2006-10-20 06:28:16 +00:00
|
|
|
|
|
|
|
#include <linux/wait.h>
|
|
|
|
#include <linux/backing-dev.h>
|
2009-09-09 07:08:54 +00:00
|
|
|
#include <linux/kthread.h>
|
|
|
|
#include <linux/freezer.h>
|
2006-10-20 06:28:16 +00:00
|
|
|
#include <linux/fs.h>
|
2009-03-17 08:35:06 +00:00
|
|
|
#include <linux/pagemap.h>
|
2009-09-09 07:08:54 +00:00
|
|
|
#include <linux/mm.h>
|
2006-10-20 06:28:16 +00:00
|
|
|
#include <linux/sched.h>
|
|
|
|
#include <linux/module.h>
|
2008-04-30 07:54:32 +00:00
|
|
|
#include <linux/writeback.h>
|
|
|
|
#include <linux/device.h>
|
2010-07-07 03:24:06 +00:00
|
|
|
#include <trace/events/writeback.h>
|
2008-04-30 07:54:32 +00:00
|
|
|
|
2010-04-22 09:37:01 +00:00
|
|
|
static atomic_long_t bdi_seq = ATOMIC_LONG_INIT(0);
|
|
|
|
|
2010-04-25 06:54:42 +00:00
|
|
|
struct backing_dev_info noop_backing_dev_info = {
|
|
|
|
.name = "noop",
|
2010-09-21 09:48:55 +00:00
|
|
|
.capabilities = BDI_CAP_NO_ACCT_AND_WRITEBACK,
|
2010-04-25 06:54:42 +00:00
|
|
|
};
|
2015-05-22 21:13:33 +00:00
|
|
|
EXPORT_SYMBOL_GPL(noop_backing_dev_info);
|
2010-04-25 06:54:42 +00:00
|
|
|
|
2008-04-30 07:54:32 +00:00
|
|
|
static struct class *bdi_class;
|
2009-09-14 11:12:40 +00:00
|
|
|
|
|
|
|
/*
|
2013-04-02 02:08:06 +00:00
|
|
|
* bdi_lock protects updates to bdi_list. bdi_list has RCU reader side
|
2009-09-14 11:12:40 +00:00
|
|
|
* locking.
|
|
|
|
*/
|
2009-09-09 07:08:54 +00:00
|
|
|
DEFINE_SPINLOCK(bdi_lock);
|
2009-09-02 07:19:46 +00:00
|
|
|
LIST_HEAD(bdi_list);
|
2009-09-09 07:08:54 +00:00
|
|
|
|
writeback: replace custom worker pool implementation with unbound workqueue
Writeback implements its own worker pool - each bdi can be associated
with a worker thread which is created and destroyed dynamically. The
worker thread for the default bdi is always present and serves as the
"forker" thread which forks off worker threads for other bdis.
there's no reason for writeback to implement its own worker pool when
using unbound workqueue instead is much simpler and more efficient.
This patch replaces custom worker pool implementation in writeback
with an unbound workqueue.
The conversion isn't too complicated but the followings are worth
mentioning.
* bdi_writeback->last_active, task and wakeup_timer are removed.
delayed_work ->dwork is added instead. Explicit timer handling is
no longer necessary. Everything works by either queueing / modding
/ flushing / canceling the delayed_work item.
* bdi_writeback_thread() becomes bdi_writeback_workfn() which runs off
bdi_writeback->dwork. On each execution, it processes
bdi->work_list and reschedules itself if there are more things to
do.
The function also handles low-mem condition, which used to be
handled by the forker thread. If the function is running off a
rescuer thread, it only writes out limited number of pages so that
the rescuer can serve other bdis too. This preserves the flusher
creation failure behavior of the forker thread.
* INIT_LIST_HEAD(&bdi->bdi_list) is used to tell
bdi_writeback_workfn() about on-going bdi unregistration so that it
always drains work_list even if it's running off the rescuer. Note
that the original code was broken in this regard. Under memory
pressure, a bdi could finish unregistration with non-empty
work_list.
* The default bdi is no longer special. It now is treated the same as
any other bdi and bdi_cap_flush_forker() is removed.
* BDI_pending is no longer used. Removed.
* Some tracepoints become non-applicable. The following TPs are
removed - writeback_nothread, writeback_wake_thread,
writeback_wake_forker_thread, writeback_thread_start,
writeback_thread_stop.
Everything, including devices coming and going away and rescuer
operation under simulated memory pressure, seems to work fine in my
test setup.
Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Fengguang Wu <fengguang.wu@intel.com>
Cc: Jeff Moyer <jmoyer@redhat.com>
2013-04-02 02:08:06 +00:00
|
|
|
/* bdi_wq serves all asynchronous writeback tasks */
|
|
|
|
struct workqueue_struct *bdi_wq;
|
|
|
|
|
2008-04-30 07:54:36 +00:00
|
|
|
#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;
|
2010-06-19 21:08:06 +00:00
|
|
|
struct bdi_writeback *wb = &bdi->wb;
|
2009-01-06 22:39:29 +00:00
|
|
|
unsigned long background_thresh;
|
|
|
|
unsigned long dirty_thresh;
|
2015-05-22 22:23:19 +00:00
|
|
|
unsigned long wb_thresh;
|
2015-02-02 05:37:00 +00:00
|
|
|
unsigned long nr_dirty, nr_io, nr_more_io, nr_dirty_time;
|
2009-05-25 07:08:21 +00:00
|
|
|
struct inode *inode;
|
|
|
|
|
2015-02-02 05:37:00 +00:00
|
|
|
nr_dirty = nr_io = nr_more_io = nr_dirty_time = 0;
|
2011-04-22 00:19:44 +00:00
|
|
|
spin_lock(&wb->list_lock);
|
2015-03-04 19:07:22 +00:00
|
|
|
list_for_each_entry(inode, &wb->b_dirty, i_io_list)
|
2010-06-19 21:08:06 +00:00
|
|
|
nr_dirty++;
|
2015-03-04 19:07:22 +00:00
|
|
|
list_for_each_entry(inode, &wb->b_io, i_io_list)
|
2010-06-19 21:08:06 +00:00
|
|
|
nr_io++;
|
2015-03-04 19:07:22 +00:00
|
|
|
list_for_each_entry(inode, &wb->b_more_io, i_io_list)
|
2010-06-19 21:08:06 +00:00
|
|
|
nr_more_io++;
|
2015-03-04 19:07:22 +00:00
|
|
|
list_for_each_entry(inode, &wb->b_dirty_time, i_io_list)
|
2015-02-02 05:37:00 +00:00
|
|
|
if (inode->i_state & I_DIRTY_TIME)
|
|
|
|
nr_dirty_time++;
|
2011-04-22 00:19:44 +00:00
|
|
|
spin_unlock(&wb->list_lock);
|
2008-04-30 07:54:36 +00:00
|
|
|
|
2010-08-11 21:17:39 +00:00
|
|
|
global_dirty_limits(&background_thresh, &dirty_thresh);
|
2015-05-22 22:23:19 +00:00
|
|
|
wb_thresh = wb_calc_thresh(wb, dirty_thresh);
|
2008-04-30 07:54:36 +00:00
|
|
|
|
|
|
|
#define K(x) ((x) << (PAGE_SHIFT - 10))
|
|
|
|
seq_printf(m,
|
2010-08-29 17:28:45 +00:00
|
|
|
"BdiWriteback: %10lu kB\n"
|
|
|
|
"BdiReclaimable: %10lu kB\n"
|
|
|
|
"BdiDirtyThresh: %10lu kB\n"
|
|
|
|
"DirtyThresh: %10lu kB\n"
|
|
|
|
"BackgroundThresh: %10lu kB\n"
|
2011-01-23 16:07:47 +00:00
|
|
|
"BdiDirtied: %10lu kB\n"
|
2010-08-29 17:28:45 +00:00
|
|
|
"BdiWritten: %10lu kB\n"
|
|
|
|
"BdiWriteBandwidth: %10lu kBps\n"
|
|
|
|
"b_dirty: %10lu\n"
|
|
|
|
"b_io: %10lu\n"
|
|
|
|
"b_more_io: %10lu\n"
|
2015-02-02 05:37:00 +00:00
|
|
|
"b_dirty_time: %10lu\n"
|
2010-08-29 17:28:45 +00:00
|
|
|
"bdi_list: %10u\n"
|
|
|
|
"state: %10lx\n",
|
2015-05-22 21:13:27 +00:00
|
|
|
(unsigned long) K(wb_stat(wb, WB_WRITEBACK)),
|
|
|
|
(unsigned long) K(wb_stat(wb, WB_RECLAIMABLE)),
|
2015-05-22 22:23:19 +00:00
|
|
|
K(wb_thresh),
|
2010-12-09 04:44:24 +00:00
|
|
|
K(dirty_thresh),
|
|
|
|
K(background_thresh),
|
2015-05-22 21:13:27 +00:00
|
|
|
(unsigned long) K(wb_stat(wb, WB_DIRTIED)),
|
|
|
|
(unsigned long) K(wb_stat(wb, WB_WRITTEN)),
|
writeback: move bandwidth related fields from backing_dev_info into bdi_writeback
Currently, a bdi (backing_dev_info) embeds single wb (bdi_writeback)
and the role of the separation is unclear. For cgroup support for
writeback IOs, a bdi will be updated to host multiple wb's where each
wb serves writeback IOs of a different cgroup on the bdi. To achieve
that, a wb should carry all states necessary for servicing writeback
IOs for a cgroup independently.
