efc17cbad1833 Mon Sep 17 00:00:00 2001
From: Jens Axboe <axboe@suse.de>
Date: Tue, 13 Jun 2006 08:26:10 +0200
Subject: [PATCH] Kill PF_SYNCWRITE flag

A process flag to indicate whether we are doing sync io is incredibly
ugly. It also causes performance problems when one does a lot of async
io and then proceeds to sync it. Part of the io will go out as async,
and the other part as sync. This causes a disconnect between the
previously submitted io and the synced io. For io schedulers such as CFQ,
this will cause us lost merges and suboptimal behaviour in scheduling.

Remove PF_SYNCWRITE completely from the fsync/msync paths, and let
the O_DIRECT path just directly indicate that the writes are sync
by using WRITE_SYNC instead.

Signed-off-by: Jens Axboe <axboe@suse.de>
---
 block/as-iosched.c                |  2 +-
 block/cfq-iosched.c               |  4 +---
 block/ll_rw_blk.c                 |  3 +++
 drivers/usb/gadget/file_storage.c |  2 --
 fs/buffer.c                       |  2 --
 fs/direct-io.c                    | 18 ++++++++----------
 fs/fs-writeback.c                 |  2 --
 include/linux/blkdev.h            |  2 ++
 include/linux/sched.h             | 11 +++++------
 mm/msync.c                        |  3 ---
 10 files changed, 20 insertions(+), 29 deletions(-)

(limited to 'block')

diff --git a/block/as-iosched.c b/block/as-iosched.c
index 9b13d72ffefa..56c99fa037df 100644
--- a/block/as-iosched.c
+++ b/block/as-iosched.c
@@ -1339,7 +1339,7 @@ static void as_add_request(request_queue_t *q, struct request *rq)
 	arq->state = AS_RQ_NEW;
 
 	if (rq_data_dir(arq->request) == READ
-			|| current->flags&PF_SYNCWRITE)
+			|| (arq->request->flags & REQ_RW_SYNC))
 		arq->is_sync = 1;
 	else
 		arq->is_sync = 0;
diff --git a/block/cfq-iosched.c b/block/cfq-iosched.c
index c88f161d3fb3..4c4e9cc3ae26 100644
--- a/block/cfq-iosched.c
+++ b/block/cfq-iosched.c
@@ -277,8 +277,6 @@ static struct cfq_queue *cfq_find_cfq_hash(struct cfq_data *, unsigned int, unsi
 static void cfq_dispatch_insert(request_queue_t *, struct cfq_rq *);
 static struct cfq_queue *cfq_get_queue(struct cfq_data *cfqd, unsigned int key, struct task_struct *tsk, gfp_t gfp_mask);
 
-#define process_sync(tsk)	((tsk)->flags & PF_SYNCWRITE)
-
 /*
  * lots of deadline iosched dupes, can be abstracted later...
  */
@@ -334,7 +332,7 @@ static int cfq_queue_empty(request_queue_t *q)
 
 static inline pid_t cfq_queue_pid(struct task_struct *task, int rw)
 {
-	if (rw == READ || process_sync(task))
+	if (rw == READ || rw == WRITE_SYNC)
 		return task->pid;
 
 	return CFQ_KEY_ASYNC;
diff --git a/block/ll_rw_blk.c b/block/ll_rw_blk.c
index 17c42ddd31db..2270bb451385 100644
--- a/block/ll_rw_blk.c
+++ b/block/ll_rw_blk.c
@@ -2827,6 +2827,9 @@ static void init_request_from_bio(struct request *req, struct bio *bio)
 	if (unlikely(bio_barrier(bio)))
 		req->flags |= (REQ_HARDBARRIER | REQ_NOMERGE);
 
+	if (bio_sync(bio))
+		req->flags |= REQ_RW_SYNC;
+
 	req->errors = 0;
 	req->hard_sector = req->sector = bio->bi_sector;
 	req->hard_nr_sectors = req->nr_sectors = bio_sectors(bio);
diff --git a/drivers/usb/gadget/file_storage.c b/drivers/usb/gadget/file_storage.c
index 6f887478b148..a43dc908ac59 100644
--- a/drivers/usb/gadget/file_storage.c
+++ b/drivers/usb/gadget/file_storage.c
@@ -1906,7 +1906,6 @@ static int fsync_sub(struct lun *curlun)
 
 	inode = filp->f_dentry->d_inode;
 	mutex_lock(&inode->i_mutex);
-	current->flags |= PF_SYNCWRITE;
 	rc = filemap_fdatawrite(inode->i_mapping);
 	err = filp->f_op->fsync(filp, filp->f_dentry, 1);
 	if (!rc)
@@ -1914,7 +1913,6 @@ static int fsync_sub(struct lun *curlun)
 	err = filemap_fdatawait(inode->i_mapping);
 	if (!rc)
 		rc = err;
-	current->flags &= ~PF_SYNCWRITE;
 	mutex_unlock(&inode->i_mutex);
 	VLDBG(curlun, "fdatasync -> %d\n", rc);
 	return rc;
diff --git a/fs/buffer.c b/fs/buffer.c
index 23f1f3a68077..373bb6292bdc 100644
--- a/fs/buffer.c
+++ b/fs/buffer.c
@@ -331,7 +331,6 @@ long do_fsync(struct file *file, int datasync)
 		goto out;
 	}
 
-	current->flags |= PF_SYNCWRITE;
 	ret = filemap_fdatawrite(mapping);
 
