linux/block/blk-throttle.c

1697 lines
46 KiB
C
Raw Normal View History

/*
* Interface for controlling IO bandwidth on a request queue
*
* Copyright (C) 2010 Vivek Goyal <vgoyal@redhat.com>
*/
#include <linux/module.h>
#include <linux/slab.h>
#include <linux/blkdev.h>
#include <linux/bio.h>
#include <linux/blktrace_api.h>
#include "blk-cgroup.h"
#include "blk.h"
/* Max dispatch from a group in 1 round */
static int throtl_grp_quantum = 8;
/* Total max dispatch from all groups in one round */
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 */
static struct blkcg_policy blkcg_policy_throtl;
/* A workqueue to queue throttle related work */
static struct workqueue_struct *kthrotld_workqueue;
blk-throttle: add throtl_qnode for dispatch fairness With flat hierarchy, there's only single level of dispatching happening and fairness beyond that point is the responsibility of the rest of the block layer and driver, which usually works out okay; however, with the planned hierarchy support, service_queue->bio_lists[] can be filled up by bios from a single source. While the limits would still be honored, it'd be very easy to starve IOs from siblings or children. To avoid such starvation, this patch implements throtl_qnode and converts service_queue->bio_lists[] to lists of per-source qnodes which in turn contains the bio's. For example, when a bio is dispatched from a child group, the bio doesn't get queued on ->bio_lists[] directly but it first gets queued on the group's qnode which in turn gets queued on service_queue->queued[]. When dispatching for the upper level, the ->queued[] list is consumed in round-robing order so that the dispatch windows is consumed fairly by all IO sources. There are two ways a bio can come to a throtl_grp - directly queued to the group or dispatched from a child. For the former throtl_grp->qnode_on_self[rw] is used. For the latter, the child's ->qnode_on_parent[rw]. Note that this means that the child which is contributing a bio to its parent should stay pinned until all its bios are dispatched to its grand-parent. This patch moves blkg refcnting from bio add/remove spots to qnode activation/deactivation so that the blkg containing an active qnode is always pinned. As child pins the parent, this is sufficient for keeping the relevant sub-tree pinned while bios are in flight. The starvation issue was spotted by Vivek Goyal. v2: The original patch used the same throtl_grp->qnode_on_self/parent for reads and writes causing RWs to be queued incorrectly if there already are outstanding IOs in the other direction. They should be throtl_grp->qnode_on_self/parent[2] so that READs and WRITEs can use different qnodes. Spotted by Vivek Goyal. Signed-off-by: Tejun Heo <tj@kernel.org> Acked-by: Vivek Goyal <vgoyal@redhat.com>
2013-05-14 22:52:38 +02:00
/*
* To implement hierarchical throttling, throtl_grps form a tree and bios
* are dispatched upwards level by level until they reach the top and get
* issued. When dispatching bios from the children and local group at each
* level, if the bios are dispatched into a single bio_list, there's a risk
* of a local or child group which can queue many bios at once filling up
* the list starving others.
*
* To avoid such starvation, dispatched bios are queued separately
* according to where they came from. When they are again dispatched to
* the parent, they're popped in round-robin order so that no single source
* hogs the dispatch window.
*
* throtl_qnode is used to keep the queued bios separated by their sources.
* Bios are queued to throtl_qnode which in turn is queued to
* throtl_service_queue and then dispatched in round-robin order.
*
* It's also used to track the reference counts on blkg's. A qnode always
* belongs to a throtl_grp and gets queued on itself or the parent, so
* incrementing the reference of the associated throtl_grp when a qnode is
* queued and decrementing when dequeued is enough to keep the whole blkg
* tree pinned while bios are in flight.
*/
struct throtl_qnode {
struct list_head node; /* service_queue->queued[] */
struct bio_list bios; /* queued bios */
struct throtl_grp *tg; /* tg this qnode belongs to */
};
struct throtl_service_queue {
struct throtl_service_queue *parent_sq; /* the parent service_queue */
/*
* Bios queued directly to this service_queue or dispatched from
* children throtl_grp's.
*/
blk-throttle: add throtl_qnode for dispatch fairness With flat hierarchy, there's only single level of dispatching happening and fairness beyond that point is the responsibility of the rest of the block layer and driver, which usually works out okay; however, with the planned hierarchy support, service_queue->bio_lists[] can be filled up by bios from a single source. While the limits would still be honored, it'd be very easy to starve IOs from siblings or children. To avoid such starvation, this patch implements throtl_qnode and converts service_queue->bio_lists[] to lists of per-source qnodes which in turn contains the bio's. For example, when a bio is dispatched from a child group, the bio doesn't get queued on ->bio_lists[] directly but it first gets queued on the group's qnode which in turn gets queued on service_queue->queued[]. When dispatching for the upper level, the ->queued[] list is consumed in round-robing order so that the dispatch windows is consumed fairly by all IO sources. There are two ways a bio can come to a throtl_grp - directly queued to the group or dispatched from a child. For the former throtl_grp->qnode_on_self[rw] is used. For the latter, the child's ->qnode_on_parent[rw]. Note that this means that the child which is contributing a bio to its parent should stay pinned until all its bios are dispatched to its grand-parent. This patch moves blkg refcnting from bio add/remove spots to qnode activation/deactivation so that the blkg containing an active qnode is always pinned. As child pins the parent, this is sufficient for keeping the relevant sub-tree pinned while bios are in flight. The starvation issue was spotted by Vivek Goyal. v2: The original patch used the same throtl_grp->qnode_on_self/parent for reads and writes causing RWs to be queued incorrectly if there already are outstanding IOs in the other direction. They should be throtl_grp->qnode_on_self/parent[2] so that READs and WRITEs can use different qnodes. Spotted by Vivek Goyal. Signed-off-by: Tejun Heo <tj@kernel.org> Acked-by: Vivek Goyal <vgoyal@redhat.com>
2013-05-14 22:52:38 +02:00
struct list_head queued[2]; /* throtl_qnode [READ/WRITE] */
unsigned int nr_queued[2]; /* number of queued bios */
/*
* RB tree of active children throtl_grp's, which are sorted by
* their ->disptime.
*/
struct rb_root pending_tree; /* RB tree of active tgs */
struct rb_node *first_pending; /* first node in the tree */
unsigned int nr_pending; /* # queued in the tree */
unsigned long first_pending_disptime; /* disptime of the first tg */
struct timer_list pending_timer; /* fires on first_pending_disptime */
};
enum tg_state_flags {
THROTL_TG_PENDING = 1 << 0, /* on parent's pending tree */
THROTL_TG_WAS_EMPTY = 1 << 1, /* bio_lists[] became non-empty */
};
#define rb_entry_tg(node) rb_entry((node), struct throtl_grp, rb_node)
/* Per-cpu group stats */
struct tg_stats_cpu {
/* total bytes transferred */
struct blkg_rwstat service_bytes;
/* total IOs serviced, post merge */
struct blkg_rwstat serviced;
};
struct throtl_grp {
/* must be the first member */
struct blkg_policy_data pd;
/* active throtl group service_queue member */
struct rb_node rb_node;
/* throtl_data this group belongs to */
struct throtl_data *td;
/* this group's service queue */
struct throtl_service_queue service_queue;
blk-throttle: add throtl_qnode for dispatch fairness With flat hierarchy, there's only single level of dispatching happening and fairness beyond that point is the responsibility of the rest of the block layer and driver, which usually works out okay; however, with the planned hierarchy support, service_queue->bio_lists[] can be filled up by bios from a single source. While the limits would still be honored, it'd be very easy to starve IOs from siblings or children. To avoid such starvation, this patch implements throtl_qnode and converts service_queue->bio_lists[] to lists of per-source qnodes which in turn contains the bio's. For example, when a bio is dispatched from a child group, the bio doesn't get queued on ->bio_lists[] directly but it first gets queued on the group's qnode which in turn gets queued on service_queue->queued[]. When dispatching for the upper level, the ->queued[] list is consumed in round-robing order so that the dispatch windows is consumed fairly by all IO sources. There are two ways a bio can come to a throtl_grp - directly queued to the group or dispatched from a child. For the former throtl_grp->qnode_on_self[rw] is used. For the latter, the child's ->qnode_on_parent[rw]. Note that this means that the child which is contributing a bio to its parent should stay pinned until all its bios are dispatched to its grand-parent. This patch moves blkg refcnting from bio add/remove spots to qnode activation/deactivation so that the blkg containing an active qnode is always pinned. As child pins the parent, this is sufficient for keeping the relevant sub-tree pinned while bios are in flight. The starvation issue was spotted by Vivek Goyal. v2: The original patch used the same throtl_grp->qnode_on_self/parent for reads and writes causing RWs to be queued incorrectly if there already are outstanding IOs in the other direction. They should be throtl_grp->qnode_on_self/parent[2] so that READs and WRITEs can use different qnodes. Spotted by Vivek Goyal. Signed-off-by: Tejun Heo <tj@kernel.org> Acked-by: Vivek Goyal <vgoyal@redhat.com>
2013-05-14 22:52:38 +02:00
/*
* qnode_on_self is used when bios are directly queued to this
* throtl_grp so that local bios compete fairly with bios
* dispatched from children. qnode_on_parent is used when bios are
* dispatched from this throtl_grp into its parent and will compete
* with the sibling qnode_on_parents and the parent's
* qnode_on_self.
*/
struct throtl_qnode qnode_on_self[2];
struct throtl_qnode qnode_on_parent[2];
/*
* Dispatch time in jiffies. This is the estimated time when group
* will unthrottle and is ready to dispatch more bio. It is used as
* key to sort active groups in service tree.
*/
unsigned long disptime;
unsigned int flags;
/* are there any throtl rules between this group and td? */
bool has_rules[2];
/* bytes per second rate limits */
uint64_t bps[2];
/* IOPS limits */
unsigned int iops[2];
/* Number of bytes disptached in current slice */
uint64_t bytes_disp[2];
/* Number of bio's dispatched in current slice */
unsigned int io_disp[2];
/* When did we start a new slice */
unsigned long slice_start[2];
unsigned long slice_end[2];
/* Per cpu stats pointer */
struct tg_stats_cpu __percpu *stats_cpu;
/* List of tgs waiting for per cpu stats memory to be allocated */
struct list_head stats_alloc_node;
};
struct throtl_data
{
/* service tree for active throtl groups */
struct throtl_service_queue service_queue;
struct request_queue *queue;
/* Total Number of queued bios on READ and WRITE lists */
unsigned int nr_queued[2];
/*
* number of total undestroyed groups
*/
unsigned int nr_undestroyed_grps;
/* Work for dispatching throttled bios */
struct work_struct dispatch_work;
};
/* list and work item to allocate percpu group stats */
static DEFINE_SPINLOCK(tg_stats_alloc_lock);
static LIST_HEAD(tg_stats_alloc_list);
static void tg_stats_alloc_fn(struct work_struct *);
static DECLARE_DELAYED_WORK(tg_stats_alloc_work, tg_stats_alloc_fn);
static void throtl_pending_timer_fn(unsigned long arg);
static inline struct throtl_grp *pd_to_tg(struct blkg_policy_data *pd)
{
return pd ? container_of(pd, struct throtl_grp, pd) : NULL;
}
static inline struct throtl_grp *blkg_to_tg(struct blkcg_gq *blkg)
{
return pd_to_tg(blkg_to_pd(blkg, &blkcg_policy_throtl));
}
static inline struct blkcg_gq *tg_to_blkg(struct throtl_grp *tg)
{
return pd_to_blkg(&tg->pd);
}
static inline struct throtl_grp *td_root_tg(struct throtl_data *td)
{
return blkg_to_tg(td->queue->root_blkg);
}
/**
* sq_to_tg - return the throl_grp the specified service queue belongs to
* @sq: the throtl_service_queue of interest
*
* Return the throtl_grp @sq belongs to. If @sq is the top-level one
* embedded in throtl_data, %NULL is returned.
*/
static struct throtl_grp *sq_to_tg(struct throtl_service_queue *sq)
{
if (sq && sq->parent_sq)
return container_of(sq, struct throtl_grp, service_queue);
else
return NULL;
}
/**
* sq_to_td - return throtl_data the specified service queue belongs to
* @sq: the throtl_service_queue of interest
*
* A service_queue can be embeded in either a throtl_grp or throtl_data.
* Determine the associated throtl_data accordingly and return it.
*/
static struct throtl_data *sq_to_td(struct throtl_service_queue *sq)
{
struct throtl_grp *tg = sq_to_tg(sq);
if (tg)
return tg->td;
else
return container_of(sq, struct throtl_data, service_queue);
}
/**
* throtl_log - log debug message via blktrace
* @sq: the service_queue being reported
* @fmt: printf format string
* @args: printf args
*
* The messages are prefixed with "throtl BLKG_NAME" if @sq belongs to a
* throtl_grp; otherwise, just "throtl".
*
* TODO: this should be made a function and name formatting should happen
* after testing whether blktrace is enabled.
*/
#define throtl_log(sq, fmt, args...) do { \
struct throtl_grp *__tg = sq_to_tg((sq)); \
struct throtl_data *__td = sq_to_td((sq)); \
\
(void)__td; \
if ((__tg)) { \
char __pbuf[128]; \
\
blkg_path(tg_to_blkg(__tg), __pbuf, sizeof(__pbuf)); \
blk_add_trace_msg(__td->queue, "throtl %s " fmt, __pbuf, ##args); \
} else { \
blk_add_trace_msg(__td->queue, "throtl " fmt, ##args); \
} \
} while (0)
block: Use u64_stats_init() to initialize seqcounts Now that seqcounts are lockdep enabled objects, we need to explicitly initialize runtime allocated seqcounts so that lockdep can track them. Without this patch, Fengguang was seeing: [ 4.127282] INFO: trying to register non-static key. [ 4.128027] the code is fine but needs lockdep annotation. [ 4.128027] turning off the locking correctness validator. [ 4.128027] CPU: 0 PID: 96 Comm: kworker/u4:1 Not tainted 3.12.0-next-20131108-10601-gbad570d #2 [ 4.128027] Hardware name: Bochs Bochs, BIOS Bochs 01/01/2011 [ ... ] [ 4.128027] Call Trace: [ 4.128027] [<7908e744>] ? console_unlock+0x353/0x380 [ 4.128027] [<79dc7cf2>] dump_stack+0x48/0x60 [ 4.128027] [<7908953e>] __lock_acquire.isra.26+0x7e3/0xceb [ 4.128027] [<7908a1c5>] lock_acquire+0x71/0x9a [ 4.128027] [<794079aa>] ? blk_throtl_bio+0x1c3/0x485 [ 4.128027] [<7940658b>] throtl_update_dispatch_stats+0x7c/0x153 [ 4.128027] [<794079aa>] ? blk_throtl_bio+0x1c3/0x485 [ 4.128027] [<794079aa>] blk_throtl_bio+0x1c3/0x485 ... Use u64_stats_init() for all affected data structures, which initializes the seqcount. Reported-and-Tested-by: Fengguang Wu <fengguang.wu@intel.com> Cc: Vivek Goyal <vgoyal@redhat.com> Cc: Jens Axboe <axboe@kernel.dk> Signed-off-by: Peter Zijlstra <peterz@infradead.org> [ Folded in another fix from the mailing list as well as a fix to that fix. Tweaked commit message. ] Signed-off-by: John Stultz <john.stultz@linaro.org> Signed-off-by: Peter Zijlstra <peterz@infradead.org> Link: http://lkml.kernel.org/r/1384314134-6895-1-git-send-email-john.stultz@linaro.org [ So I actually think that the two SOBs from PeterZ are the right depiction of the patch route. ] Signed-off-by: Ingo Molnar <mingo@kernel.org>
2013-11-13 04:42:14 +01:00
static void tg_stats_init(struct tg_stats_cpu *tg_stats)
{
blkg_rwstat_init(&tg_stats->service_bytes);
blkg_rwstat_init(&tg_stats->serviced);
}
/*
* Worker for allocating per cpu stat for tgs. This is scheduled on the
* system_wq once there are some groups on the alloc_list waiting for
* allocation.
*/
static void tg_stats_alloc_fn(struct work_struct *work)
{
static struct tg_stats_cpu *stats_cpu; /* this fn is non-reentrant */
struct delayed_work *dwork = to_delayed_work(work);
bool empty = false;
alloc_stats:
if (!stats_cpu) {
block: Use u64_stats_init() to initialize seqcounts Now that seqcounts are lockdep enabled objects, we need to explicitly initialize runtime allocated seqcounts so that lockdep can track them. Without this patch, Fengguang was seeing: [ 4.127282] INFO: trying to register non-static key. [ 4.128027] the code is fine but needs lockdep annotation. [ 4.128027] turning off the locking correctness validator. [ 4.128027] CPU: 0 PID: 96 Comm: kworker/u4:1 Not tainted 3.12.0-next-20131108-10601-gbad570d #2 [ 4.128027] Hardware name: Bochs Bochs, BIOS Bochs 01/01/2011 [ ... ] [ 4.128027] Call Trace: [ 4.128027] [<7908e744>] ? console_unlock+0x353/0x380 [ 4.128027] [<79dc7cf2>] dump_stack+0x48/0x60 [ 4.128027] [<7908953e>] __lock_acquire.isra.26+0x7e3/0xceb [ 4.128027] [<7908a1c5>] lock_acquire+0x71/0x9a [ 4.128027] [<794079aa>] ? blk_throtl_bio+0x1c3/0x485 [ 4.128027] [<7940658b>] throtl_update_dispatch_stats+0x7c/0x153 [ 4.128027] [<794079aa>] ? blk_throtl_bio+0x1c3/0x485 [ 4.128027] [<794079aa>] blk_throtl_bio+0x1c3/0x485 ... Use u64_stats_init() for all affected data structures, which initializes the seqcount. Reported-and-Tested-by: Fengguang Wu <fengguang.wu@intel.com> Cc: Vivek Goyal <vgoyal@redhat.com> Cc: Jens Axboe <axboe@kernel.dk> Signed-off-by: Peter Zijlstra <peterz@infradead.org> [ Folded in another fix from the mailing list as well as a fix to that fix. Tweaked commit message. ] Signed-off-by: John Stultz <john.stultz@linaro.org> Signed-off-by: Peter Zijlstra <peterz@infradead.org> Link: http://lkml.kernel.org/r/1384314134-6895-1-git-send-email-john.stultz@linaro.org [ So I actually think that the two SOBs from PeterZ are the right depiction of the patch route. ] Signed-off-by: Ingo Molnar <mingo@kernel.org>
2013-11-13 04:42:14 +01:00
int cpu;
stats_cpu = alloc_percpu(struct tg_stats_cpu);
if (!stats_cpu) {
/* allocation failed, try again after some time */
schedule_delayed_work(dwork, msecs_to_jiffies(10));
return;
}
block: Use u64_stats_init() to initialize seqcounts Now that seqcounts are lockdep enabled objects, we need to explicitly initialize runtime allocated seqcounts so that lockdep can track them. Without this patch, Fengguang was seeing: [ 4.127282] INFO: trying to register non-static key. [ 4.128027] the code is fine but needs lockdep annotation. [ 4.128027] turning off the locking correctness validator. [ 4.128027] CPU: 0 PID: 96 Comm: kworker/u4:1 Not tainted 3.12.0-next-20131108-10601-gbad570d #2 [ 4.128027] Hardware name: Bochs Bochs, BIOS Bochs 01/01/2011 [ ... ] [ 4.128027] Call Trace: [ 4.128027] [<7908e744>] ? console_unlock+0x353/0x380 [ 4.128027] [<79dc7cf2>] dump_stack+0x48/0x60 [ 4.128027] [<7908953e>] __lock_acquire.isra.26+0x7e3/0xceb [ 4.128027] [<7908a1c5>] lock_acquire+0x71/0x9a [ 4.128027] [<794079aa>] ? blk_throtl_bio+0x1c3/0x485 [ 4.128027] [<7940658b>] throtl_update_dispatch_stats+0x7c/0x153 [ 4.128027] [<794079aa>] ? blk_throtl_bio+0x1c3/0x485 [ 4.128027] [<794079aa>] blk_throtl_bio+0x1c3/0x485 ... Use u64_stats_init() for all affected data structures, which initializes the seqcount. Reported-and-Tested-by: Fengguang Wu <fengguang.wu@intel.com> Cc: Vivek Goyal <vgoyal@redhat.com> Cc: Jens Axboe <axboe@kernel.dk> Signed-off-by: Peter Zijlstra <peterz@infradead.org> [ Folded in another fix from the mailing list as well as a fix to that fix. Tweaked commit message. ] Signed-off-by: John Stultz <john.stultz@linaro.org> Signed-off-by: Peter Zijlstra <peterz@infradead.org> Link: http://lkml.kernel.org/r/1384314134-6895-1-git-send-email-john.stultz@linaro.org [ So I actually think that the two SOBs from PeterZ are the right depiction of the patch route. ] Signed-off-by: Ingo Molnar <mingo@kernel.org>
2013-11-13 04:42:14 +01:00
for_each_possible_cpu(cpu)
tg_stats_init(per_cpu_ptr(stats_cpu, cpu));
}
spin_lock_irq(&tg_stats_alloc_lock);
if (!list_empty(&tg_stats_alloc_list)) {
struct throtl_grp *tg = list_first_entry(&tg_stats_alloc_list,
struct throtl_grp,
stats_alloc_node);
swap(tg->stats_cpu, stats_cpu);
list_del_init(&tg->stats_alloc_node);
}
empty = list_empty(&tg_stats_alloc_list);
spin_unlock_irq(&tg_stats_alloc_lock);
if (!empty)
goto alloc_stats;
}
blk-throttle: add throtl_qnode for dispatch fairness With flat hierarchy, there's only single level of dispatching happening and fairness beyond that point is the responsibility of the rest of the block layer and driver, which usually works out okay; however, with the planned hierarchy support, service_queue->bio_lists[] can be filled up by bios from a single source. While the limits would still be honored, it'd be very easy to starve IOs from siblings or children. To avoid such starvation, this patch implements throtl_qnode and converts service_queue->bio_lists[] to lists of per-source qnodes which in turn contains the bio's. For example, when a bio is dispatched from a child group, the bio doesn't get queued on ->bio_lists[] directly but it first gets queued on the group's qnode which in turn gets queued on service_queue->queued[]. When dispatching for the upper level, the ->queued[] list is consumed in round-robing order so that the dispatch windows is consumed fairly by all IO sources. There are two ways a bio can come to a throtl_grp - directly queued to the group or dispatched from a child. For the former throtl_grp->qnode_on_self[rw] is used. For the latter, the child's ->qnode_on_parent[rw]. Note that this means that the child which is contributing a bio to its parent should stay pinned until all its bios are dispatched to its grand-parent. This patch moves blkg refcnting from bio add/remove spots to qnode activation/deactivation so that the blkg containing an active qnode is always pinned. As child pins the parent, this is sufficient for keeping the relevant sub-tree pinned while bios are in flight. The starvation issue was spotted by Vivek Goyal. v2: The original patch used the same throtl_grp->qnode_on_self/parent for reads and writes causing RWs to be queued incorrectly if there already are outstanding IOs in the other direction. They should be throtl_grp->qnode_on_self/parent[2] so that READs and WRITEs can use different qnodes. Spotted by Vivek Goyal. Signed-off-by: Tejun Heo <tj@kernel.org> Acked-by: Vivek Goyal <vgoyal@redhat.com>
2013-05-14 22:52:38 +02:00
static void throtl_qnode_init(struct throtl_qnode *qn, struct throtl_grp *tg)
{
INIT_LIST_HEAD(&qn->node);
bio_list_init(&qn->bios);
qn->tg = tg;
}
/**
* throtl_qnode_add_bio - add a bio to a throtl_qnode and activate it
* @bio: bio being added
* @qn: qnode to add bio to
* @queued: the service_queue->queued[] list @qn belongs to
*
* Add @bio to @qn and put @qn on @queued if it's not already on.
