linux/mm/backing-dev.c

776 lines
19 KiB
C
Raw Normal View History

#include <linux/wait.h>
#include <linux/backing-dev.h>
writeback: switch to per-bdi threads for flushing data This gets rid of pdflush for bdi writeout and kupdated style cleaning. pdflush writeout suffers from lack of locality and also requires more threads to handle the same workload, since it has to work in a non-blocking fashion against each queue. This also introduces lumpy behaviour and potential request starvation, since pdflush can be starved for queue access if others are accessing it. A sample ffsb workload that does random writes to files is about 8% faster here on a simple SATA drive during the benchmark phase. File layout also seems a LOT more smooth in vmstat: r b swpd free buff cache si so bi bo in cs us sy id wa 0 1 0 608848 2652 375372 0 0 0 71024 604 24 1 10 48 42 0 1 0 549644 2712 433736 0 0 0 60692 505 27 1 8 48 44 1 0 0 476928 2784 505192 0 0 4 29540 553 24 0 9 53 37 0 1 0 457972 2808 524008 0 0 0 54876 331 16 0 4 38 58 0 1 0 366128 2928 614284 0 0 4 92168 710 58 0 13 53 34 0 1 0 295092 3000 684140 0 0 0 62924 572 23 0 9 53 37 0 1 0 236592 3064 741704 0 0 4 58256 523 17 0 8 48 44 0 1 0 165608 3132 811464 0 0 0 57460 560 21 0 8 54 38 0 1 0 102952 3200 873164 0 0 4 74748 540 29 1 10 48 41 0 1 0 48604 3252 926472 0 0 0 53248 469 29 0 7 47 45 where vanilla tends to fluctuate a lot in the creation phase: r b swpd free buff cache si so bi bo in cs us sy id wa 1 1 0 678716 5792 303380 0 0 0 74064 565 50 1 11 52 36 1 0 0 662488 5864 319396 0 0 4 352 302 329 0 2 47 51 0 1 0 599312 5924 381468 0 0 0 78164 516 55 0 9 51 40 0 1 0 519952 6008 459516 0 0 4 78156 622 56 1 11 52 37 1 1 0 436640 6092 541632 0 0 0 82244 622 54 0 11 48 41 0 1 0 436640 6092 541660 0 0 0 8 152 39 0 0 51 49 0 1 0 332224 6200 644252 0 0 4 102800 728 46 1 13 49 36 1 0 0 274492 6260 701056 0 0 4 12328 459 49 0 7 50 43 0 1 0 211220 6324 763356 0 0 0 106940 515 37 1 10 51 39 1 0 0 160412 6376 813468 0 0 0 8224 415 43 0 6 49 45 1 1 0 85980 6452 886556 0 0 4 113516 575 39 1 11 54 34 0 2 0 85968 6452 886620 0 0 0 1640 158 211 0 0 46 54 A 10 disk test with btrfs performs 26% faster with per-bdi flushing. A SSD based writeback test on XFS performs over 20% better as well, with the throughput being very stable around 1GB/sec, where pdflush only manages 750MB/sec and fluctuates wildly while doing so. Random buffered writes to many files behave a lot better as well, as does random mmap'ed writes. A separate thread is added to sync the super blocks. In the long term, adding sync_supers_bdi() functionality could get rid of this thread again. Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-09-09 07:08:54 +00:00
#include <linux/kthread.h>
#include <linux/freezer.h>
#include <linux/fs.h>
#include <linux/pagemap.h>
writeback: switch to per-bdi threads for flushing data This gets rid of pdflush for bdi writeout and kupdated style cleaning. pdflush writeout suffers from lack of locality and also requires more threads to handle the same workload, since it has to work in a non-blocking fashion against each queue. This also introduces lumpy behaviour and potential request starvation, since pdflush can be starved for queue access if others are accessing it. A sample ffsb workload that does random writes to files is about 8% faster here on a simple SATA drive during the benchmark phase. File layout also seems a LOT more smooth in vmstat: r b swpd free buff cache si so bi bo in cs us sy id wa 0 1 0 608848 2652 375372 0 0 0 71024 604 24 1 10 48 42 0 1 0 549644 2712 433736 0 0 0 60692 505 27 1 8 48 44 1 0 0 476928 2784 505192 0 0 4 29540 553 24 0 9 53 37 0 1 0 457972 2808 524008 0 0 0 54876 331 16 0 4 38 58 0 1 0 366128 2928 614284 0 0 4 92168 710 58 0 13 53 34 0 1 0 295092 3000 684140 0 0 0 62924 572 23 0 9 53 37 0 1 0 236592 3064 741704 0 0 4 58256 523 17 0 8 48 44 0 1 0 165608 3132 811464 0 0 0 57460 560 21 0 8 54 38 0 1 0 102952 3200 873164 0 0 4 74748 540 29 1 10 48 41 0 1 0 48604 3252 926472 0 0 0 53248 469 29 0 7 47 45 where vanilla tends to fluctuate a lot in the creation phase: r b swpd free buff cache si so bi bo in cs us sy id wa 1 1 0 678716 5792 303380 0 0 0 74064 565 50 1 11 52 36 1 0 0 662488 5864 319396 0 0 4 352 302 329 0 2 47 51 0 1 0 599312 5924 381468 0 0 0 78164 516 55 0 9 51 40 0 1 0 519952 6008 459516 0 0 4 78156 622 56 1 11 52 37 1 1 0 436640 6092 541632 0 0 0 82244 622 54 0 11 48 41 0 1 0 436640 6092 541660 0 0 0 8 152 39 0 0 51 49 0 1 0 332224 6200 644252 0 0 4 102800 728 46 1 13 49 36 1 0 0 274492 6260 701056 0 0 4 12328 459 49 0 7 50 43 0 1 0 211220 6324 763356 0 0 0 106940 515 37 1 10 51 39 1 0 0 160412 6376 813468 0 0 0 8224 415 43 0 6 49 45 1 1 0 85980 6452 886556 0 0 4 113516 575 39 1 11 54 34 0 2 0 85968 6452 886620 0 0 0 1640 158 211 0 0 46 54 A 10 disk test with btrfs performs 26% faster with per-bdi flushing. A SSD based writeback test on XFS performs over 20% better as well, with the throughput being very stable around 1GB/sec, where pdflush only manages 750MB/sec and fluctuates wildly while doing so. Random buffered writes to many files behave a lot better as well, as does random mmap'ed writes. A separate thread is added to sync the super blocks. In the long term, adding sync_supers_bdi() functionality could get rid of this thread again. Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-09-09 07:08:54 +00:00
#include <linux/mm.h>
#include <linux/sched.h>
#include <linux/module.h>
#include <linux/writeback.h>
#include <linux/device.h>
#include <trace/events/writeback.h>
static atomic_long_t bdi_seq = ATOMIC_LONG_INIT(0);
void default_unplug_io_fn(struct backing_dev_info *bdi, struct page *page)
{
}
EXPORT_SYMBOL(default_unplug_io_fn);
struct backing_dev_info default_backing_dev_info = {
.name = "default",
.ra_pages = VM_MAX_READAHEAD * 1024 / PAGE_CACHE_SIZE,
.state = 0,
.capabilities = BDI_CAP_MAP_COPY,
.unplug_io_fn = default_unplug_io_fn,
};
EXPORT_SYMBOL_GPL(default_backing_dev_info);
struct backing_dev_info noop_backing_dev_info = {
.name = "noop",
};
EXPORT_SYMBOL_GPL(noop_backing_dev_info);
static struct class *bdi_class;
/*
* bdi_lock protects updates to bdi_list and bdi_pending_list, as well as
* reader side protection for bdi_pending_list. bdi_list has RCU reader side
* locking.
*/
writeback: switch to per-bdi threads for flushing data This gets rid of pdflush for bdi writeout and kupdated style cleaning. pdflush writeout suffers from lack of locality and also requires more threads to handle the same workload, since it has to work in a non-blocking fashion against each queue. This also introduces lumpy behaviour and potential request starvation, since pdflush can be starved for queue access if others are accessing it. A sample ffsb workload that does random writes to files is about 8% faster here on a simple SATA drive during the benchmark phase. File layout also seems a LOT more smooth in vmstat: r b swpd free buff cache si so bi bo in cs us sy id wa 0 1 0 608848 2652 375372 0 0 0 71024 604 24 1 10 48 42 0 1 0 549644 2712 433736 0 0 0 60692 505 27 1 8 48 44 1 0 0 476928 2784 505192 0 0 4 29540 553 24 0 9 53 37 0 1 0 457972 2808 524008 0 0 0 54876 331 16 0 4 38 58 0 1 0 366128 2928 614284 0 0 4 92168 710 58 0 13 53 34 0 1 0 295092 3000 684140 0 0 0 62924 572 23 0 9 53 37 0 1 0 236592 3064 741704 0 0 4 58256 523 17 0 8 48 44 0 1 0 165608 3132 811464 0 0 0 57460 560 21 0 8 54 38 0 1 0 102952 3200 873164 0 0 4 74748 540 29 1 10 48 41 0 1 0 48604 3252 926472 0 0 0 53248 469 29 0 7 47 45 where vanilla tends to fluctuate a lot in the creation phase: r b swpd free buff cache si so bi bo in cs us sy id wa 1 1 0 678716 5792 303380 0 0 0 74064 565 50 1 11 52 36 1 0 0 662488 5864 319396 0 0 4 352 302 329 0 2 47 51 0 1 0 599312 5924 381468 0 0 0 78164 516 55 0 9 51 40 0 1 0 519952 6008 459516 0 0 4 78156 622 56 1 11 52 37 1 1 0 436640 6092 541632 0 0 0 82244 622 54 0 11 48 41 0 1 0 436640 6092 541660 0 0 0 8 152 39 0 0 51 49 0 1 0 332224 6200 644252 0 0 4 102800 728 46 1 13 49 36 1 0 0 274492 6260 701056 0 0 4 12328 459 49 0 7 50 43 0 1 0 211220 6324 763356 0 0 0 106940 515 37 1 10 51 39 1 0 0 160412 6376 813468 0 0 0 8224 415 43 0 6 49 45 1 1 0 85980 6452 886556 0 0 4 113516 575 39 1 11 54 34 0 2 0 85968 6452 886620 0 0 0 1640 158 211 0 0 46 54 A 10 disk test with btrfs performs 26% faster with per-bdi flushing. A SSD based writeback test on XFS performs over 20% better as well, with the throughput being very stable around 1GB/sec, where pdflush only manages 750MB/sec and fluctuates wildly while doing so. Random buffered writes to many files behave a lot better as well, as does random mmap'ed writes. A separate thread is added to sync the super blocks. In the long term, adding sync_supers_bdi() functionality could get rid of this thread again. Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-09-09 07:08:54 +00:00
DEFINE_SPINLOCK(bdi_lock);
LIST_HEAD(bdi_list);
writeback: switch to per-bdi threads for flushing data This gets rid of pdflush for bdi writeout and kupdated style cleaning. pdflush writeout suffers from lack of locality and also requires more threads to handle the same workload, since it has to work in a non-blocking fashion against each queue. This also introduces lumpy behaviour and potential request starvation, since pdflush can be starved for queue access if others are accessing it. A sample ffsb workload that does random writes to files is about 8% faster here on a simple SATA drive during the benchmark phase. File layout also seems a LOT more smooth in vmstat: r b swpd free buff cache si so bi bo in cs us sy id wa 0 1 0 608848 2652 375372 0 0 0 71024 604 24 1 10 48 42 0 1 0 549644 2712 433736 0 0 0 60692 505 27 1 8 48 44 1 0 0 476928 2784 505192 0 0 4 29540 553 24 0 9 53 37 0 1 0 457972 2808 524008 0 0 0 54876 331 16 0 4 38 58 0 1 0 366128 2928 614284 0 0 4 92168 710 58 0 13 53 34 0 1 0 295092 3000 684140 0 0 0 62924 572 23 0 9 53 37 0 1 0 236592 3064 741704 0 0 4 58256 523 17 0 8 48 44 0 1 0 165608 3132 811464 0 0 0 57460 560 21 0 8 54 38 0 1 0 102952 3200 873164 0 0 4 74748 540 29 1 10 48 41 0 1 0 48604 3252 926472 0 0 0 53248 469 29 0 7 47 45 where vanilla tends to fluctuate a lot in the creation phase: r b swpd free buff cache si so bi bo in cs us sy id wa 1 1 0 678716 5792 303380 0 0 0 74064 565 50 1 11 52 36 1 0 0 662488 5864 319396 0 0 4 352 302 329 0 2 47 51 0 1 0 599312 5924 381468 0 0 0 78164 516 55 0 9 51 40 0 1 0 519952 6008 459516 0 0 4 78156 622 56 1 11 52 37 1 1 0 436640 6092 541632 0 0 0 82244 622 54 0 11 48 41 0 1 0 436640 6092 541660 0 0 0 8 152 39 0 0 51 49 0 1 0 332224 6200 644252 0 0 4 102800 728 46 1 13 49 36 1 0 0 274492 6260 701056 0 0 4 12328 459 49 0 7 50 43 0 1 0 211220 6324 763356 0 0 0 106940 515 37 1 10 51 39 1 0 0 160412 6376 813468 0 0 0 8224 415 43 0 6 49 45 1 1 0 85980 6452 886556 0 0 4 113516 575 39 1 11 54 34 0 2 0 85968 6452 886620 0 0 0 1640 158 211 0 0 46 54 A 10 disk test with btrfs performs 26% faster with per-bdi flushing. A SSD based writeback test on XFS performs over 20% better as well, with the throughput being very stable around 1GB/sec, where pdflush only manages 750MB/sec and fluctuates wildly while doing so. Random buffered writes to many files behave a lot better as well, as does random mmap'ed writes. A separate thread is added to sync the super blocks. In the long term, adding sync_supers_bdi() functionality could get rid of this thread again. Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-09-09 07:08:54 +00:00
LIST_HEAD(bdi_pending_list);
static struct task_struct *sync_supers_tsk;
static struct timer_list sync_supers_timer;
static int bdi_sync_supers(void *);
static void sync_supers_timer_fn(unsigned long);
#ifdef CONFIG_DEBUG_FS
#include <linux/debugfs.h>
#include <linux/seq_file.h>
static struct dentry *bdi_debug_root;
static void bdi_debug_init(void)
{
bdi_debug_root = debugfs_create_dir("bdi", NULL);
}
static int bdi_debug_stats_show(struct seq_file *m, void *v)
{
struct backing_dev_info *bdi = m->private;
struct bdi_writeback *wb = &bdi->wb;
unsigned long background_thresh;
unsigned long dirty_thresh;
unsigned long bdi_thresh;
unsigned long nr_dirty, nr_io, nr_more_io, nr_wb;
struct inode *inode;
nr_wb = nr_dirty = nr_io = nr_more_io = 0;
spin_lock(&inode_lock);
list_for_each_entry(inode, &wb->b_dirty, i_list)
nr_dirty++;
list_for_each_entry(inode, &wb->b_io, i_list)
nr_io++;
list_for_each_entry(inode, &wb->b_more_io, i_list)
nr_more_io++;
spin_unlock(&inode_lock);
get_dirty_limits(&background_thresh, &dirty_thresh, &bdi_thresh, bdi);
#define K(x) ((x) << (PAGE_SHIFT - 10))
seq_printf(m,
"BdiWriteback: %8lu kB\n"
"BdiReclaimable: %8lu kB\n"
"BdiDirtyThresh: %8lu kB\n"
"DirtyThresh: %8lu kB\n"
"BackgroundThresh: %8lu kB\n"
"b_dirty: %8lu\n"
"b_io: %8lu\n"
"b_more_io: %8lu\n"
"bdi_list: %8u\n"
"state: %8lx\n",
(unsigned long) K(bdi_stat(bdi, BDI_WRITEBACK)),
(unsigned long) K(bdi_stat(bdi, BDI_RECLAIMABLE)),
K(bdi_thresh), K(dirty_thresh),
K(background_thresh), nr_dirty, nr_io, nr_more_io,
!list_empty(&bdi->bdi_list), bdi->state);
#undef K
return 0;
}
static int bdi_debug_stats_open(struct inode *inode, struct file *file)
{
return single_open(file, bdi_debug_stats_show, inode->i_private);
}
static const struct file_operations bdi_debug_stats_fops = {
.open = bdi_debug_stats_open,
.read = seq_read,
.llseek = seq_lseek,
.release = single_release,
};
static void bdi_debug_register(struct backing_dev_info *bdi, const char *name)
{
bdi->debug_dir = debugfs_create_dir(name, bdi_debug_root);
bdi->debug_stats = debugfs_create_file("stats", 0444, bdi->debug_dir,
bdi, &bdi_debug_stats_fops);
}
static void bdi_debug_unregister(struct backing_dev_info *bdi)
{
debugfs_remove(bdi->debug_stats);
debugfs_remove(bdi->debug_dir);
}
#else
static inline void bdi_debug_init(void)
{
}
static inline void bdi_debug_register(struct backing_dev_info *bdi,
const char *name)
{
}
static inline void bdi_debug_unregister(struct backing_dev_info *bdi)
{
}
#endif
static ssize_t read_ahead_kb_store(struct device *dev,
struct device_attribute *attr,
const char *buf, size_t count)
{
struct backing_dev_info *bdi = dev_get_drvdata(dev);
char *end;
unsigned long read_ahead_kb;
ssize_t ret = -EINVAL;
read_ahead_kb = simple_strtoul(buf, &end, 10);
if (*buf && (end[0] == '\0' || (end[0] == '\n' && end[1] == '\0'))) {
bdi->ra_pages = read_ahead_kb >> (PAGE_SHIFT - 10);
ret = count;
}
return ret;
}
#define K(pages) ((pages) << (PAGE_SHIFT - 10))
#define BDI_SHOW(name, expr) \
static ssize_t name##_show(struct device *dev, \
struct device_attribute *attr, char *page) \
{ \
struct backing_dev_info *bdi = dev_get_drvdata(dev); \
\
return snprintf(page, PAGE_SIZE-1, "%lld\n", (long long)expr); \
}
BDI_SHOW(read_ahead_kb, K(bdi->ra_pages))
static ssize_t min_ratio_store(struct device *dev,
struct device_attribute *attr, const char *buf, size_t count)
{
struct backing_dev_info *bdi = dev_get_drvdata(dev);
char *end;
unsigned int ratio;
ssize_t ret = -EINVAL;
ratio = simple_strtoul(buf, &end, 10);
if (*buf && (end[0] == '\0' || (end[0] == '\n' && end[1] == '\0'))) {
ret = bdi_set_min_ratio(bdi, ratio);
if (!ret)
ret = count;
}
return ret;
}
BDI_SHOW(min_ratio, bdi->min_ratio)
static ssize_t max_ratio_store(struct device *dev,
struct device_attribute *attr, const char *buf, size_t count)
{
struct backing_dev_info *bdi = dev_get_drvdata(dev);
char *end;
unsigned int ratio;
ssize_t ret = -EINVAL;
ratio = simple_strtoul(buf, &end, 10);
if (*buf && (end[0] == '\0' || (end[0] == '\n' && end[1] == '\0'))) {
ret = bdi_set_max_ratio(bdi, ratio);
if (!ret)
ret = count;
}
return ret;
}
BDI_SHOW(max_ratio, bdi->max_ratio)
#define __ATTR_RW(attr) __ATTR(attr, 0644, attr##_show, attr##_store)
static struct device_attribute bdi_dev_attrs[] = {
__ATTR_RW(read_ahead_kb),
__ATTR_RW(min_ratio),
__ATTR_RW(max_ratio),
__ATTR_NULL,
};
static __init int bdi_class_init(void)
{
bdi_class = class_create(THIS_MODULE, "bdi");
if (IS_ERR(bdi_class))
return PTR_ERR(bdi_class);
bdi_class->dev_attrs = bdi_dev_attrs;
bdi_debug_init();
return 0;
}
postcore_initcall(bdi_class_init);
static int __init default_bdi_init(void)
{
int err;