This patch moves bandwidth related fields from backing_dev_info into
bdi_writeback.
* The moved fields are: bw_time_stamp, dirtied_stamp, written_stamp,
write_bandwidth, avg_write_bandwidth, dirty_ratelimit,
balanced_dirty_ratelimit, completions and dirty_exceeded.
* writeback_chunk_size() and over_bground_thresh() now take @wb
instead of @bdi.
* bdi_writeout_fraction(bdi, ...) -> wb_writeout_fraction(wb, ...)
bdi_dirty_limit(bdi, ...) -> wb_dirty_limit(wb, ...)
bdi_position_ration(bdi, ...) -> wb_position_ratio(wb, ...)
bdi_update_writebandwidth(bdi, ...) -> wb_update_write_bandwidth(wb, ...)
[__]bdi_update_bandwidth(bdi, ...) -> [__]wb_update_bandwidth(wb, ...)
bdi_{max|min}_pause(bdi, ...) -> wb_{max|min}_pause(wb, ...)
bdi_dirty_limits(bdi, ...) -> wb_dirty_limits(wb, ...)
* Init/exits of the relocated fields are moved to bdi_wb_init/exit()
respectively. Note that explicit zeroing is dropped in the process
as wb's are cleared in entirety anyway.
* As there's still only one bdi_writeback per backing_dev_info, all
uses of bdi->stat[] are mechanically replaced with bdi->wb.stat[]
introducing no behavior changes.
v2: Typo in description fixed as suggested by Jan.
Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Wu Fengguang <fengguang.wu@intel.com>
Cc: Jaegeuk Kim <jaegeuk@kernel.org>
Cc: Steven Whitehouse <swhiteho@redhat.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2015-05-22 21:13:28 +00:00
|
|
|
(unsigned long) K(wb->write_bandwidth),
|
2010-12-09 04:44:24 +00:00
|
|
|
nr_dirty,
|
|
|
|
nr_io,
|
|
|
|
nr_more_io,
|
2015-02-02 05:37:00 +00:00
|
|
|
nr_dirty_time,
|
2015-05-22 21:13:26 +00:00
|
|
|
!list_empty(&bdi->bdi_list), bdi->wb.state);
|
2008-04-30 07:54:36 +00:00
|
|
|
#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
|
|
|
|
|
2008-04-30 07:54:32 +00:00
|
|
|
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);
|
|
|
|
unsigned long read_ahead_kb;
|
2012-08-25 08:57:27 +00:00
|
|
|
ssize_t ret;
|
2008-04-30 07:54:32 +00:00
|
|
|
|
2012-08-25 08:57:27 +00:00
|
|
|
ret = kstrtoul(buf, 10, &read_ahead_kb);
|
|
|
|
if (ret < 0)
|
|
|
|
return ret;
|
|
|
|
|
|
|
|
bdi->ra_pages = read_ahead_kb >> (PAGE_SHIFT - 10);
|
|
|
|
|
|
|
|
return count;
|
2008-04-30 07:54:32 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
#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); \
|
2013-07-24 22:05:26 +00:00
|
|
|
} \
|
|
|
|
static DEVICE_ATTR_RW(name);
|
2008-04-30 07:54:32 +00:00
|
|
|
|
|
|
|
BDI_SHOW(read_ahead_kb, K(bdi->ra_pages))
|
|
|
|
|
2008-04-30 07:54:35 +00:00
|
|
|
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);
|
|
|
|
unsigned int ratio;
|
2012-08-25 08:57:27 +00:00
|
|
|
ssize_t ret;
|
|
|
|
|
|
|
|
ret = kstrtouint(buf, 10, &ratio);
|
|
|
|
if (ret < 0)
|
|
|
|
return ret;
|
|
|
|
|
|
|
|
ret = bdi_set_min_ratio(bdi, ratio);
|
|
|
|
if (!ret)
|
|
|
|
ret = count;
|
2008-04-30 07:54:35 +00:00
|
|
|
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
BDI_SHOW(min_ratio, bdi->min_ratio)
|
|
|
|
|
2008-04-30 07:54:36 +00:00
|
|
|
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);
|
|
|
|
unsigned int ratio;
|
2012-08-25 08:57:27 +00:00
|
|
|
ssize_t ret;
|
|
|
|
|
|
|
|
ret = kstrtouint(buf, 10, &ratio);
|
|
|
|
if (ret < 0)
|
|
|
|
return ret;
|
|
|
|
|
|
|
|
ret = bdi_set_max_ratio(bdi, ratio);
|
|
|
|
if (!ret)
|
|
|
|
ret = count;
|
2008-04-30 07:54:36 +00:00
|
|
|
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
BDI_SHOW(max_ratio, bdi->max_ratio)
|
|
|
|
|
bdi: allow block devices to say that they require stable page writes
This patchset ("stable page writes, part 2") makes some key
modifications to the original 'stable page writes' patchset. First, it
provides creators (devices and filesystems) of a backing_dev_info a flag
that declares whether or not it is necessary to ensure that page
contents cannot change during writeout. It is no longer assumed that
this is true of all devices (which was never true anyway). Second, the
flag is used to relaxed the wait_on_page_writeback calls so that wait
only occurs if the device needs it. Third, it fixes up the remaining
disk-backed filesystems to use this improved conditional-wait logic to
provide stable page writes on those filesystems.
It is hoped that (for people not using checksumming devices, anyway)
this patchset will give back unnecessary performance decreases since the
original stable page write patchset went into 3.0. Sorry about not
fixing it sooner.
Complaints were registered by several people about the long write
latencies introduced by the original stable page write patchset.
Generally speaking, the kernel ought to allocate as little extra memory
as possible to facilitate writeout, but for people who simply cannot
wait, a second page stability strategy is (re)introduced: snapshotting
page contents. The waiting behavior is still the default strategy; to
enable page snapshotting, a superblock flag (MS_SNAP_STABLE) must be
set. This flag is used to bandaid^Henable stable page writeback on
ext3[1], and is not used anywhere else.
Given that there are already a few storage devices and network FSes that
have rolled their own page stability wait/page snapshot code, it would
be nice to move towards consolidating all of these. It seems possible
that iscsi and raid5 may wish to use the new stable page write support
to enable zero-copy writeout.
Thank you to Jan Kara for helping fix a couple more filesystems.
Per Andrew Morton's request, here are the result of using dbench to measure
latencies on ext2:
3.8.0-rc3:
Operation Count AvgLat MaxLat
----------------------------------------
WriteX 109347 0.028 59.817
ReadX 347180 0.004 3.391
Flush 15514 29.828 287.283
Throughput 57.429 MB/sec 4 clients 4 procs max_latency=287.290 ms
3.8.0-rc3 + patches:
WriteX 105556 0.029 4.273
ReadX 335004 0.005 4.112
Flush 14982 30.540 298.634
Throughput 55.4496 MB/sec 4 clients 4 procs max_latency=298.650 ms
As you can see, for ext2 the maximum write latency decreases from ~60ms
on a laptop hard disk to ~4ms. I'm not sure why the flush latencies
increase, though I suspect that being able to dirty pages faster gives
the flusher more work to do.
On ext4, the average write latency decreases as well as all the maximum
latencies:
3.8.0-rc3:
WriteX 85624 0.152 33.078
ReadX 272090 0.010 61.210
Flush 12129 36.219 168.260
Throughput 44.8618 MB/sec 4 clients 4 procs max_latency=168.276 ms
3.8.0-rc3 + patches:
WriteX 86082 0.141 30.928
ReadX 273358 0.010 36.124
Flush 12214 34.800 165.689
Throughput 44.9941 MB/sec 4 clients 4 procs max_latency=165.722 ms
XFS seems to exhibit similar latency improvements as ext2:
3.8.0-rc3:
WriteX 125739 0.028 104.343
ReadX 399070 0.005 4.115
Flush 17851 25.004 131.390
Throughput 66.0024 MB/sec 4 clients 4 procs max_latency=131.406 ms
3.8.0-rc3 + patches:
WriteX 123529 0.028 6.299
ReadX 392434 0.005 4.287
Flush 17549 25.120 188.687
Throughput 64.9113 MB/sec 4 clients 4 procs max_latency=188.704 ms
...and btrfs, just to round things out, also shows some latency
decreases:
3.8.0-rc3:
WriteX 67122 0.083 82.355
ReadX 212719 0.005 2.828
Flush 9547 47.561 147.418
Throughput 35.3391 MB/sec 4 clients 4 procs max_latency=147.433 ms
3.8.0-rc3 + patches:
WriteX 64898 0.101 71.631
ReadX 206673 0.005 7.123
Flush 9190 47.963 219.034
Throughput 34.0795 MB/sec 4 clients 4 procs max_latency=219.044 ms
Before this patchset, all filesystems would block, regardless of whether
or not it was necessary. ext3 would wait, but still generate occasional
checksum errors. The network filesystems were left to do their own
thing, so they'd wait too.