* @qn->tg's reference count is bumped when @qn is activated. See the
* comment on top of throtl_qnode definition for details.
*/
static void throtl_qnode_add_bio(struct bio *bio, struct throtl_qnode *qn,
struct list_head *queued)
{
bio_list_add(&qn->bios, bio);
if (list_empty(&qn->node)) {
list_add_tail(&qn->node, queued);
blkg_get(tg_to_blkg(qn->tg));
}
}
/**
* throtl_peek_queued - peek the first bio on a qnode list
* @queued: the qnode list to peek
*/
static struct bio *throtl_peek_queued(struct list_head *queued)
{
struct throtl_qnode *qn = list_first_entry(queued, struct throtl_qnode, node);
struct bio *bio;
if (list_empty(queued))
return NULL;
bio = bio_list_peek(&qn->bios);
WARN_ON_ONCE(!bio);
return bio;
}
/**
* throtl_pop_queued - pop the first bio form a qnode list
* @queued: the qnode list to pop a bio from
* @tg_to_put: optional out argument for throtl_grp to put
*
* Pop the first bio from the qnode list @queued. After popping, the first
* qnode is removed from @queued if empty or moved to the end of @queued so
* that the popping order is round-robin.
*
* When the first qnode is removed, its associated throtl_grp should be put
* too. If @tg_to_put is NULL, this function automatically puts it;
* otherwise, *@tg_to_put is set to the throtl_grp to put and the caller is
* responsible for putting it.
*/
static struct bio *throtl_pop_queued(struct list_head *queued,
struct throtl_grp **tg_to_put)
{
struct throtl_qnode *qn = list_first_entry(queued, struct throtl_qnode, node);
struct bio *bio;
if (list_empty(queued))
return NULL;
bio = bio_list_pop(&qn->bios);
WARN_ON_ONCE(!bio);
if (bio_list_empty(&qn->bios)) {
list_del_init(&qn->node);
if (tg_to_put)
*tg_to_put = qn->tg;
else
blkg_put(tg_to_blkg(qn->tg));
} else {
list_move_tail(&qn->node, queued);
}
return bio;
}
/* init a service_queue, assumes the caller zeroed it */
static void throtl_service_queue_init(struct throtl_service_queue *sq,
struct throtl_service_queue *parent_sq)
{
blk-throttle: add throtl_qnode for dispatch fairness With flat hierarchy, there's only single level of dispatching happening and fairness beyond that point is the responsibility of the rest of the block layer and driver, which usually works out okay; however, with the planned hierarchy support, service_queue->bio_lists[] can be filled up by bios from a single source. While the limits would still be honored, it'd be very easy to starve IOs from siblings or children. To avoid such starvation, this patch implements throtl_qnode and converts service_queue->bio_lists[] to lists of per-source qnodes which in turn contains the bio's. For example, when a bio is dispatched from a child group, the bio doesn't get queued on ->bio_lists[] directly but it first gets queued on the group's qnode which in turn gets queued on service_queue->queued[]. When dispatching for the upper level, the ->queued[] list is consumed in round-robing order so that the dispatch windows is consumed fairly by all IO sources. There are two ways a bio can come to a throtl_grp - directly queued to the group or dispatched from a child. For the former throtl_grp->qnode_on_self[rw] is used. For the latter, the child's ->qnode_on_parent[rw]. Note that this means that the child which is contributing a bio to its parent should stay pinned until all its bios are dispatched to its grand-parent. This patch moves blkg refcnting from bio add/remove spots to qnode activation/deactivation so that the blkg containing an active qnode is always pinned. As child pins the parent, this is sufficient for keeping the relevant sub-tree pinned while bios are in flight. The starvation issue was spotted by Vivek Goyal. v2: The original patch used the same throtl_grp->qnode_on_self/parent for reads and writes causing RWs to be queued incorrectly if there already are outstanding IOs in the other direction. They should be throtl_grp->qnode_on_self/parent[2] so that READs and WRITEs can use different qnodes. Spotted by Vivek Goyal. Signed-off-by: Tejun Heo <tj@kernel.org> Acked-by: Vivek Goyal <vgoyal@redhat.com>
2013-05-14 22:52:38 +02:00
INIT_LIST_HEAD(&sq->queued[0]);
INIT_LIST_HEAD(&sq->queued[1]);
sq->pending_tree = RB_ROOT;
sq->parent_sq = parent_sq;
setup_timer(&sq->pending_timer, throtl_pending_timer_fn,
(unsigned long)sq);
}
static void throtl_service_queue_exit(struct throtl_service_queue *sq)
{
del_timer_sync(&sq->pending_timer);
}
static void throtl_pd_init(struct blkcg_gq *blkg)
{
struct throtl_grp *tg = blkg_to_tg(blkg);
struct throtl_data *td = blkg->q->td;
struct throtl_service_queue *parent_sq;
unsigned long flags;
blk-throttle: add throtl_qnode for dispatch fairness With flat hierarchy, there's only single level of dispatching happening and fairness beyond that point is the responsibility of the rest of the block layer and driver, which usually works out okay; however, with the planned hierarchy support, service_queue->bio_lists[] can be filled up by bios from a single source. While the limits would still be honored, it'd be very easy to starve IOs from siblings or children. To avoid such starvation, this patch implements throtl_qnode and converts service_queue->bio_lists[] to lists of per-source qnodes which in turn contains the bio's. For example, when a bio is dispatched from a child group, the bio doesn't get queued on ->bio_lists[] directly but it first gets queued on the group's qnode which in turn gets queued on service_queue->queued[]. When dispatching for the upper level, the ->queued[] list is consumed in round-robing order so that the dispatch windows is consumed fairly by all IO sources. There are two ways a bio can come to a throtl_grp - directly queued to the group or dispatched from a child. For the former throtl_grp->qnode_on_self[rw] is used. For the latter, the child's ->qnode_on_parent[rw]. Note that this means that the child which is contributing a bio to its parent should stay pinned until all its bios are dispatched to its grand-parent. This patch moves blkg refcnting from bio add/remove spots to qnode activation/deactivation so that the blkg containing an active qnode is always pinned. As child pins the parent, this is sufficient for keeping the relevant sub-tree pinned while bios are in flight. The starvation issue was spotted by Vivek Goyal. v2: The original patch used the same throtl_grp->qnode_on_self/parent for reads and writes causing RWs to be queued incorrectly if there already are outstanding IOs in the other direction. They should be throtl_grp->qnode_on_self/parent[2] so that READs and WRITEs can use different qnodes. Spotted by Vivek Goyal. Signed-off-by: Tejun Heo <tj@kernel.org> Acked-by: Vivek Goyal <vgoyal@redhat.com>
2013-05-14 22:52:38 +02:00
int rw;
blkcg: factor out blkio_group creation Currently both blk-throttle and cfq-iosched implement their own blkio_group creation code in throtl_get_tg() and cfq_get_cfqg(). This patch factors out the common code into blkg_lookup_create(), which returns ERR_PTR value so that transitional failures due to queue bypass can be distinguished from other failures. * New plkio_policy_ops methods blkio_alloc_group_fn() and blkio_link_group_fn added. Both are transitional and will be removed once the blkg management code is fully moved into blk-cgroup.c. * blkio_alloc_group_fn() allocates policy-specific blkg which is usually a larger data structure with blkg as the first entry and intiailizes it. Note that initialization of blkg proper, including percpu stats, is responsibility of blk-cgroup proper. Note that default config (weight, bps...) initialization is done from this method; otherwise, we end up violating locking order between blkcg and q locks via blkcg_get_CONF() functions. * blkio_link_group_fn() is called under queue_lock and responsible for linking the blkg to the queue. blkcg side is handled by blk-cgroup proper. * The common blkg creation function is named blkg_lookup_create() and blkiocg_lookup_group() is renamed to blkg_lookup() for consistency. Also, throtl / cfq related functions are similarly [re]named for consistency. This simplifies blkcg policy implementations and enables further cleanup. -v2: Vivek noticed that blkg_lookup_create() incorrectly tested blk_queue_dead() instead of blk_queue_bypass() leading a user of the function ending up creating a new blkg on bypassing queue. This is a bug introduced while relocating bypass patches before this one. Fixed. -v3: ERR_PTR patch folded into this one. @for_root added to blkg_lookup_create() to allow creating root group on a bypassed queue during elevator switch. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Vivek Goyal <vgoyal@redhat.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2012-03-05 22:15:06 +01:00
/*
* If sane_hierarchy is enabled, we switch to properly hierarchical
* behavior where limits on a given throtl_grp are applied to the
* whole subtree rather than just the group itself. e.g. If 16M
* read_bps limit is set on the root group, the whole system can't
* exceed 16M for the device.
*
* If sane_hierarchy is not enabled, the broken flat hierarchy
* behavior is retained where all throtl_grps are treated as if
* they're all separate root groups right below throtl_data.
* Limits of a group don't interact with limits of other groups
* regardless of the position of the group in the hierarchy.
*/
parent_sq = &td->service_queue;
if (cgroup_sane_behavior(blkg->blkcg->css.cgroup) && blkg->parent)
parent_sq = &blkg_to_tg(blkg->parent)->service_queue;
throtl_service_queue_init(&tg->service_queue, parent_sq);
blk-throttle: add throtl_qnode for dispatch fairness With flat hierarchy, there's only single level of dispatching happening and fairness beyond that point is the responsibility of the rest of the block layer and driver, which usually works out okay; however, with the planned hierarchy support, service_queue->bio_lists[] can be filled up by bios from a single source. While the limits would still be honored, it'd be very easy to starve IOs from siblings or children. To avoid such starvation, this patch implements throtl_qnode and converts service_queue->bio_lists[] to lists of per-source qnodes which in turn contains the bio's. For example, when a bio is dispatched from a child group, the bio doesn't get queued on ->bio_lists[] directly but it first gets queued on the group's qnode which in turn gets queued on service_queue->queued[]. When dispatching for the upper level, the ->queued[] list is consumed in round-robing order so that the dispatch windows is consumed fairly by all IO sources. There are two ways a bio can come to a throtl_grp - directly queued to the group or dispatched from a child. For the former throtl_grp->qnode_on_self[rw] is used. For the latter, the child's ->qnode_on_parent[rw]. Note that this means that the child which is contributing a bio to its parent should stay pinned until all its bios are dispatched to its grand-parent. This patch moves blkg refcnting from bio add/remove spots to qnode activation/deactivation so that the blkg containing an active qnode is always pinned. As child pins the parent, this is sufficient for keeping the relevant sub-tree pinned while bios are in flight. The starvation issue was spotted by Vivek Goyal. v2: The original patch used the same throtl_grp->qnode_on_self/parent for reads and writes causing RWs to be queued incorrectly if there already are outstanding IOs in the other direction. They should be throtl_grp->qnode_on_self/parent[2] so that READs and WRITEs can use different qnodes. Spotted by Vivek Goyal. Signed-off-by: Tejun Heo <tj@kernel.org> Acked-by: Vivek Goyal <vgoyal@redhat.com>
2013-05-14 22:52:38 +02:00
for (rw = READ; rw <= WRITE; rw++) {
throtl_qnode_init(&tg->qnode_on_self[rw], tg);
throtl_qnode_init(&tg->qnode_on_parent[rw], tg);
}
RB_CLEAR_NODE(&tg->rb_node);
tg->td = td;
blkcg: don't allow or retain configuration of missing devices blkcg is very peculiar in that it allows setting and remembering configurations for non-existent devices by maintaining separate data structures for configuration. This behavior is completely out of the usual norms and outright confusing; furthermore, it uses dev_t number to match the configuration to devices, which is unpredictable to begin with and becomes completely unuseable if EXT_DEVT is fully used. It is wholely unnecessary - we already have fully functional userland mechanism to program devices being hotplugged which has full access to device identification, connection topology and filesystem information. Add a new struct blkio_group_conf which contains all blkcg configurations to blkio_group and let blkio_group, which can be created iff the associated device exists and is removed when the associated device goes away, carry all configurations. Note that, after this patch, all newly created blkg's will always have the default configuration (unlimited for throttling and blkcg's weight for propio). This patch makes blkio_policy_node meaningless but doesn't remove it. The next patch will. -v2: Updated to retry after short sleep if blkg lookup/creation failed due to the queue being temporarily bypassed as indicated by -EBUSY return. Pointed out by Vivek. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Vivek Goyal <vgoyal@redhat.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2012-03-05 22:15:07 +01:00
tg->bps[READ] = -1;
tg->bps[WRITE] = -1;
tg->iops[READ] = -1;
tg->iops[WRITE] = -1;
/*
* Ugh... We need to perform per-cpu allocation for tg->stats_cpu
* but percpu allocator can't be called from IO path. Queue tg on
* tg_stats_alloc_list and allocate from work item.
*/
spin_lock_irqsave(&tg_stats_alloc_lock, flags);
list_add(&tg->stats_alloc_node, &tg_stats_alloc_list);
schedule_delayed_work(&tg_stats_alloc_work, 0);
spin_unlock_irqrestore(&tg_stats_alloc_lock, flags);
}
/*
* Set has_rules[] if @tg or any of its parents have limits configured.
* This doesn't require walking up to the top of the hierarchy as the
* parent's has_rules[] is guaranteed to be correct.
*/
static void tg_update_has_rules(struct throtl_grp *tg)
{
struct throtl_grp *parent_tg = sq_to_tg(tg->service_queue.parent_sq);
int rw;
for (rw = READ; rw <= WRITE; rw++)
tg->has_rules[rw] = (parent_tg && parent_tg->has_rules[rw]) ||
(tg->bps[rw] != -1 || tg->iops[rw] != -1);
}
static void throtl_pd_online(struct blkcg_gq *blkg)
{
/*
* We don't want new groups to escape the limits of its ancestors.
* Update has_rules[] after a new group is brought online.
*/
tg_update_has_rules(blkg_to_tg(blkg));
}
static void throtl_pd_exit(struct blkcg_gq *blkg)
{
struct throtl_grp *tg = blkg_to_tg(blkg);
unsigned long flags;
spin_lock_irqsave(&tg_stats_alloc_lock, flags);
list_del_init(&tg->stats_alloc_node);
spin_unlock_irqrestore(&tg_stats_alloc_lock, flags);
free_percpu(tg->stats_cpu);
throtl_service_queue_exit(&tg->service_queue);
}
static void throtl_pd_reset_stats(struct blkcg_gq *blkg)
{
struct throtl_grp *tg = blkg_to_tg(blkg);
int cpu;
if (tg->stats_cpu == NULL)
return;
for_each_possible_cpu(cpu) {
struct tg_stats_cpu *sc = per_cpu_ptr(tg->stats_cpu, cpu);
blkg_rwstat_reset(&sc->service_bytes);
blkg_rwstat_reset(&sc->serviced);
}
}
static struct throtl_grp *throtl_lookup_tg(struct throtl_data *td,
struct blkcg *blkcg)
{
/*
* This is the common case when there are no blkcgs. Avoid lookup
* in this case
blkcg: factor out blkio_group creation Currently both blk-throttle and cfq-iosched implement their own blkio_group creation code in throtl_get_tg() and cfq_get_cfqg(). This patch factors out the common code into blkg_lookup_create(), which returns ERR_PTR value so that transitional failures due to queue bypass can be distinguished from other failures. * New plkio_policy_ops methods blkio_alloc_group_fn() and blkio_link_group_fn added. Both are transitional and will be removed once the blkg management code is fully moved into blk-cgroup.c. * blkio_alloc_group_fn() allocates policy-specific blkg which is usually a larger data structure with blkg as the first entry and intiailizes it. Note that initialization of blkg proper, including percpu stats, is responsibility of blk-cgroup proper. Note that default config (weight, bps...) initialization is done from this method; otherwise, we end up violating locking order between blkcg and q locks via blkcg_get_CONF() functions. * blkio_link_group_fn() is called under queue_lock and responsible for linking the blkg to the queue. blkcg side is handled by blk-cgroup proper. * The common blkg creation function is named blkg_lookup_create() and blkiocg_lookup_group() is renamed to blkg_lookup() for consistency. Also, throtl / cfq related functions are similarly [re]named for consistency. This simplifies blkcg policy implementations and enables further cleanup. -v2: Vivek noticed that blkg_lookup_create() incorrectly tested blk_queue_dead() instead of blk_queue_bypass() leading a user of the function ending up creating a new blkg on bypassing queue. This is a bug introduced while relocating bypass patches before this one. Fixed. -v3: ERR_PTR patch folded into this one. @for_root added to blkg_lookup_create() to allow creating root group on a bypassed queue during elevator switch. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Vivek Goyal <vgoyal@redhat.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2012-03-05 22:15:06 +01:00
*/
if (blkcg == &blkcg_root)
return td_root_tg(td);
blkcg: unify blkg's for blkcg policies Currently, blkg is per cgroup-queue-policy combination. This is unnatural and leads to various convolutions in partially used duplicate fields in blkg, config / stat access, and general management of blkgs. This patch make blkg's per cgroup-queue and let them serve all policies. blkgs are now created and destroyed by blkcg core proper. This will allow further consolidation of common management logic into blkcg core and API with better defined semantics and layering. As a transitional step to untangle blkg management, elvswitch and policy [de]registration, all blkgs except the root blkg are being shot down during elvswitch and bypass. This patch adds blkg_root_update() to update root blkg in place on policy change. This is hacky and racy but should be good enough as interim step until we get locking simplified and switch over to proper in-place update for all blkgs. -v2: Root blkgs need to be updated on elvswitch too and blkg_alloc() comment wasn't updated according to the function change. Fixed. Both pointed out by Vivek. -v3: v2 updated blkg_destroy_all() to invoke update_root_blkg_pd() for all policies. This freed root pd during elvswitch before the last queue finished exiting and led to oops. Directly invoke update_root_blkg_pd() only on BLKIO_POLICY_PROP from cfq_exit_queue(). This also is closer to what will be done with proper in-place blkg update. Reported by Vivek. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Vivek Goyal <vgoyal@redhat.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2012-03-05 22:15:20 +01:00
return blkg_to_tg(blkg_lookup(blkcg, td->queue));
}
blkcg: factor out blkio_group creation Currently both blk-throttle and cfq-iosched implement their own blkio_group creation code in throtl_get_tg() and cfq_get_cfqg(). This patch factors out the common code into blkg_lookup_create(), which returns ERR_PTR value so that transitional failures due to queue bypass can be distinguished from other failures. * New plkio_policy_ops methods blkio_alloc_group_fn() and blkio_link_group_fn added. Both are transitional and will be removed once the blkg management code is fully moved into blk-cgroup.c. * blkio_alloc_group_fn() allocates policy-specific blkg which is usually a larger data structure with blkg as the first entry and intiailizes it. Note that initialization of blkg proper, including percpu stats, is responsibility of blk-cgroup proper. Note that default config (weight, bps...) initialization is done from this method; otherwise, we end up violating locking order between blkcg and q locks via blkcg_get_CONF() functions. * blkio_link_group_fn() is called under queue_lock and responsible for linking the blkg to the queue. blkcg side is handled by blk-cgroup proper. * The common blkg creation function is named blkg_lookup_create() and blkiocg_lookup_group() is renamed to blkg_lookup() for consistency. Also, throtl / cfq related functions are similarly [re]named for consistency. This simplifies blkcg policy implementations and enables further cleanup. -v2: Vivek noticed that blkg_lookup_create() incorrectly tested blk_queue_dead() instead of blk_queue_bypass() leading a user of the function ending up creating a new blkg on bypassing queue. This is a bug introduced while relocating bypass patches before this one. Fixed. -v3: ERR_PTR patch folded into this one. @for_root added to blkg_lookup_create() to allow creating root group on a bypassed queue during elevator switch. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Vivek Goyal <vgoyal@redhat.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2012-03-05 22:15:06 +01:00
static struct throtl_grp *throtl_lookup_create_tg(struct throtl_data *td,
struct blkcg *blkcg)
{
struct request_queue *q = td->queue;
blkcg: factor out blkio_group creation Currently both blk-throttle and cfq-iosched implement their own blkio_group creation code in throtl_get_tg() and cfq_get_cfqg(). This patch factors out the common code into blkg_lookup_create(), which returns ERR_PTR value so that transitional failures due to queue bypass can be distinguished from other failures. * New plkio_policy_ops methods blkio_alloc_group_fn() and blkio_link_group_fn added. Both are transitional and will be removed once the blkg management code is fully moved into blk-cgroup.c. * blkio_alloc_group_fn() allocates policy-specific blkg which is usually a larger data structure with blkg as the first entry and intiailizes it. Note that initialization of blkg proper, including percpu stats, is responsibility of blk-cgroup proper. Note that default config (weight, bps...) initialization is done from this method; otherwise, we end up violating locking order between blkcg and q locks via blkcg_get_CONF() functions. * blkio_link_group_fn() is called under queue_lock and responsible for linking the blkg to the queue. blkcg side is handled by blk-cgroup proper. * The common blkg creation function is named blkg_lookup_create() and blkiocg_lookup_group() is renamed to blkg_lookup() for consistency. Also, throtl / cfq related functions are similarly [re]named for consistency. This simplifies blkcg policy implementations and enables further cleanup. -v2: Vivek noticed that blkg_lookup_create() incorrectly tested blk_queue_dead() instead of blk_queue_bypass() leading a user of the function ending up creating a new blkg on bypassing queue. This is a bug introduced while relocating bypass patches before this one. Fixed. -v3: ERR_PTR patch folded into this one. @for_root added to blkg_lookup_create() to allow creating root group on a bypassed queue during elevator switch. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Vivek Goyal <vgoyal@redhat.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2012-03-05 22:15:06 +01:00
struct throtl_grp *tg = NULL;
/*
* This is the common case when there are no blkcgs. Avoid lookup
* in this case
*/
if (blkcg == &blkcg_root) {
tg = td_root_tg(td);
blkcg: factor out blkio_group creation Currently both blk-throttle and cfq-iosched implement their own blkio_group creation code in throtl_get_tg() and cfq_get_cfqg(). This patch factors out the common code into blkg_lookup_create(), which returns ERR_PTR value so that transitional failures due to queue bypass can be distinguished from other failures. * New plkio_policy_ops methods blkio_alloc_group_fn() and blkio_link_group_fn added. Both are transitional and will be removed once the blkg management code is fully moved into blk-cgroup.c. * blkio_alloc_group_fn() allocates policy-specific blkg which is usually a larger data structure with blkg as the first entry and intiailizes it. Note that initialization of blkg proper, including percpu stats, is responsibility of blk-cgroup proper. Note that default config (weight, bps...) initialization is done from this method; otherwise, we end up violating locking order between blkcg and q locks via blkcg_get_CONF() functions. * blkio_link_group_fn() is called under queue_lock and responsible for linking the blkg to the queue. blkcg side is handled by blk-cgroup proper. * The common blkg creation function is named blkg_lookup_create() and blkiocg_lookup_group() is renamed to blkg_lookup() for consistency. Also, throtl / cfq related functions are similarly [re]named for consistency. This simplifies blkcg policy implementations and enables further cleanup. -v2: Vivek noticed that blkg_lookup_create() incorrectly tested blk_queue_dead() instead of blk_queue_bypass() leading a user of the function ending up creating a new blkg on bypassing queue. This is a bug introduced while relocating bypass patches before this one. Fixed. -v3: ERR_PTR patch folded into this one. @for_root added to blkg_lookup_create() to allow creating root group on a bypassed queue during elevator switch. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Vivek Goyal <vgoyal@redhat.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2012-03-05 22:15:06 +01:00
} else {
struct blkcg_gq *blkg;
blkg = blkg_lookup_create(blkcg, q);
blkcg: factor out blkio_group creation Currently both blk-throttle and cfq-iosched implement their own blkio_group creation code in throtl_get_tg() and cfq_get_cfqg(). This patch factors out the common code into blkg_lookup_create(), which returns ERR_PTR value so that transitional failures due to queue bypass can be distinguished from other failures. * New plkio_policy_ops methods blkio_alloc_group_fn() and blkio_link_group_fn added. Both are transitional and will be removed once the blkg management code is fully moved into blk-cgroup.c. * blkio_alloc_group_fn() allocates policy-specific blkg which is usually a larger data structure with blkg as the first entry and intiailizes it. Note that initialization of blkg proper, including percpu stats, is responsibility of blk-cgroup proper. Note that default config (weight, bps...) initialization is done from this method; otherwise, we end up violating locking order between blkcg and q locks via blkcg_get_CONF() functions. * blkio_link_group_fn() is called under queue_lock and responsible for linking the blkg to the queue. blkcg side is handled by blk-cgroup proper. * The common blkg creation function is named blkg_lookup_create() and blkiocg_lookup_group() is renamed to blkg_lookup() for consistency. Also, throtl / cfq related functions are similarly [re]named for consistency. This simplifies blkcg policy implementations and enables further cleanup. -v2: Vivek noticed that blkg_lookup_create() incorrectly tested blk_queue_dead() instead of blk_queue_bypass() leading a user of the function ending up creating a new blkg on bypassing queue. This is a bug introduced while relocating bypass patches before this one. Fixed. -v3: ERR_PTR patch folded into this one. @for_root added to blkg_lookup_create() to allow creating root group on a bypassed queue during elevator switch. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Vivek Goyal <vgoyal@redhat.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2012-03-05 22:15:06 +01:00
/* if %NULL and @q is alive, fall back to root_tg */
if (!IS_ERR(blkg))
tg = blkg_to_tg(blkg);
else if (!blk_queue_dying(q))
tg = td_root_tg(td);
}
return tg;
}
static struct throtl_grp *
throtl_rb_first(struct throtl_service_queue *parent_sq)
{
/* Service tree is empty */
if (!parent_sq->nr_pending)
return NULL;
if (!parent_sq->first_pending)
parent_sq->first_pending = rb_first(&parent_sq->pending_tree);
if (parent_sq->first_pending)
return rb_entry_tg(parent_sq->first_pending);
return NULL;
}
static void rb_erase_init(struct rb_node *n, struct rb_root *root)
{
rb_erase(n, root);
RB_CLEAR_NODE(n);
}
static void throtl_rb_erase(struct rb_node *n,
struct throtl_service_queue *parent_sq)
{
if (parent_sq->first_pending == n)
parent_sq->first_pending = NULL;
rb_erase_init(n, &parent_sq->pending_tree);
--parent_sq->nr_pending;
}
static void update_min_dispatch_time(struct throtl_service_queue *parent_sq)
{
struct throtl_grp *tg;
tg = throtl_rb_first(parent_sq);
if (!tg)
return;
parent_sq->first_pending_disptime = tg->disptime;
}
static void tg_service_queue_add(struct throtl_grp *tg)
{
struct throtl_service_queue *parent_sq = tg->service_queue.parent_sq;
struct rb_node **node = &parent_sq->pending_tree.rb_node;
struct rb_node *parent = NULL;
struct throtl_grp *__tg;
unsigned long key = tg->disptime;
int left = 1;
while (*node != NULL) {
parent = *node;
__tg = rb_entry_tg(parent);
if (time_before(key, __tg->disptime))
node = &parent->rb_left;
else {
node = &parent->rb_right;
left = 0;
}
}
if (left)
parent_sq->first_pending = &tg->rb_node;
rb_link_node(&tg->rb_node, parent, node);
rb_insert_color(&tg->rb_node, &parent_sq->pending_tree);
}
static void __throtl_enqueue_tg(struct throtl_grp *tg)
{
tg_service_queue_add(tg);
tg->flags |= THROTL_TG_PENDING;
tg->service_queue.parent_sq->nr_pending++;
}
static void throtl_enqueue_tg(struct throtl_grp *tg)
{
if (!(tg->flags & THROTL_TG_PENDING))
__throtl_enqueue_tg(tg);
}
static void __throtl_dequeue_tg(struct throtl_grp *tg)
{
throtl_rb_erase(&tg->rb_node, tg->service_queue.parent_sq);
tg->flags &= ~THROTL_TG_PENDING;
}
static void throtl_dequeue_tg(struct throtl_grp *tg)
{
if (tg->flags & THROTL_TG_PENDING)
__throtl_dequeue_tg(tg);
}
/* Call with queue lock held */
static void throtl_schedule_pending_timer(struct throtl_service_queue *sq,
unsigned long expires)
{
mod_timer(&sq->pending_timer, expires);
throtl_log(sq, "schedule timer. delay=%lu jiffies=%lu",
expires - jiffies, jiffies);
}
/**
* throtl_schedule_next_dispatch - schedule the next dispatch cycle
* @sq: the service_queue to schedule dispatch for
* @force: force scheduling
*
* Arm @sq->pending_timer so that the next dispatch cycle starts on the
* dispatch time of the first pending child. Returns %true if either timer
* is armed or there's no pending child left. %false if the current
* dispatch window is still open and the caller should continue
* dispatching.