writeback: switch to per-bdi threads for flushing data This gets rid of pdflush for bdi writeout and kupdated style cleaning. pdflush writeout suffers from lack of locality and also requires more threads to handle the same workload, since it has to work in a non-blocking fashion against each queue. This also introduces lumpy behaviour and potential request starvation, since pdflush can be starved for queue access if others are accessing it. A sample ffsb workload that does random writes to files is about 8% faster here on a simple SATA drive during the benchmark phase. File layout also seems a LOT more smooth in vmstat: r b swpd free buff cache si so bi bo in cs us sy id wa 0 1 0 608848 2652 375372 0 0 0 71024 604 24 1 10 48 42 0 1 0 549644 2712 433736 0 0 0 60692 505 27 1 8 48 44 1 0 0 476928 2784 505192 0 0 4 29540 553 24 0 9 53 37 0 1 0 457972 2808 524008 0 0 0 54876 331 16 0 4 38 58 0 1 0 366128 2928 614284 0 0 4 92168 710 58 0 13 53 34 0 1 0 295092 3000 684140 0 0 0 62924 572 23 0 9 53 37 0 1 0 236592 3064 741704 0 0 4 58256 523 17 0 8 48 44 0 1 0 165608 3132 811464 0 0 0 57460 560 21 0 8 54 38 0 1 0 102952 3200 873164 0 0 4 74748 540 29 1 10 48 41 0 1 0 48604 3252 926472 0 0 0 53248 469 29 0 7 47 45 where vanilla tends to fluctuate a lot in the creation phase: r b swpd free buff cache si so bi bo in cs us sy id wa 1 1 0 678716 5792 303380 0 0 0 74064 565 50 1 11 52 36 1 0 0 662488 5864 319396 0 0 4 352 302 329 0 2 47 51 0 1 0 599312 5924 381468 0 0 0 78164 516 55 0 9 51 40 0 1 0 519952 6008 459516 0 0 4 78156 622 56 1 11 52 37 1 1 0 436640 6092 541632 0 0 0 82244 622 54 0 11 48 41 0 1 0 436640 6092 541660 0 0 0 8 152 39 0 0 51 49 0 1 0 332224 6200 644252 0 0 4 102800 728 46 1 13 49 36 1 0 0 274492 6260 701056 0 0 4 12328 459 49 0 7 50 43 0 1 0 211220 6324 763356 0 0 0 106940 515 37 1 10 51 39 1 0 0 160412 6376 813468 0 0 0 8224 415 43 0 6 49 45 1 1 0 85980 6452 886556 0 0 4 113516 575 39 1 11 54 34 0 2 0 85968 6452 886620 0 0 0 1640 158 211 0 0 46 54 A 10 disk test with btrfs performs 26% faster with per-bdi flushing. A SSD based writeback test on XFS performs over 20% better as well, with the throughput being very stable around 1GB/sec, where pdflush only manages 750MB/sec and fluctuates wildly while doing so. Random buffered writes to many files behave a lot better as well, as does random mmap'ed writes. A separate thread is added to sync the super blocks. In the long term, adding sync_supers_bdi() functionality could get rid of this thread again. Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-09-09 07:08:54 +00:00
sync_supers_tsk = kthread_run(bdi_sync_supers, NULL, "sync_supers");
BUG_ON(IS_ERR(sync_supers_tsk));
setup_timer(&sync_supers_timer, sync_supers_timer_fn, 0);
bdi_arm_supers_timer();
writeback: switch to per-bdi threads for flushing data This gets rid of pdflush for bdi writeout and kupdated style cleaning. pdflush writeout suffers from lack of locality and also requires more threads to handle the same workload, since it has to work in a non-blocking fashion against each queue. This also introduces lumpy behaviour and potential request starvation, since pdflush can be starved for queue access if others are accessing it. A sample ffsb workload that does random writes to files is about 8% faster here on a simple SATA drive during the benchmark phase. File layout also seems a LOT more smooth in vmstat: r b swpd free buff cache si so bi bo in cs us sy id wa 0 1 0 608848 2652 375372 0 0 0 71024 604 24 1 10 48 42 0 1 0 549644 2712 433736 0 0 0 60692 505 27 1 8 48 44 1 0 0 476928 2784 505192 0 0 4 29540 553 24 0 9 53 37 0 1 0 457972 2808 524008 0 0 0 54876 331 16 0 4 38 58 0 1 0 366128 2928 614284 0 0 4 92168 710 58 0 13 53 34 0 1 0 295092 3000 684140 0 0 0 62924 572 23 0 9 53 37 0 1 0 236592 3064 741704 0 0 4 58256 523 17 0 8 48 44 0 1 0 165608 3132 811464 0 0 0 57460 560 21 0 8 54 38 0 1 0 102952 3200 873164 0 0 4 74748 540 29 1 10 48 41 0 1 0 48604 3252 926472 0 0 0 53248 469 29 0 7 47 45 where vanilla tends to fluctuate a lot in the creation phase: r b swpd free buff cache si so bi bo in cs us sy id wa 1 1 0 678716 5792 303380 0 0 0 74064 565 50 1 11 52 36 1 0 0 662488 5864 319396 0 0 4 352 302 329 0 2 47 51 0 1 0 599312 5924 381468 0 0 0 78164 516 55 0 9 51 40 0 1 0 519952 6008 459516 0 0 4 78156 622 56 1 11 52 37 1 1 0 436640 6092 541632 0 0 0 82244 622 54 0 11 48 41 0 1 0 436640 6092 541660 0 0 0 8 152 39 0 0 51 49 0 1 0 332224 6200 644252 0 0 4 102800 728 46 1 13 49 36 1 0 0 274492 6260 701056 0 0 4 12328 459 49 0 7 50 43 0 1 0 211220 6324 763356 0 0 0 106940 515 37 1 10 51 39 1 0 0 160412 6376 813468 0 0 0 8224 415 43 0 6 49 45 1 1 0 85980 6452 886556 0 0 4 113516 575 39 1 11 54 34 0 2 0 85968 6452 886620 0 0 0 1640 158 211 0 0 46 54 A 10 disk test with btrfs performs 26% faster with per-bdi flushing. A SSD based writeback test on XFS performs over 20% better as well, with the throughput being very stable around 1GB/sec, where pdflush only manages 750MB/sec and fluctuates wildly while doing so. Random buffered writes to many files behave a lot better as well, as does random mmap'ed writes. A separate thread is added to sync the super blocks. In the long term, adding sync_supers_bdi() functionality could get rid of this thread again. Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-09-09 07:08:54 +00:00
err = bdi_init(&default_backing_dev_info);
if (!err)
bdi_register(&default_backing_dev_info, NULL, "default");
return err;
}
subsys_initcall(default_bdi_init);
writeback: switch to per-bdi threads for flushing data This gets rid of pdflush for bdi writeout and kupdated style cleaning. pdflush writeout suffers from lack of locality and also requires more threads to handle the same workload, since it has to work in a non-blocking fashion against each queue. This also introduces lumpy behaviour and potential request starvation, since pdflush can be starved for queue access if others are accessing it. A sample ffsb workload that does random writes to files is about 8% faster here on a simple SATA drive during the benchmark phase. File layout also seems a LOT more smooth in vmstat: r b swpd free buff cache si so bi bo in cs us sy id wa 0 1 0 608848 2652 375372 0 0 0 71024 604 24 1 10 48 42 0 1 0 549644 2712 433736 0 0 0 60692 505 27 1 8 48 44 1 0 0 476928 2784 505192 0 0 4 29540 553 24 0 9 53 37 0 1 0 457972 2808 524008 0 0 0 54876 331 16 0 4 38 58 0 1 0 366128 2928 614284 0 0 4 92168 710 58 0 13 53 34 0 1 0 295092 3000 684140 0 0 0 62924 572 23 0 9 53 37 0 1 0 236592 3064 741704 0 0 4 58256 523 17 0 8 48 44 0 1 0 165608 3132 811464 0 0 0 57460 560 21 0 8 54 38 0 1 0 102952 3200 873164 0 0 4 74748 540 29 1 10 48 41 0 1 0 48604 3252 926472 0 0 0 53248 469 29 0 7 47 45 where vanilla tends to fluctuate a lot in the creation phase: r b swpd free buff cache si so bi bo in cs us sy id wa 1 1 0 678716 5792 303380 0 0 0 74064 565 50 1 11 52 36 1 0 0 662488 5864 319396 0 0 4 352 302 329 0 2 47 51 0 1 0 599312 5924 381468 0 0 0 78164 516 55 0 9 51 40 0 1 0 519952 6008 459516 0 0 4 78156 622 56 1 11 52 37 1 1 0 436640 6092 541632 0 0 0 82244 622 54 0 11 48 41 0 1 0 436640 6092 541660 0 0 0 8 152 39 0 0 51 49 0 1 0 332224 6200 644252 0 0 4 102800 728 46 1 13 49 36 1 0 0 274492 6260 701056 0 0 4 12328 459 49 0 7 50 43 0 1 0 211220 6324 763356 0 0 0 106940 515 37 1 10 51 39 1 0 0 160412 6376 813468 0 0 0 8224 415 43 0 6 49 45 1 1 0 85980 6452 886556 0 0 4 113516 575 39 1 11 54 34 0 2 0 85968 6452 886620 0 0 0 1640 158 211 0 0 46 54 A 10 disk test with btrfs performs 26% faster with per-bdi flushing. A SSD based writeback test on XFS performs over 20% better as well, with the throughput being very stable around 1GB/sec, where pdflush only manages 750MB/sec and fluctuates wildly while doing so. Random buffered writes to many files behave a lot better as well, as does random mmap'ed writes. A separate thread is added to sync the super blocks. In the long term, adding sync_supers_bdi() functionality could get rid of this thread again. Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-09-09 07:08:54 +00:00
int bdi_has_dirty_io(struct backing_dev_info *bdi)
{
return wb_has_dirty_io(&bdi->wb);
}
static void bdi_flush_io(struct backing_dev_info *bdi)
{
struct writeback_control wbc = {
.sync_mode = WB_SYNC_NONE,
.older_than_this = NULL,
.range_cyclic = 1,
.nr_to_write = 1024,
};
writeback_inodes_wb(&bdi->wb, &wbc);
writeback: switch to per-bdi threads for flushing data This gets rid of pdflush for bdi writeout and kupdated style cleaning. pdflush writeout suffers from lack of locality and also requires more threads to handle the same workload, since it has to work in a non-blocking fashion against each queue. This also introduces lumpy behaviour and potential request starvation, since pdflush can be starved for queue access if others are accessing it. A sample ffsb workload that does random writes to files is about 8% faster here on a simple SATA drive during the benchmark phase. File layout also seems a LOT more smooth in vmstat: r b swpd free buff cache si so bi bo in cs us sy id wa 0 1 0 608848 2652 375372 0 0 0 71024 604 24 1 10 48 42 0 1 0 549644 2712 433736 0 0 0 60692 505 27 1 8 48 44 1 0 0 476928 2784 505192 0 0 4 29540 553 24 0 9 53 37 0 1 0 457972 2808 524008 0 0 0 54876 331 16 0 4 38 58 0 1 0 366128 2928 614284 0 0 4 92168 710 58 0 13 53 34 0 1 0 295092 3000 684140 0 0 0 62924 572 23 0 9 53 37 0 1 0 236592 3064 741704 0 0 4 58256 523 17 0 8 48 44 0 1 0 165608 3132 811464 0 0 0 57460 560 21 0 8 54 38 0 1 0 102952 3200 873164 0 0 4 74748 540 29 1 10 48 41 0 1 0 48604 3252 926472 0 0 0 53248 469 29 0 7 47 45 where vanilla tends to fluctuate a lot in the creation phase: r b swpd free buff cache si so bi bo in cs us sy id wa 1 1 0 678716 5792 303380 0 0 0 74064 565 50 1 11 52 36 1 0 0 662488 5864 319396 0 0 4 352 302 329 0 2 47 51 0 1 0 599312 5924 381468 0 0 0 78164 516 55 0 9 51 40 0 1 0 519952 6008 459516 0 0 4 78156 622 56 1 11 52 37 1 1 0 436640 6092 541632 0 0 0 82244 622 54 0 11 48 41 0 1 0 436640 6092 541660 0 0 0 8 152 39 0 0 51 49 0 1 0 332224 6200 644252 0 0 4 102800 728 46 1 13 49 36 1 0 0 274492 6260 701056 0 0 4 12328 459 49 0 7 50 43 0 1 0 211220 6324 763356 0 0 0 106940 515 37 1 10 51 39 1 0 0 160412 6376 813468 0 0 0 8224 415 43 0 6 49 45 1 1 0 85980 6452 886556 0 0 4 113516 575 39 1 11 54 34 0 2 0 85968 6452 886620 0 0 0 1640 158 211 0 0 46 54 A 10 disk test with btrfs performs 26% faster with per-bdi flushing. A SSD based writeback test on XFS performs over 20% better as well, with the throughput being very stable around 1GB/sec, where pdflush only manages 750MB/sec and fluctuates wildly while doing so. Random buffered writes to many files behave a lot better as well, as does random mmap'ed writes. A separate thread is added to sync the super blocks. In the long term, adding sync_supers_bdi() functionality could get rid of this thread again. Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-09-09 07:08:54 +00:00
}
/*
* kupdated() used to do this. We cannot do it from the bdi_forker_thread()
writeback: switch to per-bdi threads for flushing data This gets rid of pdflush for bdi writeout and kupdated style cleaning. pdflush writeout suffers from lack of locality and also requires more threads to handle the same workload, since it has to work in a non-blocking fashion against each queue. This also introduces lumpy behaviour and potential request starvation, since pdflush can be starved for queue access if others are accessing it. A sample ffsb workload that does random writes to files is about 8% faster here on a simple SATA drive during the benchmark phase. File layout also seems a LOT more smooth in vmstat: r b swpd free buff cache si so bi bo in cs us sy id wa 0 1 0 608848 2652 375372 0 0 0 71024 604 24 1 10 48 42 0 1 0 549644 2712 433736 0 0 0 60692 505 27 1 8 48 44 1 0 0 476928 2784 505192 0 0 4 29540 553 24 0 9 53 37 0 1 0 457972 2808 524008 0 0 0 54876 331 16 0 4 38 58 0 1 0 366128 2928 614284 0 0 4 92168 710 58 0 13 53 34 0 1 0 295092 3000 684140 0 0 0 62924 572 23 0 9 53 37 0 1 0 236592 3064 741704 0 0 4 58256 523 17 0 8 48 44 0 1 0 165608 3132 811464 0 0 0 57460 560 21 0 8 54 38 0 1 0 102952 3200 873164 0 0 4 74748 540 29 1 10 48 41 0 1 0 48604 3252 926472 0 0 0 53248 469 29 0 7 47 45 where vanilla tends to fluctuate a lot in the creation phase: r b swpd free buff cache si so bi bo in cs us sy id wa 1 1 0 678716 5792 303380 0 0 0 74064 565 50 1 11 52 36 1 0 0 662488 5864 319396 0 0 4 352 302 329 0 2 47 51 0 1 0 599312 5924 381468 0 0 0 78164 516 55 0 9 51 40 0 1 0 519952 6008 459516 0 0 4 78156 622 56 1 11 52 37 1 1 0 436640 6092 541632 0 0 0 82244 622 54 0 11 48 41 0 1 0 436640 6092 541660 0 0 0 8 152 39 0 0 51 49 0 1 0 332224 6200 644252 0 0 4 102800 728 46 1 13 49 36 1 0 0 274492 6260 701056 0 0 4 12328 459 49 0 7 50 43 0 1 0 211220 6324 763356 0 0 0 106940 515 37 1 10 51 39 1 0 0 160412 6376 813468 0 0 0 8224 415 43 0 6 49 45 1 1 0 85980 6452 886556 0 0 4 113516 575 39 1 11 54 34 0 2 0 85968 6452 886620 0 0 0 1640 158 211 0 0 46 54 A 10 disk test with btrfs performs 26% faster with per-bdi flushing. A SSD based writeback test on XFS performs over 20% better as well, with the throughput being very stable around 1GB/sec, where pdflush only manages 750MB/sec and fluctuates wildly while doing so. Random buffered writes to many files behave a lot better as well, as does random mmap'ed writes. A separate thread is added to sync the super blocks. In the long term, adding sync_supers_bdi() functionality could get rid of this thread again. Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-09-09 07:08:54 +00:00
* or we risk deadlocking on ->s_umount. The longer term solution would be
* to implement sync_supers_bdi() or similar and simply do it from the
* bdi writeback thread individually.
writeback: switch to per-bdi threads for flushing data This gets rid of pdflush for bdi writeout and kupdated style cleaning. pdflush writeout suffers from lack of locality and also requires more threads to handle the same workload, since it has to work in a non-blocking fashion against each queue. This also introduces lumpy behaviour and potential request starvation, since pdflush can be starved for queue access if others are accessing it. A sample ffsb workload that does random writes to files is about 8% faster here on a simple SATA drive during the benchmark phase. File layout also seems a LOT more smooth in vmstat: r b swpd free buff cache si so bi bo in cs us sy id wa 0 1 0 608848 2652 375372 0 0 0 71024 604 24 1 10 48 42 0 1 0 549644 2712 433736 0 0 0 60692 505 27 1 8 48 44 1 0 0 476928 2784 505192 0 0 4 29540 553 24 0 9 53 37 0 1 0 457972 2808 524008 0 0 0 54876 331 16 0 4 38 58 0 1 0 366128 2928 614284 0 0 4 92168 710 58 0 13 53 34 0 1 0 295092 3000 684140 0 0 0 62924 572 23 0 9 53 37 0 1 0 236592 3064 741704 0 0 4 58256 523 17 0 8 48 44 0 1 0 165608 3132 811464 0 0 0 57460 560 21 0 8 54 38 0 1 0 102952 3200 873164 0 0 4 74748 540 29 1 10 48 41 0 1 0 48604 3252 926472 0 0 0 53248 469 29 0 7 47 45 where vanilla tends to fluctuate a lot in the creation phase: r b swpd free buff cache si so bi bo in cs us sy id wa 1 1 0 678716 5792 303380 0 0 0 74064 565 50 1 11 52 36 1 0 0 662488 5864 319396 0 0 4 352 302 329 0 2 47 51 0 1 0 599312 5924 381468 0 0 0 78164 516 55 0 9 51 40 0 1 0 519952 6008 459516 0 0 4 78156 622 56 1 11 52 37 1 1 0 436640 6092 541632 0 0 0 82244 622 54 0 11 48 41 0 1 0 436640 6092 541660 0 0 0 8 152 39 0 0 51 49 0 1 0 332224 6200 644252 0 0 4 102800 728 46 1 13 49 36 1 0 0 274492 6260 701056 0 0 4 12328 459 49 0 7 50 43 0 1 0 211220 6324 763356 0 0 0 106940 515 37 1 10 51 39 1 0 0 160412 6376 813468 0 0 0 8224 415 43 0 6 49 45 1 1 0 85980 6452 886556 0 0 4 113516 575 39 1 11 54 34 0 2 0 85968 6452 886620 0 0 0 1640 158 211 0 0 46 54 A 10 disk test with btrfs performs 26% faster with per-bdi flushing. A SSD based writeback test on XFS performs over 20% better as well, with the throughput being very stable around 1GB/sec, where pdflush only manages 750MB/sec and fluctuates wildly while doing so. Random buffered writes to many files behave a lot better as well, as does random mmap'ed writes. A separate thread is added to sync the super blocks. In the long term, adding sync_supers_bdi() functionality could get rid of this thread again. Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-09-09 07:08:54 +00:00
*/
static int bdi_sync_supers(void *unused)
{
set_user_nice(current, 0);
while (!kthread_should_stop()) {
set_current_state(TASK_INTERRUPTIBLE);
schedule();
/*
* Do this periodically, like kupdated() did before.