After this patchset, all the disk filesystems except ext3 and btrfs will
wait only if the hardware requires it. ext3 (if necessary) snapshots
pages instead of blocking, and btrfs provides its own bdi so the mm will
never wait. Network filesystems haven't been touched, so either they
provide their own wait code, or they don't block at all. The blocking
behavior is back to what it was before 3.0 if you don't have a disk
requiring stable page writes.
This patchset has been tested on 3.8.0-rc3 on x64 with ext3, ext4, and
xfs. I've spot-checked 3.8.0-rc4 and seem to be getting the same
results as -rc3.
[1] The alternative fixes to ext3 include fixing the locking order and
page bit handling like we did for ext4 (but then why not just use
ext4?), or setting PG_writeback so early that ext3 becomes extremely
slow. I tried that, but the number of write()s I could initiate dropped
by nearly an order of magnitude. That was a bit much even for the
author of the stable page series! :)
This patch:
Creates a per-backing-device flag that tracks whether or not pages must
be held immutable during writeout. Eventually it will be used to waive
wait_for_page_writeback() if nothing requires stable pages.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Artem Bityutskiy <dedekind1@gmail.com>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Mark Fasheh <mfasheh@suse.com>
Cc: Steven Whitehouse <swhiteho@redhat.com>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Eric Van Hensbergen <ericvh@gmail.com>
Cc: Ron Minnich <rminnich@sandia.gov>
Cc: Latchesar Ionkov <lucho@ionkov.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-22 00:42:48 +00:00
|
|
|
static ssize_t stable_pages_required_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, "%d\n",
|
|
|
|
bdi_cap_stable_pages_required(bdi) ? 1 : 0);
|
|
|
|
}
|
2013-07-24 22:05:26 +00:00
|
|
|
static DEVICE_ATTR_RO(stable_pages_required);
|
|
|
|
|
|
|
|
static struct attribute *bdi_dev_attrs[] = {
|
|
|
|
&dev_attr_read_ahead_kb.attr,
|
|
|
|
&dev_attr_min_ratio.attr,
|
|
|
|
&dev_attr_max_ratio.attr,
|
|
|
|
&dev_attr_stable_pages_required.attr,
|
|
|
|
NULL,
|
2008-04-30 07:54:32 +00:00
|
|
|
};
|
2013-07-24 22:05:26 +00:00
|
|
|
ATTRIBUTE_GROUPS(bdi_dev);
|
2008-04-30 07:54:32 +00:00
|
|
|
|
|
|
|
static __init int bdi_class_init(void)
|
|
|
|
{
|
|
|
|
bdi_class = class_create(THIS_MODULE, "bdi");
|
2010-04-02 07:46:55 +00:00
|
|
|
if (IS_ERR(bdi_class))
|
|
|
|
return PTR_ERR(bdi_class);
|
|
|
|
|
2013-07-24 22:05:26 +00:00
|
|
|
bdi_class->dev_groups = bdi_dev_groups;
|
2008-04-30 07:54:36 +00:00
|
|
|
bdi_debug_init();
|
2008-04-30 07:54:32 +00:00
|
|
|
return 0;
|
|
|
|
}
|
2008-04-30 07:54:36 +00:00
|
|
|
postcore_initcall(bdi_class_init);
|
2008-04-30 07:54:32 +00:00
|
|
|
|
2009-03-17 08:35:06 +00:00
|
|
|
static int __init default_bdi_init(void)
|
|
|
|
{
|
|
|
|
int err;
|
|
|
|
|
writeback: replace custom worker pool implementation with unbound workqueue
Writeback implements its own worker pool - each bdi can be associated
with a worker thread which is created and destroyed dynamically. The
worker thread for the default bdi is always present and serves as the
"forker" thread which forks off worker threads for other bdis.
there's no reason for writeback to implement its own worker pool when
using unbound workqueue instead is much simpler and more efficient.
This patch replaces custom worker pool implementation in writeback
with an unbound workqueue.
The conversion isn't too complicated but the followings are worth
mentioning.
* bdi_writeback->last_active, task and wakeup_timer are removed.
delayed_work ->dwork is added instead. Explicit timer handling is
no longer necessary. Everything works by either queueing / modding
/ flushing / canceling the delayed_work item.
* bdi_writeback_thread() becomes bdi_writeback_workfn() which runs off
bdi_writeback->dwork. On each execution, it processes
bdi->work_list and reschedules itself if there are more things to
do.
The function also handles low-mem condition, which used to be
handled by the forker thread. If the function is running off a
rescuer thread, it only writes out limited number of pages so that
the rescuer can serve other bdis too. This preserves the flusher
creation failure behavior of the forker thread.
* INIT_LIST_HEAD(&bdi->bdi_list) is used to tell
bdi_writeback_workfn() about on-going bdi unregistration so that it
always drains work_list even if it's running off the rescuer. Note
that the original code was broken in this regard. Under memory
pressure, a bdi could finish unregistration with non-empty
work_list.
* The default bdi is no longer special. It now is treated the same as
any other bdi and bdi_cap_flush_forker() is removed.
* BDI_pending is no longer used. Removed.
* Some tracepoints become non-applicable. The following TPs are
removed - writeback_nothread, writeback_wake_thread,
writeback_wake_forker_thread, writeback_thread_start,
writeback_thread_stop.
Everything, including devices coming and going away and rescuer
operation under simulated memory pressure, seems to work fine in my
test setup.
Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Fengguang Wu <fengguang.wu@intel.com>
Cc: Jeff Moyer <jmoyer@redhat.com>
2013-04-02 02:08:06 +00:00
|
|
|
bdi_wq = alloc_workqueue("writeback", WQ_MEM_RECLAIM | WQ_FREEZABLE |
|
2013-04-02 02:08:06 +00:00
|
|
|
WQ_UNBOUND | WQ_SYSFS, 0);
|
writeback: replace custom worker pool implementation with unbound workqueue
Writeback implements its own worker pool - each bdi can be associated
with a worker thread which is created and destroyed dynamically. The
worker thread for the default bdi is always present and serves as the
"forker" thread which forks off worker threads for other bdis.
there's no reason for writeback to implement its own worker pool when
using unbound workqueue instead is much simpler and more efficient.
This patch replaces custom worker pool implementation in writeback
with an unbound workqueue.
The conversion isn't too complicated but the followings are worth
mentioning.
* bdi_writeback->last_active, task and wakeup_timer are removed.
delayed_work ->dwork is added instead. Explicit timer handling is
no longer necessary. Everything works by either queueing / modding
/ flushing / canceling the delayed_work item.
* bdi_writeback_thread() becomes bdi_writeback_workfn() which runs off
bdi_writeback->dwork. On each execution, it processes
bdi->work_list and reschedules itself if there are more things to
do.
The function also handles low-mem condition, which used to be
handled by the forker thread. If the function is running off a
rescuer thread, it only writes out limited number of pages so that
the rescuer can serve other bdis too. This preserves the flusher
creation failure behavior of the forker thread.
* INIT_LIST_HEAD(&bdi->bdi_list) is used to tell
bdi_writeback_workfn() about on-going bdi unregistration so that it
always drains work_list even if it's running off the rescuer. Note
that the original code was broken in this regard. Under memory
pressure, a bdi could finish unregistration with non-empty
work_list.
* The default bdi is no longer special. It now is treated the same as
any other bdi and bdi_cap_flush_forker() is removed.
* BDI_pending is no longer used. Removed.
* Some tracepoints become non-applicable. The following TPs are
removed - writeback_nothread, writeback_wake_thread,
writeback_wake_forker_thread, writeback_thread_start,
writeback_thread_stop.
Everything, including devices coming and going away and rescuer
operation under simulated memory pressure, seems to work fine in my
test setup.
Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Fengguang Wu <fengguang.wu@intel.com>
Cc: Jeff Moyer <jmoyer@redhat.com>
2013-04-02 02:08:06 +00:00
|
|
|
if (!bdi_wq)
|
|
|
|
return -ENOMEM;
|
|
|
|
|
2010-09-21 09:48:55 +00:00
|
|
|
err = bdi_init(&noop_backing_dev_info);
|
2009-03-17 08:35:06 +00:00
|
|
|
|
|
|
|
return err;
|
|
|
|
}
|
|
|
|
subsys_initcall(default_bdi_init);
|
|
|
|
|
2010-07-25 11:29:22 +00:00
|
|
|
/*
|
writeback: move backing_dev_info->wb_lock and ->worklist into bdi_writeback
Currently, a bdi (backing_dev_info) embeds single wb (bdi_writeback)
and the role of the separation is unclear. For cgroup support for
writeback IOs, a bdi will be updated to host multiple wb's where each
wb serves writeback IOs of a different cgroup on the bdi. To achieve
that, a wb should carry all states necessary for servicing writeback
IOs for a cgroup independently.