 	/*
@@ -346,7 +345,6 @@ long do_fsync(struct file *file, int datasync)
 	err = filemap_fdatawait(mapping);
 	if (!ret)
 		ret = err;
-	current->flags &= ~PF_SYNCWRITE;
 out:
 	return ret;
 }
diff --git a/fs/direct-io.c b/fs/direct-io.c
index b05d1b218776..538fb0418fba 100644
--- a/fs/direct-io.c
+++ b/fs/direct-io.c
@@ -162,7 +162,7 @@ static int dio_refill_pages(struct dio *dio)
 		NULL);				/* vmas */
 	up_read(&current->mm->mmap_sem);
 
-	if (ret < 0 && dio->blocks_available && (dio->rw == WRITE)) {
+	if (ret < 0 && dio->blocks_available && (dio->rw & WRITE)) {
 		struct page *page = ZERO_PAGE(dio->curr_user_address);
 		/*
 		 * A memory fault, but the filesystem has some outstanding
@@ -535,7 +535,7 @@ static int get_more_blocks(struct dio *dio)
 		map_bh->b_state = 0;
 		map_bh->b_size = fs_count << dio->inode->i_blkbits;
 
-		create = dio->rw == WRITE;
+		create = dio->rw & WRITE;
 		if (dio->lock_type == DIO_LOCKING) {
 			if (dio->block_in_file < (i_size_read(dio->inode) >>
 							dio->blkbits))
@@ -867,7 +867,7 @@ do_holes:
 				loff_t i_size_aligned;
 
 				/* AKPM: eargh, -ENOTBLK is a hack */
-				if (dio->rw == WRITE) {
+				if (dio->rw & WRITE) {
 					page_cache_release(page);
 					return -ENOTBLK;
 				}
@@ -1045,7 +1045,7 @@ direct_io_worker(int rw, struct kiocb *iocb, struct inode *inode,
 		}
 	} /* end iovec loop */
 
-	if (ret == -ENOTBLK && rw == WRITE) {
+	if (ret == -ENOTBLK && (rw & WRITE)) {
 		/*
 		 * The remaining part of the request will be
 		 * be handled by buffered I/O when we return
@@ -1089,7 +1089,7 @@ direct_io_worker(int rw, struct kiocb *iocb, struct inode *inode,
 	if (dio->is_async) {
 		int should_wait = 0;
 
-		if (dio->result < dio->size && rw == WRITE) {
+		if (dio->result < dio->size && (rw & WRITE)) {
 			dio->waiter = current;
 			should_wait = 1;
 		}
@@ -1142,7 +1142,7 @@ direct_io_worker(int rw, struct kiocb *iocb, struct inode *inode,
 			ret = transferred;
 
 		/* We could have also come here on an AIO file extend */
-		if (!is_sync_kiocb(iocb) && rw == WRITE &&
+		if (!is_sync_kiocb(iocb) && (rw & WRITE) &&
 		    ret >= 0 && dio->result == dio->size)
 			/*
 			 * For AIO writes where we have completed the
@@ -1194,7 +1194,7 @@ __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
 	int acquire_i_mutex = 0;
 
 	if (rw & WRITE)
-		current->flags |= PF_SYNCWRITE;
+		rw = WRITE_SYNC;
 
 	if (bdev)
 		bdev_blkbits = blksize_bits(bdev_hardsect_size(bdev));
@@ -1270,7 +1270,7 @@ __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
 	 * even for AIO, we need to wait for i/o to complete before
 	 * returning in this case.
 	 */
-	dio->is_async = !is_sync_kiocb(iocb) && !((rw == WRITE) &&
+	dio->is_async = !is_sync_kiocb(iocb) && !((rw & WRITE) &&
 		(end > i_size_read(inode)));
 
 	retval = direct_io_worker(rw, iocb, inode, iov, offset,
@@ -1284,8 +1284,6 @@ out:
 		mutex_unlock(&inode->i_mutex);
 	else if (acquire_i_mutex)
 		mutex_lock(&inode->i_mutex);
-	if (rw & WRITE)
-		current->flags &= ~PF_SYNCWRITE;
 	return retval;
 }
 EXPORT_SYMBOL(__blockdev_direct_IO);
diff --git a/fs/fs-writeback.c b/fs/fs-writeback.c
index 6db95cf3aaa2..031b27a4bc9a 100644
--- a/fs/fs-writeback.c
+++ b/fs/fs-writeback.c
@@ -623,7 +623,6 @@ int generic_osync_inode(struct inode *inode, struct address_space *mapping, int
 	int need_write_inode_now = 0;
 	int err2;
 
-	current->flags |= PF_SYNCWRITE;
 	if (what & OSYNC_DATA)
 		err = filemap_fdatawrite(mapping);
 	if (what & (OSYNC_METADATA|OSYNC_DATA)) {
@@ -636,7 +635,6 @@ int generic_osync_inode(struct inode *inode, struct address_space *mapping, int
 		if (!err)
 			err = err2;
 	}
-	current->flags &= ~PF_SYNCWRITE;
 
 	spin_lock(&inode_lock);
 	if ((inode->i_state & I_DIRTY) &&
diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h
index 3457e7b97363..482a21d67627 100644
--- a/include/linux/blkdev.h
+++ b/include/linux/blkdev.h
@@ -241,6 +241,7 @@ enum rq_flag_bits {
 	__REQ_PM_RESUME,	/* resume request */
 	__REQ_PM_SHUTDOWN,	/* shutdown request */
 	__REQ_ORDERED_COLOR,	/* is before or after barrier */
+	__REQ_RW_SYNC,		/* request is sync (O_DIRECT) */
 	__REQ_NR_BITS,		/* stops here */
 };
 