*/
sync_supers();
}
return 0;
}
void bdi_arm_supers_timer(void)
writeback: switch to per-bdi threads for flushing data This gets rid of pdflush for bdi writeout and kupdated style cleaning. pdflush writeout suffers from lack of locality and also requires more threads to handle the same workload, since it has to work in a non-blocking fashion against each queue. This also introduces lumpy behaviour and potential request starvation, since pdflush can be starved for queue access if others are accessing it. A sample ffsb workload that does random writes to files is about 8% faster here on a simple SATA drive during the benchmark phase. File layout also seems a LOT more smooth in vmstat: r b swpd free buff cache si so bi bo in cs us sy id wa 0 1 0 608848 2652 375372 0 0 0 71024 604 24 1 10 48 42 0 1 0 549644 2712 433736 0 0 0 60692 505 27 1 8 48 44 1 0 0 476928 2784 505192 0 0 4 29540 553 24 0 9 53 37 0 1 0 457972 2808 524008 0 0 0 54876 331 16 0 4 38 58 0 1 0 366128 2928 614284 0 0 4 92168 710 58 0 13 53 34 0 1 0 295092 3000 684140 0 0 0 62924 572 23 0 9 53 37 0 1 0 236592 3064 741704 0 0 4 58256 523 17 0 8 48 44 0 1 0 165608 3132 811464 0 0 0 57460 560 21 0 8 54 38 0 1 0 102952 3200 873164 0 0 4 74748 540 29 1 10 48 41 0 1 0 48604 3252 926472 0 0 0 53248 469 29 0 7 47 45 where vanilla tends to fluctuate a lot in the creation phase: r b swpd free buff cache si so bi bo in cs us sy id wa 1 1 0 678716 5792 303380 0 0 0 74064 565 50 1 11 52 36 1 0 0 662488 5864 319396 0 0 4 352 302 329 0 2 47 51 0 1 0 599312 5924 381468 0 0 0 78164 516 55 0 9 51 40 0 1 0 519952 6008 459516 0 0 4 78156 622 56 1 11 52 37 1 1 0 436640 6092 541632 0 0 0 82244 622 54 0 11 48 41 0 1 0 436640 6092 541660 0 0 0 8 152 39 0 0 51 49 0 1 0 332224 6200 644252 0 0 4 102800 728 46 1 13 49 36 1 0 0 274492 6260 701056 0 0 4 12328 459 49 0 7 50 43 0 1 0 211220 6324 763356 0 0 0 106940 515 37 1 10 51 39 1 0 0 160412 6376 813468 0 0 0 8224 415 43 0 6 49 45 1 1 0 85980 6452 886556 0 0 4 113516 575 39 1 11 54 34 0 2 0 85968 6452 886620 0 0 0 1640 158 211 0 0 46 54 A 10 disk test with btrfs performs 26% faster with per-bdi flushing. A SSD based writeback test on XFS performs over 20% better as well, with the throughput being very stable around 1GB/sec, where pdflush only manages 750MB/sec and fluctuates wildly while doing so. Random buffered writes to many files behave a lot better as well, as does random mmap'ed writes. A separate thread is added to sync the super blocks. In the long term, adding sync_supers_bdi() functionality could get rid of this thread again. Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-09-09 07:08:54 +00:00
{
unsigned long next;
if (!dirty_writeback_interval)
return;
writeback: switch to per-bdi threads for flushing data This gets rid of pdflush for bdi writeout and kupdated style cleaning. pdflush writeout suffers from lack of locality and also requires more threads to handle the same workload, since it has to work in a non-blocking fashion against each queue. This also introduces lumpy behaviour and potential request starvation, since pdflush can be starved for queue access if others are accessing it. A sample ffsb workload that does random writes to files is about 8% faster here on a simple SATA drive during the benchmark phase. File layout also seems a LOT more smooth in vmstat: r b swpd free buff cache si so bi bo in cs us sy id wa 0 1 0 608848 2652 375372 0 0 0 71024 604 24 1 10 48 42 0 1 0 549644 2712 433736 0 0 0 60692 505 27 1 8 48 44 1 0 0 476928 2784 505192 0 0 4 29540 553 24 0 9 53 37 0 1 0 457972 2808 524008 0 0 0 54876 331 16 0 4 38 58 0 1 0 366128 2928 614284 0 0 4 92168 710 58 0 13 53 34 0 1 0 295092 3000 684140 0 0 0 62924 572 23 0 9 53 37 0 1 0 236592 3064 741704 0 0 4 58256 523 17 0 8 48 44 0 1 0 165608 3132 811464 0 0 0 57460 560 21 0 8 54 38 0 1 0 102952 3200 873164 0 0 4 74748 540 29 1 10 48 41 0 1 0 48604 3252 926472 0 0 0 53248 469 29 0 7 47 45 where vanilla tends to fluctuate a lot in the creation phase: r b swpd free buff cache si so bi bo in cs us sy id wa 1 1 0 678716 5792 303380 0 0 0 74064 565 50 1 11 52 36 1 0 0 662488 5864 319396 0 0 4 352 302 329 0 2 47 51 0 1 0 599312 5924 381468 0 0 0 78164 516 55 0 9 51 40 0 1 0 519952 6008 459516 0 0 4 78156 622 56 1 11 52 37 1 1 0 436640 6092 541632 0 0 0 82244 622 54 0 11 48 41 0 1 0 436640 6092 541660 0 0 0 8 152 39 0 0 51 49 0 1 0 332224 6200 644252 0 0 4 102800 728 46 1 13 49 36 1 0 0 274492 6260 701056 0 0 4 12328 459 49 0 7 50 43 0 1 0 211220 6324 763356 0 0 0 106940 515 37 1 10 51 39 1 0 0 160412 6376 813468 0 0 0 8224 415 43 0 6 49 45 1 1 0 85980 6452 886556 0 0 4 113516 575 39 1 11 54 34 0 2 0 85968 6452 886620 0 0 0 1640 158 211 0 0 46 54 A 10 disk test with btrfs performs 26% faster with per-bdi flushing. A SSD based writeback test on XFS performs over 20% better as well, with the throughput being very stable around 1GB/sec, where pdflush only manages 750MB/sec and fluctuates wildly while doing so. Random buffered writes to many files behave a lot better as well, as does random mmap'ed writes. A separate thread is added to sync the super blocks. In the long term, adding sync_supers_bdi() functionality could get rid of this thread again. Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-09-09 07:08:54 +00:00
next = msecs_to_jiffies(dirty_writeback_interval * 10) + jiffies;
mod_timer(&sync_supers_timer, round_jiffies_up(next));
}
static void sync_supers_timer_fn(unsigned long unused)
{
wake_up_process(sync_supers_tsk);
bdi_arm_supers_timer();
writeback: switch to per-bdi threads for flushing data This gets rid of pdflush for bdi writeout and kupdated style cleaning. pdflush writeout suffers from lack of locality and also requires more threads to handle the same workload, since it has to work in a non-blocking fashion against each queue. This also introduces lumpy behaviour and potential request starvation, since pdflush can be starved for queue access if others are accessing it. A sample ffsb workload that does random writes to files is about 8% faster here on a simple SATA drive during the benchmark phase. File layout also seems a LOT more smooth in vmstat: r b swpd free buff cache si so bi bo in cs us sy id wa 0 1 0 608848 2652 375372 0 0 0 71024 604 24 1 10 48 42 0 1 0 549644 2712 433736 0 0 0 60692 505 27 1 8 48 44 1 0 0 476928 2784 505192 0 0 4 29540 553 24 0 9 53 37 0 1 0 457972 2808 524008 0 0 0 54876 331 16 0 4 38 58 0 1 0 366128 2928 614284 0 0 4 92168 710 58 0 13 53 34 0 1 0 295092 3000 684140 0 0 0 62924 572 23 0 9 53 37 0 1 0 236592 3064 741704 0 0 4 58256 523 17 0 8 48 44 0 1 0 165608 3132 811464 0 0 0 57460 560 21 0 8 54 38 0 1 0 102952 3200 873164 0 0 4 74748 540 29 1 10 48 41 0 1 0 48604 3252 926472 0 0 0 53248 469 29 0 7 47 45 where vanilla tends to fluctuate a lot in the creation phase: r b swpd free buff cache si so bi bo in cs us sy id wa 1 1 0 678716 5792 303380 0 0 0 74064 565 50 1 11 52 36 1 0 0 662488 5864 319396 0 0 4 352 302 329 0 2 47 51 0 1 0 599312 5924 381468 0 0 0 78164 516 55 0 9 51 40 0 1 0 519952 6008 459516 0 0 4 78156 622 56 1 11 52 37 1 1 0 436640 6092 541632 0 0 0 82244 622 54 0 11 48 41 0 1 0 436640 6092 541660 0 0 0 8 152 39 0 0 51 49 0 1 0 332224 6200 644252 0 0 4 102800 728 46 1 13 49 36 1 0 0 274492 6260 701056 0 0 4 12328 459 49 0 7 50 43 0 1 0 211220 6324 763356 0 0 0 106940 515 37 1 10 51 39 1 0 0 160412 6376 813468 0 0 0 8224 415 43 0 6 49 45 1 1 0 85980 6452 886556 0 0 4 113516 575 39 1 11 54 34 0 2 0 85968 6452 886620 0 0 0 1640 158 211 0 0 46 54 A 10 disk test with btrfs performs 26% faster with per-bdi flushing. A SSD based writeback test on XFS performs over 20% better as well, with the throughput being very stable around 1GB/sec, where pdflush only manages 750MB/sec and fluctuates wildly while doing so. Random buffered writes to many files behave a lot better as well, as does random mmap'ed writes. A separate thread is added to sync the super blocks. In the long term, adding sync_supers_bdi() functionality could get rid of this thread again. Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-09-09 07:08:54 +00:00
}
static void wakeup_timer_fn(unsigned long data)
{
struct backing_dev_info *bdi = (struct backing_dev_info *)data;
spin_lock_bh(&bdi->wb_lock);
if (bdi->wb.task) {
wake_up_process(bdi->wb.task);
} else {
/*
* When bdi tasks are inactive for long time, they are killed.
* In this case we have to wake-up the forker thread which
* should create and run the bdi thread.
*/
wake_up_process(default_backing_dev_info.wb.task);
}
spin_unlock_bh(&bdi->wb_lock);
}
/*
* This function is used when the first inode for this bdi is marked dirty. It
* wakes-up the corresponding bdi thread which should then take care of the
* periodic background write-out of dirty inodes. Since the write-out would
* starts only 'dirty_writeback_interval' centisecs from now anyway, we just
* set up a timer which wakes the bdi thread up later.
*
* Note, we wouldn't bother setting up the timer, but this function is on the
* fast-path (used by '__mark_inode_dirty()'), so we save few context switches
* by delaying the wake-up.
*/
void bdi_wakeup_thread_delayed(struct backing_dev_info *bdi)
{
unsigned long timeout;
timeout = msecs_to_jiffies(dirty_writeback_interval * 10);
mod_timer(&bdi->wb.wakeup_timer, jiffies + timeout);
}
writeback: move bdi threads exiting logic to the forker thread Currently, bdi threads can decide to exit if there were no useful activities for 5 minutes. However, this causes nasty races: we can easily oops in the 'bdi_queue_work()' if the bdi thread decides to exit while we are waking it up. And even if we do not oops, but the bdi tread exits immediately after we wake it up, we'd lose the wake-up event and have an unnecessary delay (up to 5 secs) in the bdi work processing. This patch makes the forker thread to be the central place which not only creates bdi threads, but also kills them if they were inactive long enough. This better design-wise. Another reason why this change was done is to prepare for the further changes which will prevent the bdi threads from waking up every 5 sec and wasting power. Indeed, when the task does not wake up periodically anymore, it won't be able to exit either. This patch also moves the the 'wake_up_bit()' call from the bdi thread to the forker thread as well. So now the forker thread sets the BDI_pending bit, then forks the task or kills it, then clears the bit and wakes up the waiting process. The only process which may wain on the bit is 'bdi_wb_shutdown()'. This function was changed as well - now it first removes the bdi from the 'bdi_list', then waits on the 'BDI_pending' bit. Once it wakes up, it is guaranteed that the forker thread won't race with it, because the bdi is not visible. Note, the forker thread sets the 'BDI_pending' bit under the 'bdi->wb_lock' which is essential for proper serialization. And additionally, when we change 'bdi->wb.task', we now take the 'bdi->work_lock', to make sure that we do not lose wake-ups which we otherwise would when raced with, say, 'bdi_queue_work()'. Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
2010-07-25 11:29:20 +00:00
/*
* Calculate the longest interval (jiffies) bdi threads are allowed to be
* inactive.
*/
static unsigned long bdi_longest_inactive(void)
{
unsigned long interval;
interval = msecs_to_jiffies(dirty_writeback_interval * 10);
return max(5UL * 60 * HZ, interval);
}
static int bdi_forker_thread(void *ptr)
writeback: switch to per-bdi threads for flushing data This gets rid of pdflush for bdi writeout and kupdated style cleaning. pdflush writeout suffers from lack of locality and also requires more threads to handle the same workload, since it has to work in a non-blocking fashion against each queue. This also introduces lumpy behaviour and potential request starvation, since pdflush can be starved for queue access if others are accessing it. A sample ffsb workload that does random writes to files is about 8% faster here on a simple SATA drive during the benchmark phase. File layout also seems a LOT more smooth in vmstat: r b swpd free buff cache si so bi bo in cs us sy id wa 0 1 0 608848 2652 375372 0 0 0 71024 604 24 1 10 48 42 0 1 0 549644 2712 433736 0 0 0 60692 505 27 1 8 48 44 1 0 0 476928 2784 505192 0 0 4 29540 553 24 0 9 53 37 0 1 0 457972 2808 524008 0 0 0 54876 331 16 0 4 38 58 0 1 0 366128 2928 614284 0 0 4 92168 710 58 0 13 53 34 0 1 0 295092 3000 684140 0 0 0 62924 572 23 0 9 53 37 0 1 0 236592 3064 741704 0 0 4 58256 523 17 0 8 48 44 0 1 0 165608 3132 811464 0 0 0 57460 560 21 0 8 54 38 0 1 0 102952 3200 873164 0 0 4 74748 540 29 1 10 48 41 0 1 0 48604 3252 926472 0 0 0 53248 469 29 0 7 47 45 where vanilla tends to fluctuate a lot in the creation phase: r b swpd free buff cache si so bi bo in cs us sy id wa 1 1 0 678716 5792 303380 0 0 0 74064 565 50 1 11 52 36 1 0 0 662488 5864 319396 0 0 4 352 302 329 0 2 47 51 0 1 0 599312 5924 381468 0 0 0 78164 516 55 0 9 51 40 0 1 0 519952 6008 459516 0 0 4 78156 622 56 1 11 52 37 1 1 0 436640 6092 541632 0 0 0 82244 622 54 0 11 48 41 0 1 0 436640 6092 541660 0 0 0 8 152 39 0 0 51 49 0 1 0 332224 6200 644252 0 0 4 102800 728 46 1 13 49 36 1 0 0 274492 6260 701056 0 0 4 12328 459 49 0 7 50 43 0 1 0 211220 6324 763356 0 0 0 106940 515 37 1 10 51 39 1 0 0 160412 6376 813468 0 0 0 8224 415 43 0 6 49 45 1 1 0 85980 6452 886556 0 0 4 113516 575 39 1 11 54 34 0 2 0 85968 6452 886620 0 0 0 1640 158 211 0 0 46 54 A 10 disk test with btrfs performs 26% faster with per-bdi flushing. A SSD based writeback test on XFS performs over 20% better as well, with the throughput being very stable around 1GB/sec, where pdflush only manages 750MB/sec and fluctuates wildly while doing so. Random buffered writes to many files behave a lot better as well, as does random mmap'ed writes. A separate thread is added to sync the super blocks. In the long term, adding sync_supers_bdi() functionality could get rid of this thread again. Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-09-09 07:08:54 +00:00
{
struct bdi_writeback *me = ptr;
current->flags |= PF_FLUSHER | PF_SWAPWRITE;
set_freezable();
/*
* Our parent may run at a different priority, just set us to normal
*/
set_user_nice(current, 0);
writeback: switch to per-bdi threads for flushing data This gets rid of pdflush for bdi writeout and kupdated style cleaning. pdflush writeout suffers from lack of locality and also requires more threads to handle the same workload, since it has to work in a non-blocking fashion against each queue. This also introduces lumpy behaviour and potential request starvation, since pdflush can be starved for queue access if others are accessing it. A sample ffsb workload that does random writes to files is about 8% faster here on a simple SATA drive during the benchmark phase. File layout also seems a LOT more smooth in vmstat: r b swpd free buff cache si so bi bo in cs us sy id wa 0 1 0 608848 2652 375372 0 0 0 71024 604 24 1 10 48 42 0 1 0 549644 2712 433736 0 0 0 60692 505 27 1 8 48 44 1 0 0 476928 2784 505192 0 0 4 29540 553 24 0 9 53 37 0 1 0 457972 2808 524008 0 0 0 54876 331 16 0 4 38 58 0 1 0 366128 2928 614284 0 0 4 92168 710 58 0 13 53 34 0 1 0 295092 3000 684140 0 0 0 62924 572 23 0 9 53 37 0 1 0 236592 3064 741704 0 0 4 58256 523 17 0 8 48 44 0 1 0 165608 3132 811464 0 0 0 57460 560 21 0 8 54 38 0 1 0 102952 3200 873164 0 0 4 74748 540 29 1 10 48 41 0 1 0 48604 3252 926472 0 0 0 53248 469 29 0 7 47 45 where vanilla tends to fluctuate a lot in the creation phase: r b swpd free buff cache si so bi bo in cs us sy id wa 1 1 0 678716 5792 303380 0 0 0 74064 565 50 1 11 52 36 1 0 0 662488 5864 319396 0 0 4 352 302 329 0 2 47 51 0 1 0 599312 5924 381468 0 0 0 78164 516 55 0 9 51 40 0 1 0 519952 6008 459516 0 0 4 78156 622 56 1 11 52 37 1 1 0 436640 6092 541632 0 0 0 82244 622 54 0 11 48 41 0 1 0 436640 6092 541660 0 0 0 8 152 39 0 0 51 49 0 1 0 332224 6200 644252 0 0 4 102800 728 46 1 13 49 36 1 0 0 274492 6260 701056 0 0 4 12328 459 49 0 7 50 43 0 1 0 211220 6324 763356 0 0 0 106940 515 37 1 10 51 39 1 0 0 160412 6376 813468 0 0 0 8224 415 43 0 6 49 45 1 1 0 85980 6452 886556 0 0 4 113516 575 39 1 11 54 34 0 2 0 85968 6452 886620 0 0 0 1640 158 211 0 0 46 54 A 10 disk test with btrfs performs 26% faster with per-bdi flushing. A SSD based writeback test on XFS performs over 20% better as well, with the throughput being very stable around 1GB/sec, where pdflush only manages 750MB/sec and fluctuates wildly while doing so. Random buffered writes to many files behave a lot better as well, as does random mmap'ed writes. A separate thread is added to sync the super blocks. In the long term, adding sync_supers_bdi() functionality could get rid of this thread again. Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-09-09 07:08:54 +00:00
for (;;) {
writeback: move bdi threads exiting logic to the forker thread Currently, bdi threads can decide to exit if there were no useful activities for 5 minutes. However, this causes nasty races: we can easily oops in the 'bdi_queue_work()' if the bdi thread decides to exit while we are waking it up. And even if we do not oops, but the bdi tread exits immediately after we wake it up, we'd lose the wake-up event and have an unnecessary delay (up to 5 secs) in the bdi work processing. This patch makes the forker thread to be the central place which not only creates bdi threads, but also kills them if they were inactive long enough. This better design-wise. Another reason why this change was done is to prepare for the further changes which will prevent the bdi threads from waking up every 5 sec and wasting power. Indeed, when the task does not wake up periodically anymore, it won't be able to exit either. This patch also moves the the 'wake_up_bit()' call from the bdi thread to the forker thread as well. So now the forker thread sets the BDI_pending bit, then forks the task or kills it, then clears the bit and wakes up the waiting process. The only process which may wain on the bit is 'bdi_wb_shutdown()'. This function was changed as well - now it first removes the bdi from the 'bdi_list', then waits on the 'BDI_pending' bit. Once it wakes up, it is guaranteed that the forker thread won't race with it, because the bdi is not visible. Note, the forker thread sets the 'BDI_pending' bit under the 'bdi->wb_lock' which is essential for proper serialization. And additionally, when we change 'bdi->wb.task', we now take the 'bdi->work_lock', to make sure that we do not lose wake-ups which we otherwise would when raced with, say, 'bdi_queue_work()'. Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
2010-07-25 11:29:20 +00:00
struct task_struct *task = NULL;
struct backing_dev_info *bdi;
enum {
NO_ACTION, /* Nothing to do */
FORK_THREAD, /* Fork bdi thread */