*
* If @force is %true, the dispatch timer is always scheduled and this
* function is guaranteed to return %true. This is to be used when the
* caller can't dispatch itself and needs to invoke pending_timer
* unconditionally. Note that forced scheduling is likely to induce short
* delay before dispatch starts even if @sq->first_pending_disptime is not
* in the future and thus shouldn't be used in hot paths.
*/
static bool throtl_schedule_next_dispatch(struct throtl_service_queue *sq,
bool force)
{
/* any pending children left? */
if (!sq->nr_pending)
return true;
update_min_dispatch_time(sq);
/* is the next dispatch time in the future? */
if (force || time_after(sq->first_pending_disptime, jiffies)) {
throtl_schedule_pending_timer(sq, sq->first_pending_disptime);
return true;
}
/* tell the caller to continue dispatching */
return false;
}
blk-throttle: Account for child group's start time in parent while bio climbs up With the planned proper hierarchy support, a bio will climb up the tree before actually being dispatched. This makes sure bio is also subjected to parent's throttling limits, if any. It might happen that parent is idle and when bio is transferred to parent, a new slice starts fresh. But that is incorrect as parents wait time should have started when bio was queued in child group and causes IOs to be throttled more than configured as they climb the hierarchy. Given the fact that we have not written hierarchical algorithm in a way where child's and parents time slices are synchronized, we transfer the child's start time to parent if parent was idling. If parent was busy doing dispatch of other bios all this while, this is not an issue. Child's slice start time is passed to parent. Parent looks at its last expired slice start time. If child's start time is after parents old start time, that means parent had been idle and after parent went idle, child had an IO queued. So use child's start time as parent start time. If parent's start time is after child's start time, that means, when IO got queued in child group, parent was not idle. But later it dispatched some IO, its slice got trimmed and then it went idle. After a while child's request got shifted in parent group. In this case use parent's old start time as new start time as that's the duration of slice we did not use. This logic is far from perfect as if there are multiple childs then first child transferring the bio decides the start time while a bio might have queued up even earlier in other child, which is yet to be transferred up to parent. In that case we will lose time and bandwidth in parent. This patch is just an approximation to make situation somewhat better. Signed-off-by: Vivek Goyal <vgoyal@redhat.com> Signed-off-by: Tejun Heo <tj@kernel.org>
2013-05-14 22:52:38 +02:00
static inline void throtl_start_new_slice_with_credit(struct throtl_grp *tg,
bool rw, unsigned long start)
{
tg->bytes_disp[rw] = 0;
tg->io_disp[rw] = 0;
/*
* Previous slice has expired. We must have trimmed it after last
* bio dispatch. That means since start of last slice, we never used
* that bandwidth. Do try to make use of that bandwidth while giving
* credit.
*/
if (time_after_eq(start, tg->slice_start[rw]))
tg->slice_start[rw] = start;
tg->slice_end[rw] = jiffies + throtl_slice;
throtl_log(&tg->service_queue,
"[%c] new slice with credit start=%lu end=%lu jiffies=%lu",
rw == READ ? 'R' : 'W', tg->slice_start[rw],
tg->slice_end[rw], jiffies);
}
static inline void throtl_start_new_slice(struct throtl_grp *tg, bool rw)
{
tg->bytes_disp[rw] = 0;
tg->io_disp[rw] = 0;
tg->slice_start[rw] = jiffies;
tg->slice_end[rw] = jiffies + throtl_slice;
throtl_log(&tg->service_queue,
"[%c] new slice start=%lu end=%lu jiffies=%lu",
rw == READ ? 'R' : 'W', tg->slice_start[rw],
tg->slice_end[rw], jiffies);
}
static inline void throtl_set_slice_end(struct throtl_grp *tg, bool rw,
unsigned long jiffy_end)
{
tg->slice_end[rw] = roundup(jiffy_end, throtl_slice);
}
static inline void throtl_extend_slice(struct throtl_grp *tg, bool rw,
unsigned long jiffy_end)
{
tg->slice_end[rw] = roundup(jiffy_end, throtl_slice);
throtl_log(&tg->service_queue,
"[%c] extend slice start=%lu end=%lu jiffies=%lu",
rw == READ ? 'R' : 'W', tg->slice_start[rw],
tg->slice_end[rw], jiffies);
}
/* Determine if previously allocated or extended slice is complete or not */
static bool throtl_slice_used(struct throtl_grp *tg, bool rw)
{
if (time_in_range(jiffies, tg->slice_start[rw], tg->slice_end[rw]))
return false;
return 1;
}
/* Trim the used slices and adjust slice start accordingly */
static inline void throtl_trim_slice(struct throtl_grp *tg, bool rw)
{
unsigned long nr_slices, time_elapsed, io_trim;
u64 bytes_trim, tmp;
BUG_ON(time_before(tg->slice_end[rw], tg->slice_start[rw]));
/*
* If bps are unlimited (-1), then time slice don't get
* renewed. Don't try to trim the slice if slice is used. A new
* slice will start when appropriate.
*/
if (throtl_slice_used(tg, rw))
return;
/*
* A bio has been dispatched. Also adjust slice_end. It might happen
* that initially cgroup limit was very low resulting in high
* slice_end, but later limit was bumped up and bio was dispached
* sooner, then we need to reduce slice_end. A high bogus slice_end
* is bad because it does not allow new slice to start.
*/
throtl_set_slice_end(tg, rw, jiffies + throtl_slice);
time_elapsed = jiffies - tg->slice_start[rw];
nr_slices = time_elapsed / throtl_slice;
if (!nr_slices)
return;
tmp = tg->bps[rw] * throtl_slice * nr_slices;
do_div(tmp, HZ);
bytes_trim = tmp;
io_trim = (tg->iops[rw] * throtl_slice * nr_slices)/HZ;
if (!bytes_trim && !io_trim)
return;
if (tg->bytes_disp[rw] >= bytes_trim)
tg->bytes_disp[rw] -= bytes_trim;
else
tg->bytes_disp[rw] = 0;
if (tg->io_disp[rw] >= io_trim)
tg->io_disp[rw] -= io_trim;
else
tg->io_disp[rw] = 0;
tg->slice_start[rw] += nr_slices * throtl_slice;
throtl_log(&tg->service_queue,
"[%c] trim slice nr=%lu bytes=%llu io=%lu start=%lu end=%lu jiffies=%lu",
rw == READ ? 'R' : 'W', nr_slices, bytes_trim, io_trim,
tg->slice_start[rw], tg->slice_end[rw], jiffies);
}
static bool tg_with_in_iops_limit(struct throtl_grp *tg, struct bio *bio,
unsigned long *wait)
{
bool rw = bio_data_dir(bio);
unsigned int io_allowed;
unsigned long jiffy_elapsed, jiffy_wait, jiffy_elapsed_rnd;
u64 tmp;
jiffy_elapsed = jiffy_elapsed_rnd = jiffies - tg->slice_start[rw];
/* Slice has just started. Consider one slice interval */
if (!jiffy_elapsed)
jiffy_elapsed_rnd = throtl_slice;
jiffy_elapsed_rnd = roundup(jiffy_elapsed_rnd, throtl_slice);
/*
* jiffy_elapsed_rnd should not be a big value as minimum iops can be
* 1 then at max jiffy elapsed should be equivalent of 1 second as we
* will allow dispatch after 1 second and after that slice should
* have been trimmed.
*/
tmp = (u64)tg->iops[rw] * jiffy_elapsed_rnd;
do_div(tmp, HZ);
if (tmp > UINT_MAX)
io_allowed = UINT_MAX;
else
io_allowed = tmp;
if (tg->io_disp[rw] + 1 <= io_allowed) {
if (wait)
*wait = 0;
return true;
}
/* Calc approx time to dispatch */
jiffy_wait = ((tg->io_disp[rw] + 1) * HZ)/tg->iops[rw] + 1;
if (jiffy_wait > jiffy_elapsed)
jiffy_wait = jiffy_wait - jiffy_elapsed;
else
jiffy_wait = 1;
if (wait)
*wait = jiffy_wait;
return 0;
}
static bool tg_with_in_bps_limit(struct throtl_grp *tg, struct bio *bio,
unsigned long *wait)
{
bool rw = bio_data_dir(bio);
u64 bytes_allowed, extra_bytes, tmp;
unsigned long jiffy_elapsed, jiffy_wait, jiffy_elapsed_rnd;
jiffy_elapsed = jiffy_elapsed_rnd = jiffies - tg->slice_start[rw];
/* Slice has just started. Consider one slice interval */
if (!jiffy_elapsed)
jiffy_elapsed_rnd = throtl_slice;
jiffy_elapsed_rnd = roundup(jiffy_elapsed_rnd, throtl_slice);
tmp = tg->bps[rw] * jiffy_elapsed_rnd;
do_div(tmp, HZ);
bytes_allowed = tmp;
block: Abstract out bvec iterator Immutable biovecs are going to require an explicit iterator. To implement immutable bvecs, a later patch is going to add a bi_bvec_done member to this struct; for now, this patch effectively just renames things. Signed-off-by: Kent Overstreet <kmo@daterainc.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <paulus@samba.org> Cc: "Ed L. Cashin" <ecashin@coraid.com> Cc: Nick Piggin <npiggin@kernel.dk> Cc: Lars Ellenberg <drbd-dev@lists.linbit.com> Cc: Jiri Kosina <jkosina@suse.cz> Cc: Matthew Wilcox <willy@linux.intel.com> Cc: Geoff Levand <geoff@infradead.org> Cc: Yehuda Sadeh <yehuda@inktank.com> Cc: Sage Weil <sage@inktank.com> Cc: Alex Elder <elder@inktank.com> Cc: ceph-devel@vger.kernel.org Cc: Joshua Morris <josh.h.morris@us.ibm.com> Cc: Philip Kelleher <pjk1939@linux.vnet.ibm.com> Cc: Rusty Russell <rusty@rustcorp.com.au> Cc: "Michael S. Tsirkin" <mst@redhat.com> Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com> Cc: Jeremy Fitzhardinge <jeremy@goop.org> Cc: Neil Brown <neilb@suse.de> Cc: Alasdair Kergon <agk@redhat.com> Cc: Mike Snitzer <snitzer@redhat.com> Cc: dm-devel@redhat.com Cc: Martin Schwidefsky <schwidefsky@de.ibm.com> Cc: Heiko Carstens <heiko.carstens@de.ibm.com> Cc: linux390@de.ibm.com Cc: Boaz Harrosh <bharrosh@panasas.com> Cc: Benny Halevy <bhalevy@tonian.com> Cc: "James E.J. Bottomley" <JBottomley@parallels.com> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: "Nicholas A. Bellinger" <nab@linux-iscsi.org> Cc: Alexander Viro <viro@zeniv.linux.org.uk> Cc: Chris Mason <chris.mason@fusionio.com> Cc: "Theodore Ts'o" <tytso@mit.edu> Cc: Andreas Dilger <adilger.kernel@dilger.ca> Cc: Jaegeuk Kim <jaegeuk.kim@samsung.com> Cc: Steven Whitehouse <swhiteho@redhat.com> Cc: Dave Kleikamp <shaggy@kernel.org> Cc: Joern Engel <joern@logfs.org> Cc: Prasad Joshi <prasadjoshi.linux@gmail.com> Cc: Trond Myklebust <Trond.Myklebust@netapp.com> Cc: KONISHI Ryusuke <konishi.ryusuke@lab.ntt.co.jp> Cc: Mark Fasheh <mfasheh@suse.com> Cc: Joel Becker <jlbec@evilplan.org> Cc: Ben Myers <bpm@sgi.com> Cc: xfs@oss.sgi.com Cc: Steven Rostedt <rostedt@goodmis.org> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Len Brown <len.brown@intel.com> Cc: Pavel Machek <pavel@ucw.cz> Cc: "Rafael J. Wysocki" <rjw@sisk.pl> Cc: Herton Ronaldo Krzesinski <herton.krzesinski@canonical.com> Cc: Ben Hutchings <ben@decadent.org.uk> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Guo Chao <yan@linux.vnet.ibm.com> Cc: Tejun Heo <tj@kernel.org> Cc: Asai Thambi S P <asamymuthupa@micron.com> Cc: Selvan Mani <smani@micron.com> Cc: Sam Bradshaw <sbradshaw@micron.com> Cc: Wei Yongjun <yongjun_wei@trendmicro.com.cn> Cc: "Roger Pau Monné" <roger.pau@citrix.com> Cc: Jan Beulich <jbeulich@suse.com> Cc: Stefano Stabellini <stefano.stabellini@eu.citrix.com> Cc: Ian Campbell <Ian.Campbell@citrix.com> Cc: Sebastian Ott <sebott@linux.vnet.ibm.com> Cc: Christian Borntraeger <borntraeger@de.ibm.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Jiang Liu <jiang.liu@huawei.com> Cc: Nitin Gupta <ngupta@vflare.org> Cc: Jerome Marchand <jmarchand@redhat.com> Cc: Joe Perches <joe@perches.com> Cc: Peng Tao <tao.peng@emc.com> Cc: Andy Adamson <andros@netapp.com> Cc: fanchaoting <fanchaoting@cn.fujitsu.com> Cc: Jie Liu <jeff.liu@oracle.com> Cc: Sunil Mushran <sunil.mushran@gmail.com> Cc: "Martin K. Petersen" <martin.petersen@oracle.com> Cc: Namjae Jeon <namjae.jeon@samsung.com> Cc: Pankaj Kumar <pankaj.km@samsung.com> Cc: Dan Magenheimer <dan.magenheimer@oracle.com> Cc: Mel Gorman <mgorman@suse.de>6
2013-10-12 00:44:27 +02:00
if (tg->bytes_disp[rw] + bio->bi_iter.bi_size <= bytes_allowed) {
if (wait)
*wait = 0;
return true;
}
/* Calc approx time to dispatch */
block: Abstract out bvec iterator Immutable biovecs are going to require an explicit iterator. To implement immutable bvecs, a later patch is going to add a bi_bvec_done member to this struct; for now, this patch effectively just renames things. Signed-off-by: Kent Overstreet <kmo@daterainc.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <paulus@samba.org> Cc: "Ed L. Cashin" <ecashin@coraid.com> Cc: Nick Piggin <npiggin@kernel.dk> Cc: Lars Ellenberg <drbd-dev@lists.linbit.com> Cc: Jiri Kosina <jkosina@suse.cz> Cc: Matthew Wilcox <willy@linux.intel.com> Cc: Geoff Levand <geoff@infradead.org> Cc: Yehuda Sadeh <yehuda@inktank.com> Cc: Sage Weil <sage@inktank.com> Cc: Alex Elder <elder@inktank.com> Cc: ceph-devel@vger.kernel.org Cc: Joshua Morris <josh.h.morris@us.ibm.com> Cc: Philip Kelleher <pjk1939@linux.vnet.ibm.com> Cc: Rusty Russell <rusty@rustcorp.com.au> Cc: "Michael S. Tsirkin" <mst@redhat.com> Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com> Cc: Jeremy Fitzhardinge <jeremy@goop.org> Cc: Neil Brown <neilb@suse.de> Cc: Alasdair Kergon <agk@redhat.com> Cc: Mike Snitzer <snitzer@redhat.com> Cc: dm-devel@redhat.com Cc: Martin Schwidefsky <schwidefsky@de.ibm.com> Cc: Heiko Carstens <heiko.carstens@de.ibm.com> Cc: linux390@de.ibm.com Cc: Boaz Harrosh <bharrosh@panasas.com> Cc: Benny Halevy <bhalevy@tonian.com> Cc: "James E.J. Bottomley" <JBottomley@parallels.com> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: "Nicholas A. Bellinger" <nab@linux-iscsi.org> Cc: Alexander Viro <viro@zeniv.linux.org.uk> Cc: Chris Mason <chris.mason@fusionio.com> Cc: "Theodore Ts'o" <tytso@mit.edu> Cc: Andreas Dilger <adilger.kernel@dilger.ca> Cc: Jaegeuk Kim <jaegeuk.kim@samsung.com> Cc: Steven Whitehouse <swhiteho@redhat.com> Cc: Dave Kleikamp <shaggy@kernel.org> Cc: Joern Engel <joern@logfs.org> Cc: Prasad Joshi <prasadjoshi.linux@gmail.com> Cc: Trond Myklebust <Trond.Myklebust@netapp.com> Cc: KONISHI Ryusuke <konishi.ryusuke@lab.ntt.co.jp> Cc: Mark Fasheh <mfasheh@suse.com> Cc: Joel Becker <jlbec@evilplan.org> Cc: Ben Myers <bpm@sgi.com> Cc: xfs@oss.sgi.com Cc: Steven Rostedt <rostedt@goodmis.org> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Len Brown <len.brown@intel.com> Cc: Pavel Machek <pavel@ucw.cz> Cc: "Rafael J. Wysocki" <rjw@sisk.pl> Cc: Herton Ronaldo Krzesinski <herton.krzesinski@canonical.com> Cc: Ben Hutchings <ben@decadent.org.uk> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Guo Chao <yan@linux.vnet.ibm.com> Cc: Tejun Heo <tj@kernel.org> Cc: Asai Thambi S P <asamymuthupa@micron.com> Cc: Selvan Mani <smani@micron.com> Cc: Sam Bradshaw <sbradshaw@micron.com> Cc: Wei Yongjun <yongjun_wei@trendmicro.com.cn> Cc: "Roger Pau Monné" <roger.pau@citrix.com> Cc: Jan Beulich <jbeulich@suse.com> Cc: Stefano Stabellini <stefano.stabellini@eu.citrix.com> Cc: Ian Campbell <Ian.Campbell@citrix.com> Cc: Sebastian Ott <sebott@linux.vnet.ibm.com> Cc: Christian Borntraeger <borntraeger@de.ibm.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Jiang Liu <jiang.liu@huawei.com> Cc: Nitin Gupta <ngupta@vflare.org> Cc: Jerome Marchand <jmarchand@redhat.com> Cc: Joe Perches <joe@perches.com> Cc: Peng Tao <tao.peng@emc.com> Cc: Andy Adamson <andros@netapp.com> Cc: fanchaoting <fanchaoting@cn.fujitsu.com> Cc: Jie Liu <jeff.liu@oracle.com> Cc: Sunil Mushran <sunil.mushran@gmail.com> Cc: "Martin K. Petersen" <martin.petersen@oracle.com> Cc: Namjae Jeon <namjae.jeon@samsung.com> Cc: Pankaj Kumar <pankaj.km@samsung.com> Cc: Dan Magenheimer <dan.magenheimer@oracle.com> Cc: Mel Gorman <mgorman@suse.de>6
2013-10-12 00:44:27 +02:00
extra_bytes = tg->bytes_disp[rw] + bio->bi_iter.bi_size - bytes_allowed;
jiffy_wait = div64_u64(extra_bytes * HZ, tg->bps[rw]);
if (!jiffy_wait)
jiffy_wait = 1;
/*
* This wait time is without taking into consideration the rounding
* up we did. Add that time also.