This patch moves bdi->wb_lock and ->worklist into wb.
* The lock protects bdi->worklist and bdi->wb.dwork scheduling. While
moving, rename it to wb->work_lock as wb->wb_lock is confusing.
Also, move wb->dwork downwards so that it's colocated with the new
->work_lock and ->work_list fields.
* bdi_writeback_workfn() -> wb_workfn()
bdi_wakeup_thread_delayed(bdi) -> wb_wakeup_delayed(wb)
bdi_wakeup_thread(bdi) -> wb_wakeup(wb)
bdi_queue_work(bdi, ...) -> wb_queue_work(wb, ...)
__bdi_start_writeback(bdi, ...) -> __wb_start_writeback(wb, ...)
get_next_work_item(bdi) -> get_next_work_item(wb)
* bdi_wb_shutdown() is renamed to wb_shutdown() and now takes @wb.
The function contained parts which belong to the containing bdi
rather than the wb itself - testing cap_writeback_dirty and
bdi_remove_from_list() invocation. Those are moved to
bdi_unregister().
* bdi_wb_{init|exit}() are renamed to wb_{init|exit}().
Initializations of the moved bdi->wb_lock and ->work_list are
relocated from bdi_init() to wb_init().
* As there's still only one bdi_writeback per backing_dev_info, all
uses of bdi->state are mechanically replaced with bdi->wb.state
introducing no behavior changes.
Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Wu Fengguang <fengguang.wu@intel.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2015-05-22 21:13:30 +00:00
|
|
|
* This function is used when the first inode for this wb is marked dirty. It
|
2010-07-25 11:29:22 +00:00
|
|
|
* 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.
|
2014-04-03 21:46:22 +00:00
|
|
|
*
|
|
|
|
* We have to be careful not to postpone flush work if it is scheduled for
|
|
|
|
* earlier. Thus we use queue_delayed_work().
|
2010-07-25 11:29:22 +00:00
|
|
|
*/
|
writeback: move backing_dev_info->wb_lock and ->worklist into bdi_writeback
Currently, a bdi (backing_dev_info) embeds single wb (bdi_writeback)
and the role of the separation is unclear. For cgroup support for
writeback IOs, a bdi will be updated to host multiple wb's where each
wb serves writeback IOs of a different cgroup on the bdi. To achieve
that, a wb should carry all states necessary for servicing writeback
IOs for a cgroup independently.
This patch moves bdi->wb_lock and ->worklist into wb.
* The lock protects bdi->worklist and bdi->wb.dwork scheduling. While
moving, rename it to wb->work_lock as wb->wb_lock is confusing.
Also, move wb->dwork downwards so that it's colocated with the new
->work_lock and ->work_list fields.
* bdi_writeback_workfn() -> wb_workfn()
bdi_wakeup_thread_delayed(bdi) -> wb_wakeup_delayed(wb)
bdi_wakeup_thread(bdi) -> wb_wakeup(wb)
bdi_queue_work(bdi, ...) -> wb_queue_work(wb, ...)
__bdi_start_writeback(bdi, ...) -> __wb_start_writeback(wb, ...)
get_next_work_item(bdi) -> get_next_work_item(wb)
* bdi_wb_shutdown() is renamed to wb_shutdown() and now takes @wb.
The function contained parts which belong to the containing bdi
rather than the wb itself - testing cap_writeback_dirty and
bdi_remove_from_list() invocation. Those are moved to
bdi_unregister().
* bdi_wb_{init|exit}() are renamed to wb_{init|exit}().
Initializations of the moved bdi->wb_lock and ->work_list are
relocated from bdi_init() to wb_init().
* As there's still only one bdi_writeback per backing_dev_info, all
uses of bdi->state are mechanically replaced with bdi->wb.state
introducing no behavior changes.
Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Wu Fengguang <fengguang.wu@intel.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2015-05-22 21:13:30 +00:00
|
|
|
void wb_wakeup_delayed(struct bdi_writeback *wb)
|
2010-07-25 11:29:22 +00:00
|
|
|
{
|
|
|
|
unsigned long timeout;
|
|
|
|
|
|
|
|
timeout = msecs_to_jiffies(dirty_writeback_interval * 10);
|
writeback: move backing_dev_info->wb_lock and ->worklist into bdi_writeback
Currently, a bdi (backing_dev_info) embeds single wb (bdi_writeback)
and the role of the separation is unclear. For cgroup support for
writeback IOs, a bdi will be updated to host multiple wb's where each
wb serves writeback IOs of a different cgroup on the bdi. To achieve
that, a wb should carry all states necessary for servicing writeback
IOs for a cgroup independently.
This patch moves bdi->wb_lock and ->worklist into wb.
* The lock protects bdi->worklist and bdi->wb.dwork scheduling. While
moving, rename it to wb->work_lock as wb->wb_lock is confusing.
Also, move wb->dwork downwards so that it's colocated with the new
->work_lock and ->work_list fields.
* bdi_writeback_workfn() -> wb_workfn()
bdi_wakeup_thread_delayed(bdi) -> wb_wakeup_delayed(wb)
bdi_wakeup_thread(bdi) -> wb_wakeup(wb)
bdi_queue_work(bdi, ...) -> wb_queue_work(wb, ...)
__bdi_start_writeback(bdi, ...) -> __wb_start_writeback(wb, ...)
get_next_work_item(bdi) -> get_next_work_item(wb)
* bdi_wb_shutdown() is renamed to wb_shutdown() and now takes @wb.
The function contained parts which belong to the containing bdi
rather than the wb itself - testing cap_writeback_dirty and
bdi_remove_from_list() invocation. Those are moved to
bdi_unregister().
* bdi_wb_{init|exit}() are renamed to wb_{init|exit}().
Initializations of the moved bdi->wb_lock and ->work_list are
relocated from bdi_init() to wb_init().
* As there's still only one bdi_writeback per backing_dev_info, all
uses of bdi->state are mechanically replaced with bdi->wb.state
introducing no behavior changes.
Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Wu Fengguang <fengguang.wu@intel.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2015-05-22 21:13:30 +00:00
|
|
|
spin_lock_bh(&wb->work_lock);
|
|
|
|
if (test_bit(WB_registered, &wb->state))
|
|
|
|
queue_delayed_work(bdi_wq, &wb->dwork, timeout);
|
|
|
|
spin_unlock_bh(&wb->work_lock);
|
2009-09-09 07:08:54 +00:00
|
|
|
}
|
|
|
|
|
2009-09-14 11:12:40 +00:00
|
|
|
/*
|
writeback: move bandwidth related fields from backing_dev_info into bdi_writeback
Currently, a bdi (backing_dev_info) embeds single wb (bdi_writeback)
and the role of the separation is unclear. For cgroup support for
writeback IOs, a bdi will be updated to host multiple wb's where each
wb serves writeback IOs of a different cgroup on the bdi. To achieve
that, a wb should carry all states necessary for servicing writeback
IOs for a cgroup independently.
This patch moves bandwidth related fields from backing_dev_info into
bdi_writeback.
* The moved fields are: bw_time_stamp, dirtied_stamp, written_stamp,
write_bandwidth, avg_write_bandwidth, dirty_ratelimit,
balanced_dirty_ratelimit, completions and dirty_exceeded.
* writeback_chunk_size() and over_bground_thresh() now take @wb
instead of @bdi.
* bdi_writeout_fraction(bdi, ...) -> wb_writeout_fraction(wb, ...)
bdi_dirty_limit(bdi, ...) -> wb_dirty_limit(wb, ...)
bdi_position_ration(bdi, ...) -> wb_position_ratio(wb, ...)
bdi_update_writebandwidth(bdi, ...) -> wb_update_write_bandwidth(wb, ...)
[__]bdi_update_bandwidth(bdi, ...) -> [__]wb_update_bandwidth(wb, ...)
bdi_{max|min}_pause(bdi, ...) -> wb_{max|min}_pause(wb, ...)
bdi_dirty_limits(bdi, ...) -> wb_dirty_limits(wb, ...)
* Init/exits of the relocated fields are moved to bdi_wb_init/exit()
respectively. Note that explicit zeroing is dropped in the process
as wb's are cleared in entirety anyway.
* As there's still only one bdi_writeback per backing_dev_info, all
uses of bdi->stat[] are mechanically replaced with bdi->wb.stat[]
introducing no behavior changes.
v2: Typo in description fixed as suggested by Jan.
Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Wu Fengguang <fengguang.wu@intel.com>
Cc: Jaegeuk Kim <jaegeuk@kernel.org>
Cc: Steven Whitehouse <swhiteho@redhat.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2015-05-22 21:13:28 +00:00
|
|
|
* Initial write bandwidth: 100 MB/s
|
2009-09-14 11:12:40 +00:00
|
|
|
*/
|
writeback: move bandwidth related fields from backing_dev_info into bdi_writeback
Currently, a bdi (backing_dev_info) embeds single wb (bdi_writeback)
and the role of the separation is unclear. For cgroup support for
writeback IOs, a bdi will be updated to host multiple wb's where each
wb serves writeback IOs of a different cgroup on the bdi. To achieve
that, a wb should carry all states necessary for servicing writeback
IOs for a cgroup independently.