writeback: move bdi threads exiting logic to the forker thread Currently, bdi threads can decide to exit if there were no useful activities for 5 minutes. However, this causes nasty races: we can easily oops in the 'bdi_queue_work()' if the bdi thread decides to exit while we are waking it up. And even if we do not oops, but the bdi tread exits immediately after we wake it up, we'd lose the wake-up event and have an unnecessary delay (up to 5 secs) in the bdi work processing. This patch makes the forker thread to be the central place which not only creates bdi threads, but also kills them if they were inactive long enough. This better design-wise. Another reason why this change was done is to prepare for the further changes which will prevent the bdi threads from waking up every 5 sec and wasting power. Indeed, when the task does not wake up periodically anymore, it won't be able to exit either. This patch also moves the the 'wake_up_bit()' call from the bdi thread to the forker thread as well. So now the forker thread sets the BDI_pending bit, then forks the task or kills it, then clears the bit and wakes up the waiting process. The only process which may wain on the bit is 'bdi_wb_shutdown()'. This function was changed as well - now it first removes the bdi from the 'bdi_list', then waits on the 'BDI_pending' bit. Once it wakes up, it is guaranteed that the forker thread won't race with it, because the bdi is not visible. Note, the forker thread sets the 'BDI_pending' bit under the 'bdi->wb_lock' which is essential for proper serialization. And additionally, when we change 'bdi->wb.task', we now take the 'bdi->work_lock', to make sure that we do not lose wake-ups which we otherwise would when raced with, say, 'bdi_queue_work()'. Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
2010-07-25 11:29:20 +00:00
KILL_THREAD, /* Kill inactive bdi thread */
} action = NO_ACTION;
writeback: switch to per-bdi threads for flushing data This gets rid of pdflush for bdi writeout and kupdated style cleaning. pdflush writeout suffers from lack of locality and also requires more threads to handle the same workload, since it has to work in a non-blocking fashion against each queue. This also introduces lumpy behaviour and potential request starvation, since pdflush can be starved for queue access if others are accessing it. A sample ffsb workload that does random writes to files is about 8% faster here on a simple SATA drive during the benchmark phase. File layout also seems a LOT more smooth in vmstat: r b swpd free buff cache si so bi bo in cs us sy id wa 0 1 0 608848 2652 375372 0 0 0 71024 604 24 1 10 48 42 0 1 0 549644 2712 433736 0 0 0 60692 505 27 1 8 48 44 1 0 0 476928 2784 505192 0 0 4 29540 553 24 0 9 53 37 0 1 0 457972 2808 524008 0 0 0 54876 331 16 0 4 38 58 0 1 0 366128 2928 614284 0 0 4 92168 710 58 0 13 53 34 0 1 0 295092 3000 684140 0 0 0 62924 572 23 0 9 53 37 0 1 0 236592 3064 741704 0 0 4 58256 523 17 0 8 48 44 0 1 0 165608 3132 811464 0 0 0 57460 560 21 0 8 54 38 0 1 0 102952 3200 873164 0 0 4 74748 540 29 1 10 48 41 0 1 0 48604 3252 926472 0 0 0 53248 469 29 0 7 47 45 where vanilla tends to fluctuate a lot in the creation phase: r b swpd free buff cache si so bi bo in cs us sy id wa 1 1 0 678716 5792 303380 0 0 0 74064 565 50 1 11 52 36 1 0 0 662488 5864 319396 0 0 4 352 302 329 0 2 47 51 0 1 0 599312 5924 381468 0 0 0 78164 516 55 0 9 51 40 0 1 0 519952 6008 459516 0 0 4 78156 622 56 1 11 52 37 1 1 0 436640 6092 541632 0 0 0 82244 622 54 0 11 48 41 0 1 0 436640 6092 541660 0 0 0 8 152 39 0 0 51 49 0 1 0 332224 6200 644252 0 0 4 102800 728 46 1 13 49 36 1 0 0 274492 6260 701056 0 0 4 12328 459 49 0 7 50 43 0 1 0 211220 6324 763356 0 0 0 106940 515 37 1 10 51 39 1 0 0 160412 6376 813468 0 0 0 8224 415 43 0 6 49 45 1 1 0 85980 6452 886556 0 0 4 113516 575 39 1 11 54 34 0 2 0 85968 6452 886620 0 0 0 1640 158 211 0 0 46 54 A 10 disk test with btrfs performs 26% faster with per-bdi flushing. A SSD based writeback test on XFS performs over 20% better as well, with the throughput being very stable around 1GB/sec, where pdflush only manages 750MB/sec and fluctuates wildly while doing so. Random buffered writes to many files behave a lot better as well, as does random mmap'ed writes. A separate thread is added to sync the super blocks. In the long term, adding sync_supers_bdi() functionality could get rid of this thread again. Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-09-09 07:08:54 +00:00
/*
* Temporary measure, we want to make sure we don't see
* dirty data on the default backing_dev_info
*/
if (wb_has_dirty_io(me) || !list_empty(&me->bdi->work_list)) {
del_timer(&me->wakeup_timer);
writeback: switch to per-bdi threads for flushing data This gets rid of pdflush for bdi writeout and kupdated style cleaning. pdflush writeout suffers from lack of locality and also requires more threads to handle the same workload, since it has to work in a non-blocking fashion against each queue. This also introduces lumpy behaviour and potential request starvation, since pdflush can be starved for queue access if others are accessing it. A sample ffsb workload that does random writes to files is about 8% faster here on a simple SATA drive during the benchmark phase. File layout also seems a LOT more smooth in vmstat: r b swpd free buff cache si so bi bo in cs us sy id wa 0 1 0 608848 2652 375372 0 0 0 71024 604 24 1 10 48 42 0 1 0 549644 2712 433736 0 0 0 60692 505 27 1 8 48 44 1 0 0 476928 2784 505192 0 0 4 29540 553 24 0 9 53 37 0 1 0 457972 2808 524008 0 0 0 54876 331 16 0 4 38 58 0 1 0 366128 2928 614284 0 0 4 92168 710 58 0 13 53 34 0 1 0 295092 3000 684140 0 0 0 62924 572 23 0 9 53 37 0 1 0 236592 3064 741704 0 0 4 58256 523 17 0 8 48 44 0 1 0 165608 3132 811464 0 0 0 57460 560 21 0 8 54 38 0 1 0 102952 3200 873164 0 0 4 74748 540 29 1 10 48 41 0 1 0 48604 3252 926472 0 0 0 53248 469 29 0 7 47 45 where vanilla tends to fluctuate a lot in the creation phase: r b swpd free buff cache si so bi bo in cs us sy id wa 1 1 0 678716 5792 303380 0 0 0 74064 565 50 1 11 52 36 1 0 0 662488 5864 319396 0 0 4 352 302 329 0 2 47 51 0 1 0 599312 5924 381468 0 0 0 78164 516 55 0 9 51 40 0 1 0 519952 6008 459516 0 0 4 78156 622 56 1 11 52 37 1 1 0 436640 6092 541632 0 0 0 82244 622 54 0 11 48 41 0 1 0 436640 6092 541660 0 0 0 8 152 39 0 0 51 49 0 1 0 332224 6200 644252 0 0 4 102800 728 46 1 13 49 36 1 0 0 274492 6260 701056 0 0 4 12328 459 49 0 7 50 43 0 1 0 211220 6324 763356 0 0 0 106940 515 37 1 10 51 39 1 0 0 160412 6376 813468 0 0 0 8224 415 43 0 6 49 45 1 1 0 85980 6452 886556 0 0 4 113516 575 39 1 11 54 34 0 2 0 85968 6452 886620 0 0 0 1640 158 211 0 0 46 54 A 10 disk test with btrfs performs 26% faster with per-bdi flushing. A SSD based writeback test on XFS performs over 20% better as well, with the throughput being very stable around 1GB/sec, where pdflush only manages 750MB/sec and fluctuates wildly while doing so. Random buffered writes to many files behave a lot better as well, as does random mmap'ed writes. A separate thread is added to sync the super blocks. In the long term, adding sync_supers_bdi() functionality could get rid of this thread again. Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-09-09 07:08:54 +00:00
wb_do_writeback(me, 0);
}
writeback: switch to per-bdi threads for flushing data This gets rid of pdflush for bdi writeout and kupdated style cleaning. pdflush writeout suffers from lack of locality and also requires more threads to handle the same workload, since it has to work in a non-blocking fashion against each queue. This also introduces lumpy behaviour and potential request starvation, since pdflush can be starved for queue access if others are accessing it. A sample ffsb workload that does random writes to files is about 8% faster here on a simple SATA drive during the benchmark phase. File layout also seems a LOT more smooth in vmstat: r b swpd free buff cache si so bi bo in cs us sy id wa 0 1 0 608848 2652 375372 0 0 0 71024 604 24 1 10 48 42 0 1 0 549644 2712 433736 0 0 0 60692 505 27 1 8 48 44 1 0 0 476928 2784 505192 0 0 4 29540 553 24 0 9 53 37 0 1 0 457972 2808 524008 0 0 0 54876 331 16 0 4 38 58 0 1 0 366128 2928 614284 0 0 4 92168 710 58 0 13 53 34 0 1 0 295092 3000 684140 0 0 0 62924 572 23 0 9 53 37 0 1 0 236592 3064 741704 0 0 4 58256 523 17 0 8 48 44 0 1 0 165608 3132 811464 0 0 0 57460 560 21 0 8 54 38 0 1 0 102952 3200 873164 0 0 4 74748 540 29 1 10 48 41 0 1 0 48604 3252 926472 0 0 0 53248 469 29 0 7 47 45 where vanilla tends to fluctuate a lot in the creation phase: r b swpd free buff cache si so bi bo in cs us sy id wa 1 1 0 678716 5792 303380 0 0 0 74064 565 50 1 11 52 36 1 0 0 662488 5864 319396 0 0 4 352 302 329 0 2 47 51 0 1 0 599312 5924 381468 0 0 0 78164 516 55 0 9 51 40 0 1 0 519952 6008 459516 0 0 4 78156 622 56 1 11 52 37 1 1 0 436640 6092 541632 0 0 0 82244 622 54 0 11 48 41 0 1 0 436640 6092 541660 0 0 0 8 152 39 0 0 51 49 0 1 0 332224 6200 644252 0 0 4 102800 728 46 1 13 49 36 1 0 0 274492 6260 701056 0 0 4 12328 459 49 0 7 50 43 0 1 0 211220 6324 763356 0 0 0 106940 515 37 1 10 51 39 1 0 0 160412 6376 813468 0 0 0 8224 415 43 0 6 49 45 1 1 0 85980 6452 886556 0 0 4 113516 575 39 1 11 54 34 0 2 0 85968 6452 886620 0 0 0 1640 158 211 0 0 46 54 A 10 disk test with btrfs performs 26% faster with per-bdi flushing. A SSD based writeback test on XFS performs over 20% better as well, with the throughput being very stable around 1GB/sec, where pdflush only manages 750MB/sec and fluctuates wildly while doing so. Random buffered writes to many files behave a lot better as well, as does random mmap'ed writes. A separate thread is added to sync the super blocks. In the long term, adding sync_supers_bdi() functionality could get rid of this thread again. Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-09-09 07:08:54 +00:00
spin_lock_bh(&bdi_lock);
set_current_state(TASK_INTERRUPTIBLE);
writeback: switch to per-bdi threads for flushing data This gets rid of pdflush for bdi writeout and kupdated style cleaning. pdflush writeout suffers from lack of locality and also requires more threads to handle the same workload, since it has to work in a non-blocking fashion against each queue. This also introduces lumpy behaviour and potential request starvation, since pdflush can be starved for queue access if others are accessing it. A sample ffsb workload that does random writes to files is about 8% faster here on a simple SATA drive during the benchmark phase. File layout also seems a LOT more smooth in vmstat: r b swpd free buff cache si so bi bo in cs us sy id wa 0 1 0 608848 2652 375372 0 0 0 71024 604 24 1 10 48 42 0 1 0 549644 2712 433736 0 0 0 60692 505 27 1 8 48 44 1 0 0 476928 2784 505192 0 0 4 29540 553 24 0 9 53 37 0 1 0 457972 2808 524008 0 0 0 54876 331 16 0 4 38 58 0 1 0 366128 2928 614284 0 0 4 92168 710 58 0 13 53 34 0 1 0 295092 3000 684140 0 0 0 62924 572 23 0 9 53 37 0 1 0 236592 3064 741704 0 0 4 58256 523 17 0 8 48 44 0 1 0 165608 3132 811464 0 0 0 57460 560 21 0 8 54 38 0 1 0 102952 3200 873164 0 0 4 74748 540 29 1 10 48 41 0 1 0 48604 3252 926472 0 0 0 53248 469 29 0 7 47 45 where vanilla tends to fluctuate a lot in the creation phase: r b swpd free buff cache si so bi bo in cs us sy id wa 1 1 0 678716 5792 303380 0 0 0 74064 565 50 1 11 52 36 1 0 0 662488 5864 319396 0 0 4 352 302 329 0 2 47 51 0 1 0 599312 5924 381468 0 0 0 78164 516 55 0 9 51 40 0 1 0 519952 6008 459516 0 0 4 78156 622 56 1 11 52 37 1 1 0 436640 6092 541632 0 0 0 82244 622 54 0 11 48 41 0 1 0 436640 6092 541660 0 0 0 8 152 39 0 0 51 49 0 1 0 332224 6200 644252 0 0 4 102800 728 46 1 13 49 36 1 0 0 274492 6260 701056 0 0 4 12328 459 49 0 7 50 43 0 1 0 211220 6324 763356 0 0 0 106940 515 37 1 10 51 39 1 0 0 160412 6376 813468 0 0 0 8224 415 43 0 6 49 45 1 1 0 85980 6452 886556 0 0 4 113516 575 39 1 11 54 34 0 2 0 85968 6452 886620 0 0 0 1640 158 211 0 0 46 54 A 10 disk test with btrfs performs 26% faster with per-bdi flushing. A SSD based writeback test on XFS performs over 20% better as well, with the throughput being very stable around 1GB/sec, where pdflush only manages 750MB/sec and fluctuates wildly while doing so. Random buffered writes to many files behave a lot better as well, as does random mmap'ed writes. A separate thread is added to sync the super blocks. In the long term, adding sync_supers_bdi() functionality could get rid of this thread again. Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-09-09 07:08:54 +00:00
list_for_each_entry(bdi, &bdi_list, bdi_list) {
bool have_dirty_io;
if (!bdi_cap_writeback_dirty(bdi) ||
bdi_cap_flush_forker(bdi))
writeback: switch to per-bdi threads for flushing data This gets rid of pdflush for bdi writeout and kupdated style cleaning. pdflush writeout suffers from lack of locality and also requires more threads to handle the same workload, since it has to work in a non-blocking fashion against each queue. This also introduces lumpy behaviour and potential request starvation, since pdflush can be starved for queue access if others are accessing it. A sample ffsb workload that does random writes to files is about 8% faster here on a simple SATA drive during the benchmark phase. File layout also seems a LOT more smooth in vmstat: r b swpd free buff cache si so bi bo in cs us sy id wa 0 1 0 608848 2652 375372 0 0 0 71024 604 24 1 10 48 42 0 1 0 549644 2712 433736 0 0 0 60692 505 27 1 8 48 44 1 0 0 476928 2784 505192 0 0 4 29540 553 24 0 9 53 37 0 1 0 457972 2808 524008 0 0 0 54876 331 16 0 4 38 58 0 1 0 366128 2928 614284 0 0 4 92168 710 58 0 13 53 34 0 1 0 295092 3000 684140 0 0 0 62924 572 23 0 9 53 37 0 1 0 236592 3064 741704 0 0 4 58256 523 17 0 8 48 44 0 1 0 165608 3132 811464 0 0 0 57460 560 21 0 8 54 38 0 1 0 102952 3200 873164 0 0 4 74748 540 29 1 10 48 41 0 1 0 48604 3252 926472 0 0 0 53248 469 29 0 7 47 45 where vanilla tends to fluctuate a lot in the creation phase: r b swpd free buff cache si so bi bo in cs us sy id wa 1 1 0 678716 5792 303380 0 0 0 74064 565 50 1 11 52 36 1 0 0 662488 5864 319396 0 0 4 352 302 329 0 2 47 51 0 1 0 599312 5924 381468 0 0 0 78164 516 55 0 9 51 40 0 1 0 519952 6008 459516 0 0 4 78156 622 56 1 11 52 37 1 1 0 436640 6092 541632 0 0 0 82244 622 54 0 11 48 41 0 1 0 436640 6092 541660 0 0 0 8 152 39 0 0 51 49 0 1 0 332224 6200 644252 0 0 4 102800 728 46 1 13 49 36 1 0 0 274492 6260 701056 0 0 4 12328 459 49 0 7 50 43 0 1 0 211220 6324 763356 0 0 0 106940 515 37 1 10 51 39 1 0 0 160412 6376 813468 0 0 0 8224 415 43 0 6 49 45 1 1 0 85980 6452 886556 0 0 4 113516 575 39 1 11 54 34 0 2 0 85968 6452 886620 0 0 0 1640 158 211 0 0 46 54 A 10 disk test with btrfs performs 26% faster with per-bdi flushing. A SSD based writeback test on XFS performs over 20% better as well, with the throughput being very stable around 1GB/sec, where pdflush only manages 750MB/sec and fluctuates wildly while doing so. Random buffered writes to many files behave a lot better as well, as does random mmap'ed writes. A separate thread is added to sync the super blocks. In the long term, adding sync_supers_bdi() functionality could get rid of this thread again. Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-09-09 07:08:54 +00:00
continue;
WARN(!test_bit(BDI_registered, &bdi->state),
"bdi %p/%s is not registered!\n", bdi, bdi->name);
have_dirty_io = !list_empty(&bdi->work_list) ||
wb_has_dirty_io(&bdi->wb);
/*
* If the bdi has work to do, but the thread does not
* exist - create it.
*/
if (!bdi->wb.task && have_dirty_io) {
/*
* Set the pending bit - if someone will try to
* unregister this bdi - it'll wait on this bit.
*/
set_bit(BDI_pending, &bdi->state);
action = FORK_THREAD;
break;
}
writeback: move bdi threads exiting logic to the forker thread Currently, bdi threads can decide to exit if there were no useful activities for 5 minutes. However, this causes nasty races: we can easily oops in the 'bdi_queue_work()' if the bdi thread decides to exit while we are waking it up. And even if we do not oops, but the bdi tread exits immediately after we wake it up, we'd lose the wake-up event and have an unnecessary delay (up to 5 secs) in the bdi work processing. This patch makes the forker thread to be the central place which not only creates bdi threads, but also kills them if they were inactive long enough. This better design-wise. Another reason why this change was done is to prepare for the further changes which will prevent the bdi threads from waking up every 5 sec and wasting power. Indeed, when the task does not wake up periodically anymore, it won't be able to exit either. This patch also moves the the 'wake_up_bit()' call from the bdi thread to the forker thread as well. So now the forker thread sets the BDI_pending bit, then forks the task or kills it, then clears the bit and wakes up the waiting process. The only process which may wain on the bit is 'bdi_wb_shutdown()'. This function was changed as well - now it first removes the bdi from the 'bdi_list', then waits on the 'BDI_pending' bit. Once it wakes up, it is guaranteed that the forker thread won't race with it, because the bdi is not visible. Note, the forker thread sets the 'BDI_pending' bit under the 'bdi->wb_lock' which is essential for proper serialization. And additionally, when we change 'bdi->wb.task', we now take the 'bdi->work_lock', to make sure that we do not lose wake-ups which we otherwise would when raced with, say, 'bdi_queue_work()'. Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
2010-07-25 11:29:20 +00:00
spin_lock_bh(&bdi->wb_lock);
writeback: move bdi threads exiting logic to the forker thread Currently, bdi threads can decide to exit if there were no useful activities for 5 minutes. However, this causes nasty races: we can easily oops in the 'bdi_queue_work()' if the bdi thread decides to exit while we are waking it up. And even if we do not oops, but the bdi tread exits immediately after we wake it up, we'd lose the wake-up event and have an unnecessary delay (up to 5 secs) in the bdi work processing. This patch makes the forker thread to be the central place which not only creates bdi threads, but also kills them if they were inactive long enough. This better design-wise. Another reason why this change was done is to prepare for the further changes which will prevent the bdi threads from waking up every 5 sec and wasting power. Indeed, when the task does not wake up periodically anymore, it won't be able to exit either. This patch also moves the the 'wake_up_bit()' call from the bdi thread to the forker thread as well. So now the forker thread sets the BDI_pending bit, then forks the task or kills it, then clears the bit and wakes up the waiting process. The only process which may wain on the bit is 'bdi_wb_shutdown()'. This function was changed as well - now it first removes the bdi from the 'bdi_list', then waits on the 'BDI_pending' bit. Once it wakes up, it is guaranteed that the forker thread won't race with it, because the bdi is not visible. Note, the forker thread sets the 'BDI_pending' bit under the 'bdi->wb_lock' which is essential for proper serialization. And additionally, when we change 'bdi->wb.task', we now take the 'bdi->work_lock', to make sure that we do not lose wake-ups which we otherwise would when raced with, say, 'bdi_queue_work()'. Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
2010-07-25 11:29:20 +00:00
/*
* If there is no work to do and the bdi thread was
* inactive long enough - kill it. The wb_lock is taken
* to make sure no-one adds more work to this bdi and
* wakes the bdi thread up.