*/
jiffy_wait = jiffy_wait + (jiffy_elapsed_rnd - jiffy_elapsed);
if (wait)
*wait = jiffy_wait;
return 0;
}
/*
* Returns whether one can dispatch a bio or not. Also returns approx number
* of jiffies to wait before this bio is with-in IO rate and can be dispatched
*/
static bool tg_may_dispatch(struct throtl_grp *tg, struct bio *bio,
unsigned long *wait)
{
bool rw = bio_data_dir(bio);
unsigned long bps_wait = 0, iops_wait = 0, max_wait = 0;
/*
* Currently whole state machine of group depends on first bio
* queued in the group bio list. So one should not be calling
* this function with a different bio if there are other bios
* queued.
*/
BUG_ON(tg->service_queue.nr_queued[rw] &&
blk-throttle: add throtl_qnode for dispatch fairness With flat hierarchy, there's only single level of dispatching happening and fairness beyond that point is the responsibility of the rest of the block layer and driver, which usually works out okay; however, with the planned hierarchy support, service_queue->bio_lists[] can be filled up by bios from a single source. While the limits would still be honored, it'd be very easy to starve IOs from siblings or children. To avoid such starvation, this patch implements throtl_qnode and converts service_queue->bio_lists[] to lists of per-source qnodes which in turn contains the bio's. For example, when a bio is dispatched from a child group, the bio doesn't get queued on ->bio_lists[] directly but it first gets queued on the group's qnode which in turn gets queued on service_queue->queued[]. When dispatching for the upper level, the ->queued[] list is consumed in round-robing order so that the dispatch windows is consumed fairly by all IO sources. There are two ways a bio can come to a throtl_grp - directly queued to the group or dispatched from a child. For the former throtl_grp->qnode_on_self[rw] is used. For the latter, the child's ->qnode_on_parent[rw]. Note that this means that the child which is contributing a bio to its parent should stay pinned until all its bios are dispatched to its grand-parent. This patch moves blkg refcnting from bio add/remove spots to qnode activation/deactivation so that the blkg containing an active qnode is always pinned. As child pins the parent, this is sufficient for keeping the relevant sub-tree pinned while bios are in flight. The starvation issue was spotted by Vivek Goyal. v2: The original patch used the same throtl_grp->qnode_on_self/parent for reads and writes causing RWs to be queued incorrectly if there already are outstanding IOs in the other direction. They should be throtl_grp->qnode_on_self/parent[2] so that READs and WRITEs can use different qnodes. Spotted by Vivek Goyal. Signed-off-by: Tejun Heo <tj@kernel.org> Acked-by: Vivek Goyal <vgoyal@redhat.com>
2013-05-14 22:52:38 +02:00
bio != throtl_peek_queued(&tg->service_queue.queued[rw]));
/* If tg->bps = -1, then BW is unlimited */
if (tg->bps[rw] == -1 && tg->iops[rw] == -1) {
if (wait)
*wait = 0;
return true;
}
/*
* If previous slice expired, start a new one otherwise renew/extend
* existing slice to make sure it is at least throtl_slice interval
* long since now.
*/
if (throtl_slice_used(tg, rw))
throtl_start_new_slice(tg, rw);
else {
if (time_before(tg->slice_end[rw], jiffies + throtl_slice))
throtl_extend_slice(tg, rw, jiffies + throtl_slice);
}
if (tg_with_in_bps_limit(tg, bio, &bps_wait) &&
tg_with_in_iops_limit(tg, bio, &iops_wait)) {
if (wait)
*wait = 0;
return 1;
}
max_wait = max(bps_wait, iops_wait);
if (wait)
*wait = max_wait;
if (time_before(tg->slice_end[rw], jiffies + max_wait))
throtl_extend_slice(tg, rw, jiffies + max_wait);
return 0;
}
static void throtl_update_dispatch_stats(struct blkcg_gq *blkg, u64 bytes,
int rw)
{
struct throtl_grp *tg = blkg_to_tg(blkg);
struct tg_stats_cpu *stats_cpu;
unsigned long flags;
/* If per cpu stats are not allocated yet, don't do any accounting. */
if (tg->stats_cpu == NULL)
return;
/*
* Disabling interrupts to provide mutual exclusion between two
* writes on same cpu. It probably is not needed for 64bit. Not
* optimizing that case yet.
*/
local_irq_save(flags);
stats_cpu = this_cpu_ptr(tg->stats_cpu);
blkg_rwstat_add(&stats_cpu->serviced, rw, 1);
blkg_rwstat_add(&stats_cpu->service_bytes, rw, bytes);
local_irq_restore(flags);
}
static void throtl_charge_bio(struct throtl_grp *tg, struct bio *bio)
{
bool rw = bio_data_dir(bio);
/* Charge the bio to the group */
block: Abstract out bvec iterator Immutable biovecs are going to require an explicit iterator. To implement immutable bvecs, a later patch is going to add a bi_bvec_done member to this struct; for now, this patch effectively just renames things. Signed-off-by: Kent Overstreet <kmo@daterainc.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <paulus@samba.org> Cc: "Ed L. Cashin" <ecashin@coraid.com> Cc: Nick Piggin <npiggin@kernel.dk> Cc: Lars Ellenberg <drbd-dev@lists.linbit.com> Cc: Jiri Kosina <jkosina@suse.cz> Cc: Matthew Wilcox <willy@linux.intel.com> Cc: Geoff Levand <geoff@infradead.org> Cc: Yehuda Sadeh <yehuda@inktank.com> Cc: Sage Weil <sage@inktank.com> Cc: Alex Elder <elder@inktank.com> Cc: ceph-devel@vger.kernel.org Cc: Joshua Morris <josh.h.morris@us.ibm.com> Cc: Philip Kelleher <pjk1939@linux.vnet.ibm.com> Cc: Rusty Russell <rusty@rustcorp.com.au> Cc: "Michael S. Tsirkin" <mst@redhat.com> Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com> Cc: Jeremy Fitzhardinge <jeremy@goop.org> Cc: Neil Brown <neilb@suse.de> Cc: Alasdair Kergon <agk@redhat.com> Cc: Mike Snitzer <snitzer@redhat.com> Cc: dm-devel@redhat.com Cc: Martin Schwidefsky <schwidefsky@de.ibm.com> Cc: Heiko Carstens <heiko.carstens@de.ibm.com> Cc: linux390@de.ibm.com Cc: Boaz Harrosh <bharrosh@panasas.com> Cc: Benny Halevy <bhalevy@tonian.com> Cc: "James E.J. Bottomley" <JBottomley@parallels.com> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: "Nicholas A. Bellinger" <nab@linux-iscsi.org> Cc: Alexander Viro <viro@zeniv.linux.org.uk> Cc: Chris Mason <chris.mason@fusionio.com> Cc: "Theodore Ts'o" <tytso@mit.edu> Cc: Andreas Dilger <adilger.kernel@dilger.ca> Cc: Jaegeuk Kim <jaegeuk.kim@samsung.com> Cc: Steven Whitehouse <swhiteho@redhat.com> Cc: Dave Kleikamp <shaggy@kernel.org> Cc: Joern Engel <joern@logfs.org> Cc: Prasad Joshi <prasadjoshi.linux@gmail.com> Cc: Trond Myklebust <Trond.Myklebust@netapp.com> Cc: KONISHI Ryusuke <konishi.ryusuke@lab.ntt.co.jp> Cc: Mark Fasheh <mfasheh@suse.com> Cc: Joel Becker <jlbec@evilplan.org> Cc: Ben Myers <bpm@sgi.com> Cc: xfs@oss.sgi.com Cc: Steven Rostedt <rostedt@goodmis.org> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Len Brown <len.brown@intel.com> Cc: Pavel Machek <pavel@ucw.cz> Cc: "Rafael J. Wysocki" <rjw@sisk.pl> Cc: Herton Ronaldo Krzesinski <herton.krzesinski@canonical.com> Cc: Ben Hutchings <ben@decadent.org.uk> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Guo Chao <yan@linux.vnet.ibm.com> Cc: Tejun Heo <tj@kernel.org> Cc: Asai Thambi S P <asamymuthupa@micron.com> Cc: Selvan Mani <smani@micron.com> Cc: Sam Bradshaw <sbradshaw@micron.com> Cc: Wei Yongjun <yongjun_wei@trendmicro.com.cn> Cc: "Roger Pau Monné" <roger.pau@citrix.com> Cc: Jan Beulich <jbeulich@suse.com> Cc: Stefano Stabellini <stefano.stabellini@eu.citrix.com> Cc: Ian Campbell <Ian.Campbell@citrix.com> Cc: Sebastian Ott <sebott@linux.vnet.ibm.com> Cc: Christian Borntraeger <borntraeger@de.ibm.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Jiang Liu <jiang.liu@huawei.com> Cc: Nitin Gupta <ngupta@vflare.org> Cc: Jerome Marchand <jmarchand@redhat.com> Cc: Joe Perches <joe@perches.com> Cc: Peng Tao <tao.peng@emc.com> Cc: Andy Adamson <andros@netapp.com> Cc: fanchaoting <fanchaoting@cn.fujitsu.com> Cc: Jie Liu <jeff.liu@oracle.com> Cc: Sunil Mushran <sunil.mushran@gmail.com> Cc: "Martin K. Petersen" <martin.petersen@oracle.com> Cc: Namjae Jeon <namjae.jeon@samsung.com> Cc: Pankaj Kumar <pankaj.km@samsung.com> Cc: Dan Magenheimer <dan.magenheimer@oracle.com> Cc: Mel Gorman <mgorman@suse.de>6
2013-10-12 00:44:27 +02:00
tg->bytes_disp[rw] += bio->bi_iter.bi_size;
tg->io_disp[rw]++;
/*
* REQ_THROTTLED is used to prevent the same bio to be throttled
* more than once as a throttled bio will go through blk-throtl the
* second time when it eventually gets issued. Set it when a bio
* is being charged to a tg.
*
* Dispatch stats aren't recursive and each @bio should only be
* accounted by the @tg it was originally associated with. Let's
* update the stats when setting REQ_THROTTLED for the first time
* which is guaranteed to be for the @bio's original tg.
*/
if (!(bio->bi_rw & REQ_THROTTLED)) {
bio->bi_rw |= REQ_THROTTLED;
block: Abstract out bvec iterator Immutable biovecs are going to require an explicit iterator. To implement immutable bvecs, a later patch is going to add a bi_bvec_done member to this struct; for now, this patch effectively just renames things. Signed-off-by: Kent Overstreet <kmo@daterainc.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <paulus@samba.org> Cc: "Ed L. Cashin" <ecashin@coraid.com> Cc: Nick Piggin <npiggin@kernel.dk> Cc: Lars Ellenberg <drbd-dev@lists.linbit.com> Cc: Jiri Kosina <jkosina@suse.cz> Cc: Matthew Wilcox <willy@linux.intel.com> Cc: Geoff Levand <geoff@infradead.org> Cc: Yehuda Sadeh <yehuda@inktank.com> Cc: Sage Weil <sage@inktank.com> Cc: Alex Elder <elder@inktank.com> Cc: ceph-devel@vger.kernel.org Cc: Joshua Morris <josh.h.morris@us.ibm.com> Cc: Philip Kelleher <pjk1939@linux.vnet.ibm.com> Cc: Rusty Russell <rusty@rustcorp.com.au> Cc: "Michael S. Tsirkin" <mst@redhat.com> Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com> Cc: Jeremy Fitzhardinge <jeremy@goop.org> Cc: Neil Brown <neilb@suse.de> Cc: Alasdair Kergon <agk@redhat.com> Cc: Mike Snitzer <snitzer@redhat.com> Cc: dm-devel@redhat.com Cc: Martin Schwidefsky <schwidefsky@de.ibm.com> Cc: Heiko Carstens <heiko.carstens@de.ibm.com> Cc: linux390@de.ibm.com Cc: Boaz Harrosh <bharrosh@panasas.com> Cc: Benny Halevy <bhalevy@tonian.com> Cc: "James E.J. Bottomley" <JBottomley@parallels.com> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: "Nicholas A. Bellinger" <nab@linux-iscsi.org> Cc: Alexander Viro <viro@zeniv.linux.org.uk> Cc: Chris Mason <chris.mason@fusionio.com> Cc: "Theodore Ts'o" <tytso@mit.edu> Cc: Andreas Dilger <adilger.kernel@dilger.ca> Cc: Jaegeuk Kim <jaegeuk.kim@samsung.com> Cc: Steven Whitehouse <swhiteho@redhat.com> Cc: Dave Kleikamp <shaggy@kernel.org> Cc: Joern Engel <joern@logfs.org> Cc: Prasad Joshi <prasadjoshi.linux@gmail.com> Cc: Trond Myklebust <Trond.Myklebust@netapp.com> Cc: KONISHI Ryusuke <konishi.ryusuke@lab.ntt.co.jp> Cc: Mark Fasheh <mfasheh@suse.com> Cc: Joel Becker <jlbec@evilplan.org> Cc: Ben Myers <bpm@sgi.com> Cc: xfs@oss.sgi.com Cc: Steven Rostedt <rostedt@goodmis.org> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Len Brown <len.brown@intel.com> Cc: Pavel Machek <pavel@ucw.cz> Cc: "Rafael J. Wysocki" <rjw@sisk.pl> Cc: Herton Ronaldo Krzesinski <herton.krzesinski@canonical.com> Cc: Ben Hutchings <ben@decadent.org.uk> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Guo Chao <yan@linux.vnet.ibm.com> Cc: Tejun Heo <tj@kernel.org> Cc: Asai Thambi S P <asamymuthupa@micron.com> Cc: Selvan Mani <smani@micron.com> Cc: Sam Bradshaw <sbradshaw@micron.com> Cc: Wei Yongjun <yongjun_wei@trendmicro.com.cn> Cc: "Roger Pau Monné" <roger.pau@citrix.com> Cc: Jan Beulich <jbeulich@suse.com> Cc: Stefano Stabellini <stefano.stabellini@eu.citrix.com> Cc: Ian Campbell <Ian.Campbell@citrix.com> Cc: Sebastian Ott <sebott@linux.vnet.ibm.com> Cc: Christian Borntraeger <borntraeger@de.ibm.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Jiang Liu <jiang.liu@huawei.com> Cc: Nitin Gupta <ngupta@vflare.org> Cc: Jerome Marchand <jmarchand@redhat.com> Cc: Joe Perches <joe@perches.com> Cc: Peng Tao <tao.peng@emc.com> Cc: Andy Adamson <andros@netapp.com> Cc: fanchaoting <fanchaoting@cn.fujitsu.com> Cc: Jie Liu <jeff.liu@oracle.com> Cc: Sunil Mushran <sunil.mushran@gmail.com> Cc: "Martin K. Petersen" <martin.petersen@oracle.com> Cc: Namjae Jeon <namjae.jeon@samsung.com> Cc: Pankaj Kumar <pankaj.km@samsung.com> Cc: Dan Magenheimer <dan.magenheimer@oracle.com> Cc: Mel Gorman <mgorman@suse.de>6
2013-10-12 00:44:27 +02:00
throtl_update_dispatch_stats(tg_to_blkg(tg),
bio->bi_iter.bi_size, bio->bi_rw);
}
}
blk-throttle: add throtl_qnode for dispatch fairness With flat hierarchy, there's only single level of dispatching happening and fairness beyond that point is the responsibility of the rest of the block layer and driver, which usually works out okay; however, with the planned hierarchy support, service_queue->bio_lists[] can be filled up by bios from a single source. While the limits would still be honored, it'd be very easy to starve IOs from siblings or children. To avoid such starvation, this patch implements throtl_qnode and converts service_queue->bio_lists[] to lists of per-source qnodes which in turn contains the bio's. For example, when a bio is dispatched from a child group, the bio doesn't get queued on ->bio_lists[] directly but it first gets queued on the group's qnode which in turn gets queued on service_queue->queued[]. When dispatching for the upper level, the ->queued[] list is consumed in round-robing order so that the dispatch windows is consumed fairly by all IO sources. There are two ways a bio can come to a throtl_grp - directly queued to the group or dispatched from a child. For the former throtl_grp->qnode_on_self[rw] is used. For the latter, the child's ->qnode_on_parent[rw]. Note that this means that the child which is contributing a bio to its parent should stay pinned until all its bios are dispatched to its grand-parent. This patch moves blkg refcnting from bio add/remove spots to qnode activation/deactivation so that the blkg containing an active qnode is always pinned. As child pins the parent, this is sufficient for keeping the relevant sub-tree pinned while bios are in flight. The starvation issue was spotted by Vivek Goyal. v2: The original patch used the same throtl_grp->qnode_on_self/parent for reads and writes causing RWs to be queued incorrectly if there already are outstanding IOs in the other direction. They should be throtl_grp->qnode_on_self/parent[2] so that READs and WRITEs can use different qnodes. Spotted by Vivek Goyal. Signed-off-by: Tejun Heo <tj@kernel.org> Acked-by: Vivek Goyal <vgoyal@redhat.com>
2013-05-14 22:52:38 +02:00
/**
* throtl_add_bio_tg - add a bio to the specified throtl_grp
* @bio: bio to add
* @qn: qnode to use
* @tg: the target throtl_grp
*
* Add @bio to @tg's service_queue using @qn. If @qn is not specified,
* tg->qnode_on_self[] is used.
*/
static void throtl_add_bio_tg(struct bio *bio, struct throtl_qnode *qn,
struct throtl_grp *tg)
{
struct throtl_service_queue *sq = &tg->service_queue;
bool rw = bio_data_dir(bio);
blk-throttle: add throtl_qnode for dispatch fairness With flat hierarchy, there's only single level of dispatching happening and fairness beyond that point is the responsibility of the rest of the block layer and driver, which usually works out okay; however, with the planned hierarchy support, service_queue->bio_lists[] can be filled up by bios from a single source. While the limits would still be honored, it'd be very easy to starve IOs from siblings or children. To avoid such starvation, this patch implements throtl_qnode and converts service_queue->bio_lists[] to lists of per-source qnodes which in turn contains the bio's. For example, when a bio is dispatched from a child group, the bio doesn't get queued on ->bio_lists[] directly but it first gets queued on the group's qnode which in turn gets queued on service_queue->queued[]. When dispatching for the upper level, the ->queued[] list is consumed in round-robing order so that the dispatch windows is consumed fairly by all IO sources. There are two ways a bio can come to a throtl_grp - directly queued to the group or dispatched from a child. For the former throtl_grp->qnode_on_self[rw] is used. For the latter, the child's ->qnode_on_parent[rw]. Note that this means that the child which is contributing a bio to its parent should stay pinned until all its bios are dispatched to its grand-parent. This patch moves blkg refcnting from bio add/remove spots to qnode activation/deactivation so that the blkg containing an active qnode is always pinned. As child pins the parent, this is sufficient for keeping the relevant sub-tree pinned while bios are in flight. The starvation issue was spotted by Vivek Goyal. v2: The original patch used the same throtl_grp->qnode_on_self/parent for reads and writes causing RWs to be queued incorrectly if there already are outstanding IOs in the other direction. They should be throtl_grp->qnode_on_self/parent[2] so that READs and WRITEs can use different qnodes. Spotted by Vivek Goyal. Signed-off-by: Tejun Heo <tj@kernel.org> Acked-by: Vivek Goyal <vgoyal@redhat.com>
2013-05-14 22:52:38 +02:00
if (!qn)
qn = &tg->qnode_on_self[rw];
/*
* If @tg doesn't currently have any bios queued in the same
* direction, queueing @bio can change when @tg should be
* dispatched. Mark that @tg was empty. This is automatically
* cleaered on the next tg_update_disptime().