This patch moves bandwidth related fields from backing_dev_info into
bdi_writeback.
* The moved fields are: bw_time_stamp, dirtied_stamp, written_stamp,
write_bandwidth, avg_write_bandwidth, dirty_ratelimit,
balanced_dirty_ratelimit, completions and dirty_exceeded.
* writeback_chunk_size() and over_bground_thresh() now take @wb
instead of @bdi.
* bdi_writeout_fraction(bdi, ...) -> wb_writeout_fraction(wb, ...)
bdi_dirty_limit(bdi, ...) -> wb_dirty_limit(wb, ...)
bdi_position_ration(bdi, ...) -> wb_position_ratio(wb, ...)
bdi_update_writebandwidth(bdi, ...) -> wb_update_write_bandwidth(wb, ...)
[__]bdi_update_bandwidth(bdi, ...) -> [__]wb_update_bandwidth(wb, ...)
bdi_{max|min}_pause(bdi, ...) -> wb_{max|min}_pause(wb, ...)
bdi_dirty_limits(bdi, ...) -> wb_dirty_limits(wb, ...)
* Init/exits of the relocated fields are moved to bdi_wb_init/exit()
respectively. Note that explicit zeroing is dropped in the process
as wb's are cleared in entirety anyway.
* As there's still only one bdi_writeback per backing_dev_info, all
uses of bdi->stat[] are mechanically replaced with bdi->wb.stat[]
introducing no behavior changes.
v2: Typo in description fixed as suggested by Jan.
Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Wu Fengguang <fengguang.wu@intel.com>
Cc: Jaegeuk Kim <jaegeuk@kernel.org>
Cc: Steven Whitehouse <swhiteho@redhat.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2015-05-22 21:13:28 +00:00
|
|
|
#define INIT_BW (100 << (20 - PAGE_SHIFT))
|
2009-09-14 11:12:40 +00:00
|
|
|
|
2015-05-22 21:13:34 +00:00
|
|
|
static int wb_init(struct bdi_writeback *wb, struct backing_dev_info *bdi,
|
2015-07-02 14:44:34 +00:00
|
|
|
int blkcg_id, gfp_t gfp)
|
2008-04-30 07:54:32 +00:00
|
|
|
{
|
2015-05-22 21:13:27 +00:00
|
|
|
int i, err;
|
2008-04-30 07:54:32 +00:00
|
|
|
|
2010-07-25 11:29:22 +00:00
|
|
|
memset(wb, 0, sizeof(*wb));
|
2008-12-02 18:31:50 +00:00
|
|
|
|
2010-07-25 11:29:22 +00:00
|
|
|
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);
|
2015-02-02 05:37:00 +00:00
|
|
|
INIT_LIST_HEAD(&wb->b_dirty_time);
|
2011-04-22 00:19:44 +00:00
|
|
|
spin_lock_init(&wb->list_lock);
|
2009-09-02 07:19:46 +00:00
|
|
|
|
writeback: move bandwidth related fields from backing_dev_info into bdi_writeback
Currently, a bdi (backing_dev_info) embeds single wb (bdi_writeback)
and the role of the separation is unclear. For cgroup support for
writeback IOs, a bdi will be updated to host multiple wb's where each
wb serves writeback IOs of a different cgroup on the bdi. To achieve
that, a wb should carry all states necessary for servicing writeback
IOs for a cgroup independently.
This patch moves bandwidth related fields from backing_dev_info into
bdi_writeback.
* The moved fields are: bw_time_stamp, dirtied_stamp, written_stamp,
write_bandwidth, avg_write_bandwidth, dirty_ratelimit,
balanced_dirty_ratelimit, completions and dirty_exceeded.
* writeback_chunk_size() and over_bground_thresh() now take @wb
instead of @bdi.
* bdi_writeout_fraction(bdi, ...) -> wb_writeout_fraction(wb, ...)
bdi_dirty_limit(bdi, ...) -> wb_dirty_limit(wb, ...)
bdi_position_ration(bdi, ...) -> wb_position_ratio(wb, ...)
bdi_update_writebandwidth(bdi, ...) -> wb_update_write_bandwidth(wb, ...)
[__]bdi_update_bandwidth(bdi, ...) -> [__]wb_update_bandwidth(wb, ...)
bdi_{max|min}_pause(bdi, ...) -> wb_{max|min}_pause(wb, ...)
bdi_dirty_limits(bdi, ...) -> wb_dirty_limits(wb, ...)
* Init/exits of the relocated fields are moved to bdi_wb_init/exit()
respectively. Note that explicit zeroing is dropped in the process
as wb's are cleared in entirety anyway.
* As there's still only one bdi_writeback per backing_dev_info, all
uses of bdi->stat[] are mechanically replaced with bdi->wb.stat[]
introducing no behavior changes.
v2: Typo in description fixed as suggested by Jan.
Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Wu Fengguang <fengguang.wu@intel.com>
Cc: Jaegeuk Kim <jaegeuk@kernel.org>
Cc: Steven Whitehouse <swhiteho@redhat.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2015-05-22 21:13:28 +00:00
|
|
|
wb->bw_time_stamp = jiffies;
|
|
|
|
wb->balanced_dirty_ratelimit = INIT_BW;
|
|
|
|
wb->dirty_ratelimit = INIT_BW;
|
|
|
|
wb->write_bandwidth = INIT_BW;
|
|
|
|
wb->avg_write_bandwidth = INIT_BW;
|
2008-04-30 07:54:32 +00:00
|
|
|
|
writeback: move backing_dev_info->wb_lock and ->worklist into bdi_writeback
Currently, a bdi (backing_dev_info) embeds single wb (bdi_writeback)
and the role of the separation is unclear. For cgroup support for
writeback IOs, a bdi will be updated to host multiple wb's where each
wb serves writeback IOs of a different cgroup on the bdi. To achieve
that, a wb should carry all states necessary for servicing writeback
IOs for a cgroup independently.
This patch moves bdi->wb_lock and ->worklist into wb.
* The lock protects bdi->worklist and bdi->wb.dwork scheduling. While
moving, rename it to wb->work_lock as wb->wb_lock is confusing.
Also, move wb->dwork downwards so that it's colocated with the new
->work_lock and ->work_list fields.
* bdi_writeback_workfn() -> wb_workfn()
bdi_wakeup_thread_delayed(bdi) -> wb_wakeup_delayed(wb)
bdi_wakeup_thread(bdi) -> wb_wakeup(wb)
bdi_queue_work(bdi, ...) -> wb_queue_work(wb, ...)
__bdi_start_writeback(bdi, ...) -> __wb_start_writeback(wb, ...)
get_next_work_item(bdi) -> get_next_work_item(wb)
* bdi_wb_shutdown() is renamed to wb_shutdown() and now takes @wb.
The function contained parts which belong to the containing bdi
rather than the wb itself - testing cap_writeback_dirty and
bdi_remove_from_list() invocation. Those are moved to
bdi_unregister().
* bdi_wb_{init|exit}() are renamed to wb_{init|exit}().
Initializations of the moved bdi->wb_lock and ->work_list are
relocated from bdi_init() to wb_init().
* As there's still only one bdi_writeback per backing_dev_info, all
uses of bdi->state are mechanically replaced with bdi->wb.state
introducing no behavior changes.
Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Wu Fengguang <fengguang.wu@intel.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2015-05-22 21:13:30 +00:00
|
|
|
spin_lock_init(&wb->work_lock);
|
|
|
|
INIT_LIST_HEAD(&wb->work_list);
|
|
|
|
INIT_DELAYED_WORK(&wb->dwork, wb_workfn);
|
writeback: cleanup bdi_register
This patch makes sure we first initialize everything and set the BDI_registered
flag, and only after this we add the bdi to 'bdi_list'. Current code adds the
bdi to the list too early, and as a result I the
WARN(!test_bit(BDI_registered, &bdi->state)
in bdi forker is triggered. Also, it is in general good practice to make things
visible only when they are fully initialized.
Also, this patch does few micro clean-ups:
1. Removes the 'exit' label which does not do anything, just returns. This
allows to get rid of few braces and 'ret' variable and make the code smaller.
2. If 'kthread_run()' fails, remove the error code it returns, not hard-coded
'-ENOMEM'. Theoretically, some day 'kthread_run()' can return something
else. Also, in case of failure it is not necessary to set 'bdi->wb.task' to
NULL.
Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>
Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
2010-07-25 11:29:25 +00:00
|
|
|
|
2015-07-02 14:44:34 +00:00
|
|
|
wb->congested = wb_congested_get_create(bdi, blkcg_id, gfp);
|
|
|
|
if (!wb->congested)
|
|
|
|
return -ENOMEM;
|
|
|
|
|
2015-05-22 21:13:34 +00:00
|
|
|
err = fprop_local_init_percpu(&wb->completions, gfp);
|
writeback: move bandwidth related fields from backing_dev_info into bdi_writeback
Currently, a bdi (backing_dev_info) embeds single wb (bdi_writeback)
and the role of the separation is unclear. For cgroup support for
writeback IOs, a bdi will be updated to host multiple wb's where each
wb serves writeback IOs of a different cgroup on the bdi. To achieve
that, a wb should carry all states necessary for servicing writeback
IOs for a cgroup independently.