@@ -270,6 +271,7 @@ enum rq_flag_bits {
 #define REQ_PM_RESUME	(1 << __REQ_PM_RESUME)
 #define REQ_PM_SHUTDOWN	(1 << __REQ_PM_SHUTDOWN)
 #define REQ_ORDERED_COLOR	(1 << __REQ_ORDERED_COLOR)
+#define REQ_RW_SYNC	(1 << __REQ_RW_SYNC)
 
 /*
  * State information carried for REQ_PM_SUSPEND and REQ_PM_RESUME
diff --git a/include/linux/sched.h b/include/linux/sched.h
index a9d23c7d1b25..38b4791e6a5d 100644
--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@ -941,12 +941,11 @@ static inline void put_task_struct(struct task_struct *t)
 #define PF_KSWAPD	0x00040000	/* I am kswapd */
 #define PF_SWAPOFF	0x00080000	/* I am in swapoff */
 #define PF_LESS_THROTTLE 0x00100000	/* Throttle me less: I clean memory */
-#define PF_SYNCWRITE	0x00200000	/* I am doing a sync write */
-#define PF_BORROWED_MM	0x00400000	/* I am a kthread doing use_mm */
-#define PF_RANDOMIZE	0x00800000	/* randomize virtual address space */
-#define PF_SWAPWRITE	0x01000000	/* Allowed to write to swap */
-#define PF_SPREAD_PAGE	0x04000000	/* Spread page cache over cpuset */
-#define PF_SPREAD_SLAB	0x08000000	/* Spread some slab caches over cpuset */
+#define PF_BORROWED_MM	0x00200000	/* I am a kthread doing use_mm */
+#define PF_RANDOMIZE	0x00400000	/* randomize virtual address space */
+#define PF_SWAPWRITE	0x00800000	/* Allowed to write to swap */
+#define PF_SPREAD_PAGE	0x01000000	/* Spread page cache over cpuset */
+#define PF_SPREAD_SLAB	0x02000000	/* Spread some slab caches over cpuset */
 #define PF_MEMPOLICY	0x10000000	/* Non-default NUMA mempolicy */
 
 /*
diff --git a/mm/msync.c b/mm/msync.c
index bc6c95376366..d083544df21b 100644
--- a/mm/msync.c
+++ b/mm/msync.c
@@ -170,8 +170,6 @@ asmlinkage long sys_msync(unsigned long start, size_t len, int flags)
 	 * just ignore them, but return -ENOMEM at the end.
 	 */
 	down_read(&current->mm->mmap_sem);
-	if (flags & MS_SYNC)
-		current->flags |= PF_SYNCWRITE;
 	vma = find_vma(current->mm, start);
 	if (!vma) {
 		error = -ENOMEM;
@@ -228,7 +226,6 @@ asmlinkage long sys_msync(unsigned long start, size_t len, int flags)
 		}
 	} while (vma && !done);
 out_unlock:
-	current->flags &= ~PF_SYNCWRITE;
 	up_read(&current->mm->mmap_sem);
 out:
 	return error;
-- 
cgit v1.2.3


From b17fd9bceb99610f6dc7998c9a4ed6b71520be2b Mon Sep 17 00:00:00 2001
From: Jens Axboe <axboe@suse.de>
Date: Mon, 19 Jun 2006 10:06:48 +0200
Subject: [PATCH] Make CFQ the default IO scheduler

Signed-off-by: Jens Axboe <axboe@suse.de>
---
 block/Kconfig.iosched | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'block')

diff --git a/block/Kconfig.iosched b/block/Kconfig.iosched
index f3b7753aac99..48d090e266fc 100644
--- a/block/Kconfig.iosched
+++ b/block/Kconfig.iosched
@@ -40,7 +40,7 @@ config IOSCHED_CFQ
 
 choice
 	prompt "Default I/O scheduler"
-	default DEFAULT_AS
+	default DEFAULT_CFQ
 	help
 	  Select the I/O scheduler which will be used by default for all
 	  block devices.
-- 
cgit v1.2.3


From 35e6077cb16f93517ba5a51ba849b186d2474d60 Mon Sep 17 00:00:00 2001
From: Jens Axboe <axboe@suse.de>
Date: Wed, 14 Jun 2006 09:10:45 +0200
Subject: [PATCH] cfq-iosched: correctly set ioprio on both targets

Patch originally from Vasily Tarasov <vtaras@sw.ru>

If you set io-priority of process 1 using sys_ioprio_set system call by
another process 2 (like ionice do), then cfq_init_prio_data() function
sets priority of process 2 (current) on queue of process 1 and clears
the flag, that designates change of ioprio.  So the process  1 will work
like with priority of process 2.

I propose not to call cfq_init_prio_data() on io-priority change, but
only mark queue as queue with changed prority.  Every time when new
request comes cfq-scheduler checks for this flag and atomaticaly changes
priority of queue to new value.

Signed-off-by: Jens Axboe <axboe@suse.de>
---
 block/cfq-iosched.c | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)

(limited to 'block')

diff --git a/block/cfq-iosched.c b/block/cfq-iosched.c
index 4c4e9cc3ae26..84b75f88c279 100644
--- a/block/cfq-iosched.c
+++ b/block/cfq-iosched.c
@@ -1388,10 +1388,9 @@ static inline void changed_ioprio(struct cfq_io_context *cic)
 			}
 		}
 		cfqq = cic->cfqq[SYNC];
-		if (cfqq) {
+		if (cfqq)
 			cfq_mark_cfqq_prio_changed(cfqq);
-			cfq_init_prio_data(cfqq);
-		}
+
 		spin_unlock(cfqd->queue->queue_lock);
 	}
 }
-- 
cgit v1.2.3


From caaa5f9f0a75d1dc5e812e69afdbb8720e077fd3 Mon Sep 17 00:00:00 2001
From: Jens Axboe <axboe@suse.de>
Date: Fri, 16 Jun 2006 11:23:00 +0200
Subject: [PATCH] cfq-iosched: many performance fixes

This is a collection of patches that greatly improve CFQ performance
in some circumstances.