*/
if (!sq->nr_queued[rw])
tg->flags |= THROTL_TG_WAS_EMPTY;
blk-throttle: add throtl_qnode for dispatch fairness With flat hierarchy, there's only single level of dispatching happening and fairness beyond that point is the responsibility of the rest of the block layer and driver, which usually works out okay; however, with the planned hierarchy support, service_queue->bio_lists[] can be filled up by bios from a single source. While the limits would still be honored, it'd be very easy to starve IOs from siblings or children. To avoid such starvation, this patch implements throtl_qnode and converts service_queue->bio_lists[] to lists of per-source qnodes which in turn contains the bio's. For example, when a bio is dispatched from a child group, the bio doesn't get queued on ->bio_lists[] directly but it first gets queued on the group's qnode which in turn gets queued on service_queue->queued[]. When dispatching for the upper level, the ->queued[] list is consumed in round-robing order so that the dispatch windows is consumed fairly by all IO sources. There are two ways a bio can come to a throtl_grp - directly queued to the group or dispatched from a child. For the former throtl_grp->qnode_on_self[rw] is used. For the latter, the child's ->qnode_on_parent[rw]. Note that this means that the child which is contributing a bio to its parent should stay pinned until all its bios are dispatched to its grand-parent. This patch moves blkg refcnting from bio add/remove spots to qnode activation/deactivation so that the blkg containing an active qnode is always pinned. As child pins the parent, this is sufficient for keeping the relevant sub-tree pinned while bios are in flight. The starvation issue was spotted by Vivek Goyal. v2: The original patch used the same throtl_grp->qnode_on_self/parent for reads and writes causing RWs to be queued incorrectly if there already are outstanding IOs in the other direction. They should be throtl_grp->qnode_on_self/parent[2] so that READs and WRITEs can use different qnodes. Spotted by Vivek Goyal. Signed-off-by: Tejun Heo <tj@kernel.org> Acked-by: Vivek Goyal <vgoyal@redhat.com>
2013-05-14 22:52:38 +02:00
throtl_qnode_add_bio(bio, qn, &sq->queued[rw]);
sq->nr_queued[rw]++;
throtl_enqueue_tg(tg);
}
static void tg_update_disptime(struct throtl_grp *tg)
{
struct throtl_service_queue *sq = &tg->service_queue;
unsigned long read_wait = -1, write_wait = -1, min_wait = -1, disptime;
struct bio *bio;
blk-throttle: add throtl_qnode for dispatch fairness With flat hierarchy, there's only single level of dispatching happening and fairness beyond that point is the responsibility of the rest of the block layer and driver, which usually works out okay; however, with the planned hierarchy support, service_queue->bio_lists[] can be filled up by bios from a single source. While the limits would still be honored, it'd be very easy to starve IOs from siblings or children. To avoid such starvation, this patch implements throtl_qnode and converts service_queue->bio_lists[] to lists of per-source qnodes which in turn contains the bio's. For example, when a bio is dispatched from a child group, the bio doesn't get queued on ->bio_lists[] directly but it first gets queued on the group's qnode which in turn gets queued on service_queue->queued[]. When dispatching for the upper level, the ->queued[] list is consumed in round-robing order so that the dispatch windows is consumed fairly by all IO sources. There are two ways a bio can come to a throtl_grp - directly queued to the group or dispatched from a child. For the former throtl_grp->qnode_on_self[rw] is used. For the latter, the child's ->qnode_on_parent[rw]. Note that this means that the child which is contributing a bio to its parent should stay pinned until all its bios are dispatched to its grand-parent. This patch moves blkg refcnting from bio add/remove spots to qnode activation/deactivation so that the blkg containing an active qnode is always pinned. As child pins the parent, this is sufficient for keeping the relevant sub-tree pinned while bios are in flight. The starvation issue was spotted by Vivek Goyal. v2: The original patch used the same throtl_grp->qnode_on_self/parent for reads and writes causing RWs to be queued incorrectly if there already are outstanding IOs in the other direction. They should be throtl_grp->qnode_on_self/parent[2] so that READs and WRITEs can use different qnodes. Spotted by Vivek Goyal. Signed-off-by: Tejun Heo <tj@kernel.org> Acked-by: Vivek Goyal <vgoyal@redhat.com>
2013-05-14 22:52:38 +02:00
if ((bio = throtl_peek_queued(&sq->queued[READ])))
tg_may_dispatch(tg, bio, &read_wait);
blk-throttle: add throtl_qnode for dispatch fairness With flat hierarchy, there's only single level of dispatching happening and fairness beyond that point is the responsibility of the rest of the block layer and driver, which usually works out okay; however, with the planned hierarchy support, service_queue->bio_lists[] can be filled up by bios from a single source. While the limits would still be honored, it'd be very easy to starve IOs from siblings or children. To avoid such starvation, this patch implements throtl_qnode and converts service_queue->bio_lists[] to lists of per-source qnodes which in turn contains the bio's. For example, when a bio is dispatched from a child group, the bio doesn't get queued on ->bio_lists[] directly but it first gets queued on the group's qnode which in turn gets queued on service_queue->queued[]. When dispatching for the upper level, the ->queued[] list is consumed in round-robing order so that the dispatch windows is consumed fairly by all IO sources. There are two ways a bio can come to a throtl_grp - directly queued to the group or dispatched from a child. For the former throtl_grp->qnode_on_self[rw] is used. For the latter, the child's ->qnode_on_parent[rw]. Note that this means that the child which is contributing a bio to its parent should stay pinned until all its bios are dispatched to its grand-parent. This patch moves blkg refcnting from bio add/remove spots to qnode activation/deactivation so that the blkg containing an active qnode is always pinned. As child pins the parent, this is sufficient for keeping the relevant sub-tree pinned while bios are in flight. The starvation issue was spotted by Vivek Goyal. v2: The original patch used the same throtl_grp->qnode_on_self/parent for reads and writes causing RWs to be queued incorrectly if there already are outstanding IOs in the other direction. They should be throtl_grp->qnode_on_self/parent[2] so that READs and WRITEs can use different qnodes. Spotted by Vivek Goyal. Signed-off-by: Tejun Heo <tj@kernel.org> Acked-by: Vivek Goyal <vgoyal@redhat.com>
2013-05-14 22:52:38 +02:00
if ((bio = throtl_peek_queued(&sq->queued[WRITE])))
tg_may_dispatch(tg, bio, &write_wait);
min_wait = min(read_wait, write_wait);
disptime = jiffies + min_wait;
/* Update dispatch time */
throtl_dequeue_tg(tg);
tg->disptime = disptime;
throtl_enqueue_tg(tg);
/* see throtl_add_bio_tg() */
tg->flags &= ~THROTL_TG_WAS_EMPTY;
}
blk-throttle: Account for child group's start time in parent while bio climbs up With the planned proper hierarchy support, a bio will climb up the tree before actually being dispatched. This makes sure bio is also subjected to parent's throttling limits, if any. It might happen that parent is idle and when bio is transferred to parent, a new slice starts fresh. But that is incorrect as parents wait time should have started when bio was queued in child group and causes IOs to be throttled more than configured as they climb the hierarchy. Given the fact that we have not written hierarchical algorithm in a way where child's and parents time slices are synchronized, we transfer the child's start time to parent if parent was idling. If parent was busy doing dispatch of other bios all this while, this is not an issue. Child's slice start time is passed to parent. Parent looks at its last expired slice start time. If child's start time is after parents old start time, that means parent had been idle and after parent went idle, child had an IO queued. So use child's start time as parent start time. If parent's start time is after child's start time, that means, when IO got queued in child group, parent was not idle. But later it dispatched some IO, its slice got trimmed and then it went idle. After a while child's request got shifted in parent group. In this case use parent's old start time as new start time as that's the duration of slice we did not use. This logic is far from perfect as if there are multiple childs then first child transferring the bio decides the start time while a bio might have queued up even earlier in other child, which is yet to be transferred up to parent. In that case we will lose time and bandwidth in parent. This patch is just an approximation to make situation somewhat better. Signed-off-by: Vivek Goyal <vgoyal@redhat.com> Signed-off-by: Tejun Heo <tj@kernel.org>
2013-05-14 22:52:38 +02:00
static void start_parent_slice_with_credit(struct throtl_grp *child_tg,
struct throtl_grp *parent_tg, bool rw)
{
if (throtl_slice_used(parent_tg, rw)) {
throtl_start_new_slice_with_credit(parent_tg, rw,
child_tg->slice_start[rw]);
}
}
static void tg_dispatch_one_bio(struct throtl_grp *tg, bool rw)
{
struct throtl_service_queue *sq = &tg->service_queue;
struct throtl_service_queue *parent_sq = sq->parent_sq;
struct throtl_grp *parent_tg = sq_to_tg(parent_sq);
blk-throttle: add throtl_qnode for dispatch fairness With flat hierarchy, there's only single level of dispatching happening and fairness beyond that point is the responsibility of the rest of the block layer and driver, which usually works out okay; however, with the planned hierarchy support, service_queue->bio_lists[] can be filled up by bios from a single source. While the limits would still be honored, it'd be very easy to starve IOs from siblings or children. To avoid such starvation, this patch implements throtl_qnode and converts service_queue->bio_lists[] to lists of per-source qnodes which in turn contains the bio's. For example, when a bio is dispatched from a child group, the bio doesn't get queued on ->bio_lists[] directly but it first gets queued on the group's qnode which in turn gets queued on service_queue->queued[]. When dispatching for the upper level, the ->queued[] list is consumed in round-robing order so that the dispatch windows is consumed fairly by all IO sources. There are two ways a bio can come to a throtl_grp - directly queued to the group or dispatched from a child. For the former throtl_grp->qnode_on_self[rw] is used. For the latter, the child's ->qnode_on_parent[rw]. Note that this means that the child which is contributing a bio to its parent should stay pinned until all its bios are dispatched to its grand-parent. This patch moves blkg refcnting from bio add/remove spots to qnode activation/deactivation so that the blkg containing an active qnode is always pinned. As child pins the parent, this is sufficient for keeping the relevant sub-tree pinned while bios are in flight. The starvation issue was spotted by Vivek Goyal. v2: The original patch used the same throtl_grp->qnode_on_self/parent for reads and writes causing RWs to be queued incorrectly if there already are outstanding IOs in the other direction. They should be throtl_grp->qnode_on_self/parent[2] so that READs and WRITEs can use different qnodes. Spotted by Vivek Goyal. Signed-off-by: Tejun Heo <tj@kernel.org> Acked-by: Vivek Goyal <vgoyal@redhat.com>
2013-05-14 22:52:38 +02:00
struct throtl_grp *tg_to_put = NULL;
struct bio *bio;
blk-throttle: add throtl_qnode for dispatch fairness With flat hierarchy, there's only single level of dispatching happening and fairness beyond that point is the responsibility of the rest of the block layer and driver, which usually works out okay; however, with the planned hierarchy support, service_queue->bio_lists[] can be filled up by bios from a single source. While the limits would still be honored, it'd be very easy to starve IOs from siblings or children. To avoid such starvation, this patch implements throtl_qnode and converts service_queue->bio_lists[] to lists of per-source qnodes which in turn contains the bio's. For example, when a bio is dispatched from a child group, the bio doesn't get queued on ->bio_lists[] directly but it first gets queued on the group's qnode which in turn gets queued on service_queue->queued[]. When dispatching for the upper level, the ->queued[] list is consumed in round-robing order so that the dispatch windows is consumed fairly by all IO sources. There are two ways a bio can come to a throtl_grp - directly queued to the group or dispatched from a child. For the former throtl_grp->qnode_on_self[rw] is used. For the latter, the child's ->qnode_on_parent[rw]. Note that this means that the child which is contributing a bio to its parent should stay pinned until all its bios are dispatched to its grand-parent. This patch moves blkg refcnting from bio add/remove spots to qnode activation/deactivation so that the blkg containing an active qnode is always pinned. As child pins the parent, this is sufficient for keeping the relevant sub-tree pinned while bios are in flight. The starvation issue was spotted by Vivek Goyal. v2: The original patch used the same throtl_grp->qnode_on_self/parent for reads and writes causing RWs to be queued incorrectly if there already are outstanding IOs in the other direction. They should be throtl_grp->qnode_on_self/parent[2] so that READs and WRITEs can use different qnodes. Spotted by Vivek Goyal. Signed-off-by: Tejun Heo <tj@kernel.org> Acked-by: Vivek Goyal <vgoyal@redhat.com>
2013-05-14 22:52:38 +02:00
/*
* @bio is being transferred from @tg to @parent_sq. Popping a bio
* from @tg may put its reference and @parent_sq might end up
* getting released prematurely. Remember the tg to put and put it
* after @bio is transferred to @parent_sq.
*/
bio = throtl_pop_queued(&sq->queued[rw], &tg_to_put);
sq->nr_queued[rw]--;
throtl_charge_bio(tg, bio);
/*
* If our parent is another tg, we just need to transfer @bio to
* the parent using throtl_add_bio_tg(). If our parent is
* @td->service_queue, @bio is ready to be issued. Put it on its
* bio_lists[] and decrease total number queued. The caller is
* responsible for issuing these bios.
*/
if (parent_tg) {
blk-throttle: add throtl_qnode for dispatch fairness With flat hierarchy, there's only single level of dispatching happening and fairness beyond that point is the responsibility of the rest of the block layer and driver, which usually works out okay; however, with the planned hierarchy support, service_queue->bio_lists[] can be filled up by bios from a single source. While the limits would still be honored, it'd be very easy to starve IOs from siblings or children. To avoid such starvation, this patch implements throtl_qnode and converts service_queue->bio_lists[] to lists of per-source qnodes which in turn contains the bio's. For example, when a bio is dispatched from a child group, the bio doesn't get queued on ->bio_lists[] directly but it first gets queued on the group's qnode which in turn gets queued on service_queue->queued[]. When dispatching for the upper level, the ->queued[] list is consumed in round-robing order so that the dispatch windows is consumed fairly by all IO sources. There are two ways a bio can come to a throtl_grp - directly queued to the group or dispatched from a child. For the former throtl_grp->qnode_on_self[rw] is used. For the latter, the child's ->qnode_on_parent[rw]. Note that this means that the child which is contributing a bio to its parent should stay pinned until all its bios are dispatched to its grand-parent. This patch moves blkg refcnting from bio add/remove spots to qnode activation/deactivation so that the blkg containing an active qnode is always pinned. As child pins the parent, this is sufficient for keeping the relevant sub-tree pinned while bios are in flight. The starvation issue was spotted by Vivek Goyal. v2: The original patch used the same throtl_grp->qnode_on_self/parent for reads and writes causing RWs to be queued incorrectly if there already are outstanding IOs in the other direction. They should be throtl_grp->qnode_on_self/parent[2] so that READs and WRITEs can use different qnodes. Spotted by Vivek Goyal. Signed-off-by: Tejun Heo <tj@kernel.org> Acked-by: Vivek Goyal <vgoyal@redhat.com>
2013-05-14 22:52:38 +02:00
throtl_add_bio_tg(bio, &tg->qnode_on_parent[rw], parent_tg);
blk-throttle: Account for child group's start time in parent while bio climbs up With the planned proper hierarchy support, a bio will climb up the tree before actually being dispatched. This makes sure bio is also subjected to parent's throttling limits, if any. It might happen that parent is idle and when bio is transferred to parent, a new slice starts fresh. But that is incorrect as parents wait time should have started when bio was queued in child group and causes IOs to be throttled more than configured as they climb the hierarchy. Given the fact that we have not written hierarchical algorithm in a way where child's and parents time slices are synchronized, we transfer the child's start time to parent if parent was idling. If parent was busy doing dispatch of other bios all this while, this is not an issue. Child's slice start time is passed to parent. Parent looks at its last expired slice start time. If child's start time is after parents old start time, that means parent had been idle and after parent went idle, child had an IO queued. So use child's start time as parent start time. If parent's start time is after child's start time, that means, when IO got queued in child group, parent was not idle. But later it dispatched some IO, its slice got trimmed and then it went idle. After a while child's request got shifted in parent group. In this case use parent's old start time as new start time as that's the duration of slice we did not use. This logic is far from perfect as if there are multiple childs then first child transferring the bio decides the start time while a bio might have queued up even earlier in other child, which is yet to be transferred up to parent. In that case we will lose time and bandwidth in parent. This patch is just an approximation to make situation somewhat better. Signed-off-by: Vivek Goyal <vgoyal@redhat.com> Signed-off-by: Tejun Heo <tj@kernel.org>
2013-05-14 22:52:38 +02:00
start_parent_slice_with_credit(tg, parent_tg, rw);
} else {
blk-throttle: add throtl_qnode for dispatch fairness With flat hierarchy, there's only single level of dispatching happening and fairness beyond that point is the responsibility of the rest of the block layer and driver, which usually works out okay; however, with the planned hierarchy support, service_queue->bio_lists[] can be filled up by bios from a single source. While the limits would still be honored, it'd be very easy to starve IOs from siblings or children. To avoid such starvation, this patch implements throtl_qnode and converts service_queue->bio_lists[] to lists of per-source qnodes which in turn contains the bio's. For example, when a bio is dispatched from a child group, the bio doesn't get queued on ->bio_lists[] directly but it first gets queued on the group's qnode which in turn gets queued on service_queue->queued[]. When dispatching for the upper level, the ->queued[] list is consumed in round-robing order so that the dispatch windows is consumed fairly by all IO sources. There are two ways a bio can come to a throtl_grp - directly queued to the group or dispatched from a child. For the former throtl_grp->qnode_on_self[rw] is used. For the latter, the child's ->qnode_on_parent[rw]. Note that this means that the child which is contributing a bio to its parent should stay pinned until all its bios are dispatched to its grand-parent. This patch moves blkg refcnting from bio add/remove spots to qnode activation/deactivation so that the blkg containing an active qnode is always pinned. As child pins the parent, this is sufficient for keeping the relevant sub-tree pinned while bios are in flight. The starvation issue was spotted by Vivek Goyal. v2: The original patch used the same throtl_grp->qnode_on_self/parent for reads and writes causing RWs to be queued incorrectly if there already are outstanding IOs in the other direction. They should be throtl_grp->qnode_on_self/parent[2] so that READs and WRITEs can use different qnodes. Spotted by Vivek Goyal. Signed-off-by: Tejun Heo <tj@kernel.org> Acked-by: Vivek Goyal <vgoyal@redhat.com>
2013-05-14 22:52:38 +02:00
throtl_qnode_add_bio(bio, &tg->qnode_on_parent[rw],
&parent_sq->queued[rw]);
BUG_ON(tg->td->nr_queued[rw] <= 0);
tg->td->nr_queued[rw]--;
}
throtl_trim_slice(tg, rw);
blk-throttle: add throtl_qnode for dispatch fairness With flat hierarchy, there's only single level of dispatching happening and fairness beyond that point is the responsibility of the rest of the block layer and driver, which usually works out okay; however, with the planned hierarchy support, service_queue->bio_lists[] can be filled up by bios from a single source. While the limits would still be honored, it'd be very easy to starve IOs from siblings or children. To avoid such starvation, this patch implements throtl_qnode and converts service_queue->bio_lists[] to lists of per-source qnodes which in turn contains the bio's. For example, when a bio is dispatched from a child group, the bio doesn't get queued on ->bio_lists[] directly but it first gets queued on the group's qnode which in turn gets queued on service_queue->queued[]. When dispatching for the upper level, the ->queued[] list is consumed in round-robing order so that the dispatch windows is consumed fairly by all IO sources. There are two ways a bio can come to a throtl_grp - directly queued to the group or dispatched from a child. For the former throtl_grp->qnode_on_self[rw] is used. For the latter, the child's ->qnode_on_parent[rw]. Note that this means that the child which is contributing a bio to its parent should stay pinned until all its bios are dispatched to its grand-parent. This patch moves blkg refcnting from bio add/remove spots to qnode activation/deactivation so that the blkg containing an active qnode is always pinned. As child pins the parent, this is sufficient for keeping the relevant sub-tree pinned while bios are in flight. The starvation issue was spotted by Vivek Goyal. v2: The original patch used the same throtl_grp->qnode_on_self/parent for reads and writes causing RWs to be queued incorrectly if there already are outstanding IOs in the other direction. They should be throtl_grp->qnode_on_self/parent[2] so that READs and WRITEs can use different qnodes. Spotted by Vivek Goyal. Signed-off-by: Tejun Heo <tj@kernel.org> Acked-by: Vivek Goyal <vgoyal@redhat.com>
2013-05-14 22:52:38 +02:00
if (tg_to_put)
blkg_put(tg_to_blkg(tg_to_put));
}
static int throtl_dispatch_tg(struct throtl_grp *tg)
{
struct throtl_service_queue *sq = &tg->service_queue;
unsigned int nr_reads = 0, nr_writes = 0;
unsigned int max_nr_reads = throtl_grp_quantum*3/4;
unsigned int max_nr_writes = throtl_grp_quantum - max_nr_reads;
struct bio *bio;
/* Try to dispatch 75% READS and 25% WRITES */
blk-throttle: add throtl_qnode for dispatch fairness With flat hierarchy, there's only single level of dispatching happening and fairness beyond that point is the responsibility of the rest of the block layer and driver, which usually works out okay; however, with the planned hierarchy support, service_queue->bio_lists[] can be filled up by bios from a single source. While the limits would still be honored, it'd be very easy to starve IOs from siblings or children. To avoid such starvation, this patch implements throtl_qnode and converts service_queue->bio_lists[] to lists of per-source qnodes which in turn contains the bio's. For example, when a bio is dispatched from a child group, the bio doesn't get queued on ->bio_lists[] directly but it first gets queued on the group's qnode which in turn gets queued on service_queue->queued[]. When dispatching for the upper level, the ->queued[] list is consumed in round-robing order so that the dispatch windows is consumed fairly by all IO sources. There are two ways a bio can come to a throtl_grp - directly queued to the group or dispatched from a child. For the former throtl_grp->qnode_on_self[rw] is used. For the latter, the child's ->qnode_on_parent[rw]. Note that this means that the child which is contributing a bio to its parent should stay pinned until all its bios are dispatched to its grand-parent. This patch moves blkg refcnting from bio add/remove spots to qnode activation/deactivation so that the blkg containing an active qnode is always pinned. As child pins the parent, this is sufficient for keeping the relevant sub-tree pinned while bios are in flight. The starvation issue was spotted by Vivek Goyal. v2: The original patch used the same throtl_grp->qnode_on_self/parent for reads and writes causing RWs to be queued incorrectly if there already are outstanding IOs in the other direction. They should be throtl_grp->qnode_on_self/parent[2] so that READs and WRITEs can use different qnodes. Spotted by Vivek Goyal. Signed-off-by: Tejun Heo <tj@kernel.org> Acked-by: Vivek Goyal <vgoyal@redhat.com>
2013-05-14 22:52:38 +02:00
while ((bio = throtl_peek_queued(&sq->queued[READ])) &&
tg_may_dispatch(tg, bio, NULL)) {
tg_dispatch_one_bio(tg, bio_data_dir(bio));
nr_reads++;
if (nr_reads >= max_nr_reads)
break;
}
blk-throttle: add throtl_qnode for dispatch fairness With flat hierarchy, there's only single level of dispatching happening and fairness beyond that point is the responsibility of the rest of the block layer and driver, which usually works out okay; however, with the planned hierarchy support, service_queue->bio_lists[] can be filled up by bios from a single source. While the limits would still be honored, it'd be very easy to starve IOs from siblings or children. To avoid such starvation, this patch implements throtl_qnode and converts service_queue->bio_lists[] to lists of per-source qnodes which in turn contains the bio's. For example, when a bio is dispatched from a child group, the bio doesn't get queued on ->bio_lists[] directly but it first gets queued on the group's qnode which in turn gets queued on service_queue->queued[]. When dispatching for the upper level, the ->queued[] list is consumed in round-robing order so that the dispatch windows is consumed fairly by all IO sources. There are two ways a bio can come to a throtl_grp - directly queued to the group or dispatched from a child. For the former throtl_grp->qnode_on_self[rw] is used. For the latter, the child's ->qnode_on_parent[rw]. Note that this means that the child which is contributing a bio to its parent should stay pinned until all its bios are dispatched to its grand-parent. This patch moves blkg refcnting from bio add/remove spots to qnode activation/deactivation so that the blkg containing an active qnode is always pinned. As child pins the parent, this is sufficient for keeping the relevant sub-tree pinned while bios are in flight. The starvation issue was spotted by Vivek Goyal. v2: The original patch used the same throtl_grp->qnode_on_self/parent for reads and writes causing RWs to be queued incorrectly if there already are outstanding IOs in the other direction. They should be throtl_grp->qnode_on_self/parent[2] so that READs and WRITEs can use different qnodes. Spotted by Vivek Goyal. Signed-off-by: Tejun Heo <tj@kernel.org> Acked-by: Vivek Goyal <vgoyal@redhat.com>
2013-05-14 22:52:38 +02:00
while ((bio = throtl_peek_queued(&sq->queued[WRITE])) &&
tg_may_dispatch(tg, bio, NULL)) {
tg_dispatch_one_bio(tg, bio_data_dir(bio));
nr_writes++;
if (nr_writes >= max_nr_writes)
break;
}
return nr_reads + nr_writes;
}
static int throtl_select_dispatch(struct throtl_service_queue *parent_sq)
{
unsigned int nr_disp = 0;
while (1) {
struct throtl_grp *tg = throtl_rb_first(parent_sq);
struct throtl_service_queue *sq = &tg->service_queue;
if (!tg)
break;
if (time_before(jiffies, tg->disptime))
break;
throtl_dequeue_tg(tg);
nr_disp += throtl_dispatch_tg(tg);
if (sq->nr_queued[0] || sq->nr_queued[1])
tg_update_disptime(tg);
if (nr_disp >= throtl_quantum)
break;
}
return nr_disp;
}
/**
* throtl_pending_timer_fn - timer function for service_queue->pending_timer
* @arg: the throtl_service_queue being serviced
*
* This timer is armed when a child throtl_grp with active bio's become
* pending and queued on the service_queue's pending_tree and expires when
* the first child throtl_grp should be dispatched. This function
* dispatches bio's from the children throtl_grps to the parent
* service_queue.
