From d2bf1b6723ed0eab378363649d15b7893bf14e91 Mon Sep 17 00:00:00 2001 From: Tejun Heo Date: Wed, 8 Dec 2010 20:57:36 +0100 Subject: block: move register_disk() and del_gendisk() to block/genhd.c There's no reason for register_disk() and del_gendisk() to be in fs/partitions/check.c. Move both to genhd.c. While at it, collapse unlink_gendisk(), which was artificially in a separate function due to genhd.c / check.c split, into del_gendisk(). Signed-off-by: Tejun Heo Signed-off-by: Jens Axboe --- include/linux/blkdev.h | 1 - 1 file changed, 1 deletion(-) (limited to 'include/linux/blkdev.h') diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h index aae86fd10c4f..83031bcf8366 100644 --- a/include/linux/blkdev.h +++ b/include/linux/blkdev.h @@ -643,7 +643,6 @@ static inline void rq_flush_dcache_pages(struct request *rq) extern int blk_register_queue(struct gendisk *disk); extern void blk_unregister_queue(struct gendisk *disk); -extern void register_disk(struct gendisk *dev); extern void generic_make_request(struct bio *bio); extern void blk_rq_init(struct request_queue *q, struct request *rq); extern void blk_put_request(struct request *); -- cgit v1.2.3 From 77ea887e433ad8389d416826936c110fa7910f80 Mon Sep 17 00:00:00 2001 From: Tejun Heo Date: Wed, 8 Dec 2010 20:57:37 +0100 Subject: implement in-kernel gendisk events handling Currently, media presence polling for removeable block devices is done from userland. There are several issues with this. * Polling is done by periodically opening the device. For SCSI devices, the command sequence generated by such action involves a few different commands including TEST_UNIT_READY. This behavior, while perfectly legal, is different from Windows which only issues single command, GET_EVENT_STATUS_NOTIFICATION. Unfortunately, some ATAPI devices lock up after being periodically queried such command sequences. * There is no reliable and unintrusive way for a userland program to tell whether the target device is safe for media presence polling. For example, polling for media presence during an on-going burning session can make it fail. The polling program can avoid this by opening the device with O_EXCL but then it risks making a valid exclusive user of the device fail w/ -EBUSY. * Userland polling is unnecessarily heavy and in-kernel implementation is lighter and better coordinated (workqueue, timer slack). This patch implements framework for in-kernel disk event handling, which includes media presence polling. * bdops->check_events() is added, which supercedes ->media_changed(). It should check whether there's any pending event and return if so. Currently, two events are defined - DISK_EVENT_MEDIA_CHANGE and DISK_EVENT_EJECT_REQUEST. ->check_events() is guaranteed not to be called parallelly. * gendisk->events and ->async_events are added. These should be initialized by block driver before passing the device to add_disk(). The former contains the mask of all supported events and the latter the mask of all events which the device can report without polling. /sys/block/*/events[_async] export these to userland. * Kernel parameter block.events_dfl_poll_msecs controls the system polling interval (default is 0 which means disable) and /sys/block/*/events_poll_msecs control polling intervals for individual devices (default is -1 meaning use system setting). Note that if a device can report all supported events asynchronously and its polling interval isn't explicitly set, the device won't be polled regardless of the system polling interval. * If a device is opened exclusively with write access, event checking is automatically disabled until all write exclusive accesses are released. * There are event 'clearing' events. For example, both of currently defined events are cleared after the device has been successfully opened. This information is passed to ->check_events() callback using @clearing argument as a hint. * Event checking is always performed from system_nrt_wq and timer slack is set to 25% for polling. * Nothing changes for drivers which implement ->media_changed() but not ->check_events(). Going forward, all drivers will be converted to ->check_events() and ->media_change() will be dropped. Signed-off-by: Tejun Heo Cc: Kay Sievers Cc: Jan Kara Signed-off-by: Jens Axboe --- block/genhd.c | 429 +++++++++++++++++++++++++++++++++++++++++++++++++ fs/block_dev.c | 41 ++++- include/linux/blkdev.h | 3 + include/linux/fs.h | 1 + include/linux/genhd.h | 18 ++- 5 files changed, 484 insertions(+), 8 deletions(-) (limited to 'include/linux/blkdev.h') diff --git a/block/genhd.c b/block/genhd.c index 2e5e4c0a1133..5465a824d489 100644 --- a/block/genhd.c +++ b/block/genhd.c @@ -18,6 +18,7 @@ #include #include #include +#include #include "blk.h" @@ -35,6 +36,10 @@ static DEFINE_IDR(ext_devt_idr); static struct device_type disk_type; +static void disk_add_events(struct gendisk *disk); +static void disk_del_events(struct gendisk *disk); +static void disk_release_events(struct gendisk *disk); + /** * disk_get_part - get partition * @disk: disk to look partition from @@ -609,6 +614,8 @@ void add_disk(struct gendisk *disk) retval = sysfs_create_link(&disk_to_dev(disk)->kobj, &bdi->dev->kobj, "bdi"); WARN_ON(retval); + + disk_add_events(disk); } EXPORT_SYMBOL(add_disk); @@ -617,6 +624,8 @@ void del_gendisk(struct gendisk *disk) struct disk_part_iter piter; struct hd_struct *part; + disk_del_events(disk); + /* invalidate stuff */ disk_part_iter_init(&piter, disk, DISK_PITER_INCL_EMPTY | DISK_PITER_REVERSE); @@ -1089,6 +1098,7 @@ static void disk_release(struct device *dev) { struct gendisk *disk = dev_to_disk(dev); + disk_release_events(disk); kfree(disk->random); disk_replace_part_tbl(disk, NULL); free_part_stats(&disk->part0); @@ -1350,3 +1360,422 @@ int invalidate_partition(struct gendisk *disk, int partno) } EXPORT_SYMBOL(invalidate_partition); + +/* + * Disk events - monitor disk events like media change and eject request. + */ +struct disk_events { + struct list_head node; /* all disk_event's */ + struct gendisk *disk; /* the associated disk */ + spinlock_t lock; + + int block; /* event blocking depth */ + unsigned int pending; /* events already sent out */ + unsigned int clearing; /* events being cleared */ + + long poll_msecs; /* interval, -1 for default */ + struct delayed_work dwork; +}; + +static const char *disk_events_strs[] = { + [ilog2(DISK_EVENT_MEDIA_CHANGE)] = "media_change", + [ilog2(DISK_EVENT_EJECT_REQUEST)] = "eject_request", +}; + +static char *disk_uevents[] = { + [ilog2(DISK_EVENT_MEDIA_CHANGE)] = "DISK_MEDIA_CHANGE=1", + [ilog2(DISK_EVENT_EJECT_REQUEST)] = "DISK_EJECT_REQUEST=1", +}; + +/* list of all disk_events */ +static DEFINE_MUTEX(disk_events_mutex); +static LIST_HEAD(disk_events); + +/* disable in-kernel polling by default */ +static unsigned long disk_events_dfl_poll_msecs = 0; + +static unsigned long disk_events_poll_jiffies(struct gendisk *disk) +{ + struct disk_events *ev = disk->ev; + long intv_msecs = 0; + + /* + * If device-specific poll interval is set, always use it. If + * the default is being used, poll iff there are events which + * can't be monitored asynchronously. + */ + if (ev->poll_msecs >= 0) + intv_msecs = ev->poll_msecs; + else if (disk->events & ~disk->async_events) + intv_msecs = disk_events_dfl_poll_msecs; + + return msecs_to_jiffies(intv_msecs); +} + +static void __disk_block_events(struct gendisk *disk, bool sync) +{ + struct disk_events *ev = disk->ev; + unsigned long flags; + bool cancel; + + spin_lock_irqsave(&ev->lock, flags); + cancel = !ev->block++; + spin_unlock_irqrestore(&ev->lock, flags); + + if (cancel) { + if (sync) + cancel_delayed_work_sync(&disk->ev->dwork); + else + cancel_delayed_work(&disk->ev->dwork); + } +} + +static void __disk_unblock_events(struct gendisk *disk, bool check_now) +{ + struct disk_events *ev = disk->ev; + unsigned long intv; + unsigned long flags; + + spin_lock_irqsave(&ev->lock, flags); + + if (WARN_ON_ONCE(ev->block <= 0)) + goto out_unlock; + + if (--ev->block) + goto out_unlock; + + /* + * Not exactly a latency critical operation, set poll timer + * slack to 25% and kick event check. + */ + intv = disk_events_poll_jiffies(disk); + set_timer_slack(&ev->dwork.timer, intv / 4); + if (check_now) + queue_delayed_work(system_nrt_wq, &ev->dwork, 0); + else if (intv) + queue_delayed_work(system_nrt_wq, &ev->dwork, intv); +out_unlock: + spin_unlock_irqrestore(&ev->lock, flags); +} + +/** + * disk_block_events - block and flush disk event checking + * @disk: disk to block events for + * + * On return from this function, it is guaranteed that event checking + * isn't in progress and won't happen until unblocked by + * disk_unblock_events(). Events blocking is counted and the actual + * unblocking happens after the matching number of unblocks are done. + * + * Note that this intentionally does not block event checking from + * disk_clear_events(). + * + * CONTEXT: + * Might sleep. + */ +void disk_block_events(struct gendisk *disk) +{ + if (disk->ev) + __disk_block_events(disk, true); +} + +/** + * disk_unblock_events - unblock disk event checking + * @disk: disk to unblock events for + * + * Undo disk_block_events(). When the block count reaches zero, it + * starts events polling if configured. + * + * CONTEXT: + * Don't care. Safe to call from irq context. + */ +void disk_unblock_events(struct gendisk *disk) +{ + if (disk->ev) + __disk_unblock_events(disk, true); +} + +/** + * disk_check_events - schedule immediate event checking + * @disk: disk to check events for + * + * Schedule immediate event checking on @disk if not blocked. + * + * CONTEXT: + * Don't care. Safe to call from irq context. + */ +void disk_check_events(struct gendisk *disk) +{ + if (disk->ev) { + __disk_block_events(disk, false); + __disk_unblock_events(disk, true); + } +} +EXPORT_SYMBOL_GPL(disk_check_events); + +/** + * disk_clear_events - synchronously check, clear and return pending events + * @disk: disk to fetch and clear events from + * @mask: mask of events to be fetched and clearted + * + * Disk events are synchronously checked and pending events in @mask + * are cleared and returned. This ignores the block count. + * + * CONTEXT: + * Might sleep. + */ +unsigned int disk_clear_events(struct gendisk *disk, unsigned int mask) +{ + const struct block_device_operations *bdops = disk->fops; + struct disk_events *ev = disk->ev; + unsigned int pending; + + if (!ev) { + /* for drivers still using the old ->media_changed method */ + if ((mask & DISK_EVENT_MEDIA_CHANGE) && + bdops->media_changed && bdops->media_changed(disk)) + return