- Change the idling logic to only kick in after a request is done and we
  are deciding what to do. Before the idling included the request service
  time, so it was hard to adjust. Now it's true think/idle time.

- Take advantage of TCQ/NCQ/queueing for seeky sync workloads, but keep
  it in control for sync and sequential (or close to) workloads.

- Expire queues immediately and move on to other busy queues, if we are
  not going to idle after the current one finishes.

- Don't rearm idle timer if there are no busy queues. Just leave the
  system idle.

Signed-off-by: Jens Axboe <axboe@suse.de>
---
 block/cfq-iosched.c | 116 ++++++++++++++++++++++++++++++++++------------------
 1 file changed, 76 insertions(+), 40 deletions(-)

(limited to 'block')

diff --git a/block/cfq-iosched.c b/block/cfq-iosched.c
index 84b75f88c279..13c4793fdf5f 100644
--- a/block/cfq-iosched.c
+++ b/block/cfq-iosched.c
@@ -26,7 +26,7 @@ static const int cfq_back_penalty = 2;		/* penalty of a backwards seek */
 static const int cfq_slice_sync = HZ / 10;
 static int cfq_slice_async = HZ / 25;
 static const int cfq_slice_async_rq = 2;
-static int cfq_slice_idle = HZ / 70;
+static int cfq_slice_idle = HZ / 125;
 
 #define CFQ_IDLE_GRACE		(HZ / 10)
 #define CFQ_SLICE_SCALE		(5)
@@ -906,6 +906,8 @@ static struct cfq_queue *cfq_set_active_queue(struct cfq_data *cfqd)
 	return cfqq;
 }
 
+#define CIC_SEEKY(cic) ((cic)->seek_mean > (128 * 1024))
+
 static int cfq_arm_slice_timer(struct cfq_data *cfqd, struct cfq_queue *cfqq)
 
 {
@@ -939,7 +941,7 @@ static int cfq_arm_slice_timer(struct cfq_data *cfqd, struct cfq_queue *cfqq)
 	 * fair distribution of slice time for a process doing back-to-back
 	 * seeks. so allow a little bit of time for him to submit a new rq
 	 */
-	if (sample_valid(cic->seek_samples) && cic->seek_mean > 131072)
+	if (sample_valid(cic->seek_samples) && CIC_SEEKY(cic))
 		sl = 2;
 
 	mod_timer(&cfqd->idle_slice_timer, jiffies + sl);
@@ -1038,8 +1040,10 @@ static struct cfq_queue *cfq_select_queue(struct cfq_data *cfqd)
 	 */
 	if (!RB_EMPTY(&cfqq->sort_list))
 		goto keep_queue;
-	else if (cfq_cfqq_class_sync(cfqq) &&
-		 time_before(now, cfqq->slice_end)) {
+	else if (cfq_cfqq_dispatched(cfqq)) {
+		cfqq = NULL;
+		goto keep_queue;
+	} else if (cfq_cfqq_class_sync(cfqq)) {
 		if (cfq_arm_slice_timer(cfqd, cfqq))
 			return NULL;
 	}
@@ -1088,8 +1092,7 @@ __cfq_dispatch_requests(struct cfq_data *cfqd, struct cfq_queue *cfqq,
 	} while (dispatched < max_dispatch);
 
 	/*
-	 * if slice end isn't set yet, set it. if at least one request was
-	 * sync, use the sync time slice value
+	 * if slice end isn't set yet, set it.
 	 */
 	if (!cfqq->slice_end)
 		cfq_set_prio_slice(cfqd, cfqq);
@@ -1100,7 +1103,8 @@ __cfq_dispatch_requests(struct cfq_data *cfqd, struct cfq_queue *cfqq,
 	 */
 	if ((!cfq_cfqq_sync(cfqq) &&
 	    cfqd->dispatch_slice >= cfq_prio_to_maxrq(cfqd, cfqq)) ||
-	    cfq_class_idle(cfqq))
+	    cfq_class_idle(cfqq) ||
+	    !cfq_cfqq_idle_window(cfqq))
 		cfq_slice_expired(cfqd, 0);
 
 	return dispatched;
@@ -1109,10 +1113,11 @@ __cfq_dispatch_requests(struct cfq_data *cfqd, struct cfq_queue *cfqq,
 static int
 cfq_forced_dispatch_cfqqs(struct list_head *list)
 {
-	int dispatched = 0;
 	struct cfq_queue *cfqq, *next;
 	struct cfq_rq *crq;
+	int dispatched;
 
+	dispatched = 0;
 	list_for_each_entry_safe(cfqq, next, list, cfq_list) {
 		while ((crq = cfqq->next_crq)) {
 			cfq_dispatch_insert(cfqq->cfqd->queue, crq);
@@ -1120,6 +1125,7 @@ cfq_forced_dispatch_cfqqs(struct list_head *list)
 		}
 		BUG_ON(!list_empty(&cfqq->fifo));
 	}
+
 	return dispatched;
 }
 
@@ -1146,7 +1152,8 @@ static int
 cfq_dispatch_requests(request_queue_t *q, int force)
 {
 	struct cfq_data *cfqd = q->elevator->elevator_data;
-	struct cfq_queue *cfqq;
+	struct cfq_queue *cfqq, *prev_cfqq;
+	int dispatched;
 
 	if (!cfqd->busy_queues)
 		return 0;
@@ -1154,10 +1161,17 @@ cfq_dispatch_requests(request_queue_t *q, int force)
 	if (unlikely(force))
 		return cfq_forced_dispatch(cfqd);
 