*
* If the parent's parent is another throtl_grp, dispatching is propagated
* by either arming its pending_timer or repeating dispatch directly. If
* the top-level service_tree is reached, throtl_data->dispatch_work is
* kicked so that the ready bio's are issued.
*/
static void throtl_pending_timer_fn(unsigned long arg)
{
struct throtl_service_queue *sq = (void *)arg;
struct throtl_grp *tg = sq_to_tg(sq);
struct throtl_data *td = sq_to_td(sq);
struct request_queue *q = td->queue;
struct throtl_service_queue *parent_sq;
bool dispatched;
int ret;
spin_lock_irq(q->queue_lock);
again:
parent_sq = sq->parent_sq;
dispatched = false;
while (true) {
throtl_log(sq, "dispatch nr_queued=%u read=%u write=%u",
sq->nr_queued[READ] + sq->nr_queued[WRITE],
sq->nr_queued[READ], sq->nr_queued[WRITE]);
ret = throtl_select_dispatch(sq);
if (ret) {
throtl_log(sq, "bios disp=%u", ret);
dispatched = true;
}
if (throtl_schedule_next_dispatch(sq, false))
break;
/* this dispatch windows is still open, relax and repeat */
spin_unlock_irq(q->queue_lock);
cpu_relax();
spin_lock_irq(q->queue_lock);
}
if (!dispatched)
goto out_unlock;
if (parent_sq) {
/* @parent_sq is another throl_grp, propagate dispatch */
if (tg->flags & THROTL_TG_WAS_EMPTY) {
tg_update_disptime(tg);
if (!throtl_schedule_next_dispatch(parent_sq, false)) {
/* window is already open, repeat dispatching */
sq = parent_sq;
tg = sq_to_tg(sq);
goto again;
}
}
} else {
/* reached the top-level, queue issueing */
queue_work(kthrotld_workqueue, &td->dispatch_work);
}
out_unlock:
spin_unlock_irq(q->queue_lock);
}
/**
* blk_throtl_dispatch_work_fn - work function for throtl_data->dispatch_work
* @work: work item being executed
*
* This function is queued for execution when bio's reach the bio_lists[]
* of throtl_data->service_queue. Those bio's are ready and issued by this
* function.
*/
static void blk_throtl_dispatch_work_fn(struct work_struct *work)
{
struct throtl_data *td = container_of(work, struct throtl_data,
dispatch_work);
struct throtl_service_queue *td_sq = &td->service_queue;
struct request_queue *q = td->queue;
struct bio_list bio_list_on_stack;
struct bio *bio;
struct blk_plug plug;
int rw;
bio_list_init(&bio_list_on_stack);
spin_lock_irq(q->queue_lock);
blk-throttle: add throtl_qnode for dispatch fairness With flat hierarchy, there's only single level of dispatching happening and fairness beyond that point is the responsibility of the rest of the block layer and driver, which usually works out okay; however, with the planned hierarchy support, service_queue->bio_lists[] can be filled up by bios from a single source. While the limits would still be honored, it'd be very easy to starve IOs from siblings or children. To avoid such starvation, this patch implements throtl_qnode and converts service_queue->bio_lists[] to lists of per-source qnodes which in turn contains the bio's. For example, when a bio is dispatched from a child group, the bio doesn't get queued on ->bio_lists[] directly but it first gets queued on the group's qnode which in turn gets queued on service_queue->queued[]. When dispatching for the upper level, the ->queued[] list is consumed in round-robing order so that the dispatch windows is consumed fairly by all IO sources. There are two ways a bio can come to a throtl_grp - directly queued to the group or dispatched from a child. For the former throtl_grp->qnode_on_self[rw] is used. For the latter, the child's ->qnode_on_parent[rw]. Note that this means that the child which is contributing a bio to its parent should stay pinned until all its bios are dispatched to its grand-parent. This patch moves blkg refcnting from bio add/remove spots to qnode activation/deactivation so that the blkg containing an active qnode is always pinned. As child pins the parent, this is sufficient for keeping the relevant sub-tree pinned while bios are in flight. The starvation issue was spotted by Vivek Goyal. v2: The original patch used the same throtl_grp->qnode_on_self/parent for reads and writes causing RWs to be queued incorrectly if there already are outstanding IOs in the other direction. They should be throtl_grp->qnode_on_self/parent[2] so that READs and WRITEs can use different qnodes. Spotted by Vivek Goyal. Signed-off-by: Tejun Heo <tj@kernel.org> Acked-by: Vivek Goyal <vgoyal@redhat.com>
2013-05-14 22:52:38 +02:00
for (rw = READ; rw <= WRITE; rw++)
while ((bio = throtl_pop_queued(&td_sq->queued[rw], NULL)))
bio_list_add(&bio_list_on_stack, bio);
spin_unlock_irq(q->queue_lock);
if (!bio_list_empty(&bio_list_on_stack)) {
blk_start_plug(&plug);
while((bio = bio_list_pop(&bio_list_on_stack)))
generic_make_request(bio);
blk_finish_plug(&plug);
}
}
static u64 tg_prfill_cpu_rwstat(struct seq_file *sf,
struct blkg_policy_data *pd, int off)
{
struct throtl_grp *tg = pd_to_tg(pd);
struct blkg_rwstat rwstat = { }, tmp;
int i, cpu;
for_each_possible_cpu(cpu) {
struct tg_stats_cpu *sc = per_cpu_ptr(tg->stats_cpu, cpu);
tmp = blkg_rwstat_read((void *)sc + off);
for (i = 0; i < BLKG_RWSTAT_NR; i++)
rwstat.cnt[i] += tmp.cnt[i];
}
return __blkg_prfill_rwstat(sf, pd, &rwstat);
}
static int tg_print_cpu_rwstat(struct seq_file *sf, void *v)
{
blkcg_print_blkgs(sf, css_to_blkcg(seq_css(sf)), tg_prfill_cpu_rwstat,
&blkcg_policy_throtl, seq_cft(sf)->private, true);
return 0;
}
static u64 tg_prfill_conf_u64(struct seq_file *sf, struct blkg_policy_data *pd,
int off)
{
struct throtl_grp *tg = pd_to_tg(pd);
u64 v = *(u64 *)((void *)tg + off);
if (v == -1)
return 0;
return __blkg_prfill_u64(sf, pd, v);
}
static u64 tg_prfill_conf_uint(struct seq_file *sf, struct blkg_policy_data *pd,
int off)
{
struct throtl_grp *tg = pd_to_tg(pd);
unsigned int v = *(unsigned int *)((void *)tg + off);
if (v == -1)
return 0;
return __blkg_prfill_u64(sf, pd, v);
}
static int tg_print_conf_u64(struct seq_file *sf, void *v)
{
blkcg_print_blkgs(sf, css_to_blkcg(seq_css(sf)), tg_prfill_conf_u64,
&blkcg_policy_throtl, seq_cft(sf)->private, false);
return 0;
}
static int tg_print_conf_uint(struct seq_file *sf, void *v)
{
blkcg_print_blkgs(sf, css_to_blkcg(seq_css(sf)), tg_prfill_conf_uint,
&blkcg_policy_throtl, seq_cft(sf)->private, false);
return 0;
}
static ssize_t tg_set_conf(struct kernfs_open_file *of,
char *buf, size_t nbytes, loff_t off, bool is_u64)
{
struct blkcg *blkcg = css_to_blkcg(of_css(of));
struct blkg_conf_ctx ctx;
struct throtl_grp *tg;
struct throtl_service_queue *sq;
struct blkcg_gq *blkg;
2013-08-09 02:11:25 +02:00
struct cgroup_subsys_state *pos_css;
int ret;
ret = blkg_conf_prep(blkcg, &blkcg_policy_throtl, buf, &ctx);
if (ret)
return ret;
tg = blkg_to_tg(ctx.blkg);
sq = &tg->service_queue;
if (!ctx.v)
ctx.v = -1;
if (is_u64)
*(u64 *)((void *)tg + of_cft(of)->private) = ctx.v;
else
*(unsigned int *)((void *)tg + of_cft(of)->private) = ctx.v;
throtl_log(&tg->service_queue,
"limit change rbps=%llu wbps=%llu riops=%u wiops=%u",
tg->bps[READ], tg->bps[WRITE],
tg->iops[READ], tg->iops[WRITE]);
/*
* Update has_rules[] flags for the updated tg's subtree. A tg is
* considered to have rules if either the tg itself or any of its
* ancestors has rules. This identifies groups without any
* restrictions in the whole hierarchy and allows them to bypass
* blk-throttle.
*/
2013-08-09 02:11:25 +02:00
blkg_for_each_descendant_pre(blkg, pos_css, ctx.blkg)
tg_update_has_rules(blkg_to_tg(blkg));
/*
* We're already holding queue_lock and know @tg is valid. Let's
* apply the new config directly.
*
* Restart the slices for both READ and WRITES. It might happen
* that a group's limit are dropped suddenly and we don't want to
* account recently dispatched IO with new low rate.
*/
throtl_start_new_slice(tg, 0);
throtl_start_new_slice(tg, 1);
if (tg->flags & THROTL_TG_PENDING) {
tg_update_disptime(tg);
throtl_schedule_next_dispatch(sq->parent_sq, true);
}
blkg_conf_finish(&ctx);
return nbytes;
}
static ssize_t tg_set_conf_u64(struct kernfs_open_file *of,
char *buf, size_t nbytes, loff_t off)
{
return tg_set_conf(of, buf, nbytes, off, true);
}
static ssize_t tg_set_conf_uint(struct kernfs_open_file *of,
char *buf, size_t nbytes, loff_t off)
{
return tg_set_conf(of, buf, nbytes, off, false);
}
static struct cftype throtl_files[] = {
{
.name = "throttle.read_bps_device",
.private = offsetof(struct throtl_grp, bps[READ]),
.seq_show = tg_print_conf_u64,
.write = tg_set_conf_u64,
},
{
.name = "throttle.write_bps_device",
.private = offsetof(struct throtl_grp, bps[WRITE]),
.seq_show = tg_print_conf_u64,
.write = tg_set_conf_u64,
},
{
.name = "throttle.read_iops_device",
.private = offsetof(struct throtl_grp, iops[READ]),
.seq_show = tg_print_conf_uint,
.write = tg_set_conf_uint,
},
{
.name = "throttle.write_iops_device",
.private = offsetof(struct throtl_grp, iops[WRITE]),
.seq_show = tg_print_conf_uint,
.write = tg_set_conf_uint,
},
{
.name = "throttle.io_service_bytes",
.private = offsetof(struct tg_stats_cpu, service_bytes),
.seq_show = tg_print_cpu_rwstat,
},
{
.name = "throttle.io_serviced",
.private = offsetof(struct tg_stats_cpu, serviced),
.seq_show = tg_print_cpu_rwstat,
},
{ } /* terminate */
};
block: Move blk_throtl_exit() call to blk_cleanup_queue() Move blk_throtl_exit() in blk_cleanup_queue() as blk_throtl_exit() is written in such a way that it needs queue lock. In blk_release_queue() there is no gurantee that ->queue_lock is still around. Initially blk_throtl_exit() was in blk_cleanup_queue() but Ingo reported one problem. https://lkml.org/lkml/2010/10/23/86 And a quick fix moved blk_throtl_exit() to blk_release_queue(). commit 7ad58c028652753814054f4e3ac58f925e7343f4 Author: Jens Axboe <jaxboe@fusionio.com> Date: Sat Oct 23 20:40:26 2010 +0200 block: fix use-after-free bug in blk throttle code This patch reverts above change and does not try to shutdown the throtl work in blk_sync_queue(). By avoiding call to throtl_shutdown_timer_wq() from blk_sync_queue(), we should also avoid the problem reported by Ingo. blk_sync_queue() seems to be used only by md driver and it seems to be using it to make sure q->unplug_fn is not called as md registers its own unplug functions and it is about to free up the data structures used by unplug_fn(). Block throttle does not call back into unplug_fn() or into md. So there is no need to cancel blk throttle work. In fact I think cancelling block throttle work is bad because it might happen that some bios are throttled and scheduled to be dispatched later with the help of pending work and if work is cancelled, these bios might never be dispatched. Block layer also uses blk_sync_queue() during blk_cleanup_queue() and blk_release_queue() time. That should be safe as we are also calling blk_throtl_exit() which should make sure all the throttling related data structures are cleaned up. Signed-off-by: Vivek Goyal <vgoyal@redhat.com> Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
2011-03-03 01:05:33 +01:00
static void throtl_shutdown_wq(struct request_queue *q)
{
struct throtl_data *td = q->td;
cancel_work_sync(&td->dispatch_work);
}
static struct blkcg_policy blkcg_policy_throtl = {
.pd_size = sizeof(struct throtl_grp),
.cftypes = throtl_files,
.pd_init_fn = throtl_pd_init,
.pd_online_fn = throtl_pd_online,
.pd_exit_fn = throtl_pd_exit,
.pd_reset_stats_fn = throtl_pd_reset_stats,
};
bool blk_throtl_bio(struct request_queue *q, struct bio *bio)
{
struct throtl_data *td = q->td;
blk-throttle: add throtl_qnode for dispatch fairness With flat hierarchy, there's only single level of dispatching happening and fairness beyond that point is the responsibility of the rest of the block layer and driver, which usually works out okay; however, with the planned hierarchy support, service_queue->bio_lists[] can be filled up by bios from a single source. While the limits would still be honored, it'd be very easy to starve IOs from siblings or children. To avoid such starvation, this patch implements throtl_qnode and converts service_queue->bio_lists[] to lists of per-source qnodes which in turn contains the bio's. For example, when a bio is dispatched from a child group, the bio doesn't get queued on ->bio_lists[] directly but it first gets queued on the group's qnode which in turn gets queued on service_queue->queued[]. When dispatching for the upper level, the ->queued[] list is consumed in round-robing order so that the dispatch windows is consumed fairly by all IO sources. There are two ways a bio can come to a throtl_grp - directly queued to the group or dispatched from a child. For the former throtl_grp->qnode_on_self[rw] is used. For the latter, the child's ->qnode_on_parent[rw]. Note that this means that the child which is contributing a bio to its parent should stay pinned until all its bios are dispatched to its grand-parent. This patch moves blkg refcnting from bio add/remove spots to qnode activation/deactivation so that the blkg containing an active qnode is always pinned. As child pins the parent, this is sufficient for keeping the relevant sub-tree pinned while bios are in flight. The starvation issue was spotted by Vivek Goyal. v2: The original patch used the same throtl_grp->qnode_on_self/parent for reads and writes causing RWs to be queued incorrectly if there already are outstanding IOs in the other direction. They should be throtl_grp->qnode_on_self/parent[2] so that READs and WRITEs can use different qnodes. Spotted by Vivek Goyal. Signed-off-by: Tejun Heo <tj@kernel.org> Acked-by: Vivek Goyal <vgoyal@redhat.com>
2013-05-14 22:52:38 +02:00
struct throtl_qnode *qn = NULL;
struct throtl_grp *tg;
struct throtl_service_queue *sq;
bool rw = bio_data_dir(bio);
struct blkcg *blkcg;
bool throttled = false;
/* see throtl_charge_bio() */
if (bio->bi_rw & REQ_THROTTLED)
goto out;
/*
* A throtl_grp pointer retrieved under rcu can be used to access
* basic fields like stats and io rates. If a group has no rules,
* just update the dispatch stats in lockless manner and return.
*/
rcu_read_lock();
blkcg = bio_blkcg(bio);
blkcg: factor out blkio_group creation Currently both blk-throttle and cfq-iosched implement their own blkio_group creation code in throtl_get_tg() and cfq_get_cfqg(). This patch factors out the common code into blkg_lookup_create(), which returns ERR_PTR value so that transitional failures due to queue bypass can be distinguished from other failures. * New plkio_policy_ops methods blkio_alloc_group_fn() and blkio_link_group_fn added. Both are transitional and will be removed once the blkg management code is fully moved into blk-cgroup.c. * blkio_alloc_group_fn() allocates policy-specific blkg which is usually a larger data structure with blkg as the first entry and intiailizes it. Note that initialization of blkg proper, including percpu stats, is responsibility of blk-cgroup proper. Note that default config (weight, bps...) initialization is done from this method; otherwise, we end up violating locking order between blkcg and q locks via blkcg_get_CONF() functions. * blkio_link_group_fn() is called under queue_lock and responsible for linking the blkg to the queue. blkcg side is handled by blk-cgroup proper. * The common blkg creation function is named blkg_lookup_create() and blkiocg_lookup_group() is renamed to blkg_lookup() for consistency. Also, throtl / cfq related functions are similarly [re]named for consistency. This simplifies blkcg policy implementations and enables further cleanup. -v2: Vivek noticed that blkg_lookup_create() incorrectly tested blk_queue_dead() instead of blk_queue_bypass() leading a user of the function ending up creating a new blkg on bypassing queue. This is a bug introduced while relocating bypass patches before this one. Fixed. -v3: ERR_PTR patch folded into this one. @for_root added to blkg_lookup_create() to allow creating root group on a bypassed queue during elevator switch. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Vivek Goyal <vgoyal@redhat.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2012-03-05 22:15:06 +01:00
tg = throtl_lookup_tg(td, blkcg);
if (tg) {
if (!tg->has_rules[rw]) {
throtl_update_dispatch_stats(tg_to_blkg(tg),
block: Abstract out bvec iterator Immutable biovecs are going to require an explicit iterator. To implement immutable bvecs, a later patch is going to add a bi_bvec_done member to this struct; for now, this patch effectively just renames things. Signed-off-by: Kent Overstreet <kmo@daterainc.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <paulus@samba.org> Cc: "Ed L. Cashin" <ecashin@coraid.com> Cc: Nick Piggin <npiggin@kernel.dk> Cc: Lars Ellenberg <drbd-dev@lists.linbit.com> Cc: Jiri Kosina <jkosina@suse.cz> Cc: Matthew Wilcox <willy@linux.intel.com> Cc: Geoff Levand <geoff@infradead.org> Cc: Yehuda Sadeh <yehuda@inktank.com> Cc: Sage Weil <sage@inktank.com> Cc: Alex Elder <elder@inktank.com> Cc: ceph-devel@vger.kernel.org Cc: Joshua Morris <josh.h.morris@us.ibm.com> Cc: Philip Kelleher <pjk1939@linux.vnet.ibm.com> Cc: Rusty Russell <rusty@rustcorp.com.au> Cc: "Michael S. Tsirkin" <mst@redhat.com> Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com> Cc: Jeremy Fitzhardinge <jeremy@goop.org> Cc: Neil Brown <neilb@suse.de> Cc: Alasdair Kergon <agk@redhat.com> Cc: Mike Snitzer <snitzer@redhat.com> Cc: dm-devel@redhat.com Cc: Martin Schwidefsky <schwidefsky@de.ibm.com> Cc: Heiko Carstens <heiko.carstens@de.ibm.com> Cc: linux390@de.ibm.com Cc: Boaz Harrosh <bharrosh@panasas.com> Cc: Benny Halevy <bhalevy@tonian.com> Cc: "James E.J. Bottomley" <JBottomley@parallels.com> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: "Nicholas A. Bellinger" <nab@linux-iscsi.org> Cc: Alexander Viro <viro@zeniv.linux.org.uk> Cc: Chris Mason <chris.mason@fusionio.com> Cc: "Theodore Ts'o" <tytso@mit.edu> Cc: Andreas Dilger <adilger.kernel@dilger.ca> Cc: Jaegeuk Kim <jaegeuk.kim@samsung.com> Cc: Steven Whitehouse <swhiteho@redhat.com> Cc: Dave Kleikamp <shaggy@kernel.org> Cc: Joern Engel <joern@logfs.org> Cc: Prasad Joshi <prasadjoshi.linux@gmail.com> Cc: Trond Myklebust <Trond.Myklebust@netapp.com> Cc: KONISHI Ryusuke <konishi.ryusuke@lab.ntt.co.jp> Cc: Mark Fasheh <mfasheh@suse.com> Cc: Joel Becker <jlbec@evilplan.org> Cc: Ben Myers <bpm@sgi.com> Cc: xfs@oss.sgi.com Cc: Steven Rostedt <rostedt@goodmis.org> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Len Brown <len.brown@intel.com> Cc: Pavel Machek <pavel@ucw.cz> Cc: "Rafael J. Wysocki" <rjw@sisk.pl> Cc: Herton Ronaldo Krzesinski <herton.krzesinski@canonical.com> Cc: Ben Hutchings <ben@decadent.org.uk> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Guo Chao <yan@linux.vnet.ibm.com> Cc: Tejun Heo <tj@kernel.org> Cc: Asai Thambi S P <asamymuthupa@micron.com> Cc: Selvan Mani <smani@micron.com> Cc: Sam Bradshaw <sbradshaw@micron.com> Cc: Wei Yongjun <yongjun_wei@trendmicro.com.cn> Cc: "Roger Pau Monné" <roger.pau@citrix.com> Cc: Jan Beulich <jbeulich@suse.com> Cc: Stefano Stabellini <stefano.stabellini@eu.citrix.com> Cc: Ian Campbell <Ian.Campbell@citrix.com> Cc: Sebastian Ott <sebott@linux.vnet.ibm.com> Cc: Christian Borntraeger <borntraeger@de.ibm.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Jiang Liu <jiang.liu@huawei.com> Cc: Nitin Gupta <ngupta@vflare.org> Cc: Jerome Marchand <jmarchand@redhat.com> Cc: Joe Perches <joe@perches.com> Cc: Peng Tao <tao.peng@emc.com> Cc: Andy Adamson <andros@netapp.com> Cc: fanchaoting <fanchaoting@cn.fujitsu.com> Cc: Jie Liu <jeff.liu@oracle.com> Cc: Sunil Mushran <sunil.mushran@gmail.com> Cc: "Martin K. Petersen" <martin.petersen@oracle.com> Cc: Namjae Jeon <namjae.jeon@samsung.com> Cc: Pankaj Kumar <pankaj.km@samsung.com> Cc: Dan Magenheimer <dan.magenheimer@oracle.com> Cc: Mel Gorman <mgorman@suse.de>6
2013-10-12 00:44:27 +02:00
bio->bi_iter.bi_size, bio->bi_rw);
goto out_unlock_rcu;
}
}
/*
* Either group has not been allocated yet or it is not an unlimited
* IO group
*/
spin_lock_irq(q->queue_lock);
blkcg: factor out blkio_group creation Currently both blk-throttle and cfq-iosched implement their own blkio_group creation code in throtl_get_tg() and cfq_get_cfqg(). This patch factors out the common code into blkg_lookup_create(), which returns ERR_PTR value so that transitional failures due to queue bypass can be distinguished from other failures. * New plkio_policy_ops methods blkio_alloc_group_fn() and blkio_link_group_fn added. Both are transitional and will be removed once the blkg management code is fully moved into blk-cgroup.c. * blkio_alloc_group_fn() allocates policy-specific blkg which is usually a larger data structure with blkg as the first entry and intiailizes it. Note that initialization of blkg proper, including percpu stats, is responsibility of blk-cgroup proper. Note that default config (weight, bps...) initialization is done from this method; otherwise, we end up violating locking order between blkcg and q locks via blkcg_get_CONF() functions. * blkio_link_group_fn() is called under queue_lock and responsible for linking the blkg to the queue. blkcg side is handled by blk-cgroup proper. * The common blkg creation function is named blkg_lookup_create() and blkiocg_lookup_group() is renamed to blkg_lookup() for consistency. Also, throtl / cfq related functions are similarly [re]named for consistency. This simplifies blkcg policy implementations and enables further cleanup. -v2: Vivek noticed that blkg_lookup_create() incorrectly tested blk_queue_dead() instead of blk_queue_bypass() leading a user of the function ending up creating a new blkg on bypassing queue. This is a bug introduced while relocating bypass patches before this one. Fixed. -v3: ERR_PTR patch folded into this one. @for_root added to blkg_lookup_create() to allow creating root group on a bypassed queue during elevator switch. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Vivek Goyal <vgoyal@redhat.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2012-03-05 22:15:06 +01:00
tg = throtl_lookup_create_tg(td, blkcg);
if (unlikely(!tg))
goto out_unlock;
sq = &tg->service_queue;
while (true) {
/* throtl is FIFO - if bios are already queued, should queue */
if (sq->nr_queued[rw])
break;
/* if above limits, break to queue */
if (!tg_may_dispatch(tg, bio, NULL))
break;
/* within limits, let's charge and dispatch directly */
throtl_charge_bio(tg, bio);
/*
* We need to trim slice even when bios are not being queued
* otherwise it might happen that a bio is not queued for
* a long time and slice keeps on extending and trim is not
* called for a long time. Now if limits are reduced suddenly
* we take into account all the IO dispatched so far at new
* low rate and * newly queued IO gets a really long dispatch
* time.