DISK_EVENT_MEDIA_CHANGE; + return 0; + } + + /* tell the workfn about the events being cleared */ + spin_lock_irq(&ev->lock); + ev->clearing |= mask; + spin_unlock_irq(&ev->lock); + + /* uncondtionally schedule event check and wait for it to finish */ + __disk_block_events(disk, true); + queue_delayed_work(system_nrt_wq, &ev->dwork, 0); + flush_delayed_work(&ev->dwork); + __disk_unblock_events(disk, false); + + /* then, fetch and clear pending events */ + spin_lock_irq(&ev->lock); + WARN_ON_ONCE(ev->clearing & mask); /* cleared by workfn */ + pending = ev->pending & mask; + ev->pending &= ~mask; + spin_unlock_irq(&ev->lock); + + return pending; +} + +static void disk_events_workfn(struct work_struct *work) +{ + struct delayed_work *dwork = to_delayed_work(work); + struct disk_events *ev = container_of(dwork, struct disk_events, dwork); + struct gendisk *disk = ev->disk; + char *envp[ARRAY_SIZE(disk_uevents) + 1] = { }; + unsigned int clearing = ev->clearing; + unsigned int events; + unsigned long intv; + int nr_events = 0, i; + + /* check events */ + events = disk->fops->check_events(disk, clearing); + + /* accumulate pending events and schedule next poll if necessary */ + spin_lock_irq(&ev->lock); + + events &= ~ev->pending; + ev->pending |= events; + ev->clearing &= ~clearing; + + intv = disk_events_poll_jiffies(disk); + if (!ev->block && intv) + queue_delayed_work(system_nrt_wq, &ev->dwork, intv); + + spin_unlock_irq(&ev->lock); + + /* tell userland about new events */ + for (i = 0; i < ARRAY_SIZE(disk_uevents); i++) + if (events & (1 << i)) + envp[nr_events++] = disk_uevents[i]; + + if (nr_events) + kobject_uevent_env(&disk_to_dev(disk)->kobj, KOBJ_CHANGE, envp); +} + +/* + * A disk events enabled device has the following sysfs nodes under + * its /sys/block/X/ directory. + * + * events : list of all supported events + * events_async : list of events which can be detected w/o polling + * events_poll_msecs : polling interval, 0: disable, -1: system default + */ +static ssize_t __disk_events_show(unsigned int events, char *buf) +{ + const char *delim = ""; + ssize_t pos = 0; + int i; + + for (i = 0; i < ARRAY_SIZE(disk_events_strs); i++) + if (events & (1 << i)) { + pos += sprintf(buf + pos, "%s%s", + delim, disk_events_strs[i]); + delim = " "; + } + if (pos) + pos += sprintf(buf + pos, "\n"); + return pos; +} + +static ssize_t disk_events_show(struct device *dev, + struct device_attribute *attr, char *buf) +{ + struct gendisk *disk = dev_to_disk(dev); + + return __disk_events_show(disk->events, buf); +} + +static ssize_t disk_events_async_show(struct device *dev, + struct device_attribute *attr, char *buf) +{ + struct gendisk *disk = dev_to_disk(dev); + + return __disk_events_show(disk->async_events, buf); +} + +static ssize_t disk_events_poll_msecs_show(struct device *dev, + struct device_attribute *attr, + char *buf) +{ + struct gendisk *disk = dev_to_disk(dev); + + return sprintf(buf, "%ld\n", disk->ev->poll_msecs); +} + +static ssize_t disk_events_poll_msecs_store(struct device *dev, + struct device_attribute *attr, + const char *buf, size_t count) +{ + struct gendisk *disk = dev_to_disk(dev); + long intv; + + if (!count || !sscanf(buf, "%ld", &intv)) + return -EINVAL; + + if (intv < 0 && intv != -1) + return -EINVAL; + + __disk_block_events(disk, true); + disk->ev->poll_msecs = intv; + __disk_unblock_events(disk, true); + + return count; +} + +static const DEVICE_ATTR(events, S_IRUGO, disk_events_show, NULL); +static const DEVICE_ATTR(events_async, S_IRUGO, disk_events_async_show, NULL); +static const DEVICE_ATTR(events_poll_msecs, S_IRUGO|S_IWUSR, + disk_events_poll_msecs_show, + disk_events_poll_msecs_store); + +static const struct attribute *disk_events_attrs[] = { + &dev_attr_events.attr, + &dev_attr_events_async.attr, + &dev_attr_events_poll_msecs.attr, + NULL, +}; + +/* + * The default polling interval can be specified by the kernel + * parameter block.events_dfl_poll_msecs which defaults to 0 + * (disable). This can also be modified runtime by writing to + * /sys/module/block/events_dfl_poll_msecs. + */ +static int disk_events_set_dfl_poll_msecs(const char *val, + const struct kernel_param *kp) +{ + struct disk_events *ev; + int ret; + + ret = param_set_ulong(val, kp); + if (ret < 0) + return ret; + + mutex_lock(&disk_events_mutex); + + list_for_each_entry(ev, &disk_events, node) + disk_check_events(ev->disk); + + mutex_unlock(&disk_events_mutex); + + return 0; +} + +static const struct kernel_param_ops disk_events_dfl_poll_msecs_param_ops = { + .set = disk_events_set_dfl_poll_msecs, + .get = param_get_ulong, +}; + +#undef MODULE_PARAM_PREFIX +#define MODULE_PARAM_PREFIX "block." + +module_param_cb(events_dfl_poll_msecs, &disk_events_dfl_poll_msecs_param_ops, + &disk_events_dfl_poll_msecs, 0644); + +/* + * disk_{add|del|release}_events - initialize and destroy disk_events. + */ +static void disk_add_events(struct gendisk *disk) +{ + struct disk_events *ev; + + if (!disk->fops->check_events || !