-	cfqq = cfq_select_queue(cfqd);
-	if (cfqq) {
+	dispatched = 0;
+	prev_cfqq = NULL;
+	while ((cfqq = cfq_select_queue(cfqd)) != NULL) {
 		int max_dispatch;
 
+		/*
+		 * Don't repeat dispatch from the previous queue.
+		 */
+		if (prev_cfqq == cfqq)
+			break;
+
 		cfq_clear_cfqq_must_dispatch(cfqq);
 		cfq_clear_cfqq_wait_request(cfqq);
 		del_timer(&cfqd->idle_slice_timer);
@@ -1166,10 +1180,19 @@ cfq_dispatch_requests(request_queue_t *q, int force)
 		if (cfq_class_idle(cfqq))
 			max_dispatch = 1;
 
-		return __cfq_dispatch_requests(cfqd, cfqq, max_dispatch);
+		dispatched += __cfq_dispatch_requests(cfqd, cfqq, max_dispatch);
+
+		/*
+		 * If the dispatch cfqq has idling enabled and is still
+		 * the active queue, break out.
+		 */
+		if (cfq_cfqq_idle_window(cfqq) && cfqd->active_queue)
+			break;
+
+		prev_cfqq = cfqq;
 	}
 
-	return 0;
+	return dispatched;
 }
 
 /*
@@ -1375,24 +1398,28 @@ static inline void changed_ioprio(struct cfq_io_context *cic)
 {
 	struct cfq_data *cfqd = cic->key;
 	struct cfq_queue *cfqq;
-	if (cfqd) {
-		spin_lock(cfqd->queue->queue_lock);
-		cfqq = cic->cfqq[ASYNC];
-		if (cfqq) {
-			struct cfq_queue *new_cfqq;
-			new_cfqq = cfq_get_queue(cfqd, CFQ_KEY_ASYNC,
-						cic->ioc->task, GFP_ATOMIC);
-			if (new_cfqq) {
-				cic->cfqq[ASYNC] = new_cfqq;
-				cfq_put_queue(cfqq);
-			}
-		}
-		cfqq = cic->cfqq[SYNC];
-		if (cfqq)
-			cfq_mark_cfqq_prio_changed(cfqq);
 
-		spin_unlock(cfqd->queue->queue_lock);
+	if (unlikely(!cfqd))
+		return;
+
+	spin_lock(cfqd->queue->queue_lock);
+
+	cfqq = cic->cfqq[ASYNC];
+	if (cfqq) {
+		struct cfq_queue *new_cfqq;
+		new_cfqq = cfq_get_queue(cfqd, CFQ_KEY_ASYNC, cic->ioc->task,
+					 GFP_ATOMIC);
+		if (new_cfqq) {
+			cic->cfqq[ASYNC] = new_cfqq;
+			cfq_put_queue(cfqq);
+		}
 	}
+
+	cfqq = cic->cfqq[SYNC];
+	if (cfqq)
+		cfq_mark_cfqq_prio_changed(cfqq);
+
+	spin_unlock(cfqd->queue->queue_lock);
 }
 
 /*
@@ -1461,8 +1488,7 @@ retry:
 		 * set ->slice_left to allow preemption for a new process
 		 */
 		cfqq->slice_left = 2 * cfqd->cfq_slice_idle;
-		if (!cfqd->hw_tag)
-			cfq_mark_cfqq_idle_window(cfqq);
+		cfq_mark_cfqq_idle_window(cfqq);
 		cfq_mark_cfqq_prio_changed(cfqq);
 		cfq_init_prio_data(cfqq);
 	}
@@ -1653,7 +1679,8 @@ cfq_update_idle_window(struct cfq_data *cfqd, struct cfq_queue *cfqq,
 {
 	int enable_idle = cfq_cfqq_idle_window(cfqq);
 
-	if (!cic->ioc->task || !cfqd->cfq_slice_idle || cfqd->hw_tag)
+	if (!cic->ioc->task || !cfqd->cfq_slice_idle ||
+	    (cfqd->hw_tag && CIC_SEEKY(cic)))
 		enable_idle = 0;
 	else if (sample_valid(cic->ttime_samples)) {
 		if (cic->ttime_mean > cfqd->cfq_slice_idle)
@@ -1683,7 +1710,7 @@ cfq_should_preempt(struct cfq_data *cfqd, struct cfq_queue *new_cfqq,
 		return 0;
 
 	if (!cfqq)
-		return 1;
+		return 0;
 
 	if (cfq_class_idle(cfqq))
 		return 1;
@@ -1715,7 +1742,7 @@ static void cfq_preempt_queue(struct cfq_data *cfqd, struct cfq_queue *cfqq)
 		cfqq->slice_left = cfq_prio_to_slice(cfqd, cfqq) / 2;
 
 	cfqq->slice_end = cfqq->slice_left + jiffies;
-	__cfq_slice_expired(cfqd, cfqq, 1);
+	cfq_slice_expired(cfqd, 1);
 	__cfq_set_active_queue(cfqd, cfqq);
 }
 
@@ -1834,11 +1861,23 @@ static void cfq_completed_request(request_queue_t *q, struct request *rq)
 			cfqq->service_last = now;
 			cfq_resort_rr_list(cfqq, 0);
 		}
-		cfq_schedule_dispatch(cfqd);
 	}
 