This patch moves bandwidth related fields from backing_dev_info into
bdi_writeback.
* The moved fields are: bw_time_stamp, dirtied_stamp, written_stamp,
write_bandwidth, avg_write_bandwidth, dirty_ratelimit,
balanced_dirty_ratelimit, completions and dirty_exceeded.
* writeback_chunk_size() and over_bground_thresh() now take @wb
instead of @bdi.
* bdi_writeout_fraction(bdi, ...) -> wb_writeout_fraction(wb, ...)
bdi_dirty_limit(bdi, ...) -> wb_dirty_limit(wb, ...)
bdi_position_ration(bdi, ...) -> wb_position_ratio(wb, ...)
bdi_update_writebandwidth(bdi, ...) -> wb_update_write_bandwidth(wb, ...)
[__]bdi_update_bandwidth(bdi, ...) -> [__]wb_update_bandwidth(wb, ...)
bdi_{max|min}_pause(bdi, ...) -> wb_{max|min}_pause(wb, ...)
bdi_dirty_limits(bdi, ...) -> wb_dirty_limits(wb, ...)
* Init/exits of the relocated fields are moved to bdi_wb_init/exit()
respectively. Note that explicit zeroing is dropped in the process
as wb's are cleared in entirety anyway.
* As there's still only one bdi_writeback per backing_dev_info, all
uses of bdi->stat[] are mechanically replaced with bdi->wb.stat[]
introducing no behavior changes.
v2: Typo in description fixed as suggested by Jan.
Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Wu Fengguang <fengguang.wu@intel.com>
Cc: Jaegeuk Kim <jaegeuk@kernel.org>
Cc: Steven Whitehouse <swhiteho@redhat.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2015-05-22 21:13:28 +00:00
|
|
|
if (err)
|
2015-07-02 14:44:34 +00:00
|
|
|
goto out_put_cong;
|
writeback: cleanup bdi_register
This patch makes sure we first initialize everything and set the BDI_registered
flag, and only after this we add the bdi to 'bdi_list'. Current code adds the
bdi to the list too early, and as a result I the
WARN(!test_bit(BDI_registered, &bdi->state)
in bdi forker is triggered. Also, it is in general good practice to make things
visible only when they are fully initialized.
Also, this patch does few micro clean-ups:
1. Removes the 'exit' label which does not do anything, just returns. This
allows to get rid of few braces and 'ret' variable and make the code smaller.
2. If 'kthread_run()' fails, remove the error code it returns, not hard-coded
'-ENOMEM'. Theoretically, some day 'kthread_run()' can return something
else. Also, in case of failure it is not necessary to set 'bdi->wb.task' to
NULL.
Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>
Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
2010-07-25 11:29:25 +00:00
|
|
|
|
2015-05-22 21:13:27 +00:00
|
|
|
for (i = 0; i < NR_WB_STAT_ITEMS; i++) {
|
2015-05-22 21:13:34 +00:00
|
|
|
err = percpu_counter_init(&wb->stat[i], 0, gfp);
|
2015-07-02 14:44:34 +00:00
|
|
|
if (err)
|
|
|
|
goto out_destroy_stat;
|
2015-05-22 21:13:27 +00:00
|
|
|
}
|
2008-04-30 07:54:32 +00:00
|
|
|
|
2015-05-22 21:13:27 +00:00
|
|
|
return 0;
|
2015-07-02 14:44:34 +00:00
|
|
|
|
|
|
|
out_destroy_stat:
|
|
|
|
while (--i)
|
|
|
|
percpu_counter_destroy(&wb->stat[i]);
|
|
|
|
fprop_local_destroy_percpu(&wb->completions);
|
|
|
|
out_put_cong:
|
|
|
|
wb_congested_put(wb->congested);
|
|
|
|
return err;
|
2008-04-30 07:54:32 +00:00
|
|
|
}
|
|
|
|
|
2009-09-09 07:08:54 +00:00
|
|
|
/*
|
|
|
|
* Remove bdi from the global list and shutdown any threads we have running
|
|
|
|
*/
|
2015-05-22 21:13:31 +00:00
|
|
|
static void wb_shutdown(struct bdi_writeback *wb)
|
2009-09-02 07:19:46 +00:00
|
|
|
{
|
2015-01-20 21:05:00 +00:00
|
|
|
/* Make sure nobody queues further work */
|
2015-05-22 21:13:31 +00:00
|
|
|
spin_lock_bh(&wb->work_lock);
|
|
|
|
if (!test_and_clear_bit(WB_registered, &wb->state)) {
|
|
|
|
spin_unlock_bh(&wb->work_lock);
|
2009-09-09 07:08:54 +00:00
|
|
|
return;
|
2015-01-20 21:05:00 +00:00
|
|
|
}
|
2015-05-22 21:13:31 +00:00
|
|
|
spin_unlock_bh(&wb->work_lock);
|
2009-09-09 07:08:54 +00:00
|
|
|
|
|
|
|
/*
|
2015-05-22 21:13:31 +00:00
|
|
|
* Drain work list and shutdown the delayed_work. !WB_registered
|
|
|
|
* tells wb_workfn() that @wb is dying and its work_list needs to
|
|
|
|
* be drained no matter what.
|
2009-09-09 07:08:54 +00:00
|
|
|
*/
|
2015-05-22 21:13:31 +00:00
|
|
|
mod_delayed_work(bdi_wq, &wb->dwork, 0);
|
|
|
|
flush_delayed_work(&wb->dwork);
|
|
|
|
WARN_ON(!list_empty(&wb->work_list));
|
|
|
|
}
|
|
|
|
|
writeback: move backing_dev_info->wb_lock and ->worklist into bdi_writeback
Currently, a bdi (backing_dev_info) embeds single wb (bdi_writeback)
and the role of the separation is unclear. For cgroup support for
writeback IOs, a bdi will be updated to host multiple wb's where each
wb serves writeback IOs of a different cgroup on the bdi. To achieve
that, a wb should carry all states necessary for servicing writeback
IOs for a cgroup independently.
This patch moves bdi->wb_lock and ->worklist into wb.
* The lock protects bdi->worklist and bdi->wb.dwork scheduling. While
moving, rename it to wb->work_lock as wb->wb_lock is confusing.
Also, move wb->dwork downwards so that it's colocated with the new
->work_lock and ->work_list fields.
* bdi_writeback_workfn() -> wb_workfn()
bdi_wakeup_thread_delayed(bdi) -> wb_wakeup_delayed(wb)
bdi_wakeup_thread(bdi) -> wb_wakeup(wb)
bdi_queue_work(bdi, ...) -> wb_queue_work(wb, ...)
__bdi_start_writeback(bdi, ...) -> __wb_start_writeback(wb, ...)
get_next_work_item(bdi) -> get_next_work_item(wb)
* bdi_wb_shutdown() is renamed to wb_shutdown() and now takes @wb.
The function contained parts which belong to the containing bdi
rather than the wb itself - testing cap_writeback_dirty and
bdi_remove_from_list() invocation. Those are moved to
bdi_unregister().
* bdi_wb_{init|exit}() are renamed to wb_{init|exit}().
Initializations of the moved bdi->wb_lock and ->work_list are
relocated from bdi_init() to wb_init().
* As there's still only one bdi_writeback per backing_dev_info, all
uses of bdi->state are mechanically replaced with bdi->wb.state
introducing no behavior changes.
Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Wu Fengguang <fengguang.wu@intel.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2015-05-22 21:13:30 +00:00
|
|
|
static void wb_exit(struct bdi_writeback *wb)
|
2015-05-22 21:13:27 +00:00
|
|
|
{
|
|
|
|
int i;
|
|
|
|
|
|
|
|
WARN_ON(delayed_work_pending(&wb->dwork));
|
|
|
|
|
|
|
|
for (i = 0; i < NR_WB_STAT_ITEMS; i++)
|
|
|
|
percpu_counter_destroy(&wb->stat[i]);
|
2010-07-25 11:29:22 +00:00
|
|
|
|
writeback: move bandwidth related fields from backing_dev_info into bdi_writeback
Currently, a bdi (backing_dev_info) embeds single wb (bdi_writeback)
and the role of the separation is unclear. For cgroup support for
writeback IOs, a bdi will be updated to host multiple wb's where each
wb serves writeback IOs of a different cgroup on the bdi. To achieve
that, a wb should carry all states necessary for servicing writeback
IOs for a cgroup independently.
This patch moves bandwidth related fields from backing_dev_info into
bdi_writeback.
* The moved fields are: bw_time_stamp, dirtied_stamp, written_stamp,
write_bandwidth, avg_write_bandwidth, dirty_ratelimit,
balanced_dirty_ratelimit, completions and dirty_exceeded.