*/
if (bdi->wb.task && !have_dirty_io &&
time_after(jiffies, bdi->wb.last_active +
bdi_longest_inactive())) {
task = bdi->wb.task;
bdi->wb.task = NULL;
spin_unlock(&bdi->wb_lock);
set_bit(BDI_pending, &bdi->state);
action = KILL_THREAD;
break;
}
spin_unlock_bh(&bdi->wb_lock);
writeback: switch to per-bdi threads for flushing data This gets rid of pdflush for bdi writeout and kupdated style cleaning. pdflush writeout suffers from lack of locality and also requires more threads to handle the same workload, since it has to work in a non-blocking fashion against each queue. This also introduces lumpy behaviour and potential request starvation, since pdflush can be starved for queue access if others are accessing it. A sample ffsb workload that does random writes to files is about 8% faster here on a simple SATA drive during the benchmark phase. File layout also seems a LOT more smooth in vmstat: r b swpd free buff cache si so bi bo in cs us sy id wa 0 1 0 608848 2652 375372 0 0 0 71024 604 24 1 10 48 42 0 1 0 549644 2712 433736 0 0 0 60692 505 27 1 8 48 44 1 0 0 476928 2784 505192 0 0 4 29540 553 24 0 9 53 37 0 1 0 457972 2808 524008 0 0 0 54876 331 16 0 4 38 58 0 1 0 366128 2928 614284 0 0 4 92168 710 58 0 13 53 34 0 1 0 295092 3000 684140 0 0 0 62924 572 23 0 9 53 37 0 1 0 236592 3064 741704 0 0 4 58256 523 17 0 8 48 44 0 1 0 165608 3132 811464 0 0 0 57460 560 21 0 8 54 38 0 1 0 102952 3200 873164 0 0 4 74748 540 29 1 10 48 41 0 1 0 48604 3252 926472 0 0 0 53248 469 29 0 7 47 45 where vanilla tends to fluctuate a lot in the creation phase: r b swpd free buff cache si so bi bo in cs us sy id wa 1 1 0 678716 5792 303380 0 0 0 74064 565 50 1 11 52 36 1 0 0 662488 5864 319396 0 0 4 352 302 329 0 2 47 51 0 1 0 599312 5924 381468 0 0 0 78164 516 55 0 9 51 40 0 1 0 519952 6008 459516 0 0 4 78156 622 56 1 11 52 37 1 1 0 436640 6092 541632 0 0 0 82244 622 54 0 11 48 41 0 1 0 436640 6092 541660 0 0 0 8 152 39 0 0 51 49 0 1 0 332224 6200 644252 0 0 4 102800 728 46 1 13 49 36 1 0 0 274492 6260 701056 0 0 4 12328 459 49 0 7 50 43 0 1 0 211220 6324 763356 0 0 0 106940 515 37 1 10 51 39 1 0 0 160412 6376 813468 0 0 0 8224 415 43 0 6 49 45 1 1 0 85980 6452 886556 0 0 4 113516 575 39 1 11 54 34 0 2 0 85968 6452 886620 0 0 0 1640 158 211 0 0 46 54 A 10 disk test with btrfs performs 26% faster with per-bdi flushing. A SSD based writeback test on XFS performs over 20% better as well, with the throughput being very stable around 1GB/sec, where pdflush only manages 750MB/sec and fluctuates wildly while doing so. Random buffered writes to many files behave a lot better as well, as does random mmap'ed writes. A separate thread is added to sync the super blocks. In the long term, adding sync_supers_bdi() functionality could get rid of this thread again. Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-09-09 07:08:54 +00:00
}
spin_unlock_bh(&bdi_lock);
writeback: switch to per-bdi threads for flushing data This gets rid of pdflush for bdi writeout and kupdated style cleaning. pdflush writeout suffers from lack of locality and also requires more threads to handle the same workload, since it has to work in a non-blocking fashion against each queue. This also introduces lumpy behaviour and potential request starvation, since pdflush can be starved for queue access if others are accessing it. A sample ffsb workload that does random writes to files is about 8% faster here on a simple SATA drive during the benchmark phase. File layout also seems a LOT more smooth in vmstat: r b swpd free buff cache si so bi bo in cs us sy id wa 0 1 0 608848 2652 375372 0 0 0 71024 604 24 1 10 48 42 0 1 0 549644 2712 433736 0 0 0 60692 505 27 1 8 48 44 1 0 0 476928 2784 505192 0 0 4 29540 553 24 0 9 53 37 0 1 0 457972 2808 524008 0 0 0 54876 331 16 0 4 38 58 0 1 0 366128 2928 614284 0 0 4 92168 710 58 0 13 53 34 0 1 0 295092 3000 684140 0 0 0 62924 572 23 0 9 53 37 0 1 0 236592 3064 741704 0 0 4 58256 523 17 0 8 48 44 0 1 0 165608 3132 811464 0 0 0 57460 560 21 0 8 54 38 0 1 0 102952 3200 873164 0 0 4 74748 540 29 1 10 48 41 0 1 0 48604 3252 926472 0 0 0 53248 469 29 0 7 47 45 where vanilla tends to fluctuate a lot in the creation phase: r b swpd free buff cache si so bi bo in cs us sy id wa 1 1 0 678716 5792 303380 0 0 0 74064 565 50 1 11 52 36 1 0 0 662488 5864 319396 0 0 4 352 302 329 0 2 47 51 0 1 0 599312 5924 381468 0 0 0 78164 516 55 0 9 51 40 0 1 0 519952 6008 459516 0 0 4 78156 622 56 1 11 52 37 1 1 0 436640 6092 541632 0 0 0 82244 622 54 0 11 48 41 0 1 0 436640 6092 541660 0 0 0 8 152 39 0 0 51 49 0 1 0 332224 6200 644252 0 0 4 102800 728 46 1 13 49 36 1 0 0 274492 6260 701056 0 0 4 12328 459 49 0 7 50 43 0 1 0 211220 6324 763356 0 0 0 106940 515 37 1 10 51 39 1 0 0 160412 6376 813468 0 0 0 8224 415 43 0 6 49 45 1 1 0 85980 6452 886556 0 0 4 113516 575 39 1 11 54 34 0 2 0 85968 6452 886620 0 0 0 1640 158 211 0 0 46 54 A 10 disk test with btrfs performs 26% faster with per-bdi flushing. A SSD based writeback test on XFS performs over 20% better as well, with the throughput being very stable around 1GB/sec, where pdflush only manages 750MB/sec and fluctuates wildly while doing so. Random buffered writes to many files behave a lot better as well, as does random mmap'ed writes. A separate thread is added to sync the super blocks. In the long term, adding sync_supers_bdi() functionality could get rid of this thread again. Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-09-09 07:08:54 +00:00
/* Keep working if default bdi still has things to do */
if (!list_empty(&me->bdi->work_list))
__set_current_state(TASK_RUNNING);
switch (action) {
case FORK_THREAD:
__set_current_state(TASK_RUNNING);
task = kthread_run(bdi_writeback_thread, &bdi->wb, "flush-%s",
dev_name(bdi->dev));
if (IS_ERR(task)) {
/*
* If thread creation fails, force writeout of
* the bdi from the thread.
*/
bdi_flush_io(bdi);
writeback: move bdi threads exiting logic to the forker thread Currently, bdi threads can decide to exit if there were no useful activities for 5 minutes. However, this causes nasty races: we can easily oops in the 'bdi_queue_work()' if the bdi thread decides to exit while we are waking it up. And even if we do not oops, but the bdi tread exits immediately after we wake it up, we'd lose the wake-up event and have an unnecessary delay (up to 5 secs) in the bdi work processing. This patch makes the forker thread to be the central place which not only creates bdi threads, but also kills them if they were inactive long enough. This better design-wise. Another reason why this change was done is to prepare for the further changes which will prevent the bdi threads from waking up every 5 sec and wasting power. Indeed, when the task does not wake up periodically anymore, it won't be able to exit either. This patch also moves the the 'wake_up_bit()' call from the bdi thread to the forker thread as well. So now the forker thread sets the BDI_pending bit, then forks the task or kills it, then clears the bit and wakes up the waiting process. The only process which may wain on the bit is 'bdi_wb_shutdown()'. This function was changed as well - now it first removes the bdi from the 'bdi_list', then waits on the 'BDI_pending' bit. Once it wakes up, it is guaranteed that the forker thread won't race with it, because the bdi is not visible. Note, the forker thread sets the 'BDI_pending' bit under the 'bdi->wb_lock' which is essential for proper serialization. And additionally, when we change 'bdi->wb.task', we now take the 'bdi->work_lock', to make sure that we do not lose wake-ups which we otherwise would when raced with, say, 'bdi_queue_work()'. Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
2010-07-25 11:29:20 +00:00
} else {
/*
* The spinlock makes sure we do not lose
* wake-ups when racing with 'bdi_queue_work()'.
*/
spin_lock_bh(&bdi->wb_lock);
bdi->wb.task = task;
spin_unlock_bh(&bdi->wb_lock);
writeback: move bdi threads exiting logic to the forker thread Currently, bdi threads can decide to exit if there were no useful activities for 5 minutes. However, this causes nasty races: we can easily oops in the 'bdi_queue_work()' if the bdi thread decides to exit while we are waking it up. And even if we do not oops, but the bdi tread exits immediately after we wake it up, we'd lose the wake-up event and have an unnecessary delay (up to 5 secs) in the bdi work processing. This patch makes the forker thread to be the central place which not only creates bdi threads, but also kills them if they were inactive long enough. This better design-wise. Another reason why this change was done is to prepare for the further changes which will prevent the bdi threads from waking up every 5 sec and wasting power. Indeed, when the task does not wake up periodically anymore, it won't be able to exit either. This patch also moves the the 'wake_up_bit()' call from the bdi thread to the forker thread as well. So now the forker thread sets the BDI_pending bit, then forks the task or kills it, then clears the bit and wakes up the waiting process. The only process which may wain on the bit is 'bdi_wb_shutdown()'. This function was changed as well - now it first removes the bdi from the 'bdi_list', then waits on the 'BDI_pending' bit. Once it wakes up, it is guaranteed that the forker thread won't race with it, because the bdi is not visible. Note, the forker thread sets the 'BDI_pending' bit under the 'bdi->wb_lock' which is essential for proper serialization. And additionally, when we change 'bdi->wb.task', we now take the 'bdi->work_lock', to make sure that we do not lose wake-ups which we otherwise would when raced with, say, 'bdi_queue_work()'. Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
2010-07-25 11:29:20 +00:00
}
break;
case KILL_THREAD:
__set_current_state(TASK_RUNNING);
kthread_stop(task);
break;
writeback: switch to per-bdi threads for flushing data This gets rid of pdflush for bdi writeout and kupdated style cleaning. pdflush writeout suffers from lack of locality and also requires more threads to handle the same workload, since it has to work in a non-blocking fashion against each queue. This also introduces lumpy behaviour and potential request starvation, since pdflush can be starved for queue access if others are accessing it. A sample ffsb workload that does random writes to files is about 8% faster here on a simple SATA drive during the benchmark phase. File layout also seems a LOT more smooth in vmstat: r b swpd free buff cache si so bi bo in cs us sy id wa 0 1 0 608848 2652 375372 0 0 0 71024 604 24 1 10 48 42 0 1 0 549644 2712 433736 0 0 0 60692 505 27 1 8 48 44 1 0 0 476928 2784 505192 0 0 4 29540 553 24 0 9 53 37 0 1 0 457972 2808 524008 0 0 0 54876 331 16 0 4 38 58 0 1 0 366128 2928 614284 0 0 4 92168 710 58 0 13 53 34 0 1 0 295092 3000 684140 0 0 0 62924 572 23 0 9 53 37 0 1 0 236592 3064 741704 0 0 4 58256 523 17 0 8 48 44 0 1 0 165608 3132 811464 0 0 0 57460 560 21 0 8 54 38 0 1 0 102952 3200 873164 0 0 4 74748 540 29 1 10 48 41 0 1 0 48604 3252 926472 0 0 0 53248 469 29 0 7 47 45 where vanilla tends to fluctuate a lot in the creation phase: r b swpd free buff cache si so bi bo in cs us sy id wa 1 1 0 678716 5792 303380 0 0 0 74064 565 50 1 11 52 36 1 0 0 662488 5864 319396 0 0 4 352 302 329 0 2 47 51 0 1 0 599312 5924 381468 0 0 0 78164 516 55 0 9 51 40 0 1 0 519952 6008 459516 0 0 4 78156 622 56 1 11 52 37 1 1 0 436640 6092 541632 0 0 0 82244 622 54 0 11 48 41 0 1 0 436640 6092 541660 0 0 0 8 152 39 0 0 51 49 0 1 0 332224 6200 644252 0 0 4 102800 728 46 1 13 49 36 1 0 0 274492 6260 701056 0 0 4 12328 459 49 0 7 50 43 0 1 0 211220 6324 763356 0 0 0 106940 515 37 1 10 51 39 1 0 0 160412 6376 813468 0 0 0 8224 415 43 0 6 49 45 1 1 0 85980 6452 886556 0 0 4 113516 575 39 1 11 54 34 0 2 0 85968 6452 886620 0 0 0 1640 158 211 0 0 46 54 A 10 disk test with btrfs performs 26% faster with per-bdi flushing. A SSD based writeback test on XFS performs over 20% better as well, with the throughput being very stable around 1GB/sec, where pdflush only manages 750MB/sec and fluctuates wildly while doing so. Random buffered writes to many files behave a lot better as well, as does random mmap'ed writes. A separate thread is added to sync the super blocks. In the long term, adding sync_supers_bdi() functionality could get rid of this thread again. Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-09-09 07:08:54 +00:00
case NO_ACTION:
if (!wb_has_dirty_io(me) || !dirty_writeback_interval)
/*
* There are no dirty data. The only thing we
* should now care about is checking for
* inactive bdi threads and killing them. Thus,
* let's sleep for longer time, save energy and
* be friendly for battery-driven devices.
*/
schedule_timeout(bdi_longest_inactive());
else
schedule_timeout(msecs_to_jiffies(dirty_writeback_interval * 10));
writeback: switch to per-bdi threads for flushing data This gets rid of pdflush for bdi writeout and kupdated style cleaning. pdflush writeout suffers from lack of locality and also requires more threads to handle the same workload, since it has to work in a non-blocking fashion against each queue. This also introduces lumpy behaviour and potential request starvation, since pdflush can be starved for queue access if others are accessing it. A sample ffsb workload that does random writes to files is about 8% faster here on a simple SATA drive during the benchmark phase. File layout also seems a LOT more smooth in vmstat: r b swpd free buff cache si so bi bo in cs us sy id wa 0 1 0 608848 2652 375372 0 0 0 71024 604 24 1 10 48 42 0 1 0 549644 2712 433736 0 0 0 60692 505 27 1 8 48 44 1 0 0 476928 2784 505192 0 0 4 29540 553 24 0 9 53 37 0 1 0 457972 2808 524008 0 0 0 54876 331 16 0 4 38 58 0 1 0 366128 2928 614284 0 0 4 92168 710 58 0 13 53 34 0 1 0 295092 3000 684140 0 0 0 62924 572 23 0 9 53 37 0 1 0 236592 3064 741704 0 0 4 58256 523 17 0 8 48 44 0 1 0 165608 3132 811464 0 0 0 57460 560 21 0 8 54 38 0 1 0 102952 3200 873164 0 0 4 74748 540 29 1 10 48 41 0 1 0 48604 3252 926472 0 0 0 53248 469 29 0 7 47 45 where vanilla tends to fluctuate a lot in the creation phase: r b swpd free buff cache si so bi bo in cs us sy id wa 1 1 0 678716 5792 303380 0 0 0 74064 565 50 1 11 52 36 1 0 0 662488 5864 319396 0 0 4 352 302 329 0 2 47 51 0 1 0 599312 5924 381468 0 0 0 78164 516 55 0 9 51 40 0 1 0 519952 6008 459516 0 0 4 78156 622 56 1 11 52 37 1 1 0 436640 6092 541632 0 0 0 82244 622 54 0 11 48 41 0 1 0 436640 6092 541660 0 0 0 8 152 39 0 0 51 49 0 1 0 332224 6200 644252 0 0 4 102800 728 46 1 13 49 36 1 0 0 274492 6260 701056 0 0 4 12328 459 49 0 7 50 43 0 1 0 211220 6324 763356 0 0 0 106940 515 37 1 10 51 39 1 0 0 160412 6376 813468 0 0 0 8224 415 43 0 6 49 45 1 1 0 85980 6452 886556 0 0 4 113516 575 39 1 11 54 34 0 2 0 85968 6452 886620 0 0 0 1640 158 211 0 0 46 54 A 10 disk test with btrfs performs 26% faster with per-bdi flushing. A SSD based writeback test on XFS performs over 20% better as well, with the throughput being very stable around 1GB/sec, where pdflush only manages 750MB/sec and fluctuates wildly while doing so. Random buffered writes to many files behave a lot better as well, as does random mmap'ed writes. A separate thread is added to sync the super blocks. In the long term, adding sync_supers_bdi() functionality could get rid of this thread again. Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-09-09 07:08:54 +00:00
try_to_freeze();
/* Back to the main loop */
writeback: switch to per-bdi threads for flushing data This gets rid of pdflush for bdi writeout and kupdated style cleaning. pdflush writeout suffers from lack of locality and also requires more threads to handle the same workload, since it has to work in a non-blocking fashion against each queue. This also introduces lumpy behaviour and potential request starvation, since pdflush can be starved for queue access if others are accessing it. A sample ffsb workload that does random writes to files is about 8% faster here on a simple SATA drive during the benchmark phase. File layout also seems a LOT more smooth in vmstat: r b swpd free buff cache si so bi bo in cs us sy id wa 0 1 0 608848 2652 375372 0 0 0 71024 604 24 1 10 48 42 0 1 0 549644 2712 433736 0 0 0 60692 505 27 1 8 48 44 1 0 0 476928 2784 505192 0 0 4 29540 553 24 0 9 53 37 0 1 0 457972 2808 524008 0 0 0 54876 331 16 0 4 38 58 0 1 0 366128 2928 614284 0 0 4 92168 710 58 0 13 53 34 0 1 0 295092 3000 684140 0 0 0 62924 572 23 0 9 53 37 0 1 0 236592 3064 741704 0 0 4 58256 523 17 0 8 48 44 0 1 0 165608 3132 811464 0 0 0 57460 560 21 0 8 54 38 0 1 0 102952 3200 873164 0 0 4 74748 540 29 1 10 48 41 0 1 0 48604 3252 926472 0 0 0 53248 469 29 0 7 47 45 where vanilla tends to fluctuate a lot in the creation phase: r b swpd free buff cache si so bi bo in cs us sy id wa 1 1 0 678716 5792 303380 0 0 0 74064 565 50 1 11 52 36 1 0 0 662488 5864 319396 0 0 4 352 302 329 0 2 47 51 0 1 0 599312 5924 381468 0 0 0 78164 516 55 0 9 51 40 0 1 0 519952 6008 459516 0 0 4 78156 622 56 1 11 52 37 1 1 0 436640 6092 541632 0 0 0 82244 622 54 0 11 48 41 0 1 0 436640 6092 541660 0 0 0 8 152 39 0 0 51 49 0 1 0 332224 6200 644252 0 0 4 102800 728 46 1 13 49 36 1 0 0 274492 6260 701056 0 0 4 12328 459 49 0 7 50 43 0 1 0 211220 6324 763356 0 0 0 106940 515 37 1 10 51 39 1 0 0 160412 6376 813468 0 0 0 8224 415 43 0 6 49 45 1 1 0 85980 6452 886556 0 0 4 113516 575 39 1 11 54 34 0 2 0 85968 6452 886620 0 0 0 1640 158 211 0 0 46 54 A 10 disk test with btrfs performs 26% faster with per-bdi flushing. A SSD based writeback test on XFS performs over 20% better as well, with the throughput being very stable around 1GB/sec, where pdflush only manages 750MB/sec and fluctuates wildly while doing so. Random buffered writes to many files behave a lot better as well, as does random mmap'ed writes. A separate thread is added to sync the super blocks. In the long term, adding sync_supers_bdi() functionality could get rid of this thread again. Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-09-09 07:08:54 +00:00
continue;
}
writeback: move bdi threads exiting logic to the forker thread Currently, bdi threads can decide to exit if there were no useful activities for 5 minutes. However, this causes nasty races: we can easily oops in the 'bdi_queue_work()' if the bdi thread decides to exit while we are waking it up. And even if we do not oops, but the bdi tread exits immediately after we wake it up, we'd lose the wake-up event and have an unnecessary delay (up to 5 secs) in the bdi work processing. This patch makes the forker thread to be the central place which not only creates bdi threads, but also kills them if they were inactive long enough. This better design-wise. Another reason why this change was done is to prepare for the further changes which will prevent the bdi threads from waking up every 5 sec and wasting power. Indeed, when the task does not wake up periodically anymore, it won't be able to exit either. This patch also moves the the 'wake_up_bit()' call from the bdi thread to the forker thread as well. So now the forker thread sets the BDI_pending bit, then forks the task or kills it, then clears the bit and wakes up the waiting process. The only process which may wain on the bit is 'bdi_wb_shutdown()'. This function was changed as well - now it first removes the bdi from the 'bdi_list', then waits on the 'BDI_pending' bit. Once it wakes up, it is guaranteed that the forker thread won't race with it, because the bdi is not visible. Note, the forker thread sets the 'BDI_pending' bit under the 'bdi->wb_lock' which is essential for proper serialization. And additionally, when we change 'bdi->wb.task', we now take the 'bdi->work_lock', to make sure that we do not lose wake-ups which we otherwise would when raced with, say, 'bdi_queue_work()'. Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
2010-07-25 11:29:20 +00:00
/*
* Clear pending bit and wakeup anybody waiting to tear us down.
*/
clear_bit(BDI_pending, &bdi->state);
smp_mb__after_clear_bit();
wake_up_bit(&bdi->state, BDI_pending);
writeback: switch to per-bdi threads for flushing data This gets rid of pdflush for bdi writeout and kupdated style cleaning. pdflush writeout suffers from lack of locality and also requires more threads to handle the same workload, since it has to work in a non-blocking fashion against each queue. This also introduces lumpy behaviour and potential request starvation, since pdflush can be starved for queue access if others are accessing it. A sample ffsb workload that does random writes to files is about 8% faster here on a simple SATA drive during the benchmark phase. File layout also seems a LOT more smooth in vmstat: r b swpd free buff cache si so bi bo in cs us sy id wa 0 1 0 608848 2652 375372 0 0 0 71024 604 24 1 10 48 42 0 1 0 549644 2712 433736 0 0 0 60692 505 27 1 8 48 44 1 0 0 476928 2784 505192 0 0 4 29540 553 24 0 9 53 37 0 1 0 457972 2808 524008 0 0 0 54876 331 16 0 4 38 58 0 1 0 366128 2928 614284 0 0 4 92168 710 58 0 13 53 34 0 1 0 295092 3000 684140 0 0 0 62924 572 23 0 9 53 37 0 1 0 236592 3064 741704 0 0 4 58256 523 17 0 8 48 44 0 1 0 165608 3132 811464 0 0 0 57460 560 21 0 8 54 38 0 1 0 102952 3200 873164 0 0 4 74748 540 29 1 10 48 41 0 1 0 48604 3252 926472 0 0 0 53248 469 29 0 7 47 45 where vanilla tends to fluctuate a lot in the creation phase: r b swpd free buff cache si so bi bo in cs us sy id wa 1 1 0 678716 5792 303380 0 0 0 74064 565 50 1 11 52 36 1 0 0 662488 5864 319396 0 0 4 352 302 329 0 2 47 51 0 1 0 599312 5924 381468 0 0 0 78164 516 55 0 9 51 40 0 1 0 519952 6008 459516 0 0 4 78156 622 56 1 11 52 37 1 1 0 436640 6092 541632 0 0 0 82244 622 54 0 11 48 41 0 1 0 436640 6092 541660 0 0 0 8 152 39 0 0 51 49 0 1 0 332224 6200 644252 0 0 4 102800 728 46 1 13 49 36 1 0 0 274492 6260 701056 0 0 4 12328 459 49 0 7 50 43 0 1 0 211220 6324 763356 0 0 0 106940 515 37 1 10 51 39 1 0 0 160412 6376 813468 0 0 0 8224 415 43 0 6 49 45 1 1 0 85980 6452 886556 0 0 4 113516 575 39 1 11 54 34 0 2 0 85968 6452 886620 0 0 0 1640 158 211 0 0 46 54 A 10 disk test with btrfs performs 26% faster with per-bdi flushing. A SSD based writeback test on XFS performs over 20% better as well, with the throughput being very stable around 1GB/sec, where pdflush only manages 750MB/sec and fluctuates wildly while doing so. Random buffered writes to many files behave a lot better as well, as does random mmap'ed writes. A separate thread is added to sync the super blocks. In the long term, adding sync_supers_bdi() functionality could get rid of this thread again. Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-09-09 07:08:54 +00:00
}
return 0;
}
/*
* Remove bdi from bdi_list, and ensure that it is no longer visible
*/
static void bdi_remove_from_list(struct backing_dev_info *bdi)
{
spin_lock_bh(&bdi_lock);
list_del_rcu(&bdi->bdi_list);
spin_unlock_bh(&bdi_lock);
synchronize_rcu();
}
int bdi_register(struct backing_dev_info *bdi, struct device *parent,
const char *fmt, ...)