-	if (cfq_crq_is_sync(crq))
+	if (sync)
 		crq->io_context->last_end_request = now;
+
+	/*
+	 * If this is the active queue, check if it needs to be expired,
+	 * or if we want to idle in case it has no pending requests.
+	 */
+	if (cfqd->active_queue == cfqq) {
+		if (time_after(now, cfqq->slice_end))
+			cfq_slice_expired(cfqd, 0);
+		else if (sync && RB_EMPTY(&cfqq->sort_list)) {
+			if (!cfq_arm_slice_timer(cfqd, cfqq))
+				cfq_schedule_dispatch(cfqd);
+		}
+	}
 }
 
 static struct request *
@@ -2106,11 +2145,8 @@ static void cfq_idle_slice_timer(unsigned long data)
 		 * only expire and reinvoke request handler, if there are
 		 * other queues with pending requests
 		 */
-		if (!cfqd->busy_queues) {
-			cfqd->idle_slice_timer.expires = min(now + cfqd->cfq_slice_idle, cfqq->slice_end);
-			add_timer(&cfqd->idle_slice_timer);
+		if (!cfqd->busy_queues)
 			goto out_cont;
-		}
 
 		/*
 		 * not expired and it has a request pending, let it dispatch
-- 
cgit v1.2.3


From fd61af0384014ca29428ace7c17a978b755aeddd Mon Sep 17 00:00:00 2001
From: Jens Axboe <axboe@suse.de>
Date: Fri, 16 Jun 2006 15:35:39 +0200
Subject: [PATCH] cfq-iosched: rq update fixes

- Remember to set ->last_sector so that the cfq_choose_req() logic
  works correctly.

- Remove redundant call to cfq_choose_req()

Signed-off-by: Jens Axboe <axboe@suse.de>
---
 block/cfq-iosched.c | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

(limited to 'block')

diff --git a/block/cfq-iosched.c b/block/cfq-iosched.c
index 13c4793fdf5f..940364edf2b9 100644
--- a/block/cfq-iosched.c
+++ b/block/cfq-iosched.c
@@ -952,11 +952,15 @@ static void cfq_dispatch_insert(request_queue_t *q, struct cfq_rq *crq)
 {
 	struct cfq_data *cfqd = q->elevator->elevator_data;
 	struct cfq_queue *cfqq = crq->cfq_queue;
+	struct request *rq;
 
 	cfqq->next_crq = cfq_find_next_crq(cfqd, cfqq, crq);
 	cfq_remove_request(crq->request);
 	cfqq->on_dispatch[cfq_crq_is_sync(crq)]++;
 	elv_dispatch_sort(q, crq->request);
+
+	rq = list_entry(q->queue_head.prev, struct request, queuelist);
+	cfqd->last_sector = rq->sector + rq->nr_sectors;
 }
 
 /*
@@ -1767,11 +1771,7 @@ static void
 cfq_crq_enqueued(struct cfq_data *cfqd, struct cfq_queue *cfqq,
 		 struct cfq_rq *crq)
 {
-	struct cfq_io_context *cic;
-
-	cfqq->next_crq = cfq_choose_req(cfqd, cfqq->next_crq, crq);
-
-	cic = crq->io_context;
+	struct cfq_io_context *cic = crq->io_context;
 
 	/*
 	 * we never wait for an async request and we don't allow preemption
-- 
cgit v1.2.3


From dd67d051529387f6e44d22d1d5540ef281965fdd Mon Sep 17 00:00:00 2001
From: Jens Axboe <axboe@suse.de>
Date: Wed, 21 Jun 2006 09:36:18 +0200
Subject: [PATCH] rbtree: support functions used by the io schedulers

They all duplicate macros to check for empty root and/or node, and
clearing a node. So put those in rbtree.h.

Signed-off-by: Jens Axboe <axboe@suse.de>
---
 block/as-iosched.c       | 17 +++++++----------
 block/cfq-iosched.c      | 22 ++++++++--------------
 block/deadline-iosched.c | 13 +++++--------
 include/linux/rbtree.h   |  4 ++++
 4 files changed, 24 insertions(+), 32 deletions(-)

(limited to 'block')

diff --git a/block/as-iosched.c b/block/as-iosched.c
index 56c99fa037df..1ec5df466708 100644
--- a/block/as-iosched.c
+++ b/block/as-iosched.c
@@ -347,9 +347,6 @@ static struct request *as_find_arq_hash(struct as_data *ad, sector_t offset)
 /*
  * rb tree support functions
  */
-#define RB_EMPTY(root)	((root)->rb_node == NULL)
-#define ON_RB(node)	(rb_parent(node) != node)
-#define RB_CLEAR(node)	(rb_set_parent(node, node))
 #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
@@ -418,13 +415,13 @@ static void as_add_arq_rb(struct as_data *ad, struct as_rq *arq)
 
 static inline void as_del_arq_rb(struct as_data *ad, struct as_rq *arq)
 {
-	if (!ON_RB(&arq->rb_node)) {
+	if (!RB_EMPTY_NODE(&arq->rb_node)) {
 		WARN_ON(1);
 		return;
 	}
 
 	rb_erase(&arq->rb_node, ARQ_RB_ROOT(ad, arq));
-	RB_CLEAR(&arq->rb_node);
+	RB_CLEAR_NODE(&arq->rb_node);
 }
 
 static struct request *
@@ -545,7 +542,7 @@ static struct as_rq *as_find_next_arq(struct as_data *ad, struct as_rq *last)
 	struct rb_node *rbprev = rb_prev(&last->rb_node);
 	struct as_rq *arq_next, *arq_prev;
 
-	BUG_ON(!ON_RB(&last->rb_node));
+	BUG_ON(!RB_EMPTY_NODE(&last->rb_node));
 
 	if (rbprev)
 		arq_prev = rb_entry_arq(rbprev);
@@ -1122,7 +1119,7 @@ static void as_move_to_dispatch(struct as_data *ad, struct as_rq *arq)
 	struct request *rq = arq->request;
 	const int data_dir = arq->is_sync;
 