* writeback_chunk_size() and over_bground_thresh() now take @wb
instead of @bdi.
* bdi_writeout_fraction(bdi, ...) -> wb_writeout_fraction(wb, ...)
bdi_dirty_limit(bdi, ...) -> wb_dirty_limit(wb, ...)
bdi_position_ration(bdi, ...) -> wb_position_ratio(wb, ...)
bdi_update_writebandwidth(bdi, ...) -> wb_update_write_bandwidth(wb, ...)
[__]bdi_update_bandwidth(bdi, ...) -> [__]wb_update_bandwidth(wb, ...)
bdi_{max|min}_pause(bdi, ...) -> wb_{max|min}_pause(wb, ...)
bdi_dirty_limits(bdi, ...) -> wb_dirty_limits(wb, ...)
* Init/exits of the relocated fields are moved to bdi_wb_init/exit()
respectively. Note that explicit zeroing is dropped in the process
as wb's are cleared in entirety anyway.
* As there's still only one bdi_writeback per backing_dev_info, all
uses of bdi->stat[] are mechanically replaced with bdi->wb.stat[]
introducing no behavior changes.
v2: Typo in description fixed as suggested by Jan.
Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Wu Fengguang <fengguang.wu@intel.com>
Cc: Jaegeuk Kim <jaegeuk@kernel.org>
Cc: Steven Whitehouse <swhiteho@redhat.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2015-05-22 21:13:28 +00:00
|
|
|
fprop_local_destroy_percpu(&wb->completions);
|
2015-07-02 14:44:34 +00:00
|
|
|
wb_congested_put(wb->congested);
|
writeback: move bandwidth related fields from backing_dev_info into bdi_writeback
Currently, a bdi (backing_dev_info) embeds single wb (bdi_writeback)
and the role of the separation is unclear. For cgroup support for
writeback IOs, a bdi will be updated to host multiple wb's where each
wb serves writeback IOs of a different cgroup on the bdi. To achieve
that, a wb should carry all states necessary for servicing writeback
IOs for a cgroup independently.
This patch moves bandwidth related fields from backing_dev_info into
bdi_writeback.
* The moved fields are: bw_time_stamp, dirtied_stamp, written_stamp,
write_bandwidth, avg_write_bandwidth, dirty_ratelimit,
balanced_dirty_ratelimit, completions and dirty_exceeded.
* writeback_chunk_size() and over_bground_thresh() now take @wb
instead of @bdi.
* bdi_writeout_fraction(bdi, ...) -> wb_writeout_fraction(wb, ...)
bdi_dirty_limit(bdi, ...) -> wb_dirty_limit(wb, ...)
bdi_position_ration(bdi, ...) -> wb_position_ratio(wb, ...)
bdi_update_writebandwidth(bdi, ...) -> wb_update_write_bandwidth(wb, ...)
[__]bdi_update_bandwidth(bdi, ...) -> [__]wb_update_bandwidth(wb, ...)
bdi_{max|min}_pause(bdi, ...) -> wb_{max|min}_pause(wb, ...)
bdi_dirty_limits(bdi, ...) -> wb_dirty_limits(wb, ...)
* Init/exits of the relocated fields are moved to bdi_wb_init/exit()
respectively. Note that explicit zeroing is dropped in the process
as wb's are cleared in entirety anyway.
* As there's still only one bdi_writeback per backing_dev_info, all
uses of bdi->stat[] are mechanically replaced with bdi->wb.stat[]
introducing no behavior changes.
v2: Typo in description fixed as suggested by Jan.
Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Wu Fengguang <fengguang.wu@intel.com>
Cc: Jaegeuk Kim <jaegeuk@kernel.org>
Cc: Steven Whitehouse <swhiteho@redhat.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2015-05-22 21:13:28 +00:00
|
|
|
}
|
2010-08-29 17:22:30 +00:00
|
|
|
|
2015-05-22 21:13:37 +00:00
|
|
|
#ifdef CONFIG_CGROUP_WRITEBACK
|
|
|
|
|
|
|
|
#include <linux/memcontrol.h>
|
|
|
|
|
|
|
|
/*
|
|
|
|
* cgwb_lock protects bdi->cgwb_tree, bdi->cgwb_congested_tree,
|
|
|
|
* blkcg->cgwb_list, and memcg->cgwb_list. bdi->cgwb_tree is also RCU
|
|
|
|
* protected. cgwb_release_wait is used to wait for the completion of cgwb
|
|
|
|
* releases from bdi destruction path.
|
|
|
|
*/
|
|
|
|
static DEFINE_SPINLOCK(cgwb_lock);
|
|
|
|
static DECLARE_WAIT_QUEUE_HEAD(cgwb_release_wait);
|
|
|
|
|
|
|
|
/**
|
|
|
|
* wb_congested_get_create - get or create a wb_congested
|
|
|
|
* @bdi: associated bdi
|
|
|
|
* @blkcg_id: ID of the associated blkcg
|
|
|
|
* @gfp: allocation mask
|
|
|
|
*
|
|
|
|
* Look up the wb_congested for @blkcg_id on @bdi. If missing, create one.
|
|
|
|
* The returned wb_congested has its reference count incremented. Returns
|
|
|
|
* NULL on failure.
|
|
|
|
*/
|
|
|
|
struct bdi_writeback_congested *
|
|
|
|
wb_congested_get_create(struct backing_dev_info *bdi, int blkcg_id, gfp_t gfp)
|
|
|
|
{
|
|
|
|
struct bdi_writeback_congested *new_congested = NULL, *congested;
|
|
|
|
struct rb_node **node, *parent;
|
|
|
|
unsigned long flags;
|
|
|
|
retry:
|
|
|
|
spin_lock_irqsave(&cgwb_lock, flags);
|
|
|
|
|
|
|
|
node = &bdi->cgwb_congested_tree.rb_node;
|
|
|
|
parent = NULL;
|
|
|
|
|
|
|
|
while (*node != NULL) {
|
|
|
|
parent = *node;
|
|
|
|
congested = container_of(parent, struct bdi_writeback_congested,
|
|
|
|
rb_node);
|
|
|
|
if (congested->blkcg_id < blkcg_id)
|
|
|
|
node = &parent->rb_left;
|
|
|
|
else if (congested->blkcg_id > blkcg_id)
|
|
|
|
node = &parent->rb_right;
|
|
|
|
else
|
|
|
|
goto found;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (new_congested) {
|
|
|
|
/* !found and storage for new one already allocated, insert */
|
|
|
|
congested = new_congested;
|
|
|
|
new_congested = NULL;
|
|
|
|
rb_link_node(&congested->rb_node, parent, node);
|
|
|
|
rb_insert_color(&congested->rb_node, &bdi->cgwb_congested_tree);
|
|
|
|
goto found;
|
|
|
|
}
|
|
|
|
|
|
|
|
spin_unlock_irqrestore(&cgwb_lock, flags);
|
|
|
|
|
|
|
|
/* allocate storage for new one and retry */
|
|
|
|
new_congested = kzalloc(sizeof(*new_congested), gfp);
|
|
|
|
if (!new_congested)
|
|
|
|
return NULL;
|
|
|
|
|
|
|
|
atomic_set(&new_congested->refcnt, 0);
|
|
|
|
new_congested->bdi = bdi;
|
|
|
|
new_congested->blkcg_id = blkcg_id;
|
|
|
|
goto retry;
|
|
|
|
|
|
|
|
found:
|
|
|
|
atomic_inc(&congested->refcnt);
|
|
|
|
spin_unlock_irqrestore(&cgwb_lock, flags);
|
|
|
|
kfree(new_congested);
|
|
|
|
return congested;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* wb_congested_put - put a wb_congested
|
|
|
|
* @congested: wb_congested to put
|
|
|
|
*
|
|
|
|
* Put @congested and destroy it if the refcnt reaches zero.