{
va_list args;
int ret = 0;
struct device *dev;
if (bdi->dev) /* The driver needs to use separate queues per device */
goto exit;
va_start(args, fmt);
dev = device_create_vargs(bdi_class, parent, MKDEV(0, 0), bdi, fmt, args);
va_end(args);
if (IS_ERR(dev)) {
ret = PTR_ERR(dev);
goto exit;
}
spin_lock_bh(&bdi_lock);
list_add_tail_rcu(&bdi->bdi_list, &bdi_list);
spin_unlock_bh(&bdi_lock);
bdi->dev = dev;
writeback: switch to per-bdi threads for flushing data This gets rid of pdflush for bdi writeout and kupdated style cleaning. pdflush writeout suffers from lack of locality and also requires more threads to handle the same workload, since it has to work in a non-blocking fashion against each queue. This also introduces lumpy behaviour and potential request starvation, since pdflush can be starved for queue access if others are accessing it. A sample ffsb workload that does random writes to files is about 8% faster here on a simple SATA drive during the benchmark phase. File layout also seems a LOT more smooth in vmstat: r b swpd free buff cache si so bi bo in cs us sy id wa 0 1 0 608848 2652 375372 0 0 0 71024 604 24 1 10 48 42 0 1 0 549644 2712 433736 0 0 0 60692 505 27 1 8 48 44 1 0 0 476928 2784 505192 0 0 4 29540 553 24 0 9 53 37 0 1 0 457972 2808 524008 0 0 0 54876 331 16 0 4 38 58 0 1 0 366128 2928 614284 0 0 4 92168 710 58 0 13 53 34 0 1 0 295092 3000 684140 0 0 0 62924 572 23 0 9 53 37 0 1 0 236592 3064 741704 0 0 4 58256 523 17 0 8 48 44 0 1 0 165608 3132 811464 0 0 0 57460 560 21 0 8 54 38 0 1 0 102952 3200 873164 0 0 4 74748 540 29 1 10 48 41 0 1 0 48604 3252 926472 0 0 0 53248 469 29 0 7 47 45 where vanilla tends to fluctuate a lot in the creation phase: r b swpd free buff cache si so bi bo in cs us sy id wa 1 1 0 678716 5792 303380 0 0 0 74064 565 50 1 11 52 36 1 0 0 662488 5864 319396 0 0 4 352 302 329 0 2 47 51 0 1 0 599312 5924 381468 0 0 0 78164 516 55 0 9 51 40 0 1 0 519952 6008 459516 0 0 4 78156 622 56 1 11 52 37 1 1 0 436640 6092 541632 0 0 0 82244 622 54 0 11 48 41 0 1 0 436640 6092 541660 0 0 0 8 152 39 0 0 51 49 0 1 0 332224 6200 644252 0 0 4 102800 728 46 1 13 49 36 1 0 0 274492 6260 701056 0 0 4 12328 459 49 0 7 50 43 0 1 0 211220 6324 763356 0 0 0 106940 515 37 1 10 51 39 1 0 0 160412 6376 813468 0 0 0 8224 415 43 0 6 49 45 1 1 0 85980 6452 886556 0 0 4 113516 575 39 1 11 54 34 0 2 0 85968 6452 886620 0 0 0 1640 158 211 0 0 46 54 A 10 disk test with btrfs performs 26% faster with per-bdi flushing. A SSD based writeback test on XFS performs over 20% better as well, with the throughput being very stable around 1GB/sec, where pdflush only manages 750MB/sec and fluctuates wildly while doing so. Random buffered writes to many files behave a lot better as well, as does random mmap'ed writes. A separate thread is added to sync the super blocks. In the long term, adding sync_supers_bdi() functionality could get rid of this thread again. Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-09-09 07:08:54 +00:00
/*
* Just start the forker thread for our default backing_dev_info,
* and add other bdi's to the list. They will get a thread created
* on-demand when they need it.
*/
if (bdi_cap_flush_forker(bdi)) {
struct bdi_writeback *wb = &bdi->wb;
wb->task = kthread_run(bdi_forker_thread, wb, "bdi-%s",
writeback: switch to per-bdi threads for flushing data This gets rid of pdflush for bdi writeout and kupdated style cleaning. pdflush writeout suffers from lack of locality and also requires more threads to handle the same workload, since it has to work in a non-blocking fashion against each queue. This also introduces lumpy behaviour and potential request starvation, since pdflush can be starved for queue access if others are accessing it. A sample ffsb workload that does random writes to files is about 8% faster here on a simple SATA drive during the benchmark phase. File layout also seems a LOT more smooth in vmstat: r b swpd free buff cache si so bi bo in cs us sy id wa 0 1 0 608848 2652 375372 0 0 0 71024 604 24 1 10 48 42 0 1 0 549644 2712 433736 0 0 0 60692 505 27 1 8 48 44 1 0 0 476928 2784 505192 0 0 4 29540 553 24 0 9 53 37 0 1 0 457972 2808 524008 0 0 0 54876 331 16 0 4 38 58 0 1 0 366128 2928 614284 0 0 4 92168 710 58 0 13 53 34 0 1 0 295092 3000 684140 0 0 0 62924 572 23 0 9 53 37 0 1 0 236592 3064 741704 0 0 4 58256 523 17 0 8 48 44 0 1 0 165608 3132 811464 0 0 0 57460 560 21 0 8 54 38 0 1 0 102952 3200 873164 0 0 4 74748 540 29 1 10 48 41 0 1 0 48604 3252 926472 0 0 0 53248 469 29 0 7 47 45 where vanilla tends to fluctuate a lot in the creation phase: r b swpd free buff cache si so bi bo in cs us sy id wa 1 1 0 678716 5792 303380 0 0 0 74064 565 50 1 11 52 36 1 0 0 662488 5864 319396 0 0 4 352 302 329 0 2 47 51 0 1 0 599312 5924 381468 0 0 0 78164 516 55 0 9 51 40 0 1 0 519952 6008 459516 0 0 4 78156 622 56 1 11 52 37 1 1 0 436640 6092 541632 0 0 0 82244 622 54 0 11 48 41 0 1 0 436640 6092 541660 0 0 0 8 152 39 0 0 51 49 0 1 0 332224 6200 644252 0 0 4 102800 728 46 1 13 49 36 1 0 0 274492 6260 701056 0 0 4 12328 459 49 0 7 50 43 0 1 0 211220 6324 763356 0 0 0 106940 515 37 1 10 51 39 1 0 0 160412 6376 813468 0 0 0 8224 415 43 0 6 49 45 1 1 0 85980 6452 886556 0 0 4 113516 575 39 1 11 54 34 0 2 0 85968 6452 886620 0 0 0 1640 158 211 0 0 46 54 A 10 disk test with btrfs performs 26% faster with per-bdi flushing. A SSD based writeback test on XFS performs over 20% better as well, with the throughput being very stable around 1GB/sec, where pdflush only manages 750MB/sec and fluctuates wildly while doing so. Random buffered writes to many files behave a lot better as well, as does random mmap'ed writes. A separate thread is added to sync the super blocks. In the long term, adding sync_supers_bdi() functionality could get rid of this thread again. Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-09-09 07:08:54 +00:00
dev_name(dev));
if (IS_ERR(wb->task)) {
wb->task = NULL;
ret = -ENOMEM;
bdi_remove_from_list(bdi);
writeback: switch to per-bdi threads for flushing data This gets rid of pdflush for bdi writeout and kupdated style cleaning. pdflush writeout suffers from lack of locality and also requires more threads to handle the same workload, since it has to work in a non-blocking fashion against each queue. This also introduces lumpy behaviour and potential request starvation, since pdflush can be starved for queue access if others are accessing it. A sample ffsb workload that does random writes to files is about 8% faster here on a simple SATA drive during the benchmark phase. File layout also seems a LOT more smooth in vmstat: r b swpd free buff cache si so bi bo in cs us sy id wa 0 1 0 608848 2652 375372 0 0 0 71024 604 24 1 10 48 42 0 1 0 549644 2712 433736 0 0 0 60692 505 27 1 8 48 44 1 0 0 476928 2784 505192 0 0 4 29540 553 24 0 9 53 37 0 1 0 457972 2808 524008 0 0 0 54876 331 16 0 4 38 58 0 1 0 366128 2928 614284 0 0 4 92168 710 58 0 13 53 34 0 1 0 295092 3000 684140 0 0 0 62924 572 23 0 9 53 37 0 1 0 236592 3064 741704 0 0 4 58256 523 17 0 8 48 44 0 1 0 165608 3132 811464 0 0 0 57460 560 21 0 8 54 38 0 1 0 102952 3200 873164 0 0 4 74748 540 29 1 10 48 41 0 1 0 48604 3252 926472 0 0 0 53248 469 29 0 7 47 45 where vanilla tends to fluctuate a lot in the creation phase: r b swpd free buff cache si so bi bo in cs us sy id wa 1 1 0 678716 5792 303380 0 0 0 74064 565 50 1 11 52 36 1 0 0 662488 5864 319396 0 0 4 352 302 329 0 2 47 51 0 1 0 599312 5924 381468 0 0 0 78164 516 55 0 9 51 40 0 1 0 519952 6008 459516 0 0 4 78156 622 56 1 11 52 37 1 1 0 436640 6092 541632 0 0 0 82244 622 54 0 11 48 41 0 1 0 436640 6092 541660 0 0 0 8 152 39 0 0 51 49 0 1 0 332224 6200 644252 0 0 4 102800 728 46 1 13 49 36 1 0 0 274492 6260 701056 0 0 4 12328 459 49 0 7 50 43 0 1 0 211220 6324 763356 0 0 0 106940 515 37 1 10 51 39 1 0 0 160412 6376 813468 0 0 0 8224 415 43 0 6 49 45 1 1 0 85980 6452 886556 0 0 4 113516 575 39 1 11 54 34 0 2 0 85968 6452 886620 0 0 0 1640 158 211 0 0 46 54 A 10 disk test with btrfs performs 26% faster with per-bdi flushing. A SSD based writeback test on XFS performs over 20% better as well, with the throughput being very stable around 1GB/sec, where pdflush only manages 750MB/sec and fluctuates wildly while doing so. Random buffered writes to many files behave a lot better as well, as does random mmap'ed writes. A separate thread is added to sync the super blocks. In the long term, adding sync_supers_bdi() functionality could get rid of this thread again. Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-09-09 07:08:54 +00:00
goto exit;
}
}
bdi_debug_register(bdi, dev_name(dev));
set_bit(BDI_registered, &bdi->state);
trace_writeback_bdi_register(bdi);
exit:
return ret;
}
EXPORT_SYMBOL(bdi_register);
int bdi_register_dev(struct backing_dev_info *bdi, dev_t dev)
{
return bdi_register(bdi, NULL, "%u:%u", MAJOR(dev), MINOR(dev));
}
EXPORT_SYMBOL(bdi_register_dev);
writeback: switch to per-bdi threads for flushing data This gets rid of pdflush for bdi writeout and kupdated style cleaning. pdflush writeout suffers from lack of locality and also requires more threads to handle the same workload, since it has to work in a non-blocking fashion against each queue. This also introduces lumpy behaviour and potential request starvation, since pdflush can be starved for queue access if others are accessing it. A sample ffsb workload that does random writes to files is about 8% faster here on a simple SATA drive during the benchmark phase. File layout also seems a LOT more smooth in vmstat: r b swpd free buff cache si so bi bo in cs us sy id wa 0 1 0 608848 2652 375372 0 0 0 71024 604 24 1 10 48 42 0 1 0 549644 2712 433736 0 0 0 60692 505 27 1 8 48 44 1 0 0 476928 2784 505192 0 0 4 29540 553 24 0 9 53 37 0 1 0 457972 2808 524008 0 0 0 54876 331 16 0 4 38 58 0 1 0 366128 2928 614284 0 0 4 92168 710 58 0 13 53 34 0 1 0 295092 3000 684140 0 0 0 62924 572 23 0 9 53 37 0 1 0 236592 3064 741704 0 0 4 58256 523 17 0 8 48 44 0 1 0 165608 3132 811464 0 0 0 57460 560 21 0 8 54 38 0 1 0 102952 3200 873164 0 0 4 74748 540 29 1 10 48 41 0 1 0 48604 3252 926472 0 0 0 53248 469 29 0 7 47 45 where vanilla tends to fluctuate a lot in the creation phase: r b swpd free buff cache si so bi bo in cs us sy id wa 1 1 0 678716 5792 303380 0 0 0 74064 565 50 1 11 52 36 1 0 0 662488 5864 319396 0 0 4 352 302 329 0 2 47 51 0 1 0 599312 5924 381468 0 0 0 78164 516 55 0 9 51 40 0 1 0 519952 6008 459516 0 0 4 78156 622 56 1 11 52 37 1 1 0 436640 6092 541632 0 0 0 82244 622 54 0 11 48 41 0 1 0 436640 6092 541660 0 0 0 8 152 39 0 0 51 49 0 1 0 332224 6200 644252 0 0 4 102800 728 46 1 13 49 36 1 0 0 274492 6260 701056 0 0 4 12328 459 49 0 7 50 43 0 1 0 211220 6324 763356 0 0 0 106940 515 37 1 10 51 39 1 0 0 160412 6376 813468 0 0 0 8224 415 43 0 6 49 45 1 1 0 85980 6452 886556 0 0 4 113516 575 39 1 11 54 34 0 2 0 85968 6452 886620 0 0 0 1640 158 211 0 0 46 54 A 10 disk test with btrfs performs 26% faster with per-bdi flushing. A SSD based writeback test on XFS performs over 20% better as well, with the throughput being very stable around 1GB/sec, where pdflush only manages 750MB/sec and fluctuates wildly while doing so. Random buffered writes to many files behave a lot better as well, as does random mmap'ed writes. A separate thread is added to sync the super blocks. In the long term, adding sync_supers_bdi() functionality could get rid of this thread again. Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-09-09 07:08:54 +00:00
/*
* Remove bdi from the global list and shutdown any threads we have running
*/
static void bdi_wb_shutdown(struct backing_dev_info *bdi)
{
writeback: switch to per-bdi threads for flushing data This gets rid of pdflush for bdi writeout and kupdated style cleaning. pdflush writeout suffers from lack of locality and also requires more threads to handle the same workload, since it has to work in a non-blocking fashion against each queue. This also introduces lumpy behaviour and potential request starvation, since pdflush can be starved for queue access if others are accessing it. A sample ffsb workload that does random writes to files is about 8% faster here on a simple SATA drive during the benchmark phase. File layout also seems a LOT more smooth in vmstat: r b swpd free buff cache si so bi bo in cs us sy id wa 0 1 0 608848 2652 375372 0 0 0 71024 604 24 1 10 48 42 0 1 0 549644 2712 433736 0 0 0 60692 505 27 1 8 48 44 1 0 0 476928 2784 505192 0 0 4 29540 553 24 0 9 53 37 0 1 0 457972 2808 524008 0 0 0 54876 331 16 0 4 38 58 0 1 0 366128 2928 614284 0 0 4 92168 710 58 0 13 53 34 0 1 0 295092 3000 684140 0 0 0 62924 572 23 0 9 53 37 0 1 0 236592 3064 741704 0 0 4 58256 523 17 0 8 48 44 0 1 0 165608 3132 811464 0 0 0 57460 560 21 0 8 54 38 0 1 0 102952 3200 873164 0 0 4 74748 540 29 1 10 48 41 0 1 0 48604 3252 926472 0 0 0 53248 469 29 0 7 47 45 where vanilla tends to fluctuate a lot in the creation phase: r b swpd free buff cache si so bi bo in cs us sy id wa 1 1 0 678716 5792 303380 0 0 0 74064 565 50 1 11 52 36 1 0 0 662488 5864 319396 0 0 4 352 302 329 0 2 47 51 0 1 0 599312 5924 381468 0 0 0 78164 516 55 0 9 51 40 0 1 0 519952 6008 459516 0 0 4 78156 622 56 1 11 52 37 1 1 0 436640 6092 541632 0 0 0 82244 622 54 0 11 48 41 0 1 0 436640 6092 541660 0 0 0 8 152 39 0 0 51 49 0 1 0 332224 6200 644252 0 0 4 102800 728 46 1 13 49 36 1 0 0 274492 6260 701056 0 0 4 12328 459 49 0 7 50 43 0 1 0 211220 6324 763356 0 0 0 106940 515 37 1 10 51 39 1 0 0 160412 6376 813468 0 0 0 8224 415 43 0 6 49 45 1 1 0 85980 6452 886556 0 0 4 113516 575 39 1 11 54 34 0 2 0 85968 6452 886620 0 0 0 1640 158 211 0 0 46 54 A 10 disk test with btrfs performs 26% faster with per-bdi flushing. A SSD based writeback test on XFS performs over 20% better as well, with the throughput being very stable around 1GB/sec, where pdflush only manages 750MB/sec and fluctuates wildly while doing so. Random buffered writes to many files behave a lot better as well, as does random mmap'ed writes. A separate thread is added to sync the super blocks. In the long term, adding sync_supers_bdi() functionality could get rid of this thread again. Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-09-09 07:08:54 +00:00
if (!bdi_cap_writeback_dirty(bdi))
return;
/*
writeback: move bdi threads exiting logic to the forker thread Currently, bdi threads can decide to exit if there were no useful activities for 5 minutes. However, this causes nasty races: we can easily oops in the 'bdi_queue_work()' if the bdi thread decides to exit while we are waking it up. And even if we do not oops, but the bdi tread exits immediately after we wake it up, we'd lose the wake-up event and have an unnecessary delay (up to 5 secs) in the bdi work processing. This patch makes the forker thread to be the central place which not only creates bdi threads, but also kills them if they were inactive long enough. This better design-wise. Another reason why this change was done is to prepare for the further changes which will prevent the bdi threads from waking up every 5 sec and wasting power. Indeed, when the task does not wake up periodically anymore, it won't be able to exit either. This patch also moves the the 'wake_up_bit()' call from the bdi thread to the forker thread as well. So now the forker thread sets the BDI_pending bit, then forks the task or kills it, then clears the bit and wakes up the waiting process. The only process which may wain on the bit is 'bdi_wb_shutdown()'. This function was changed as well - now it first removes the bdi from the 'bdi_list', then waits on the 'BDI_pending' bit. Once it wakes up, it is guaranteed that the forker thread won't race with it, because the bdi is not visible. Note, the forker thread sets the 'BDI_pending' bit under the 'bdi->wb_lock' which is essential for proper serialization. And additionally, when we change 'bdi->wb.task', we now take the 'bdi->work_lock', to make sure that we do not lose wake-ups which we otherwise would when raced with, say, 'bdi_queue_work()'. Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
2010-07-25 11:29:20 +00:00
* Make sure nobody finds us on the bdi_list anymore