-	BUG_ON(!ON_RB(&arq->rb_node));
+	BUG_ON(!RB_EMPTY_NODE(&arq->rb_node));
 
 	as_antic_stop(ad);
 	ad->antic_status = ANTIC_OFF;
@@ -1247,7 +1244,7 @@ static int as_dispatch_request(request_queue_t *q, int force)
 	 */
 
 	if (reads) {
-		BUG_ON(RB_EMPTY(&ad->sort_list[REQ_SYNC]));
+		BUG_ON(RB_EMPTY_ROOT(&ad->sort_list[REQ_SYNC]));
 
 		if (writes && ad->batch_data_dir == REQ_SYNC)
 			/*
@@ -1271,7 +1268,7 @@ static int as_dispatch_request(request_queue_t *q, int force)
 
 	if (writes) {
 dispatch_writes:
-		BUG_ON(RB_EMPTY(&ad->sort_list[REQ_ASYNC]));
+		BUG_ON(RB_EMPTY_ROOT(&ad->sort_list[REQ_ASYNC]));
 
 		if (ad->batch_data_dir == REQ_SYNC) {
 			ad->changed_batch = 1;
@@ -1591,7 +1588,7 @@ static int as_set_request(request_queue_t *q, struct request *rq,
 
 	if (arq) {
 		memset(arq, 0, sizeof(*arq));
-		RB_CLEAR(&arq->rb_node);
+		RB_CLEAR_NODE(&arq->rb_node);
 		arq->request = rq;
 		arq->state = AS_RQ_PRESCHED;
 		arq->io_context = NULL;
diff --git a/block/cfq-iosched.c b/block/cfq-iosched.c
index 940364edf2b9..e25223e147a2 100644
--- a/block/cfq-iosched.c
+++ b/block/cfq-iosched.c
@@ -60,11 +60,6 @@ static DEFINE_SPINLOCK(cfq_exit_lock);
 /*
  * rb-tree defines
  */
-#define RB_EMPTY(node)		((node)->rb_node == NULL)
-#define RB_CLEAR(node)		do {	\
-		memset(node, 0, sizeof(*node)); \
-} while (0)
-#define RB_CLEAR_ROOT(root)	((root)->rb_node = NULL)
 #define rb_entry_crq(node)	rb_entry((node), struct cfq_rq, rb_node)
 #define rq_rb_key(rq)		(rq)->sector
 
@@ -559,7 +554,7 @@ static inline void cfq_del_crq_rb(struct cfq_rq *crq)
 
 	rb_erase(&crq->rb_node, &cfqq->sort_list);
 
-	if (cfq_cfqq_on_rr(cfqq) && RB_EMPTY(&cfqq->sort_list))
+	if (cfq_cfqq_on_rr(cfqq) && RB_EMPTY_ROOT(&cfqq->sort_list))
 		cfq_del_cfqq_rr(cfqd, cfqq);
 }
 
@@ -914,7 +909,7 @@ static int cfq_arm_slice_timer(struct cfq_data *cfqd, struct cfq_queue *cfqq)
 	struct cfq_io_context *cic;
 	unsigned long sl;
 
-	WARN_ON(!RB_EMPTY(&cfqq->sort_list));
+	WARN_ON(!RB_EMPTY_ROOT(&cfqq->sort_list));
 	WARN_ON(cfqq != cfqd->active_queue);
 
 	/*
@@ -1042,7 +1037,7 @@ static struct cfq_queue *cfq_select_queue(struct cfq_data *cfqd)
 	 * if queue has requests, dispatch one. if not, check if
 	 * enough slice is left to wait for one
 	 */
-	if (!RB_EMPTY(&cfqq->sort_list))
+	if (!RB_EMPTY_ROOT(&cfqq->sort_list))
 		goto keep_queue;
 	else if (cfq_cfqq_dispatched(cfqq)) {
 		cfqq = NULL;
@@ -1066,7 +1061,7 @@ __cfq_dispatch_requests(struct cfq_data *cfqd, struct cfq_queue *cfqq,
 {
 	int dispatched = 0;
 
-	BUG_ON(RB_EMPTY(&cfqq->sort_list));
+	BUG_ON(RB_EMPTY_ROOT(&cfqq->sort_list));
 
 	do {
 		struct cfq_rq *crq;
@@ -1090,7 +1085,7 @@ __cfq_dispatch_requests(struct cfq_data *cfqd, struct cfq_queue *cfqq,
 			cfqd->active_cic = crq->io_context;
 		}
 
-		if (RB_EMPTY(&cfqq->sort_list))
+		if (RB_EMPTY_ROOT(&cfqq->sort_list))
 			break;
 
 	} while (dispatched < max_dispatch);
@@ -1480,7 +1475,6 @@ retry:
 
 		INIT_HLIST_NODE(&cfqq->cfq_hash);
 		INIT_LIST_HEAD(&cfqq->cfq_list);
-		RB_CLEAR_ROOT(&cfqq->sort_list);
 		INIT_LIST_HEAD(&cfqq->fifo);
 
 		cfqq->key = key;
@@ -1873,7 +1867,7 @@ static void cfq_completed_request(request_queue_t *q, struct request *rq)
 	if (cfqd->active_queue == cfqq) {
 		if (time_after(now, cfqq->slice_end))
 			cfq_slice_expired(cfqd, 0);
-		else if (sync && RB_EMPTY(&cfqq->sort_list)) {
+		else if (sync && RB_EMPTY_ROOT(&cfqq->sort_list)) {
 			if (!cfq_arm_slice_timer(cfqd, cfqq))
 				cfq_schedule_dispatch(cfqd);
 		}
@@ -2059,7 +2053,7 @@ cfq_set_request(request_queue_t *q, struct request *rq, struct bio *bio,
 