*
* So keep on trimming slice even if bio is not queued.
*/
throtl_trim_slice(tg, rw);
/*
* @bio passed through this layer without being throttled.
* Climb up the ladder. If we''re already at the top, it
* can be executed directly.
*/
blk-throttle: add throtl_qnode for dispatch fairness With flat hierarchy, there's only single level of dispatching happening and fairness beyond that point is the responsibility of the rest of the block layer and driver, which usually works out okay; however, with the planned hierarchy support, service_queue->bio_lists[] can be filled up by bios from a single source. While the limits would still be honored, it'd be very easy to starve IOs from siblings or children. To avoid such starvation, this patch implements throtl_qnode and converts service_queue->bio_lists[] to lists of per-source qnodes which in turn contains the bio's. For example, when a bio is dispatched from a child group, the bio doesn't get queued on ->bio_lists[] directly but it first gets queued on the group's qnode which in turn gets queued on service_queue->queued[]. When dispatching for the upper level, the ->queued[] list is consumed in round-robing order so that the dispatch windows is consumed fairly by all IO sources. There are two ways a bio can come to a throtl_grp - directly queued to the group or dispatched from a child. For the former throtl_grp->qnode_on_self[rw] is used. For the latter, the child's ->qnode_on_parent[rw]. Note that this means that the child which is contributing a bio to its parent should stay pinned until all its bios are dispatched to its grand-parent. This patch moves blkg refcnting from bio add/remove spots to qnode activation/deactivation so that the blkg containing an active qnode is always pinned. As child pins the parent, this is sufficient for keeping the relevant sub-tree pinned while bios are in flight. The starvation issue was spotted by Vivek Goyal. v2: The original patch used the same throtl_grp->qnode_on_self/parent for reads and writes causing RWs to be queued incorrectly if there already are outstanding IOs in the other direction. They should be throtl_grp->qnode_on_self/parent[2] so that READs and WRITEs can use different qnodes. Spotted by Vivek Goyal. Signed-off-by: Tejun Heo <tj@kernel.org> Acked-by: Vivek Goyal <vgoyal@redhat.com>
2013-05-14 22:52:38 +02:00
qn = &tg->qnode_on_parent[rw];
sq = sq->parent_sq;
tg = sq_to_tg(sq);
if (!tg)
goto out_unlock;
}
/* out-of-limit, queue to @tg */
throtl_log(sq, "[%c] bio. bdisp=%llu sz=%u bps=%llu iodisp=%u iops=%u queued=%d/%d",
rw == READ ? 'R' : 'W',
block: Abstract out bvec iterator Immutable biovecs are going to require an explicit iterator. To implement immutable bvecs, a later patch is going to add a bi_bvec_done member to this struct; for now, this patch effectively just renames things. Signed-off-by: Kent Overstreet <kmo@daterainc.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <paulus@samba.org> Cc: "Ed L. Cashin" <ecashin@coraid.com> Cc: Nick Piggin <npiggin@kernel.dk> Cc: Lars Ellenberg <drbd-dev@lists.linbit.com> Cc: Jiri Kosina <jkosina@suse.cz> Cc: Matthew Wilcox <willy@linux.intel.com> Cc: Geoff Levand <geoff@infradead.org> Cc: Yehuda Sadeh <yehuda@inktank.com> Cc: Sage Weil <sage@inktank.com> Cc: Alex Elder <elder@inktank.com> Cc: ceph-devel@vger.kernel.org Cc: Joshua Morris <josh.h.morris@us.ibm.com> Cc: Philip Kelleher <pjk1939@linux.vnet.ibm.com> Cc: Rusty Russell <rusty@rustcorp.com.au> Cc: "Michael S. Tsirkin" <mst@redhat.com> Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com> Cc: Jeremy Fitzhardinge <jeremy@goop.org> Cc: Neil Brown <neilb@suse.de> Cc: Alasdair Kergon <agk@redhat.com> Cc: Mike Snitzer <snitzer@redhat.com> Cc: dm-devel@redhat.com Cc: Martin Schwidefsky <schwidefsky@de.ibm.com> Cc: Heiko Carstens <heiko.carstens@de.ibm.com> Cc: linux390@de.ibm.com Cc: Boaz Harrosh <bharrosh@panasas.com> Cc: Benny Halevy <bhalevy@tonian.com> Cc: "James E.J. Bottomley" <JBottomley@parallels.com> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: "Nicholas A. Bellinger" <nab@linux-iscsi.org> Cc: Alexander Viro <viro@zeniv.linux.org.uk> Cc: Chris Mason <chris.mason@fusionio.com> Cc: "Theodore Ts'o" <tytso@mit.edu> Cc: Andreas Dilger <adilger.kernel@dilger.ca> Cc: Jaegeuk Kim <jaegeuk.kim@samsung.com> Cc: Steven Whitehouse <swhiteho@redhat.com> Cc: Dave Kleikamp <shaggy@kernel.org> Cc: Joern Engel <joern@logfs.org> Cc: Prasad Joshi <prasadjoshi.linux@gmail.com> Cc: Trond Myklebust <Trond.Myklebust@netapp.com> Cc: KONISHI Ryusuke <konishi.ryusuke@lab.ntt.co.jp> Cc: Mark Fasheh <mfasheh@suse.com> Cc: Joel Becker <jlbec@evilplan.org> Cc: Ben Myers <bpm@sgi.com> Cc: xfs@oss.sgi.com Cc: Steven Rostedt <rostedt@goodmis.org> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Len Brown <len.brown@intel.com> Cc: Pavel Machek <pavel@ucw.cz> Cc: "Rafael J. Wysocki" <rjw@sisk.pl> Cc: Herton Ronaldo Krzesinski <herton.krzesinski@canonical.com> Cc: Ben Hutchings <ben@decadent.org.uk> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Guo Chao <yan@linux.vnet.ibm.com> Cc: Tejun Heo <tj@kernel.org> Cc: Asai Thambi S P <asamymuthupa@micron.com> Cc: Selvan Mani <smani@micron.com> Cc: Sam Bradshaw <sbradshaw@micron.com> Cc: Wei Yongjun <yongjun_wei@trendmicro.com.cn> Cc: "Roger Pau Monné" <roger.pau@citrix.com> Cc: Jan Beulich <jbeulich@suse.com> Cc: Stefano Stabellini <stefano.stabellini@eu.citrix.com> Cc: Ian Campbell <Ian.Campbell@citrix.com> Cc: Sebastian Ott <sebott@linux.vnet.ibm.com> Cc: Christian Borntraeger <borntraeger@de.ibm.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Jiang Liu <jiang.liu@huawei.com> Cc: Nitin Gupta <ngupta@vflare.org> Cc: Jerome Marchand <jmarchand@redhat.com> Cc: Joe Perches <joe@perches.com> Cc: Peng Tao <tao.peng@emc.com> Cc: Andy Adamson <andros@netapp.com> Cc: fanchaoting <fanchaoting@cn.fujitsu.com> Cc: Jie Liu <jeff.liu@oracle.com> Cc: Sunil Mushran <sunil.mushran@gmail.com> Cc: "Martin K. Petersen" <martin.petersen@oracle.com> Cc: Namjae Jeon <namjae.jeon@samsung.com> Cc: Pankaj Kumar <pankaj.km@samsung.com> Cc: Dan Magenheimer <dan.magenheimer@oracle.com> Cc: Mel Gorman <mgorman@suse.de>6
2013-10-12 00:44:27 +02:00
tg->bytes_disp[rw], bio->bi_iter.bi_size, tg->bps[rw],
tg->io_disp[rw], tg->iops[rw],
sq->nr_queued[READ], sq->nr_queued[WRITE]);
bio_associate_current(bio);
tg->td->nr_queued[rw]++;
blk-throttle: add throtl_qnode for dispatch fairness With flat hierarchy, there's only single level of dispatching happening and fairness beyond that point is the responsibility of the rest of the block layer and driver, which usually works out okay; however, with the planned hierarchy support, service_queue->bio_lists[] can be filled up by bios from a single source. While the limits would still be honored, it'd be very easy to starve IOs from siblings or children. To avoid such starvation, this patch implements throtl_qnode and converts service_queue->bio_lists[] to lists of per-source qnodes which in turn contains the bio's. For example, when a bio is dispatched from a child group, the bio doesn't get queued on ->bio_lists[] directly but it first gets queued on the group's qnode which in turn gets queued on service_queue->queued[]. When dispatching for the upper level, the ->queued[] list is consumed in round-robing order so that the dispatch windows is consumed fairly by all IO sources. There are two ways a bio can come to a throtl_grp - directly queued to the group or dispatched from a child. For the former throtl_grp->qnode_on_self[rw] is used. For the latter, the child's ->qnode_on_parent[rw]. Note that this means that the child which is contributing a bio to its parent should stay pinned until all its bios are dispatched to its grand-parent. This patch moves blkg refcnting from bio add/remove spots to qnode activation/deactivation so that the blkg containing an active qnode is always pinned. As child pins the parent, this is sufficient for keeping the relevant sub-tree pinned while bios are in flight. The starvation issue was spotted by Vivek Goyal. v2: The original patch used the same throtl_grp->qnode_on_self/parent for reads and writes causing RWs to be queued incorrectly if there already are outstanding IOs in the other direction. They should be throtl_grp->qnode_on_self/parent[2] so that READs and WRITEs can use different qnodes. Spotted by Vivek Goyal. Signed-off-by: Tejun Heo <tj@kernel.org> Acked-by: Vivek Goyal <vgoyal@redhat.com>
2013-05-14 22:52:38 +02:00
throtl_add_bio_tg(bio, qn, tg);
throttled = true;
/*
* Update @tg's dispatch time and force schedule dispatch if @tg
* was empty before @bio. The forced scheduling isn't likely to
* cause undue delay as @bio is likely to be dispatched directly if
* its @tg's disptime is not in the future.
*/
if (tg->flags & THROTL_TG_WAS_EMPTY) {
tg_update_disptime(tg);
throtl_schedule_next_dispatch(tg->service_queue.parent_sq, true);
}
out_unlock:
spin_unlock_irq(q->queue_lock);
out_unlock_rcu:
rcu_read_unlock();
out:
/*
* As multiple blk-throtls may stack in the same issue path, we
* don't want bios to leave with the flag set. Clear the flag if
* being issued.
*/
if (!throttled)
bio->bi_rw &= ~REQ_THROTTLED;
return throttled;
}
/*
* Dispatch all bios from all children tg's queued on @parent_sq. On
* return, @parent_sq is guaranteed to not have any active children tg's
* and all bios from previously active tg's are on @parent_sq->bio_lists[].
*/
static void tg_drain_bios(struct throtl_service_queue *parent_sq)
{
struct throtl_grp *tg;
while ((tg = throtl_rb_first(parent_sq))) {
struct throtl_service_queue *sq = &tg->service_queue;
struct bio *bio;
throtl_dequeue_tg(tg);
blk-throttle: add throtl_qnode for dispatch fairness With flat hierarchy, there's only single level of dispatching happening and fairness beyond that point is the responsibility of the rest of the block layer and driver, which usually works out okay; however, with the planned hierarchy support, service_queue->bio_lists[] can be filled up by bios from a single source. While the limits would still be honored, it'd be very easy to starve IOs from siblings or children. To avoid such starvation, this patch implements throtl_qnode and converts service_queue->bio_lists[] to lists of per-source qnodes which in turn contains the bio's. For example, when a bio is dispatched from a child group, the bio doesn't get queued on ->bio_lists[] directly but it first gets queued on the group's qnode which in turn gets queued on service_queue->queued[]. When dispatching for the upper level, the ->queued[] list is consumed in round-robing order so that the dispatch windows is consumed fairly by all IO sources. There are two ways a bio can come to a throtl_grp - directly queued to the group or dispatched from a child. For the former throtl_grp->qnode_on_self[rw] is used. For the latter, the child's ->qnode_on_parent[rw]. Note that this means that the child which is contributing a bio to its parent should stay pinned until all its bios are dispatched to its grand-parent. This patch moves blkg refcnting from bio add/remove spots to qnode activation/deactivation so that the blkg containing an active qnode is always pinned. As child pins the parent, this is sufficient for keeping the relevant sub-tree pinned while bios are in flight. The starvation issue was spotted by Vivek Goyal. v2: The original patch used the same throtl_grp->qnode_on_self/parent for reads and writes causing RWs to be queued incorrectly if there already are outstanding IOs in the other direction. They should be throtl_grp->qnode_on_self/parent[2] so that READs and WRITEs can use different qnodes. Spotted by Vivek Goyal. Signed-off-by: Tejun Heo <tj@kernel.org> Acked-by: Vivek Goyal <vgoyal@redhat.com>
2013-05-14 22:52:38 +02:00
while ((bio = throtl_peek_queued(&sq->queued[READ])))
tg_dispatch_one_bio(tg, bio_data_dir(bio));
blk-throttle: add throtl_qnode for dispatch fairness With flat hierarchy, there's only single level of dispatching happening and fairness beyond that point is the responsibility of the rest of the block layer and driver, which usually works out okay; however, with the planned hierarchy support, service_queue->bio_lists[] can be filled up by bios from a single source. While the limits would still be honored, it'd be very easy to starve IOs from siblings or children. To avoid such starvation, this patch implements throtl_qnode and converts service_queue->bio_lists[] to lists of per-source qnodes which in turn contains the bio's. For example, when a bio is dispatched from a child group, the bio doesn't get queued on ->bio_lists[] directly but it first gets queued on the group's qnode which in turn gets queued on service_queue->queued[]. When dispatching for the upper level, the ->queued[] list is consumed in round-robing order so that the dispatch windows is consumed fairly by all IO sources. There are two ways a bio can come to a throtl_grp - directly queued to the group or dispatched from a child. For the former throtl_grp->qnode_on_self[rw] is used. For the latter, the child's ->qnode_on_parent[rw]. Note that this means that the child which is contributing a bio to its parent should stay pinned until all its bios are dispatched to its grand-parent. This patch moves blkg refcnting from bio add/remove spots to qnode activation/deactivation so that the blkg containing an active qnode is always pinned. As child pins the parent, this is sufficient for keeping the relevant sub-tree pinned while bios are in flight. The starvation issue was spotted by Vivek Goyal. v2: The original patch used the same throtl_grp->qnode_on_self/parent for reads and writes causing RWs to be queued incorrectly if there already are outstanding IOs in the other direction. They should be throtl_grp->qnode_on_self/parent[2] so that READs and WRITEs can use different qnodes. Spotted by Vivek Goyal. Signed-off-by: Tejun Heo <tj@kernel.org> Acked-by: Vivek Goyal <vgoyal@redhat.com>
2013-05-14 22:52:38 +02:00
while ((bio = throtl_peek_queued(&sq->queued[WRITE])))
tg_dispatch_one_bio(tg, bio_data_dir(bio));
}
}
block: fix request_queue lifetime handling by making blk_queue_cleanup() properly shutdown request_queue is refcounted but actually depdends on lifetime management from the queue owner - on blk_cleanup_queue(), block layer expects that there's no request passing through request_queue and no new one will. This is fundamentally broken. The queue owner (e.g. SCSI layer) doesn't have a way to know whether there are other active users before calling blk_cleanup_queue() and other users (e.g. bsg) don't have any guarantee that the queue is and would stay valid while it's holding a reference. With delay added in blk_queue_bio() before queue_lock is grabbed, the following oops can be easily triggered when a device is removed with in-flight IOs. sd 0:0:1:0: [sdb] Stopping disk ata1.01: disabled general protection fault: 0000 [#1] PREEMPT SMP CPU 2 Modules linked in: Pid: 648, comm: test_rawio Not tainted 3.1.0-rc3-work+ #56 Bochs Bochs RIP: 0010:[<ffffffff8137d651>] [<ffffffff8137d651>] elv_rqhash_find+0x61/0x100 ... Process test_rawio (pid: 648, threadinfo ffff880019efa000, task ffff880019ef8a80) ... Call Trace: [<ffffffff8137d774>] elv_merge+0x84/0xe0 [<ffffffff81385b54>] blk_queue_bio+0xf4/0x400 [<ffffffff813838ea>] generic_make_request+0xca/0x100 [<ffffffff81383994>] submit_bio+0x74/0x100 [<ffffffff811c53ec>] dio_bio_submit+0xbc/0xc0 [<ffffffff811c610e>] __blockdev_direct_IO+0x92e/0xb40 [<ffffffff811c39f7>] blkdev_direct_IO+0x57/0x60 [<ffffffff8113b1c5>] generic_file_aio_read+0x6d5/0x760 [<ffffffff8118c1ca>] do_sync_read+0xda/0x120 [<ffffffff8118ce55>] vfs_read+0xc5/0x180 [<ffffffff8118cfaa>] sys_pread64+0x9a/0xb0 [<ffffffff81afaf6b>] system_call_fastpath+0x16/0x1b This happens because blk_queue_cleanup() destroys the queue and elevator whether IOs are in progress or not and DEAD tests are sprinkled in the request processing path without proper synchronization. Similar problem exists for blk-throtl. On queue cleanup, blk-throtl is shutdown whether it has requests in it or not. Depending on timing, it either oopses or throttled bios are lost putting tasks which are waiting for bio completion into eternal D state. The way it should work is having the usual clear distinction between shutdown and release. Shutdown drains all currently pending requests, marks the queue dead, and performs partial teardown of the now unnecessary part of the queue. Even after shutdown is complete, reference holders are still allowed to issue requests to the queue although they will be immmediately failed. The rest of teardown happens on release. This patch makes the following changes to make blk_queue_cleanup() behave as proper shutdown. * QUEUE_FLAG_DEAD is now set while holding both q->exit_mutex and queue_lock. * Unsynchronized DEAD check in generic_make_request_checks() removed. This couldn't make any meaningful difference as the queue could die after the check. * blk_drain_queue() updated such that it can drain all requests and is now called during cleanup. * blk_throtl updated such that it checks DEAD on grabbing queue_lock, drains all throttled bios during cleanup and free td when queue is released. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Vivek Goyal <vgoyal@redhat.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2011-10-19 14:42:16 +02:00
/**
* blk_throtl_drain - drain throttled bios
* @q: request_queue to drain throttled bios for
*
* Dispatch all currently throttled bios on @q through ->make_request_fn().