writeback: switch to per-bdi threads for flushing data This gets rid of pdflush for bdi writeout and kupdated style cleaning. pdflush writeout suffers from lack of locality and also requires more threads to handle the same workload, since it has to work in a non-blocking fashion against each queue. This also introduces lumpy behaviour and potential request starvation, since pdflush can be starved for queue access if others are accessing it. A sample ffsb workload that does random writes to files is about 8% faster here on a simple SATA drive during the benchmark phase. File layout also seems a LOT more smooth in vmstat: r b swpd free buff cache si so bi bo in cs us sy id wa 0 1 0 608848 2652 375372 0 0 0 71024 604 24 1 10 48 42 0 1 0 549644 2712 433736 0 0 0 60692 505 27 1 8 48 44 1 0 0 476928 2784 505192 0 0 4 29540 553 24 0 9 53 37 0 1 0 457972 2808 524008 0 0 0 54876 331 16 0 4 38 58 0 1 0 366128 2928 614284 0 0 4 92168 710 58 0 13 53 34 0 1 0 295092 3000 684140 0 0 0 62924 572 23 0 9 53 37 0 1 0 236592 3064 741704 0 0 4 58256 523 17 0 8 48 44 0 1 0 165608 3132 811464 0 0 0 57460 560 21 0 8 54 38 0 1 0 102952 3200 873164 0 0 4 74748 540 29 1 10 48 41 0 1 0 48604 3252 926472 0 0 0 53248 469 29 0 7 47 45 where vanilla tends to fluctuate a lot in the creation phase: r b swpd free buff cache si so bi bo in cs us sy id wa 1 1 0 678716 5792 303380 0 0 0 74064 565 50 1 11 52 36 1 0 0 662488 5864 319396 0 0 4 352 302 329 0 2 47 51 0 1 0 599312 5924 381468 0 0 0 78164 516 55 0 9 51 40 0 1 0 519952 6008 459516 0 0 4 78156 622 56 1 11 52 37 1 1 0 436640 6092 541632 0 0 0 82244 622 54 0 11 48 41 0 1 0 436640 6092 541660 0 0 0 8 152 39 0 0 51 49 0 1 0 332224 6200 644252 0 0 4 102800 728 46 1 13 49 36 1 0 0 274492 6260 701056 0 0 4 12328 459 49 0 7 50 43 0 1 0 211220 6324 763356 0 0 0 106940 515 37 1 10 51 39 1 0 0 160412 6376 813468 0 0 0 8224 415 43 0 6 49 45 1 1 0 85980 6452 886556 0 0 4 113516 575 39 1 11 54 34 0 2 0 85968 6452 886620 0 0 0 1640 158 211 0 0 46 54 A 10 disk test with btrfs performs 26% faster with per-bdi flushing. A SSD based writeback test on XFS performs over 20% better as well, with the throughput being very stable around 1GB/sec, where pdflush only manages 750MB/sec and fluctuates wildly while doing so. Random buffered writes to many files behave a lot better as well, as does random mmap'ed writes. A separate thread is added to sync the super blocks. In the long term, adding sync_supers_bdi() functionality could get rid of this thread again. Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-09-09 07:08:54 +00:00
*/
writeback: move bdi threads exiting logic to the forker thread Currently, bdi threads can decide to exit if there were no useful activities for 5 minutes. However, this causes nasty races: we can easily oops in the 'bdi_queue_work()' if the bdi thread decides to exit while we are waking it up. And even if we do not oops, but the bdi tread exits immediately after we wake it up, we'd lose the wake-up event and have an unnecessary delay (up to 5 secs) in the bdi work processing. This patch makes the forker thread to be the central place which not only creates bdi threads, but also kills them if they were inactive long enough. This better design-wise. Another reason why this change was done is to prepare for the further changes which will prevent the bdi threads from waking up every 5 sec and wasting power. Indeed, when the task does not wake up periodically anymore, it won't be able to exit either. This patch also moves the the 'wake_up_bit()' call from the bdi thread to the forker thread as well. So now the forker thread sets the BDI_pending bit, then forks the task or kills it, then clears the bit and wakes up the waiting process. The only process which may wain on the bit is 'bdi_wb_shutdown()'. This function was changed as well - now it first removes the bdi from the 'bdi_list', then waits on the 'BDI_pending' bit. Once it wakes up, it is guaranteed that the forker thread won't race with it, because the bdi is not visible. Note, the forker thread sets the 'BDI_pending' bit under the 'bdi->wb_lock' which is essential for proper serialization. And additionally, when we change 'bdi->wb.task', we now take the 'bdi->work_lock', to make sure that we do not lose wake-ups which we otherwise would when raced with, say, 'bdi_queue_work()'. Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
2010-07-25 11:29:20 +00:00
bdi_remove_from_list(bdi);
writeback: switch to per-bdi threads for flushing data This gets rid of pdflush for bdi writeout and kupdated style cleaning. pdflush writeout suffers from lack of locality and also requires more threads to handle the same workload, since it has to work in a non-blocking fashion against each queue. This also introduces lumpy behaviour and potential request starvation, since pdflush can be starved for queue access if others are accessing it. A sample ffsb workload that does random writes to files is about 8% faster here on a simple SATA drive during the benchmark phase. File layout also seems a LOT more smooth in vmstat: r b swpd free buff cache si so bi bo in cs us sy id wa 0 1 0 608848 2652 375372 0 0 0 71024 604 24 1 10 48 42 0 1 0 549644 2712 433736 0 0 0 60692 505 27 1 8 48 44 1 0 0 476928 2784 505192 0 0 4 29540 553 24 0 9 53 37 0 1 0 457972 2808 524008 0 0 0 54876 331 16 0 4 38 58 0 1 0 366128 2928 614284 0 0 4 92168 710 58 0 13 53 34 0 1 0 295092 3000 684140 0 0 0 62924 572 23 0 9 53 37 0 1 0 236592 3064 741704 0 0 4 58256 523 17 0 8 48 44 0 1 0 165608 3132 811464 0 0 0 57460 560 21 0 8 54 38 0 1 0 102952 3200 873164 0 0 4 74748 540 29 1 10 48 41 0 1 0 48604 3252 926472 0 0 0 53248 469 29 0 7 47 45 where vanilla tends to fluctuate a lot in the creation phase: r b swpd free buff cache si so bi bo in cs us sy id wa 1 1 0 678716 5792 303380 0 0 0 74064 565 50 1 11 52 36 1 0 0 662488 5864 319396 0 0 4 352 302 329 0 2 47 51 0 1 0 599312 5924 381468 0 0 0 78164 516 55 0 9 51 40 0 1 0 519952 6008 459516 0 0 4 78156 622 56 1 11 52 37 1 1 0 436640 6092 541632 0 0 0 82244 622 54 0 11 48 41 0 1 0 436640 6092 541660 0 0 0 8 152 39 0 0 51 49 0 1 0 332224 6200 644252 0 0 4 102800 728 46 1 13 49 36 1 0 0 274492 6260 701056 0 0 4 12328 459 49 0 7 50 43 0 1 0 211220 6324 763356 0 0 0 106940 515 37 1 10 51 39 1 0 0 160412 6376 813468 0 0 0 8224 415 43 0 6 49 45 1 1 0 85980 6452 886556 0 0 4 113516 575 39 1 11 54 34 0 2 0 85968 6452 886620 0 0 0 1640 158 211 0 0 46 54 A 10 disk test with btrfs performs 26% faster with per-bdi flushing. A SSD based writeback test on XFS performs over 20% better as well, with the throughput being very stable around 1GB/sec, where pdflush only manages 750MB/sec and fluctuates wildly while doing so. Random buffered writes to many files behave a lot better as well, as does random mmap'ed writes. A separate thread is added to sync the super blocks. In the long term, adding sync_supers_bdi() functionality could get rid of this thread again. Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-09-09 07:08:54 +00:00
/*
writeback: move bdi threads exiting logic to the forker thread Currently, bdi threads can decide to exit if there were no useful activities for 5 minutes. However, this causes nasty races: we can easily oops in the 'bdi_queue_work()' if the bdi thread decides to exit while we are waking it up. And even if we do not oops, but the bdi tread exits immediately after we wake it up, we'd lose the wake-up event and have an unnecessary delay (up to 5 secs) in the bdi work processing. This patch makes the forker thread to be the central place which not only creates bdi threads, but also kills them if they were inactive long enough. This better design-wise. Another reason why this change was done is to prepare for the further changes which will prevent the bdi threads from waking up every 5 sec and wasting power. Indeed, when the task does not wake up periodically anymore, it won't be able to exit either. This patch also moves the the 'wake_up_bit()' call from the bdi thread to the forker thread as well. So now the forker thread sets the BDI_pending bit, then forks the task or kills it, then clears the bit and wakes up the waiting process. The only process which may wain on the bit is 'bdi_wb_shutdown()'. This function was changed as well - now it first removes the bdi from the 'bdi_list', then waits on the 'BDI_pending' bit. Once it wakes up, it is guaranteed that the forker thread won't race with it, because the bdi is not visible. Note, the forker thread sets the 'BDI_pending' bit under the 'bdi->wb_lock' which is essential for proper serialization. And additionally, when we change 'bdi->wb.task', we now take the 'bdi->work_lock', to make sure that we do not lose wake-ups which we otherwise would when raced with, say, 'bdi_queue_work()'. Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
2010-07-25 11:29:20 +00:00
* If setup is pending, wait for that to complete first
writeback: switch to per-bdi threads for flushing data This gets rid of pdflush for bdi writeout and kupdated style cleaning. pdflush writeout suffers from lack of locality and also requires more threads to handle the same workload, since it has to work in a non-blocking fashion against each queue. This also introduces lumpy behaviour and potential request starvation, since pdflush can be starved for queue access if others are accessing it. A sample ffsb workload that does random writes to files is about 8% faster here on a simple SATA drive during the benchmark phase. File layout also seems a LOT more smooth in vmstat: r b swpd free buff cache si so bi bo in cs us sy id wa 0 1 0 608848 2652 375372 0 0 0 71024 604 24 1 10 48 42 0 1 0 549644 2712 433736 0 0 0 60692 505 27 1 8 48 44 1 0 0 476928 2784 505192 0 0 4 29540 553 24 0 9 53 37 0 1 0 457972 2808 524008 0 0 0 54876 331 16 0 4 38 58 0 1 0 366128 2928 614284 0 0 4 92168 710 58 0 13 53 34 0 1 0 295092 3000 684140 0 0 0 62924 572 23 0 9 53 37 0 1 0 236592 3064 741704 0 0 4 58256 523 17 0 8 48 44 0 1 0 165608 3132 811464 0 0 0 57460 560 21 0 8 54 38 0 1 0 102952 3200 873164 0 0 4 74748 540 29 1 10 48 41 0 1 0 48604 3252 926472 0 0 0 53248 469 29 0 7 47 45 where vanilla tends to fluctuate a lot in the creation phase: r b swpd free buff cache si so bi bo in cs us sy id wa 1 1 0 678716 5792 303380 0 0 0 74064 565 50 1 11 52 36 1 0 0 662488 5864 319396 0 0 4 352 302 329 0 2 47 51 0 1 0 599312 5924 381468 0 0 0 78164 516 55 0 9 51 40 0 1 0 519952 6008 459516 0 0 4 78156 622 56 1 11 52 37 1 1 0 436640 6092 541632 0 0 0 82244 622 54 0 11 48 41 0 1 0 436640 6092 541660 0 0 0 8 152 39 0 0 51 49 0 1 0 332224 6200 644252 0 0 4 102800 728 46 1 13 49 36 1 0 0 274492 6260 701056 0 0 4 12328 459 49 0 7 50 43 0 1 0 211220 6324 763356 0 0 0 106940 515 37 1 10 51 39 1 0 0 160412 6376 813468 0 0 0 8224 415 43 0 6 49 45 1 1 0 85980 6452 886556 0 0 4 113516 575 39 1 11 54 34 0 2 0 85968 6452 886620 0 0 0 1640 158 211 0 0 46 54 A 10 disk test with btrfs performs 26% faster with per-bdi flushing. A SSD based writeback test on XFS performs over 20% better as well, with the throughput being very stable around 1GB/sec, where pdflush only manages 750MB/sec and fluctuates wildly while doing so. Random buffered writes to many files behave a lot better as well, as does random mmap'ed writes. A separate thread is added to sync the super blocks. In the long term, adding sync_supers_bdi() functionality could get rid of this thread again. Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-09-09 07:08:54 +00:00
*/
writeback: move bdi threads exiting logic to the forker thread Currently, bdi threads can decide to exit if there were no useful activities for 5 minutes. However, this causes nasty races: we can easily oops in the 'bdi_queue_work()' if the bdi thread decides to exit while we are waking it up. And even if we do not oops, but the bdi tread exits immediately after we wake it up, we'd lose the wake-up event and have an unnecessary delay (up to 5 secs) in the bdi work processing. This patch makes the forker thread to be the central place which not only creates bdi threads, but also kills them if they were inactive long enough. This better design-wise. Another reason why this change was done is to prepare for the further changes which will prevent the bdi threads from waking up every 5 sec and wasting power. Indeed, when the task does not wake up periodically anymore, it won't be able to exit either. This patch also moves the the 'wake_up_bit()' call from the bdi thread to the forker thread as well. So now the forker thread sets the BDI_pending bit, then forks the task or kills it, then clears the bit and wakes up the waiting process. The only process which may wain on the bit is 'bdi_wb_shutdown()'. This function was changed as well - now it first removes the bdi from the 'bdi_list', then waits on the 'BDI_pending' bit. Once it wakes up, it is guaranteed that the forker thread won't race with it, because the bdi is not visible. Note, the forker thread sets the 'BDI_pending' bit under the 'bdi->wb_lock' which is essential for proper serialization. And additionally, when we change 'bdi->wb.task', we now take the 'bdi->work_lock', to make sure that we do not lose wake-ups which we otherwise would when raced with, say, 'bdi_queue_work()'. Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
2010-07-25 11:29:20 +00:00
wait_on_bit(&bdi->state, BDI_pending, bdi_sched_wait,
TASK_UNINTERRUPTIBLE);
writeback: switch to per-bdi threads for flushing data This gets rid of pdflush for bdi writeout and kupdated style cleaning. pdflush writeout suffers from lack of locality and also requires more threads to handle the same workload, since it has to work in a non-blocking fashion against each queue. This also introduces lumpy behaviour and potential request starvation, since pdflush can be starved for queue access if others are accessing it. A sample ffsb workload that does random writes to files is about 8% faster here on a simple SATA drive during the benchmark phase. File layout also seems a LOT more smooth in vmstat: r b swpd free buff cache si so bi bo in cs us sy id wa 0 1 0 608848 2652 375372 0 0 0 71024 604 24 1 10 48 42 0 1 0 549644 2712 433736 0 0 0 60692 505 27 1 8 48 44 1 0 0 476928 2784 505192 0 0 4 29540 553 24 0 9 53 37 0 1 0 457972 2808 524008 0 0 0 54876 331 16 0 4 38 58 0 1 0 366128 2928 614284 0 0 4 92168 710 58 0 13 53 34 0 1 0 295092 3000 684140 0 0 0 62924 572 23 0 9 53 37 0 1 0 236592 3064 741704 0 0 4 58256 523 17 0 8 48 44 0 1 0 165608 3132 811464 0 0 0 57460 560 21 0 8 54 38 0 1 0 102952 3200 873164 0 0 4 74748 540 29 1 10 48 41 0 1 0 48604 3252 926472 0 0 0 53248 469 29 0 7 47 45 where vanilla tends to fluctuate a lot in the creation phase: r b swpd free buff cache si so bi bo in cs us sy id wa 1 1 0 678716 5792 303380 0 0 0 74064 565 50 1 11 52 36 1 0 0 662488 5864 319396 0 0 4 352 302 329 0 2 47 51 0 1 0 599312 5924 381468 0 0 0 78164 516 55 0 9 51 40 0 1 0 519952 6008 459516 0 0 4 78156 622 56 1 11 52 37 1 1 0 436640 6092 541632 0 0 0 82244 622 54 0 11 48 41 0 1 0 436640 6092 541660 0 0 0 8 152 39 0 0 51 49 0 1 0 332224 6200 644252 0 0 4 102800 728 46 1 13 49 36 1 0 0 274492 6260 701056 0 0 4 12328 459 49 0 7 50 43 0 1 0 211220 6324 763356 0 0 0 106940 515 37 1 10 51 39 1 0 0 160412 6376 813468 0 0 0 8224 415 43 0 6 49 45 1 1 0 85980 6452 886556 0 0 4 113516 575 39 1 11 54 34 0 2 0 85968 6452 886620 0 0 0 1640 158 211 0 0 46 54 A 10 disk test with btrfs performs 26% faster with per-bdi flushing. A SSD based writeback test on XFS performs over 20% better as well, with the throughput being very stable around 1GB/sec, where pdflush only manages 750MB/sec and fluctuates wildly while doing so. Random buffered writes to many files behave a lot better as well, as does random mmap'ed writes. A separate thread is added to sync the super blocks. In the long term, adding sync_supers_bdi() functionality could get rid of this thread again. Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-09-09 07:08:54 +00:00
/*
* Finally, kill the kernel thread. We don't need to be RCU
* safe anymore, since the bdi is gone from visibility. Force
* unfreeze of the thread before calling kthread_stop(), otherwise
* it would never exet if it is currently stuck in the refrigerator.