(disk->events | disk->async_events)) + return; + + ev = kzalloc(sizeof(*ev), GFP_KERNEL); + if (!ev) { + pr_warn("%s: failed to initialize events\n", disk->disk_name); + return; + } + + if (sysfs_create_files(&disk_to_dev(disk)->kobj, + disk_events_attrs) < 0) { + pr_warn("%s: failed to create sysfs files for events\n", + disk->disk_name); + kfree(ev); + return; + } + + disk->ev = ev; + + INIT_LIST_HEAD(&ev->node); + ev->disk = disk; + spin_lock_init(&ev->lock); + ev->block = 1; + ev->poll_msecs = -1; + INIT_DELAYED_WORK(&ev->dwork, disk_events_workfn); + + mutex_lock(&disk_events_mutex); + list_add_tail(&ev->node, &disk_events); + mutex_unlock(&disk_events_mutex); + + /* + * Block count is initialized to 1 and the following initial + * unblock kicks it into action. + */ + __disk_unblock_events(disk, true); +} + +static void disk_del_events(struct gendisk *disk) +{ + if (!disk->ev) + return; + + __disk_block_events(disk, true); + + mutex_lock(&disk_events_mutex); + list_del_init(&disk->ev->node); + mutex_unlock(&disk_events_mutex); + + sysfs_remove_files(&disk_to_dev(disk)->kobj, disk_events_attrs); +} + +static void disk_release_events(struct gendisk *disk) +{ + /* the block count should be 1 from disk_del_events() */ + WARN_ON_ONCE(disk->ev && disk->ev->block != 1); + kfree(disk->ev); +} diff --git a/fs/block_dev.c b/fs/block_dev.c index c1c1b8c3fb99..6017389711ee 100644 --- a/fs/block_dev.c +++ b/fs/block_dev.c @@ -948,10 +948,11 @@ int check_disk_change(struct block_device *bdev) { struct gendisk *disk = bdev->bd_disk; const struct block_device_operations *bdops = disk->fops; + unsigned int events; - if (!bdops->media_changed) - return 0; - if (!bdops->media_changed(bdev->bd_disk)) + events = disk_clear_events(disk, DISK_EVENT_MEDIA_CHANGE | + DISK_EVENT_EJECT_REQUEST); + if (!(events & DISK_EVENT_MEDIA_CHANGE)) return 0; flush_disk(bdev); @@ -1158,9 +1159,10 @@ int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder) if (whole) { /* finish claiming */ + mutex_lock(&bdev->bd_mutex); spin_lock(&bdev_lock); - if (res == 0) { + if (!res) { BUG_ON(!bd_may_claim(bdev, whole, holder)); /* * Note that for a whole device bd_holders @@ -1180,6 +1182,20 @@ int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder) wake_up_bit(&whole->bd_claiming, 0); spin_unlock(&bdev_lock); + + /* + * Block event polling for write claims. Any write + * holder makes the write_holder state stick until all + * are released. This is good enough and tracking + * individual writeable reference is too fragile given + * the way @mode is used in blkdev_get/put(). + */ + if (!res && (mode & FMODE_WRITE) && !bdev->bd_write_holder) { + bdev->bd_write_holder = true; + disk_block_events(bdev->bd_disk); + } + + mutex_unlock(&bdev->bd_mutex); bdput(whole); } @@ -1353,12 +1369,23 @@ int blkdev_put(struct block_device *bdev, fmode_t mode) spin_unlock(&bdev_lock); - /* if this was the last claim, holder link should go too */ - if (bdev_free) + /* + * If this was the last claim, remove holder link and + * unblock evpoll if it was a write holder. + */ + if (bdev_free) { bd_unlink_disk_holder(bdev); + if (bdev->bd_write_holder) { + disk_unblock_events(bdev->bd_disk); + bdev->bd_write_holder = false; + } else + disk_check_events(bdev->bd_disk); + } mutex_unlock(&bdev->bd_mutex); - } + } else + disk_check_events(bdev->bd_disk); + return __blkdev_put(bdev, mode, 0); } EXPORT_SYMBOL(blkdev_put); diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h index 83031bcf8366..05667e6989f1 100644 --- a/include/linux/blkdev.h +++ b/include/linux/blkdev.h @@ -1251,6 +1251,9 @@ struct block_device_operations { int (*compat_ioctl) (struct block_device *, fmode_t, unsigned, unsigned long); int (*direct_access) (struct block_device *, sector_t, void **, unsigned long *); + unsigned int (*check_events) (struct gendisk *disk, + unsigned int clearing); + /* ->media_changed() is DEPRECATED, use ->check_events() instead */ int (*media_changed) (struct gendisk *); void (*unlock_native_capacity) (struct gendisk *); int (*revalidate_disk) (struct gendisk *); diff --git a/include/linux/fs.h b/include/linux/fs.h index f48501563917..997d22efdef0 100644 --- a/include/linux/fs.h +++ b/include/linux/fs.h @@ -662,6 +662,7 @@ struct block_device { void * bd_claiming; void * bd_holder; int bd_holders; + bool bd_write_holder; #ifdef CONFIG_SYSFS struct gendisk * bd_holder_disk; /* for sysfs slave linkng */ #endif diff --git a/include/linux/genhd.h b/include/linux/genhd.h index 56e17ed24816..13893aa2ac9d 100644 --- a/include/linux/genhd.h +++ b/include/linux/genhd.h @@ -127,6 +127,11 @@ struct hd_struct { #define GENHD_FL_EXT_DEVT 64 /* allow extended devt */ #define GENHD_FL_NATIVE_CAPACITY 128 +enum { + DISK_EVENT_MEDIA_CHANGE = 1 << 0, /* media changed */ + DISK_EVENT_EJECT_REQUEST = 1 << 1, /* eject requested */ +}; + #define BLK_SCSI_MAX_CMDS (256) #define BLK_SCSI_CMD_PER_LONG (BLK_SCSI_MAX_CMDS / (sizeof(long) * 8)) @@ -143,6 +148,8 @@ struct disk_part_tbl { struct hd_struct __rcu *part[]; }; +struct disk_events; + struct gendisk { /* major, first_minor and minors are input parameters only, * don't use directly. Use disk_devt() and disk_max_parts(). @@ -154,6 +161,10 @@ struct gendisk { char disk_name[DISK_NAME_LEN]; /* name of major driver */ char *(*devnode)(struct gendisk *gd, mode_t *mode); + + unsigned int events; /* supported events */ + unsigned int async_events; /* async events, subset of all */ + /* Array of pointers to partitions indexed by partno. * Protected with matching bdev lock but stat and other * non-critical accesses use RCU. Always access through @@ -171,8 +182,8 @@ struct gendisk { struct kobject *slave_dir; struct timer_rand_state *random; - atomic_t sync_io; /* RAID */ + struct disk_events *ev; #ifdef CONFIG_BLK_DEV_INTEGRITY struct blk_integrity *integrity; #endif @@ -405,6 +416,11 @@ static inline int get_disk_ro(struct gendisk *disk) return disk->part0.policy; } +extern void disk_block_events(struct gendisk *disk); +extern void disk_unblock_events(struct gendisk *disk); +extern void disk_check_events(struct gendisk *disk); +extern unsigned int disk_clear_events(struct gendisk *disk, unsigned int mask); + /* drivers/char/random.c */ extern void add_disk_randomness(struct gendisk *disk); extern void rand_initialize_disk(struct gendisk *disk); -- cgit v1.2.3 From 09e099d4bafea3b15be003d548bdf94b4b6e0e17 Mon Sep 17 00:00:00 2001 From: Jerome Marchand Date: Wed, 5 Jan 2011 16:57:38 +0100 Subject: block: fix accounting bug on cross partition merges /proc/diskstats would display a strange output as follows. $ cat /proc/diskstats |grep sda 8 0 sda 90524 7579 102154 20464 0 0 0 0 0 14096 20089 8 1 sda1 19085 1352 21841 4209 0 0 0 0 4294967064 15689 4293424691 ~~~~~~~~~~ 8 2 sda2 71252 3624 74891 15950 0 0 0 0 232 23995 1562390 8 3 sda3 54 487 2188 92 0 0 0 0 0 88 92 8 4 sda4 4 0 8 0 0 0 0 0 0 0 0 8 5 sda5 81 2027 2130 138 0 0 0 0 0 87 137 Its reason is the wrong way of accounting hd_struct->in_flight. When a bio is merged into a request belongs to different partition by ELEVATOR_FRONT_MERGE. The detailed root cause is as follows. Assuming that there are two partition, sda1 and sda2. 1. A request for sda2 is in request_queue. Hence sda1's hd_struct->in_flight is 0 and sda2's one is 1. | hd_struct->in_flight --------------------------- sda1 | 0 sda2 | 1 --------------------------- 2. A bio belongs to sda1 is issued and is merged into the request mentioned on step1 by ELEVATOR_BACK_MERGE. The first sector of the request is changed from sda2 region to sda1 region. However the two partition's hd_struct->in_flight are not changed. | hd_struct->in_flight --------------------------- sda1 | 0 sda2 | 1 --------------------------- 3. The request is finished and blk_account_io_done() is called. In this case, sda2's hd_struct->in_flight, not a sda1's one, is decremented. | hd_struct->in_flight --------------------------- sda1 | -1 sda2 | 1 --------------------------- The patch fixes the problem by caching the partition lookup inside the request structure, hence making sure that the increment and decrement will always happen on the same partition struct. This also speeds up IO with accounting enabled, since it cuts down on the number of lookups we have to do. Also add a refcount to struct hd_struct to keep the partition in memory as long as users exist. We use kref_test_and_get() to ensure we don't add a reference to a partition which is going away. Signed-off-by: Jerome Marchand Signed-off-by: Yasuaki Ishimatsu Cc: stable@kernel.org Signed-off-by: Jens Axboe --- block/blk-core.c | 26 +++++++++++++++++++++----- block/blk-merge.c | 3 ++- block/genhd.c | 1 + fs/partitions/check.c | 10 +++++++++- include/linux/blkdev.h | 1 + include/linux/genhd.h | 2 ++ 6 files changed, 36 insertions(+), 7 deletions(-) (limited to 'include/linux/blkdev.h') diff --git a/block/blk-core.c b/block/blk-core.c index 3689319a5974..500c080a6a6b 100644 --- a/block/blk-core.c +++ b/block/blk-core.c @@ -64,13 +64,27 @@ static void drive_stat_acct(struct request *rq, int new_io) return; cpu = part_stat_lock(); - part = disk_map_sector_rcu(rq->rq_disk, blk_rq_pos(rq)); - if (!new_io) + if (!new_io) { + part = rq->part; part_stat_inc(cpu, part, merges[rw]); - else { + } else { + part = disk_map_sector_rcu(rq->rq_disk, blk_rq_pos(rq)); + if (!kref_test_and_get(&part->ref)) { + /* + * The partition is already being removed, + * the request will be accounted on the disk only + * + * We take a reference on disk->part0 although that + * partition will never be deleted, so we can treat + * it as any other partition. + */ + part = &rq->rq_disk->part0; + kref_get(&part->ref); + } part_round_stats(cpu, part); part_inc_in_flight(part, rw); + rq->part = part; } part_stat_unlock(); @@ -128,6 +142,7 @@ void blk_rq_init(struct request_queue *q, struct request *rq) rq->ref_count = 1; rq->start_time = jiffies; set_start_time_ns(rq); + rq->part = NULL; } EXPORT_SYMBOL(blk_rq_init); @@ -1776,7 +1791,7 @@ static void blk_account_io_completion(struct request *req, unsigned int bytes) int cpu; cpu = part_stat_lock(); - part = disk_map_sector_rcu(req->rq_disk, blk_rq_pos(req)); + part = req->part; part_stat_add(cpu, part, sectors[rw], bytes >> 9); part_stat_unlock(); } @@ -1796,13 +1811,14 @@ static void blk_account_io_done(struct request *req) int cpu; cpu = part_stat_lock(); - part = disk_map_sector_rcu(req->rq_disk, blk_rq_pos(req)); + part = req->part; part_stat_inc(cpu, part, ios[rw]); part_stat_add(cpu, part, ticks[rw], duration); part_round_stats(cpu, part); part_dec_in_flight(part, rw); + kref_put(&part->ref, __delete_partition); part_stat_unlock(); } } diff --git a/block/blk-merge.c b/block/blk-merge.c index 77b7c26df6b5..b06b83b89d89 100644 --- a/block/blk-merge.c +++ b/block/blk-merge.c @@ -351,11 +351,12 @@ static void blk_account_io_merge(struct request *req) int cpu; cpu = part_stat_lock(); - part = disk_map_sector_rcu(req->rq_disk, blk_rq_pos(req)); + part = req->part; part_round_stats(cpu, part); part_dec_in_flight(part, rq_data_dir(req)); + kref_put(&part->ref, __delete_partition); part_stat_unlock(); } } diff --git a/block/genhd.c b/block/genhd.c index 16ccc0d2d5d3..85c150598830 100644 --- a/block/genhd.c +++ b/block/genhd.c @@ -1192,6 +1192,7 @@ struct gendisk *alloc_disk_node(int minors, int node_id) return NULL; } disk->part_tbl->part[0] = &disk->part0; + kref_init(&disk->part0.ref); disk->minors = minors; rand_initialize_disk(disk); diff --git a/fs/partitions/check.c b/fs/partitions/check.c index bdf8d3cc95a4..48209f58522b 100644 --- a/fs/partitions/check.c +++ b/fs/partitions/check.c @@ -381,6 +381,13 @@ static void delete_partition_rcu_cb(struct rcu_head *head) put_device(part_to_dev(part)); } +void __delete_partition(struct kref *ref) +{ + struct hd_struct *part = container_of(ref, struct hd_struct, ref); + + call_rcu(&part->rcu_head, delete_partition_rcu_cb); +} + void delete_partition(struct gendisk *disk, int partno) { struct disk_part_tbl *ptbl = disk->part_tbl; @@ -399,7 +406,7 @@ void delete_partition(struct gendisk *disk, int partno) kobject_put(part->holder_dir); device_del(part_to_dev(part)); - call_rcu(&part->rcu_head, delete_partition_rcu_cb); + kref_put(&part->ref, __delete_partition); } static ssize_t whole_disk_show(struct device *dev, @@ -498,6 +505,7 @@ struct hd_struct *add_partition(struct gendisk *disk, int partno, if (!dev_get_uevent_suppress(ddev)) kobject_uevent(&pdev->kobj, KOBJ_ADD); + kref_init(&p->ref); return p; out_free_info: diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h index aae86fd10c4f..482a7fd48831 100644 --- a/include/linux/blkdev.h +++ b/include/linux/blkdev.h @@ -115,6 +115,7 @@ struct request { void *elevator_private3; struct gendisk *rq_disk; + struct hd_struct *part; unsigned long start_time; #ifdef CONFIG_BLK_CGROUP unsigned long long start_time_ns; diff --git a/include/linux/genhd.h b/include/linux/genhd.h index 7a7b9c1644e4..2ba2792a3dd4 100644 --- a/include/linux/genhd.h +++ b/include/linux/genhd.h @@ -116,6 +116,7 @@ struct hd_struct { struct disk_stats dkstats; #endif struct rcu_head rcu_head; + struct kref ref; }; #define GENHD_FL_REMOVABLE 1 @@ -583,6 +584,7 @@ extern struct hd_struct * __must_check add_partition(struct gendisk *disk, sector_t len, int flags, struct partition_meta_info *info); +extern void __delete_partition(struct kref *ref); extern void delete_partition(struct gendisk *, int); extern void printk_all_partitions(void); -- cgit v1.2.3 From 450adcbe518ab3a3953d8475309525d22de77cba Mon Sep 17 00:00:00 2001 From: Vivek Goyal Date: Tue, 1 Mar 2011 13:40:54 -0500 Subject: blk-throttle: Do not use kblockd workqueue for throtl work o Dominik Klein reported a system hang issue while doing some blkio throttling testing. https://lkml.org/lkml/2011/2/24/173 o Some tracing revealed that CFQ was not dispatching any more jobs as queue unplug was not happening. And queue unplug was not happening because unplug work was not being called as there was one throttling work on same cpu which as not finished yet. And throttling work had not finished as it was tyring to dispatch a bio to CFQ but all the request descriptors were consume to it was put to sleep. o So basically it is a cyclic dependecny between CFQ unplug work and throtl dispatch work. Tejun suggested that use separate workqueue for such cases. o This patch uses a separate workqueue for throttle related work and does not rely on kblockd workqueue anymore. Cc: stable@kernel.org Reported-by: Dominik Klein Signed-off-by: Vivek Goyal Acked-by: Tejun Heo Signed-off-by: Jens Axboe --- block/blk-core.c | 7 ------- block/blk-throttle.c | 29 ++++++++++++++++++----------- include/linux/blkdev.h | 3 --- 3 files changed, 18 insertions(+), 21 deletions(-) (limited to 'include/linux/blkdev.h') diff --git a/block/blk-core.c b/block/blk-core.c index 2f4002f79a24..792ece276160 100644 --- a/block/blk-core.c +++ b/block/blk-core.c @@ -2610,13 +2610,6 @@ int kblockd_schedule_work(struct request_queue *q, struct work_struct *work) } EXPORT_SYMBOL(kblockd_schedule_work); -int kblockd_schedule_delayed_work(struct request_queue *q, - struct delayed_work *dwork, unsigned long delay) -{ - return queue_delayed_work(kblockd_workqueue, dwork, delay); -} -EXPORT_SYMBOL(kblockd_schedule_delayed_work); - int __init blk_dev_init(void) { BUILD_BUG_ON(__REQ_NR_BITS > 8 * diff --git a/block/blk-throttle.c b/block/blk-throttle.c index a89043a3caa4..e36cc10a346c 100644 --- a/block/blk-throttle.c +++ b/block/blk-throttle.c @@ -20,6 +20,11 @@ static int throtl_quantum = 32; /* Throttling is performed over 100ms slice and after that slice is renewed */ static unsigned long throtl_slice = HZ/10; /* 100 ms */ +/* A workqueue to queue throttle related work */ +static struct workqueue_struct *kthrotld_workqueue; +static void throtl_schedule_delayed_work(struct throtl_data *td, + unsigned long delay); + struct throtl_rb_root { struct rb_root rb; struct rb_node *left; @@ -345,10 +350,9 @@ static void throtl_schedule_next_dispatch(struct throtl_data *td) update_min_dispatch_time(st); if (time_before_eq(st->min_disptime, jiffies)) - throtl_schedule_delayed_work(td->queue, 0); + throtl_schedule_delayed_work(td, 0); else - throtl_schedule_delayed_work(td->queue, - (st->min_disptime - jiffies)); + throtl_schedule_delayed_work(td, (st->min_disptime - jiffies)); } static inline void @@ -815,10 +819,10 @@ void blk_throtl_work(struct work_struct *work) } /* Call with queue lock held */ -void throtl_schedule_delayed_work(struct request_queue *q, unsigned long delay) +static void +throtl_schedule_delayed_work(struct throtl_data *td, unsigned long delay) { - struct throtl_data *td = q->td; struct delayed_work *dwork = &td->throtl_work; if (total_nr_queued(td) > 0) { @@ -827,12 +831,11 @@ void throtl_schedule_delayed_work(struct request_queue *q, unsigned long delay) * Cancel that and schedule a new one. */ __cancel_delayed_work(dwork); - kblockd_schedule_delayed_work(q, dwork, delay); + queue_delayed_work(kthrotld_workqueue, dwork, delay); throtl_log(td, "schedule work. delay=%lu jiffies=%lu", delay, jiffies); } } -EXPORT_SYMBOL(throtl_schedule_delayed_work); static void throtl_destroy_tg(struct throtl_data *td, struct throtl_grp *tg) @@ -920,7 +923,7 @@ static void throtl_update_blkio_group_read_bps(void *key, smp_mb__after_atomic_inc(); /* Schedule a work now to process the limit change */ - throtl_schedule_delayed_work(td->queue, 0); + throtl_schedule_delayed_work(td, 0); } static void throtl_update_blkio_group_write_bps(void *key, @@ -934,7 +937,7 @@ static void throtl_update_blkio_group_write_bps(void *key, smp_mb__before_atomic_inc(); atomic_inc(&td->limits_changed); smp_mb__after_atomic_inc(); - throtl_schedule_delayed_work(td->queue, 0); + throtl_schedule_delayed_work(td, 0); } static void throtl_update_blkio_group_read_iops(void *key, @@ -948,7 +951,7 @@ static void throtl_update_blkio_group_read_iops(void *key, smp_mb__before_atomic_inc(); atomic_inc(&td->limits_changed); smp_mb__after_atomic_inc(); - throtl_schedule_delayed_work(td->queue, 0); + throtl_schedule_delayed_work(td, 0); } static void throtl_update_blkio_group_write_iops(void *key, @@ -962,7 +965,7 @@ static void throtl_update_blkio_group_write_iops(void *key, smp_mb__before_atomic_inc(); atomic_inc(&td->limits_changed); smp_mb__after_atomic_inc(); - throtl_schedule_delayed_work(td->queue, 0); + throtl_schedule_delayed_work(td, 0); } void throtl_shutdown_timer_wq(struct request_queue *q) @@ -1135,6 +1138,10 @@ void blk_throtl_exit(struct request_queue *q) static int __init throtl_init(void) { + kthrotld_workqueue = alloc_workqueue("kthrotld", WQ_MEM_RECLAIM, 0); + if (!kthrotld_workqueue) + panic("Failed to create kthrotld\n"); + blkio_policy_register(&blkio_policy_throtl); return 0; } diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h index 4d18ff34670a..dd8cd0f47e3a 100644 --- a/include/linux/blkdev.h +++ b/include/linux/blkdev.h @@ -1088,7 +1088,6 @@ static inline void put_dev_sector(Sector p) struct work_struct; int kblockd_schedule_work(struct request_queue *q, struct work_struct *work); -int kblockd_schedule_delayed_work(struct request_queue *q, struct delayed_work *dwork, unsigned long delay); #ifdef CONFIG_BLK_CGROUP /* @@ -1136,7 +1135,6 @@ static inline uint64_t rq_io_start_time_ns(struct request *req) extern int blk_throtl_init(struct request_queue *q); extern void blk_throtl_exit(struct request_queue *q); extern int blk_throtl_bio(struct request_queue *q, struct bio **bio); -extern void throtl_schedule_delayed_work(struct request_queue *q, unsigned long delay); extern void throtl_shutdown_timer_wq(struct request_queue *q); #else /* CONFIG_BLK_DEV_THROTTLING */ static inline int blk_throtl_bio(struct request_queue *q, struct bio **bio) @@ -1146,7 +1144,6 @@ static inline int blk_throtl_bio(struct request_queue *q, struct bio **bio) static inline int blk_throtl_init(struct request_queue *q) { return 0; } static inline int blk_throtl_exit(struct request_queue *q) { return 0; } -static inline void throtl_schedule_delayed_work(struct request_queue *q, unsigned long delay) {} static inline void throtl_shutdown_timer_wq(struct request_queue *q) {} #endif /* CONFIG_BLK_DEV_THROTTLING */ -- cgit v1.2.3 From 1654e7411a1ad4999fe7890ef51d2a2bbb1fcf76 Mon Sep 17 00:00:00 2001 From: Tejun Heo Date: Wed, 2 Mar 2011 08:48:05 -0500 Subject: block: add @force_kblockd to __blk_run_queue() __blk_run_queue() automatically either calls q->request_fn() directly or schedules kblockd depending on whether the function is recursed. blk-flush implementation needs to be able to explicitly choose kblockd. Add @force_kblockd. All the current users are converted to specify %false for the parameter and this patch doesn't introduce any behavior change. stable: This is prerequisite for fixing ide oops caused by the new blk-flush implementation. Signed-off-by: Tejun Heo Cc: Jan Beulich Cc: James Bottomley Cc: stable@kernel.org Signed-off-by: Jens Axboe --- block/blk-core.c | 11 ++++++----- block/blk-flush.c | 2 +- block/cfq-iosched.c | 6 +++--- block/elevator.c | 4 ++-- drivers/scsi/scsi_lib.c | 2 +- drivers/scsi/scsi_transport_fc.c | 2 +- include/linux/blkdev.h | 2 +- 7 files changed, 15 insertions(+), 14 deletions(-) (limited to 'include/linux/blkdev.h') diff --git a/block/blk-core.c b/block/blk-core.c index 792ece276160..518dd423a5fe 100644 --- a/block/blk-core.c +++ b/block/blk-core.c @@ -352,7 +352,7 @@ void blk_start_queue(struct request_queue *q) WARN_ON(!irqs_disabled()); queue_flag_clear(QUEUE_FLAG_STOPPED, q); - __blk_run_queue(q); + __blk_run_queue(q, false); } EXPORT_SYMBOL(blk_start_queue); @@ -403,13 +403,14 @@ EXPORT_SYMBOL(blk_sync_queue); /** * __blk_run_queue - run a single device queue * @q: The queue to run + * @force_kblockd: Don't run @q->request_fn directly. Use kblockd. * * Description: * See @blk_run_queue. This variant must be called with the queue lock * held and interrupts disabled. * */ -void __blk_run_queue(struct request_queue *q) +void __blk_run_queue(struct request_queue *q, bool force_kblockd) { blk_remove_plug(q); @@ -423,7 +424,7 @@ void __blk_run_queue(struct request_queue *q) * Only recurse once to avoid overrunning the stack, let the unplug * handling reinvoke the handler shortly if we already got there. */ - if (!queue_flag_test_and_set(QUEUE_FLAG_REENTER, q)) { + if (!force_kblockd && !queue_flag_test_and_set(QUEUE_FLAG_REENTER, q)) { q->request_fn(q); queue_flag_clear(QUEUE_FLAG_REENTER, q); } else { @@ -446,7 +447,7 @@ void blk_run_queue(struct request_queue *q) unsigned long flags; spin_lock_irqsave(q->queue_lock, flags); - __blk_run_queue(q); + __blk_run_queue(q, false); spin_unlock_irqrestore(q->queue_lock, flags); } EXPORT_SYMBOL(blk_run_queue); @@ -1053,7 +1054,7 @@ void blk_insert_request(struct request_queue *q, struct request *rq, drive_stat_acct(rq, 1); __elv_add_request(q, rq, where, 0); - __blk_run_queue(q); + __blk_run_queue(q, false); spin_unlock_irqrestore(q->queue_lock, flags); } EXPORT_SYMBOL(blk_insert_request); diff --git a/block/blk-flush.c b/block/blk-flush.c index 54b123d6563e..56adaa8d55cd 100644 --- a/block/blk-flush.c +++ b/block/blk-flush.c @@ -69,7 +69,7 @@ static void blk_flush_complete_seq_end_io(struct request_queue *q, * queue. Kick the queue in those cases. */ if (was_empty && next_rq) - __blk_run_queue(q); + __blk_run_queue(q, false); } static void pre_flush_end_io(struct request *rq, int error) diff --git a/block/cfq-iosched.c b/block/cfq-iosched.c index 7be4c7959625..ea83a4f0c27d 100644 --- a/block/cfq-iosched.c +++ b/block/cfq-iosched.c @@ -3355,7 +3355,7 @@ cfq_rq_enqueued(struct cfq_data *cfqd, struct cfq_queue *cfqq, cfqd->busy_queues > 1) { cfq_del_timer(cfqd, cfqq); cfq_clear_cfqq_wait_request(cfqq); - __blk_run_queue(cfqd->queue); + __blk_run_queue(cfqd->queue, false); } else { cfq_blkiocg_update_idle_time_stats( &cfqq->cfqg->blkg); @@ -3370,7 +3370,7 @@ cfq_rq_enqueued(struct cfq_data *cfqd, struct cfq_queue *cfqq, * this new queue is RT and the current one is BE */ cfq_preempt_queue(cfqd, cfqq); - __blk_run_queue(cfqd->queue); + __blk_run_queue(cfqd->queue, false); } } @@ -3731,7 +3731,7 @@ static void cfq_kick_queue(struct work_struct *work) struct request_queue *q = cfqd->queue; spin_lock_irq(q->queue_lock); - __blk_run_queue(cfqd->queue); + __blk_run_queue(cfqd->queue, false); spin_unlock_irq(q->queue_lock); } diff --git a/block/elevator.c b/block/elevator.c index 2569512830d3..236e93c1f46c 100644 --- a/block/elevator.c +++ b/block/elevator.c @@ -602,7 +602,7 @@ void elv_quiesce_start(struct request_queue *q) */ elv_drain_elevator(q); while (q->rq.elvpriv) { - __blk_run_queue(q); + __blk_run_queue(q, false); spin_unlock_irq(q->queue_lock); msleep(10); spin_lock_irq(q->queue_lock); @@ -651,7 +651,7 @@ void elv_insert(struct request_queue *q, struct request *rq, int where) * with anything. There's no point in delaying queue * processing. */ - __blk_run_queue(q); + __blk_run_queue(q, false); break; case ELEVATOR_INSERT_SORT: diff --git a/drivers/scsi/scsi_lib.c b/drivers/scsi/scsi_lib.c index 9045c52abd25..fb2bb35c62cb 100644 --- a/drivers/scsi/scsi_lib.c +++ b/drivers/scsi/scsi_lib.c @@ -443,7 +443,7 @@ static void scsi_run_queue(struct request_queue *q) &sdev->request_queue->queue_flags); if (flagset) queue_flag_set(QUEUE_FLAG_REENTER, sdev->request_queue); - __blk_run_queue(sdev->request_queue); + __blk_run_queue(sdev->request_queue, false); if (flagset) queue_flag_clear(QUEUE_FLAG_REENTER, sdev->request_queue); spin_unlock(sdev->request_queue->queue_lock); diff --git a/drivers/scsi/scsi_transport_fc.c b/drivers/scsi/scsi_transport_fc.c index 998c01be3234..5c3ccfc6b622 100644 --- a/drivers/scsi/scsi_transport_fc.c +++ b/drivers/scsi/scsi_transport_fc.c @@ -3829,7 +3829,7 @@ fc_bsg_goose_queue(struct fc_rport *rport) !test_bit(QUEUE_FLAG_REENTER, &rport->rqst_q->queue_flags); if (flagset) queue_flag_set(QUEUE_FLAG_REENTER, rport->rqst_q); - __blk_run_queue(rport->rqst_q); + __blk_run_queue(rport->rqst_q, false); if (flagset) queue_flag_clear(QUEUE_FLAG_REENTER, rport->rqst_q); spin_unlock_irqrestore(rport->rqst_q->queue_lock, flags); diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h index dd8cd0f47e3a..d5063e1b5555 100644 --- a/include/linux/blkdev.h +++ b/include/linux/blkdev.h @@ -699,7 +699,7 @@ extern void blk_start_queue(struct request_queue *q); extern void blk_stop_queue(struct request_queue *q); extern void blk_sync_queue(struct request_queue *q); extern void __blk_stop_queue(struct request_queue *q); -extern void __blk_run_queue(struct request_queue *); +extern void __blk_run_queue(struct request_queue *q, bool force_kblockd); extern void blk_run_queue(struct request_queue *); extern int blk_rq_map_user(struct request_queue *, struct request *, struct rq_map_data *, void __user *, unsigned long, -- cgit v1.2.3