 	crq = mempool_alloc(cfqd->crq_pool, gfp_mask);
 	if (crq) {
-		RB_CLEAR(&crq->rb_node);
+		RB_CLEAR_NODE(&crq->rb_node);
 		crq->rb_key = 0;
 		crq->request = rq;
 		INIT_HLIST_NODE(&crq->hash);
@@ -2151,7 +2145,7 @@ static void cfq_idle_slice_timer(unsigned long data)
 		/*
 		 * not expired and it has a request pending, let it dispatch
 		 */
-		if (!RB_EMPTY(&cfqq->sort_list)) {
+		if (!RB_EMPTY_ROOT(&cfqq->sort_list)) {
 			cfq_mark_cfqq_must_dispatch(cfqq);
 			goto out_kick;
 		}
diff --git a/block/deadline-iosched.c b/block/deadline-iosched.c
index e5bccaaed563..4469dd84623c 100644
--- a/block/deadline-iosched.c
+++ b/block/deadline-iosched.c
@@ -159,9 +159,6 @@ deadline_find_drq_hash(struct deadline_data *dd, sector_t offset)
 /*
  * rb tree support functions
  */
-#define RB_EMPTY(root)	((root)->rb_node == NULL)
-#define ON_RB(node)	(rb_parent(node) != node)
-#define RB_CLEAR(node)	(rb_set_parent(node, node))
 #define rb_entry_drq(node)	rb_entry((node), struct deadline_rq, rb_node)
 #define DRQ_RB_ROOT(dd, drq)	(&(dd)->sort_list[rq_data_dir((drq)->request)])
 #define rq_rb_key(rq)		(rq)->sector
@@ -220,9 +217,9 @@ deadline_del_drq_rb(struct deadline_data *dd, struct deadline_rq *drq)
 			dd->next_drq[data_dir] = rb_entry_drq(rbnext);
 	}
 
-	BUG_ON(!ON_RB(&drq->rb_node));
+	BUG_ON(!RB_EMPTY_NODE(&drq->rb_node));
 	rb_erase(&drq->rb_node, DRQ_RB_ROOT(dd, drq));
-	RB_CLEAR(&drq->rb_node);
+	RB_CLEAR_NODE(&drq->rb_node);
 }
 
 static struct request *
@@ -496,7 +493,7 @@ static int deadline_dispatch_requests(request_queue_t *q, int force)
 	 */
 
 	if (reads) {
-		BUG_ON(RB_EMPTY(&dd->sort_list[READ]));
+		BUG_ON(RB_EMPTY_ROOT(&dd->sort_list[READ]));
 
 		if (writes && (dd->starved++ >= dd->writes_starved))
 			goto dispatch_writes;
@@ -512,7 +509,7 @@ static int deadline_dispatch_requests(request_queue_t *q, int force)
 
 	if (writes) {
 dispatch_writes:
-		BUG_ON(RB_EMPTY(&dd->sort_list[WRITE]));
+		BUG_ON(RB_EMPTY_ROOT(&dd->sort_list[WRITE]));
 
 		dd->starved = 0;
 
@@ -668,7 +665,7 @@ deadline_set_request(request_queue_t *q, struct request *rq, struct bio *bio,
 	drq = mempool_alloc(dd->drq_pool, gfp_mask);
 	if (drq) {
 		memset(drq, 0, sizeof(*drq));
-		RB_CLEAR(&drq->rb_node);
+		RB_CLEAR_NODE(&drq->rb_node);
 		drq->request = rq;
 
 		INIT_HLIST_NODE(&drq->hash);
diff --git a/include/linux/rbtree.h b/include/linux/rbtree.h
index f37006f21664..8d5382e62c08 100644
--- a/include/linux/rbtree.h
+++ b/include/linux/rbtree.h
@@ -132,6 +132,10 @@ static inline void rb_set_color(struct rb_node *rb, int color)
 #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_CLEAR_NODE(node)	(rb_set_parent(node, node))
+
 extern void rb_insert_color(struct rb_node *, struct rb_root *);
 extern void rb_erase(struct rb_node *, struct rb_root *);
 
-- 
cgit v1.2.3


From 8269730b389d4793348d521140f9c76fb1828249 Mon Sep 17 00:00:00 2001
From: Andi Kleen <ak@suse.de>
Date: Wed, 21 Jun 2006 14:48:09 +0200
Subject: [BLOCK] Fix bounce limit address check

Do a safer check for when to enable DMA. Currently we enable ISA DMA
for cases that do not need it, resulting in OOM conditions when ZONE_DMA
runs out of space.

Signed-off-by: Jens Axboe <axboe@suse.de>
---
 block/ll_rw_blk.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'block')

diff --git a/block/ll_rw_blk.c b/block/ll_rw_blk.c
index 2270bb451385..0603ab2f3692 100644
--- a/block/ll_rw_blk.c
+++ b/block/ll_rw_blk.c
@@ -638,7 +638,7 @@ void blk_queue_bounce_limit(request_queue_t *q, u64 dma_addr)
 	/* Assume anything <= 4GB can be handled by IOMMU.
 	   Actually some IOMMUs can handle everything, but I don't
 	   know of a way to test this here. */
-	if (bounce_pfn < (0xffffffff>>PAGE_SHIFT))
+	if (bounce_pfn < (min_t(u64,0xffffffff,BLK_BOUNCE_HIGH) >> PAGE_SHIFT))
 		dma = 1;
 	q->bounce_pfn = max_low_pfn;
 #else
-- 
cgit v1.2.3