writeback: switch to per-bdi threads for flushing data This gets rid of pdflush for bdi writeout and kupdated style cleaning. pdflush writeout suffers from lack of locality and also requires more threads to handle the same workload, since it has to work in a non-blocking fashion against each queue. This also introduces lumpy behaviour and potential request starvation, since pdflush can be starved for queue access if others are accessing it. A sample ffsb workload that does random writes to files is about 8% faster here on a simple SATA drive during the benchmark phase. File layout also seems a LOT more smooth in vmstat: r b swpd free buff cache si so bi bo in cs us sy id wa 0 1 0 608848 2652 375372 0 0 0 71024 604 24 1 10 48 42 0 1 0 549644 2712 433736 0 0 0 60692 505 27 1 8 48 44 1 0 0 476928 2784 505192 0 0 4 29540 553 24 0 9 53 37 0 1 0 457972 2808 524008 0 0 0 54876 331 16 0 4 38 58 0 1 0 366128 2928 614284 0 0 4 92168 710 58 0 13 53 34 0 1 0 295092 3000 684140 0 0 0 62924 572 23 0 9 53 37 0 1 0 236592 3064 741704 0 0 4 58256 523 17 0 8 48 44 0 1 0 165608 3132 811464 0 0 0 57460 560 21 0 8 54 38 0 1 0 102952 3200 873164 0 0 4 74748 540 29 1 10 48 41 0 1 0 48604 3252 926472 0 0 0 53248 469 29 0 7 47 45 where vanilla tends to fluctuate a lot in the creation phase: r b swpd free buff cache si so bi bo in cs us sy id wa 1 1 0 678716 5792 303380 0 0 0 74064 565 50 1 11 52 36 1 0 0 662488 5864 319396 0 0 4 352 302 329 0 2 47 51 0 1 0 599312 5924 381468 0 0 0 78164 516 55 0 9 51 40 0 1 0 519952 6008 459516 0 0 4 78156 622 56 1 11 52 37 1 1 0 436640 6092 541632 0 0 0 82244 622 54 0 11 48 41 0 1 0 436640 6092 541660 0 0 0 8 152 39 0 0 51 49 0 1 0 332224 6200 644252 0 0 4 102800 728 46 1 13 49 36 1 0 0 274492 6260 701056 0 0 4 12328 459 49 0 7 50 43 0 1 0 211220 6324 763356 0 0 0 106940 515 37 1 10 51 39 1 0 0 160412 6376 813468 0 0 0 8224 415 43 0 6 49 45 1 1 0 85980 6452 886556 0 0 4 113516 575 39 1 11 54 34 0 2 0 85968 6452 886620 0 0 0 1640 158 211 0 0 46 54 A 10 disk test with btrfs performs 26% faster with per-bdi flushing. A SSD based writeback test on XFS performs over 20% better as well, with the throughput being very stable around 1GB/sec, where pdflush only manages 750MB/sec and fluctuates wildly while doing so. Random buffered writes to many files behave a lot better as well, as does random mmap'ed writes. A separate thread is added to sync the super blocks. In the long term, adding sync_supers_bdi() functionality could get rid of this thread again. Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-09-09 07:08:54 +00:00
*/
if (bdi->wb.task) {
thaw_process(bdi->wb.task);
kthread_stop(bdi->wb.task);
}
}
/*
* This bdi is going away now, make sure that no super_blocks point to it
*/
static void bdi_prune_sb(struct backing_dev_info *bdi)
{
struct super_block *sb;
spin_lock(&sb_lock);
list_for_each_entry(sb, &super_blocks, s_list) {
if (sb->s_bdi == bdi)
sb->s_bdi = NULL;
}
spin_unlock(&sb_lock);
}
void bdi_unregister(struct backing_dev_info *bdi)
{
if (bdi->dev) {
trace_writeback_bdi_unregister(bdi);
bdi_prune_sb(bdi);
del_timer_sync(&bdi->wb.wakeup_timer);
writeback: switch to per-bdi threads for flushing data This gets rid of pdflush for bdi writeout and kupdated style cleaning. pdflush writeout suffers from lack of locality and also requires more threads to handle the same workload, since it has to work in a non-blocking fashion against each queue. This also introduces lumpy behaviour and potential request starvation, since pdflush can be starved for queue access if others are accessing it. A sample ffsb workload that does random writes to files is about 8% faster here on a simple SATA drive during the benchmark phase. File layout also seems a LOT more smooth in vmstat: r b swpd free buff cache si so bi bo in cs us sy id wa 0 1 0 608848 2652 375372 0 0 0 71024 604 24 1 10 48 42 0 1 0 549644 2712 433736 0 0 0 60692 505 27 1 8 48 44 1 0 0 476928 2784 505192 0 0 4 29540 553 24 0 9 53 37 0 1 0 457972 2808 524008 0 0 0 54876 331 16 0 4 38 58 0 1 0 366128 2928 614284 0 0 4 92168 710 58 0 13 53 34 0 1 0 295092 3000 684140 0 0 0 62924 572 23 0 9 53 37 0 1 0 236592 3064 741704 0 0 4 58256 523 17 0 8 48 44 0 1 0 165608 3132 811464 0 0 0 57460 560 21 0 8 54 38 0 1 0 102952 3200 873164 0 0 4 74748 540 29 1 10 48 41 0 1 0 48604 3252 926472 0 0 0 53248 469 29 0 7 47 45 where vanilla tends to fluctuate a lot in the creation phase: r b swpd free buff cache si so bi bo in cs us sy id wa 1 1 0 678716 5792 303380 0 0 0 74064 565 50 1 11 52 36 1 0 0 662488 5864 319396 0 0 4 352 302 329 0 2 47 51 0 1 0 599312 5924 381468 0 0 0 78164 516 55 0 9 51 40 0 1 0 519952 6008 459516 0 0 4 78156 622 56 1 11 52 37 1 1 0 436640 6092 541632 0 0 0 82244 622 54 0 11 48 41 0 1 0 436640 6092 541660 0 0 0 8 152 39 0 0 51 49 0 1 0 332224 6200 644252 0 0 4 102800 728 46 1 13 49 36 1 0 0 274492 6260 701056 0 0 4 12328 459 49 0 7 50 43 0 1 0 211220 6324 763356 0 0 0 106940 515 37 1 10 51 39 1 0 0 160412 6376 813468 0 0 0 8224 415 43 0 6 49 45 1 1 0 85980 6452 886556 0 0 4 113516 575 39 1 11 54 34 0 2 0 85968 6452 886620 0 0 0 1640 158 211 0 0 46 54 A 10 disk test with btrfs performs 26% faster with per-bdi flushing. A SSD based writeback test on XFS performs over 20% better as well, with the throughput being very stable around 1GB/sec, where pdflush only manages 750MB/sec and fluctuates wildly while doing so. Random buffered writes to many files behave a lot better as well, as does random mmap'ed writes. A separate thread is added to sync the super blocks. In the long term, adding sync_supers_bdi() functionality could get rid of this thread again. Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-09-09 07:08:54 +00:00
if (!bdi_cap_flush_forker(bdi))
bdi_wb_shutdown(bdi);
bdi_debug_unregister(bdi);
device_unregister(bdi->dev);
bdi->dev = NULL;
}
}
EXPORT_SYMBOL(bdi_unregister);
static void bdi_wb_init(struct bdi_writeback *wb, struct backing_dev_info *bdi)
{
memset(wb, 0, sizeof(*wb));
wb->bdi = bdi;
wb->last_old_flush = jiffies;
INIT_LIST_HEAD(&wb->b_dirty);
INIT_LIST_HEAD(&wb->b_io);
INIT_LIST_HEAD(&wb->b_more_io);
setup_timer(&wb->wakeup_timer, wakeup_timer_fn, (unsigned long)bdi);
}
int bdi_init(struct backing_dev_info *bdi)
{
writeback: switch to per-bdi threads for flushing data This gets rid of pdflush for bdi writeout and kupdated style cleaning. pdflush writeout suffers from lack of locality and also requires more threads to handle the same workload, since it has to work in a non-blocking fashion against each queue. This also introduces lumpy behaviour and potential request starvation, since pdflush can be starved for queue access if others are accessing it. A sample ffsb workload that does random writes to files is about 8% faster here on a simple SATA drive during the benchmark phase. File layout also seems a LOT more smooth in vmstat: r b swpd free buff cache si so bi bo in cs us sy id wa 0 1 0 608848 2652 375372 0 0 0 71024 604 24 1 10 48 42 0 1 0 549644 2712 433736 0 0 0 60692 505 27 1 8 48 44 1 0 0 476928 2784 505192 0 0 4 29540 553 24 0 9 53 37 0 1 0 457972 2808 524008 0 0 0 54876 331 16 0 4 38 58 0 1 0 366128 2928 614284 0 0 4 92168 710 58 0 13 53 34 0 1 0 295092 3000 684140 0 0 0 62924 572 23 0 9 53 37 0 1 0 236592 3064 741704 0 0 4 58256 523 17 0 8 48 44 0 1 0 165608 3132 811464 0 0 0 57460 560 21 0 8 54 38 0 1 0 102952 3200 873164 0 0 4 74748 540 29 1 10 48 41 0 1 0 48604 3252 926472 0 0 0 53248 469 29 0 7 47 45 where vanilla tends to fluctuate a lot in the creation phase: r b swpd free buff cache si so bi bo in cs us sy id wa 1 1 0 678716 5792 303380 0 0 0 74064 565 50 1 11 52 36 1 0 0 662488 5864 319396 0 0 4 352 302 329 0 2 47 51 0 1 0 599312 5924 381468 0 0 0 78164 516 55 0 9 51 40 0 1 0 519952 6008 459516 0 0 4 78156 622 56 1 11 52 37 1 1 0 436640 6092 541632 0 0 0 82244 622 54 0 11 48 41 0 1 0 436640 6092 541660 0 0 0 8 152 39 0 0 51 49 0 1 0 332224 6200 644252 0 0 4 102800 728 46 1 13 49 36 1 0 0 274492 6260 701056 0 0 4 12328 459 49 0 7 50 43 0 1 0 211220 6324 763356 0 0 0 106940 515 37 1 10 51 39 1 0 0 160412 6376 813468 0 0 0 8224 415 43 0 6 49 45 1 1 0 85980 6452 886556 0 0 4 113516 575 39 1 11 54 34 0 2 0 85968 6452 886620 0 0 0 1640 158 211 0 0 46 54 A 10 disk test with btrfs performs 26% faster with per-bdi flushing. A SSD based writeback test on XFS performs over 20% better as well, with the throughput being very stable around 1GB/sec, where pdflush only manages 750MB/sec and fluctuates wildly while doing so. Random buffered writes to many files behave a lot better as well, as does random mmap'ed writes. A separate thread is added to sync the super blocks. In the long term, adding sync_supers_bdi() functionality could get rid of this thread again. Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-09-09 07:08:54 +00:00
int i, err;
bdi->dev = NULL;
bdi->min_ratio = 0;
bdi->max_ratio = 100;
bdi->max_prop_frac = PROP_FRAC_BASE;
writeback: switch to per-bdi threads for flushing data This gets rid of pdflush for bdi writeout and kupdated style cleaning. pdflush writeout suffers from lack of locality and also requires more threads to handle the same workload, since it has to work in a non-blocking fashion against each queue. This also introduces lumpy behaviour and potential request starvation, since pdflush can be starved for queue access if others are accessing it. A sample ffsb workload that does random writes to files is about 8% faster here on a simple SATA drive during the benchmark phase. File layout also seems a LOT more smooth in vmstat: r b swpd free buff cache si so bi bo in cs us sy id wa 0 1 0 608848 2652 375372 0 0 0 71024 604 24 1 10 48 42 0 1 0 549644 2712 433736 0 0 0 60692 505 27 1 8 48 44 1 0 0 476928 2784 505192 0 0 4 29540 553 24 0 9 53 37 0 1 0 457972 2808 524008 0 0 0 54876 331 16 0 4 38 58 0 1 0 366128 2928 614284 0 0 4 92168 710 58 0 13 53 34 0 1 0 295092 3000 684140 0 0 0 62924 572 23 0 9 53 37 0 1 0 236592 3064 741704 0 0 4 58256 523 17 0 8 48 44 0 1 0 165608 3132 811464 0 0 0 57460 560 21 0 8 54 38 0 1 0 102952 3200 873164 0 0 4 74748 540 29 1 10 48 41 0 1 0 48604 3252 926472 0 0 0 53248 469 29 0 7 47 45 where vanilla tends to fluctuate a lot in the creation phase: r b swpd free buff cache si so bi bo in cs us sy id wa 1 1 0 678716 5792 303380 0 0 0 74064 565 50 1 11 52 36 1 0 0 662488 5864 319396 0 0 4 352 302 329 0 2 47 51 0 1 0 599312 5924 381468 0 0 0 78164 516 55 0 9 51 40 0 1 0 519952 6008 459516 0 0 4 78156 622 56 1 11 52 37 1 1 0 436640 6092 541632 0 0 0 82244 622 54 0 11 48 41 0 1 0 436640 6092 541660 0 0 0 8 152 39 0 0 51 49 0 1 0 332224 6200 644252 0 0 4 102800 728 46 1 13 49 36 1 0 0 274492 6260 701056 0 0 4 12328 459 49 0 7 50 43 0 1 0 211220 6324 763356 0 0 0 106940 515 37 1 10 51 39 1 0 0 160412 6376 813468 0 0 0 8224 415 43 0 6 49 45 1 1 0 85980 6452 886556 0 0 4 113516 575 39 1 11 54 34 0 2 0 85968 6452 886620 0 0 0 1640 158 211 0 0 46 54 A 10 disk test with btrfs performs 26% faster with per-bdi flushing. A SSD based writeback test on XFS performs over 20% better as well, with the throughput being very stable around 1GB/sec, where pdflush only manages 750MB/sec and fluctuates wildly while doing so. Random buffered writes to many files behave a lot better as well, as does random mmap'ed writes. A separate thread is added to sync the super blocks. In the long term, adding sync_supers_bdi() functionality could get rid of this thread again. Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-09-09 07:08:54 +00:00
spin_lock_init(&bdi->wb_lock);
INIT_LIST_HEAD(&bdi->bdi_list);
writeback: switch to per-bdi threads for flushing data This gets rid of pdflush for bdi writeout and kupdated style cleaning. pdflush writeout suffers from lack of locality and also requires more threads to handle the same workload, since it has to work in a non-blocking fashion against each queue. This also introduces lumpy behaviour and potential request starvation, since pdflush can be starved for queue access if others are accessing it. A sample ffsb workload that does random writes to files is about 8% faster here on a simple SATA drive during the benchmark phase. File layout also seems a LOT more smooth in vmstat: r b swpd free buff cache si so bi bo in cs us sy id wa 0 1 0 608848 2652 375372 0 0 0 71024 604 24 1 10 48 42 0 1 0 549644 2712 433736 0 0 0 60692 505 27 1 8 48 44 1 0 0 476928 2784 505192 0 0 4 29540 553 24 0 9 53 37 0 1 0 457972 2808 524008 0 0 0 54876 331 16 0 4 38 58 0 1 0 366128 2928 614284 0 0 4 92168 710 58 0 13 53 34 0 1 0 295092 3000 684140 0 0 0 62924 572 23 0 9 53 37 0 1 0 236592 3064 741704 0 0 4 58256 523 17 0 8 48 44 0 1 0 165608 3132 811464 0 0 0 57460 560 21 0 8 54 38 0 1 0 102952 3200 873164 0 0 4 74748 540 29 1 10 48 41 0 1 0 48604 3252 926472 0 0 0 53248 469 29 0 7 47 45 where vanilla tends to fluctuate a lot in the creation phase: r b swpd free buff cache si so bi bo in cs us sy id wa 1 1 0 678716 5792 303380 0 0 0 74064 565 50 1 11 52 36 1 0 0 662488 5864 319396 0 0 4 352 302 329 0 2 47 51 0 1 0 599312 5924 381468 0 0 0 78164 516 55 0 9 51 40 0 1 0 519952 6008 459516 0 0 4 78156 622 56 1 11 52 37 1 1 0 436640 6092 541632 0 0 0 82244 622 54 0 11 48 41 0 1 0 436640 6092 541660 0 0 0 8 152 39 0 0 51 49 0 1 0 332224 6200 644252 0 0 4 102800 728 46 1 13 49 36 1 0 0 274492 6260 701056 0 0 4 12328 459 49 0 7 50 43 0 1 0 211220 6324 763356 0 0 0 106940 515 37 1 10 51 39 1 0 0 160412 6376 813468 0 0 0 8224 415 43 0 6 49 45 1 1 0 85980 6452 886556 0 0 4 113516 575 39 1 11 54 34 0 2 0 85968 6452 886620 0 0 0 1640 158 211 0 0 46 54 A 10 disk test with btrfs performs 26% faster with per-bdi flushing. A SSD based writeback test on XFS performs over 20% better as well, with the throughput being very stable around 1GB/sec, where pdflush only manages 750MB/sec and fluctuates wildly while doing so. Random buffered writes to many files behave a lot better as well, as does random mmap'ed writes. A separate thread is added to sync the super blocks. In the long term, adding sync_supers_bdi() functionality could get rid of this thread again. Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2009-09-09 07:08:54 +00:00
INIT_LIST_HEAD(&bdi->work_list);
bdi_wb_init(&bdi->wb, bdi);
for (i = 0; i < NR_BDI_STAT_ITEMS; i++) {
err = percpu_counter_init(&bdi->bdi_stat[i], 0);
mm: per device dirty threshold Scale writeback cache per backing device, proportional to its writeout speed. By decoupling the BDI dirty thresholds a number of problems we currently have will go away, namely: - mutual interference starvation (for any number of BDIs); - deadlocks with stacked BDIs (loop, FUSE and local NFS mounts). It might be that all dirty pages are for a single BDI while other BDIs are idling. By giving each BDI a 'fair' share of the dirty limit, each one can have dirty pages outstanding and make progress. A global threshold also creates a deadlock for stacked BDIs; when A writes to B, and A generates enough dirty pages to get throttled, B will never start writeback until the dirty pages go away. Again, by giving each BDI its own 'independent' dirty limit, this problem is avoided. So the problem is to determine how to distribute the total dirty limit across the BDIs fairly and efficiently. A DBI that has a large dirty limit but does not have any dirty pages outstanding is a waste. What is done is to keep a floating proportion between the DBIs based on writeback completions. This way faster/more active devices get a larger share than slower/idle devices. [akpm@linux-foundation.org: fix warnings] [hugh@veritas.com: Fix occasional hang when a task couldn't get out of balance_dirty_pages] Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Signed-off-by: Hugh Dickins <hugh@veritas.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-17 06:25:50 +00:00
if (err)
goto err;
}
bdi->dirty_exceeded = 0;
err = prop_local_init_percpu(&bdi->completions);
if (err) {
err:
while (i--)
mm: per device dirty threshold Scale writeback cache per backing device, proportional to its writeout speed. By decoupling the BDI dirty thresholds a number of problems we currently have will go away, namely: - mutual interference starvation (for any number of BDIs); - deadlocks with stacked BDIs (loop, FUSE and local NFS mounts). It might be that all dirty pages are for a single BDI while other BDIs are idling. By giving each BDI a 'fair' share of the dirty limit, each one can have dirty pages outstanding and make progress. A global threshold also creates a deadlock for stacked BDIs; when A writes to B, and A generates enough dirty pages to get throttled, B will never start writeback until the dirty pages go away. Again, by giving each BDI its own 'independent' dirty limit, this problem is avoided. So the problem is to determine how to distribute the total dirty limit across the BDIs fairly and efficiently. A DBI that has a large dirty limit but does not have any dirty pages outstanding is a waste. What is done is to keep a floating proportion between the DBIs based on writeback completions. This way faster/more active devices get a larger share than slower/idle devices. [akpm@linux-foundation.org: fix warnings] [hugh@veritas.com: Fix occasional hang when a task couldn't get out of balance_dirty_pages] Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Signed-off-by: Hugh Dickins <hugh@veritas.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-17 06:25:50 +00:00
percpu_counter_destroy(&bdi->bdi_stat[i]);
}
return err;
}
EXPORT_SYMBOL(bdi_init);
void bdi_destroy(struct backing_dev_info *bdi)
{
int i;
/*
* Splice our entries to the default_backing_dev_info, if this
* bdi disappears
*/
if (bdi_has_dirty_io(bdi)) {
struct bdi_writeback *dst = &default_backing_dev_info.wb;
spin_lock(&inode_lock);
list_splice(&bdi->wb.b_dirty, &dst->b_dirty);
list_splice(&bdi->wb.b_io, &dst->b_io);
list_splice(&bdi->wb.b_more_io, &dst->b_more_io);
spin_unlock(&inode_lock);
}
bdi_unregister(bdi);
for (i = 0; i < NR_BDI_STAT_ITEMS; i++)
percpu_counter_destroy(&bdi->bdi_stat[i]);
mm: per device dirty threshold Scale writeback cache per backing device, proportional to its writeout speed. By decoupling the BDI dirty thresholds a number of problems we currently have will go away, namely: - mutual interference starvation (for any number of BDIs); - deadlocks with stacked BDIs (loop, FUSE and local NFS mounts). It might be that all dirty pages are for a single BDI while other BDIs are idling. By giving each BDI a 'fair' share of the dirty limit, each one can have dirty pages outstanding and make progress. A global threshold also creates a deadlock for stacked BDIs; when A writes to B, and A generates enough dirty pages to get throttled, B will never start writeback until the dirty pages go away. Again, by giving each BDI its own 'independent' dirty limit, this problem is avoided. So the problem is to determine how to distribute the total dirty limit across the BDIs fairly and efficiently. A DBI that has a large dirty limit but does not have any dirty pages outstanding is a waste. What is done is to keep a floating proportion between the DBIs based on writeback completions. This way faster/more active devices get a larger share than slower/idle devices. [akpm@linux-foundation.org: fix warnings] [hugh@veritas.com: Fix occasional hang when a task couldn't get out of balance_dirty_pages] Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Signed-off-by: Hugh Dickins <hugh@veritas.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-17 06:25:50 +00:00
prop_local_destroy_percpu(&bdi->completions);
}
EXPORT_SYMBOL(bdi_destroy);
/*
* For use from filesystems to quickly init and register a bdi associated
* with dirty writeback
*/
int bdi_setup_and_register(struct backing_dev_info *bdi, char *name,
unsigned int cap)
{
char tmp[32];
int err;
bdi->name = name;
bdi->capabilities = cap;
err = bdi_init(bdi);
if (err)
return err;
sprintf(tmp, "%.28s%s", name, "-%d");
err = bdi_register(bdi, NULL, tmp, atomic_long_inc_return(&bdi_seq));
if (err) {
bdi_destroy(bdi);
return err;
}
return 0;
}
EXPORT_SYMBOL(bdi_setup_and_register);
static wait_queue_head_t congestion_wqh[2] = {
__WAIT_QUEUE_HEAD_INITIALIZER(congestion_wqh[0]),
__WAIT_QUEUE_HEAD_INITIALIZER(congestion_wqh[1])
};
void clear_bdi_congested(struct backing_dev_info *bdi, int sync)
{
enum bdi_state bit;
wait_queue_head_t *wqh = &congestion_wqh[sync];
bit = sync ? BDI_sync_congested : BDI_async_congested;
clear_bit(bit, &bdi->state);
smp_mb__after_clear_bit();
if (waitqueue_active(wqh))
wake_up(wqh);
}
EXPORT_SYMBOL(clear_bdi_congested);
void set_bdi_congested(struct backing_dev_info *bdi, int sync)
{
enum bdi_state bit;
bit = sync ? BDI_sync_congested : BDI_async_congested;
set_bit(bit, &bdi->state);
}
EXPORT_SYMBOL(set_bdi_congested);
/**
* congestion_wait - wait for a backing_dev to become uncongested
* @sync: SYNC or ASYNC IO
* @timeout: timeout in jiffies
*
* Waits for up to @timeout jiffies for a backing_dev (any backing_dev) to exit
* write congestion. If no backing_devs are congested then just wait for the
* next write to be completed.
*/
long congestion_wait(int sync, long timeout)
{
long ret;
DEFINE_WAIT(wait);
wait_queue_head_t *wqh = &congestion_wqh[sync];
prepare_to_wait(wqh, &wait, TASK_UNINTERRUPTIBLE);
ret = io_schedule_timeout(timeout);
finish_wait(wqh, &wait);
return ret;
}
EXPORT_SYMBOL(congestion_wait);
mm: per device dirty threshold Scale writeback cache per backing device, proportional to its writeout speed. By decoupling the BDI dirty thresholds a number of problems we currently have will go away, namely: - mutual interference starvation (for any number of BDIs); - deadlocks with stacked BDIs (loop, FUSE and local NFS mounts). It might be that all dirty pages are for a single BDI while other BDIs are idling. By giving each BDI a 'fair' share of the dirty limit, each one can have dirty pages outstanding and make progress. A global threshold also creates a deadlock for stacked BDIs; when A writes to B, and A generates enough dirty pages to get throttled, B will never start writeback until the dirty pages go away. Again, by giving each BDI its own 'independent' dirty limit, this problem is avoided. So the problem is to determine how to distribute the total dirty limit across the BDIs fairly and efficiently. A DBI that has a large dirty limit but does not have any dirty pages outstanding is a waste. What is done is to keep a floating proportion between the DBIs based on writeback completions. This way faster/more active devices get a larger share than slower/idle devices. [akpm@linux-foundation.org: fix warnings] [hugh@veritas.com: Fix occasional hang when a task couldn't get out of balance_dirty_pages] Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Signed-off-by: Hugh Dickins <hugh@veritas.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-17 06:25:50 +00:00