*/
void blk_throtl_drain(struct request_queue *q)
__releases(q->queue_lock) __acquires(q->queue_lock)
{
struct throtl_data *td = q->td;
struct blkcg_gq *blkg;
2013-08-09 02:11:25 +02:00
struct cgroup_subsys_state *pos_css;
block: fix request_queue lifetime handling by making blk_queue_cleanup() properly shutdown request_queue is refcounted but actually depdends on lifetime management from the queue owner - on blk_cleanup_queue(), block layer expects that there's no request passing through request_queue and no new one will. This is fundamentally broken. The queue owner (e.g. SCSI layer) doesn't have a way to know whether there are other active users before calling blk_cleanup_queue() and other users (e.g. bsg) don't have any guarantee that the queue is and would stay valid while it's holding a reference. With delay added in blk_queue_bio() before queue_lock is grabbed, the following oops can be easily triggered when a device is removed with in-flight IOs. sd 0:0:1:0: [sdb] Stopping disk ata1.01: disabled general protection fault: 0000 [#1] PREEMPT SMP CPU 2 Modules linked in: Pid: 648, comm: test_rawio Not tainted 3.1.0-rc3-work+ #56 Bochs Bochs RIP: 0010:[<ffffffff8137d651>] [<ffffffff8137d651>] elv_rqhash_find+0x61/0x100 ... Process test_rawio (pid: 648, threadinfo ffff880019efa000, task ffff880019ef8a80) ... Call Trace: [<ffffffff8137d774>] elv_merge+0x84/0xe0 [<ffffffff81385b54>] blk_queue_bio+0xf4/0x400 [<ffffffff813838ea>] generic_make_request+0xca/0x100 [<ffffffff81383994>] submit_bio+0x74/0x100 [<ffffffff811c53ec>] dio_bio_submit+0xbc/0xc0 [<ffffffff811c610e>] __blockdev_direct_IO+0x92e/0xb40 [<ffffffff811c39f7>] blkdev_direct_IO+0x57/0x60 [<ffffffff8113b1c5>] generic_file_aio_read+0x6d5/0x760 [<ffffffff8118c1ca>] do_sync_read+0xda/0x120 [<ffffffff8118ce55>] vfs_read+0xc5/0x180 [<ffffffff8118cfaa>] sys_pread64+0x9a/0xb0 [<ffffffff81afaf6b>] system_call_fastpath+0x16/0x1b This happens because blk_queue_cleanup() destroys the queue and elevator whether IOs are in progress or not and DEAD tests are sprinkled in the request processing path without proper synchronization. Similar problem exists for blk-throtl. On queue cleanup, blk-throtl is shutdown whether it has requests in it or not. Depending on timing, it either oopses or throttled bios are lost putting tasks which are waiting for bio completion into eternal D state. The way it should work is having the usual clear distinction between shutdown and release. Shutdown drains all currently pending requests, marks the queue dead, and performs partial teardown of the now unnecessary part of the queue. Even after shutdown is complete, reference holders are still allowed to issue requests to the queue although they will be immmediately failed. The rest of teardown happens on release. This patch makes the following changes to make blk_queue_cleanup() behave as proper shutdown. * QUEUE_FLAG_DEAD is now set while holding both q->exit_mutex and queue_lock. * Unsynchronized DEAD check in generic_make_request_checks() removed. This couldn't make any meaningful difference as the queue could die after the check. * blk_drain_queue() updated such that it can drain all requests and is now called during cleanup. * blk_throtl updated such that it checks DEAD on grabbing queue_lock, drains all throttled bios during cleanup and free td when queue is released. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Vivek Goyal <vgoyal@redhat.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2011-10-19 14:42:16 +02:00
struct bio *bio;
int rw;
block: fix request_queue lifetime handling by making blk_queue_cleanup() properly shutdown request_queue is refcounted but actually depdends on lifetime management from the queue owner - on blk_cleanup_queue(), block layer expects that there's no request passing through request_queue and no new one will. This is fundamentally broken. The queue owner (e.g. SCSI layer) doesn't have a way to know whether there are other active users before calling blk_cleanup_queue() and other users (e.g. bsg) don't have any guarantee that the queue is and would stay valid while it's holding a reference. With delay added in blk_queue_bio() before queue_lock is grabbed, the following oops can be easily triggered when a device is removed with in-flight IOs. sd 0:0:1:0: [sdb] Stopping disk ata1.01: disabled general protection fault: 0000 [#1] PREEMPT SMP CPU 2 Modules linked in: Pid: 648, comm: test_rawio Not tainted 3.1.0-rc3-work+ #56 Bochs Bochs RIP: 0010:[<ffffffff8137d651>] [<ffffffff8137d651>] elv_rqhash_find+0x61/0x100 ... Process test_rawio (pid: 648, threadinfo ffff880019efa000, task ffff880019ef8a80) ... Call Trace: [<ffffffff8137d774>] elv_merge+0x84/0xe0 [<ffffffff81385b54>] blk_queue_bio+0xf4/0x400 [<ffffffff813838ea>] generic_make_request+0xca/0x100 [<ffffffff81383994>] submit_bio+0x74/0x100 [<ffffffff811c53ec>] dio_bio_submit+0xbc/0xc0 [<ffffffff811c610e>] __blockdev_direct_IO+0x92e/0xb40 [<ffffffff811c39f7>] blkdev_direct_IO+0x57/0x60 [<ffffffff8113b1c5>] generic_file_aio_read+0x6d5/0x760 [<ffffffff8118c1ca>] do_sync_read+0xda/0x120 [<ffffffff8118ce55>] vfs_read+0xc5/0x180 [<ffffffff8118cfaa>] sys_pread64+0x9a/0xb0 [<ffffffff81afaf6b>] system_call_fastpath+0x16/0x1b This happens because blk_queue_cleanup() destroys the queue and elevator whether IOs are in progress or not and DEAD tests are sprinkled in the request processing path without proper synchronization. Similar problem exists for blk-throtl. On queue cleanup, blk-throtl is shutdown whether it has requests in it or not. Depending on timing, it either oopses or throttled bios are lost putting tasks which are waiting for bio completion into eternal D state. The way it should work is having the usual clear distinction between shutdown and release. Shutdown drains all currently pending requests, marks the queue dead, and performs partial teardown of the now unnecessary part of the queue. Even after shutdown is complete, reference holders are still allowed to issue requests to the queue although they will be immmediately failed. The rest of teardown happens on release. This patch makes the following changes to make blk_queue_cleanup() behave as proper shutdown. * QUEUE_FLAG_DEAD is now set while holding both q->exit_mutex and queue_lock. * Unsynchronized DEAD check in generic_make_request_checks() removed. This couldn't make any meaningful difference as the queue could die after the check. * blk_drain_queue() updated such that it can drain all requests and is now called during cleanup. * blk_throtl updated such that it checks DEAD on grabbing queue_lock, drains all throttled bios during cleanup and free td when queue is released. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Vivek Goyal <vgoyal@redhat.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2011-10-19 14:42:16 +02:00
queue_lockdep_assert_held(q);
rcu_read_lock();
block: fix request_queue lifetime handling by making blk_queue_cleanup() properly shutdown request_queue is refcounted but actually depdends on lifetime management from the queue owner - on blk_cleanup_queue(), block layer expects that there's no request passing through request_queue and no new one will. This is fundamentally broken. The queue owner (e.g. SCSI layer) doesn't have a way to know whether there are other active users before calling blk_cleanup_queue() and other users (e.g. bsg) don't have any guarantee that the queue is and would stay valid while it's holding a reference. With delay added in blk_queue_bio() before queue_lock is grabbed, the following oops can be easily triggered when a device is removed with in-flight IOs. sd 0:0:1:0: [sdb] Stopping disk ata1.01: disabled general protection fault: 0000 [#1] PREEMPT SMP CPU 2 Modules linked in: Pid: 648, comm: test_rawio Not tainted 3.1.0-rc3-work+ #56 Bochs Bochs RIP: 0010:[<ffffffff8137d651>] [<ffffffff8137d651>] elv_rqhash_find+0x61/0x100 ... Process test_rawio (pid: 648, threadinfo ffff880019efa000, task ffff880019ef8a80) ... Call Trace: [<ffffffff8137d774>] elv_merge+0x84/0xe0 [<ffffffff81385b54>] blk_queue_bio+0xf4/0x400 [<ffffffff813838ea>] generic_make_request+0xca/0x100 [<ffffffff81383994>] submit_bio+0x74/0x100 [<ffffffff811c53ec>] dio_bio_submit+0xbc/0xc0 [<ffffffff811c610e>] __blockdev_direct_IO+0x92e/0xb40 [<ffffffff811c39f7>] blkdev_direct_IO+0x57/0x60 [<ffffffff8113b1c5>] generic_file_aio_read+0x6d5/0x760 [<ffffffff8118c1ca>] do_sync_read+0xda/0x120 [<ffffffff8118ce55>] vfs_read+0xc5/0x180 [<ffffffff8118cfaa>] sys_pread64+0x9a/0xb0 [<ffffffff81afaf6b>] system_call_fastpath+0x16/0x1b This happens because blk_queue_cleanup() destroys the queue and elevator whether IOs are in progress or not and DEAD tests are sprinkled in the request processing path without proper synchronization. Similar problem exists for blk-throtl. On queue cleanup, blk-throtl is shutdown whether it has requests in it or not. Depending on timing, it either oopses or throttled bios are lost putting tasks which are waiting for bio completion into eternal D state. The way it should work is having the usual clear distinction between shutdown and release. Shutdown drains all currently pending requests, marks the queue dead, and performs partial teardown of the now unnecessary part of the queue. Even after shutdown is complete, reference holders are still allowed to issue requests to the queue although they will be immmediately failed. The rest of teardown happens on release. This patch makes the following changes to make blk_queue_cleanup() behave as proper shutdown. * QUEUE_FLAG_DEAD is now set while holding both q->exit_mutex and queue_lock. * Unsynchronized DEAD check in generic_make_request_checks() removed. This couldn't make any meaningful difference as the queue could die after the check. * blk_drain_queue() updated such that it can drain all requests and is now called during cleanup. * blk_throtl updated such that it checks DEAD on grabbing queue_lock, drains all throttled bios during cleanup and free td when queue is released. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Vivek Goyal <vgoyal@redhat.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2011-10-19 14:42:16 +02:00
/*
* Drain each tg while doing post-order walk on the blkg tree, so
* that all bios are propagated to td->service_queue. It'd be
* better to walk service_queue tree directly but blkg walk is
* easier.
*/
2013-08-09 02:11:25 +02:00
blkg_for_each_descendant_post(blkg, pos_css, td->queue->root_blkg)
tg_drain_bios(&blkg_to_tg(blkg)->service_queue);
/* finally, transfer bios from top-level tg's into the td */
tg_drain_bios(&td->service_queue);
rcu_read_unlock();
block: fix request_queue lifetime handling by making blk_queue_cleanup() properly shutdown request_queue is refcounted but actually depdends on lifetime management from the queue owner - on blk_cleanup_queue(), block layer expects that there's no request passing through request_queue and no new one will. This is fundamentally broken. The queue owner (e.g. SCSI layer) doesn't have a way to know whether there are other active users before calling blk_cleanup_queue() and other users (e.g. bsg) don't have any guarantee that the queue is and would stay valid while it's holding a reference. With delay added in blk_queue_bio() before queue_lock is grabbed, the following oops can be easily triggered when a device is removed with in-flight IOs. sd 0:0:1:0: [sdb] Stopping disk ata1.01: disabled general protection fault: 0000 [#1] PREEMPT SMP CPU 2 Modules linked in: Pid: 648, comm: test_rawio Not tainted 3.1.0-rc3-work+ #56 Bochs Bochs RIP: 0010:[<ffffffff8137d651>] [<ffffffff8137d651>] elv_rqhash_find+0x61/0x100 ... Process test_rawio (pid: 648, threadinfo ffff880019efa000, task ffff880019ef8a80) ... Call Trace: [<ffffffff8137d774>] elv_merge+0x84/0xe0 [<ffffffff81385b54>] blk_queue_bio+0xf4/0x400 [<ffffffff813838ea>] generic_make_request+0xca/0x100 [<ffffffff81383994>] submit_bio+0x74/0x100 [<ffffffff811c53ec>] dio_bio_submit+0xbc/0xc0 [<ffffffff811c610e>] __blockdev_direct_IO+0x92e/0xb40 [<ffffffff811c39f7>] blkdev_direct_IO+0x57/0x60 [<ffffffff8113b1c5>] generic_file_aio_read+0x6d5/0x760 [<ffffffff8118c1ca>] do_sync_read+0xda/0x120 [<ffffffff8118ce55>] vfs_read+0xc5/0x180 [<ffffffff8118cfaa>] sys_pread64+0x9a/0xb0 [<ffffffff81afaf6b>] system_call_fastpath+0x16/0x1b This happens because blk_queue_cleanup() destroys the queue and elevator whether IOs are in progress or not and DEAD tests are sprinkled in the request processing path without proper synchronization. Similar problem exists for blk-throtl. On queue cleanup, blk-throtl is shutdown whether it has requests in it or not. Depending on timing, it either oopses or throttled bios are lost putting tasks which are waiting for bio completion into eternal D state. The way it should work is having the usual clear distinction between shutdown and release. Shutdown drains all currently pending requests, marks the queue dead, and performs partial teardown of the now unnecessary part of the queue. Even after shutdown is complete, reference holders are still allowed to issue requests to the queue although they will be immmediately failed. The rest of teardown happens on release. This patch makes the following changes to make blk_queue_cleanup() behave as proper shutdown. * QUEUE_FLAG_DEAD is now set while holding both q->exit_mutex and queue_lock. * Unsynchronized DEAD check in generic_make_request_checks() removed. This couldn't make any meaningful difference as the queue could die after the check. * blk_drain_queue() updated such that it can drain all requests and is now called during cleanup. * blk_throtl updated such that it checks DEAD on grabbing queue_lock, drains all throttled bios during cleanup and free td when queue is released. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Vivek Goyal <vgoyal@redhat.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2011-10-19 14:42:16 +02:00
spin_unlock_irq(q->queue_lock);
/* all bios now should be in td->service_queue, issue them */
for (rw = READ; rw <= WRITE; rw++)
blk-throttle: add throtl_qnode for dispatch fairness With flat hierarchy, there's only single level of dispatching happening and fairness beyond that point is the responsibility of the rest of the block layer and driver, which usually works out okay; however, with the planned hierarchy support, service_queue->bio_lists[] can be filled up by bios from a single source. While the limits would still be honored, it'd be very easy to starve IOs from siblings or children. To avoid such starvation, this patch implements throtl_qnode and converts service_queue->bio_lists[] to lists of per-source qnodes which in turn contains the bio's. For example, when a bio is dispatched from a child group, the bio doesn't get queued on ->bio_lists[] directly but it first gets queued on the group's qnode which in turn gets queued on service_queue->queued[]. When dispatching for the upper level, the ->queued[] list is consumed in round-robing order so that the dispatch windows is consumed fairly by all IO sources. There are two ways a bio can come to a throtl_grp - directly queued to the group or dispatched from a child. For the former throtl_grp->qnode_on_self[rw] is used. For the latter, the child's ->qnode_on_parent[rw]. Note that this means that the child which is contributing a bio to its parent should stay pinned until all its bios are dispatched to its grand-parent. This patch moves blkg refcnting from bio add/remove spots to qnode activation/deactivation so that the blkg containing an active qnode is always pinned. As child pins the parent, this is sufficient for keeping the relevant sub-tree pinned while bios are in flight. The starvation issue was spotted by Vivek Goyal. v2: The original patch used the same throtl_grp->qnode_on_self/parent for reads and writes causing RWs to be queued incorrectly if there already are outstanding IOs in the other direction. They should be throtl_grp->qnode_on_self/parent[2] so that READs and WRITEs can use different qnodes. Spotted by Vivek Goyal. Signed-off-by: Tejun Heo <tj@kernel.org> Acked-by: Vivek Goyal <vgoyal@redhat.com>
2013-05-14 22:52:38 +02:00
while ((bio = throtl_pop_queued(&td->service_queue.queued[rw],
NULL)))
generic_make_request(bio);
block: fix request_queue lifetime handling by making blk_queue_cleanup() properly shutdown request_queue is refcounted but actually depdends on lifetime management from the queue owner - on blk_cleanup_queue(), block layer expects that there's no request passing through request_queue and no new one will. This is fundamentally broken. The queue owner (e.g. SCSI layer) doesn't have a way to know whether there are other active users before calling blk_cleanup_queue() and other users (e.g. bsg) don't have any guarantee that the queue is and would stay valid while it's holding a reference. With delay added in blk_queue_bio() before queue_lock is grabbed, the following oops can be easily triggered when a device is removed with in-flight IOs. sd 0:0:1:0: [sdb] Stopping disk ata1.01: disabled general protection fault: 0000 [#1] PREEMPT SMP CPU 2 Modules linked in: Pid: 648, comm: test_rawio Not tainted 3.1.0-rc3-work+ #56 Bochs Bochs RIP: 0010:[<ffffffff8137d651>] [<ffffffff8137d651>] elv_rqhash_find+0x61/0x100 ... Process test_rawio (pid: 648, threadinfo ffff880019efa000, task ffff880019ef8a80) ... Call Trace: [<ffffffff8137d774>] elv_merge+0x84/0xe0 [<ffffffff81385b54>] blk_queue_bio+0xf4/0x400 [<ffffffff813838ea>] generic_make_request+0xca/0x100 [<ffffffff81383994>] submit_bio+0x74/0x100 [<ffffffff811c53ec>] dio_bio_submit+0xbc/0xc0 [<ffffffff811c610e>] __blockdev_direct_IO+0x92e/0xb40 [<ffffffff811c39f7>] blkdev_direct_IO+0x57/0x60 [<ffffffff8113b1c5>] generic_file_aio_read+0x6d5/0x760 [<ffffffff8118c1ca>] do_sync_read+0xda/0x120 [<ffffffff8118ce55>] vfs_read+0xc5/0x180 [<ffffffff8118cfaa>] sys_pread64+0x9a/0xb0 [<ffffffff81afaf6b>] system_call_fastpath+0x16/0x1b This happens because blk_queue_cleanup() destroys the queue and elevator whether IOs are in progress or not and DEAD tests are sprinkled in the request processing path without proper synchronization. Similar problem exists for blk-throtl. On queue cleanup, blk-throtl is shutdown whether it has requests in it or not. Depending on timing, it either oopses or throttled bios are lost putting tasks which are waiting for bio completion into eternal D state. The way it should work is having the usual clear distinction between shutdown and release. Shutdown drains all currently pending requests, marks the queue dead, and performs partial teardown of the now unnecessary part of the queue. Even after shutdown is complete, reference holders are still allowed to issue requests to the queue although they will be immmediately failed. The rest of teardown happens on release. This patch makes the following changes to make blk_queue_cleanup() behave as proper shutdown. * QUEUE_FLAG_DEAD is now set while holding both q->exit_mutex and queue_lock. * Unsynchronized DEAD check in generic_make_request_checks() removed. This couldn't make any meaningful difference as the queue could die after the check. * blk_drain_queue() updated such that it can drain all requests and is now called during cleanup. * blk_throtl updated such that it checks DEAD on grabbing queue_lock, drains all throttled bios during cleanup and free td when queue is released. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Vivek Goyal <vgoyal@redhat.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2011-10-19 14:42:16 +02:00
spin_lock_irq(q->queue_lock);
}
int blk_throtl_init(struct request_queue *q)
{
struct throtl_data *td;
int ret;
td = kzalloc_node(sizeof(*td), GFP_KERNEL, q->node);
if (!td)
return -ENOMEM;
INIT_WORK(&td->dispatch_work, blk_throtl_dispatch_work_fn);
throtl_service_queue_init(&td->service_queue, NULL);
blkcg: factor out blkio_group creation Currently both blk-throttle and cfq-iosched implement their own blkio_group creation code in throtl_get_tg() and cfq_get_cfqg(). This patch factors out the common code into blkg_lookup_create(), which returns ERR_PTR value so that transitional failures due to queue bypass can be distinguished from other failures. * New plkio_policy_ops methods blkio_alloc_group_fn() and blkio_link_group_fn added. Both are transitional and will be removed once the blkg management code is fully moved into blk-cgroup.c. * blkio_alloc_group_fn() allocates policy-specific blkg which is usually a larger data structure with blkg as the first entry and intiailizes it. Note that initialization of blkg proper, including percpu stats, is responsibility of blk-cgroup proper. Note that default config (weight, bps...) initialization is done from this method; otherwise, we end up violating locking order between blkcg and q locks via blkcg_get_CONF() functions. * blkio_link_group_fn() is called under queue_lock and responsible for linking the blkg to the queue. blkcg side is handled by blk-cgroup proper. * The common blkg creation function is named blkg_lookup_create() and blkiocg_lookup_group() is renamed to blkg_lookup() for consistency. Also, throtl / cfq related functions are similarly [re]named for consistency. This simplifies blkcg policy implementations and enables further cleanup. -v2: Vivek noticed that blkg_lookup_create() incorrectly tested blk_queue_dead() instead of blk_queue_bypass() leading a user of the function ending up creating a new blkg on bypassing queue. This is a bug introduced while relocating bypass patches before this one. Fixed. -v3: ERR_PTR patch folded into this one. @for_root added to blkg_lookup_create() to allow creating root group on a bypassed queue during elevator switch. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Vivek Goyal <vgoyal@redhat.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2012-03-05 22:15:06 +01:00
q->td = td;
td->queue = q;
/* activate policy */
ret = blkcg_activate_policy(q, &blkcg_policy_throtl);
if (ret)
kfree(td);
return ret;
}
void blk_throtl_exit(struct request_queue *q)
{
BUG_ON(!q->td);
block: Move blk_throtl_exit() call to blk_cleanup_queue() Move blk_throtl_exit() in blk_cleanup_queue() as blk_throtl_exit() is written in such a way that it needs queue lock. In blk_release_queue() there is no gurantee that ->queue_lock is still around. Initially blk_throtl_exit() was in blk_cleanup_queue() but Ingo reported one problem. https://lkml.org/lkml/2010/10/23/86 And a quick fix moved blk_throtl_exit() to blk_release_queue(). commit 7ad58c028652753814054f4e3ac58f925e7343f4 Author: Jens Axboe <jaxboe@fusionio.com> Date: Sat Oct 23 20:40:26 2010 +0200 block: fix use-after-free bug in blk throttle code This patch reverts above change and does not try to shutdown the throtl work in blk_sync_queue(). By avoiding call to throtl_shutdown_timer_wq() from blk_sync_queue(), we should also avoid the problem reported by Ingo. blk_sync_queue() seems to be used only by md driver and it seems to be using it to make sure q->unplug_fn is not called as md registers its own unplug functions and it is about to free up the data structures used by unplug_fn(). Block throttle does not call back into unplug_fn() or into md. So there is no need to cancel blk throttle work. In fact I think cancelling block throttle work is bad because it might happen that some bios are throttled and scheduled to be dispatched later with the help of pending work and if work is cancelled, these bios might never be dispatched. Block layer also uses blk_sync_queue() during blk_cleanup_queue() and blk_release_queue() time. That should be safe as we are also calling blk_throtl_exit() which should make sure all the throttling related data structures are cleaned up. Signed-off-by: Vivek Goyal <vgoyal@redhat.com> Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
2011-03-03 01:05:33 +01:00
throtl_shutdown_wq(q);
blkcg_deactivate_policy(q, &blkcg_policy_throtl);
block: fix request_queue lifetime handling by making blk_queue_cleanup() properly shutdown request_queue is refcounted but actually depdends on lifetime management from the queue owner - on blk_cleanup_queue(), block layer expects that there's no request passing through request_queue and no new one will. This is fundamentally broken. The queue owner (e.g. SCSI layer) doesn't have a way to know whether there are other active users before calling blk_cleanup_queue() and other users (e.g. bsg) don't have any guarantee that the queue is and would stay valid while it's holding a reference. With delay added in blk_queue_bio() before queue_lock is grabbed, the following oops can be easily triggered when a device is removed with in-flight IOs. sd 0:0:1:0: [sdb] Stopping disk ata1.01: disabled general protection fault: 0000 [#1] PREEMPT SMP CPU 2 Modules linked in: Pid: 648, comm: test_rawio Not tainted 3.1.0-rc3-work+ #56 Bochs Bochs RIP: 0010:[<ffffffff8137d651>] [<ffffffff8137d651>] elv_rqhash_find+0x61/0x100 ... Process test_rawio (pid: 648, threadinfo ffff880019efa000, task ffff880019ef8a80) ... Call Trace: [<ffffffff8137d774>] elv_merge+0x84/0xe0 [<ffffffff81385b54>] blk_queue_bio+0xf4/0x400 [<ffffffff813838ea>] generic_make_request+0xca/0x100 [<ffffffff81383994>] submit_bio+0x74/0x100 [<ffffffff811c53ec>] dio_bio_submit+0xbc/0xc0 [<ffffffff811c610e>] __blockdev_direct_IO+0x92e/0xb40 [<ffffffff811c39f7>] blkdev_direct_IO+0x57/0x60 [<ffffffff8113b1c5>] generic_file_aio_read+0x6d5/0x760 [<ffffffff8118c1ca>] do_sync_read+0xda/0x120 [<ffffffff8118ce55>] vfs_read+0xc5/0x180 [<ffffffff8118cfaa>] sys_pread64+0x9a/0xb0 [<ffffffff81afaf6b>] system_call_fastpath+0x16/0x1b This happens because blk_queue_cleanup() destroys the queue and elevator whether IOs are in progress or not and DEAD tests are sprinkled in the request processing path without proper synchronization. Similar problem exists for blk-throtl. On queue cleanup, blk-throtl is shutdown whether it has requests in it or not. Depending on timing, it either oopses or throttled bios are lost putting tasks which are waiting for bio completion into eternal D state. The way it should work is having the usual clear distinction between shutdown and release. Shutdown drains all currently pending requests, marks the queue dead, and performs partial teardown of the now unnecessary part of the queue. Even after shutdown is complete, reference holders are still allowed to issue requests to the queue although they will be immmediately failed. The rest of teardown happens on release. This patch makes the following changes to make blk_queue_cleanup() behave as proper shutdown. * QUEUE_FLAG_DEAD is now set while holding both q->exit_mutex and queue_lock. * Unsynchronized DEAD check in generic_make_request_checks() removed. This couldn't make any meaningful difference as the queue could die after the check. * blk_drain_queue() updated such that it can drain all requests and is now called during cleanup. * blk_throtl updated such that it checks DEAD on grabbing queue_lock, drains all throttled bios during cleanup and free td when queue is released. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Vivek Goyal <vgoyal@redhat.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2011-10-19 14:42:16 +02:00
kfree(q->td);
}
static int __init throtl_init(void)
{
kthrotld_workqueue = alloc_workqueue("kthrotld", WQ_MEM_RECLAIM, 0);
if (!kthrotld_workqueue)
panic("Failed to create kthrotld\n");
return blkcg_policy_register(&blkcg_policy_throtl);
}
module_init(throtl_init);