|
|
|
|
*/
|
|
|
|
void wb_congested_put(struct bdi_writeback_congested *congested)
|
|
|
|
{
|
|
|
|
unsigned long flags;
|
|
|
|
|
|
|
|
local_irq_save(flags);
|
|
|
|
if (!atomic_dec_and_lock(&congested->refcnt, &cgwb_lock)) {
|
|
|
|
local_irq_restore(flags);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
2015-07-02 00:53:37 +00:00
|
|
|
/* bdi might already have been destroyed leaving @congested unlinked */
|
|
|
|
if (congested->bdi) {
|
|
|
|
rb_erase(&congested->rb_node,
|
|
|
|
&congested->bdi->cgwb_congested_tree);
|
|
|
|
congested->bdi = NULL;
|
|
|
|
}
|
|
|
|
|
2015-05-22 21:13:37 +00:00
|
|
|
spin_unlock_irqrestore(&cgwb_lock, flags);
|
|
|
|
kfree(congested);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void cgwb_release_workfn(struct work_struct *work)
|
|
|
|
{
|
|
|
|
struct bdi_writeback *wb = container_of(work, struct bdi_writeback,
|
|
|
|
release_work);
|
|
|
|
struct backing_dev_info *bdi = wb->bdi;
|
|
|
|
|
2015-10-02 18:47:05 +00:00
|
|
|
spin_lock_irq(&cgwb_lock);
|
|
|
|
list_del_rcu(&wb->bdi_node);
|
|
|
|
spin_unlock_irq(&cgwb_lock);
|
|
|
|
|
2015-05-22 21:13:37 +00:00
|
|
|
wb_shutdown(wb);
|
|
|
|
|
|
|
|
css_put(wb->memcg_css);
|
|
|
|
css_put(wb->blkcg_css);
|
|
|
|
|
2015-05-22 22:23:33 +00:00
|
|
|
fprop_local_destroy_percpu(&wb->memcg_completions);
|
2015-05-22 21:13:37 +00:00
|
|
|
percpu_ref_exit(&wb->refcnt);
|
|
|
|
wb_exit(wb);
|
|
|
|
kfree_rcu(wb, rcu);
|
|
|
|
|
|
|
|
if (atomic_dec_and_test(&bdi->usage_cnt))
|
|
|
|
wake_up_all(&cgwb_release_wait);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void cgwb_release(struct percpu_ref *refcnt)
|
|
|
|
{
|
|
|
|
struct bdi_writeback *wb = container_of(refcnt, struct bdi_writeback,
|
|
|
|
refcnt);
|
|
|
|
schedule_work(&wb->release_work);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void cgwb_kill(struct bdi_writeback *wb)
|
|
|
|
{
|
|
|
|
lockdep_assert_held(&cgwb_lock);
|
|
|
|
|
|
|
|
WARN_ON(!radix_tree_delete(&wb->bdi->cgwb_tree, wb->memcg_css->id));
|
|
|
|
list_del(&wb->memcg_node);
|
|
|
|
list_del(&wb->blkcg_node);
|
|
|
|
percpu_ref_kill(&wb->refcnt);
|
|
|
|
}
|
|
|
|
|
|
|
|
static int cgwb_create(struct backing_dev_info *bdi,
|
|
|
|
struct cgroup_subsys_state *memcg_css, gfp_t gfp)
|
|
|
|
{
|
|
|
|
struct mem_cgroup *memcg;
|
|
|
|
struct cgroup_subsys_state *blkcg_css;
|
|
|
|
struct blkcg *blkcg;
|
|
|
|
struct list_head *memcg_cgwb_list, *blkcg_cgwb_list;
|
|
|
|
struct bdi_writeback *wb;
|
|
|
|
unsigned long flags;
|
|
|
|
int ret = 0;
|
|
|
|
|
|
|
|
memcg = mem_cgroup_from_css(memcg_css);
|
blkcg: rename subsystem name from blkio to io
blkio interface has become messy over time and is currently the
largest. In addition to the inconsistent naming scheme, it has
multiple stat files which report more or less the same thing, a number
of debug stat files which expose internal details which shouldn't have
been part of the public interface in the first place, recursive and
non-recursive stats and leaf and non-leaf knobs.
Both recursive vs. non-recursive and leaf vs. non-leaf distinctions
don't make any sense on the unified hierarchy as only leaf cgroups can
contain processes. cgroups is going through a major interface
revision with the unified hierarchy involving significant fundamental
usage changes and given that a significant portion of the interface
doesn't make sense anymore, it's a good time to reorganize the
interface.
As the first step, this patch renames the external visible subsystem
name from "blkio" to "io". This is more concise, matches the other
two major subsystem names, "cpu" and "memory", and better suited as
blkcg will be involved in anything writeback related too whether an
actual block device is involved or not.
As the subsystem legacy_name is set to "blkio", the only userland
visible change outside the unified hierarchy is that blkcg is reported
as "io" instead of "blkio" in the subsystem initialized message during
boot. On the unified hierarchy, blkcg now appears as "io".
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Li Zefan <lizefan@huawei.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: cgroups@vger.kernel.org
Signed-off-by: Jens Axboe <axboe@fb.com>
2015-08-18 21:55:29 +00:00
|
|
|
blkcg_css = cgroup_get_e_css(memcg_css->cgroup, &io_cgrp_subsys);
|
2015-05-22 21:13:37 +00:00
|
|
|
blkcg = css_to_blkcg(blkcg_css);
|
|
|
|
memcg_cgwb_list = mem_cgroup_cgwb_list(memcg);
|
|
|
|
blkcg_cgwb_list = &blkcg->cgwb_list;
|
|
|
|
|
|
|
|
/* look up again under lock and discard on blkcg mismatch */
|
|
|
|
spin_lock_irqsave(&cgwb_lock, flags);
|
|
|
|
wb = radix_tree_lookup(&bdi->cgwb_tree, memcg_css->id);
|
|
|
|
if (wb && wb->blkcg_css != blkcg_css) {
|
|
|
|
cgwb_kill(wb);
|
|
|
|
wb = NULL;
|
|
|
|
}
|
|
|
|
spin_unlock_irqrestore(&cgwb_lock, flags);
|
|
|
|
if (wb)
|
|
|
|
goto out_put;
|
|
|
|
|
|
|
|
/* need to create a new one */
|
|
|
|
wb = kmalloc(sizeof(*wb), gfp);
|
|
|
|
if (!wb)
|
|
|
|
return -ENOMEM;
|
|
|
|
|
2015-07-02 14:44:34 +00:00
|
|
|
ret = wb_init(wb, bdi, blkcg_css->id, gfp);
|
2015-05-22 21:13:37 +00:00
|
|
|
if (ret)
|
|
|
|
goto err_free;
|
|
|
|
|
|
|
|
ret = percpu_ref_init(&wb->refcnt, cgwb_release, 0, gfp);
|
|
|
|
if (ret)
|
|
|
|
goto err_wb_exit;
|
|
|
|
|
2015-05-22 22:23:33 +00:00
|
|
|
ret = fprop_local_init_percpu(&wb->memcg_completions, gfp);
|
|
|
|
if (ret)
|
|
|
|
goto err_ref_exit;
|
|
|
|
|
2015-05-22 21:13:37 +00:00
|
|
|
wb->memcg_css = memcg_css;
|
|
|
|
wb->blkcg_css = blkcg_css;
|
|
|
|
INIT_WORK(&wb->release_work, cgwb_release_workfn);
|
|
|
|
set_bit(WB_registered, &wb->state);
|
2009-09-09 07:08:54 +00:00
|
|
|
|
|
|
|
/*
|
2015-05-22 21:13:37 +00:00
|
|
|
* The root wb determines the registered state of the whole bdi and
|
|
|
|
* memcg_cgwb_list and blkcg_cgwb_list's next pointers indicate
|
|
|
|
* whether they're still online. Don't link @wb if any is dead.
|
|
|
|
* See wb_memcg_offline() and wb_blkcg_offline().
|
2009-09-09 07:08:54 +00:00
|
|
|
*/
|
2015-05-22 21:13:37 +00:00
|
|
|
ret = -ENODEV;
|
|
|
|
spin_lock_irqsave(&cgwb_lock, flags);
|
|
|
|
if (test_bit(WB_registered, &bdi->wb.state) &&
|
|
|
|
blkcg_cgwb_list->next && memcg_cgwb_list->next) {
|
|
|
|
/* we might have raced another instance of this function */
|
|
|
|
ret = radix_tree_insert(&bdi->cgwb_tree, memcg_css->id, wb);
|
|
|
|
if (!ret) {
|
|
|
|
atomic_inc(&bdi->usage_cnt);
|
2015-10-02 18:47:05 +00:00
|
|
|
list_add_tail_rcu(&wb->bdi_node, &bdi->wb_list);
|
2015-05-22 21:13:37 +00:00
|
|
|
list_add(&wb->memcg_node, memcg_cgwb_list);
|
|
|
|
list_add(&wb->blkcg_node, blkcg_cgwb_list);
|
|
|
|
css_get(memcg_css);
|
|
|
|
css_get(blkcg_css);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
spin_unlock_irqrestore(&cgwb_lock, flags);
|
|
|
|
if (ret) {
|
|
|
|
if (ret == -EEXIST)
|
|
|
|
ret = 0;
|
2015-07-02 14:44:34 +00:00
|
|
|
goto err_fprop_exit;
|
2015-05-22 21:13:37 +00:00
|
|
|
}
|
|
|
|
goto out_put;
|
|
|
|
|
2015-05-22 22:23:33 +00:00
|
|
|
err_fprop_exit:
|
|
|
|
fprop_local_destroy_percpu(&wb->memcg_completions);
|
2015-05-22 21:13:37 +00:00
|
|
|
err_ref_exit:
|
|
|
|
percpu_ref_exit(&wb->refcnt);
|
|
|
|
err_wb_exit:
|
|
|
|
wb_exit(wb);
|
|
|
|
err_free:
|
|
|
|
kfree(wb);
|
|
|
|
out_put:
|
|
|
|
css_put(blkcg_css);
|
|
|
|
return ret;
|
2009-09-02 07:19:46 +00:00
|
|
|
}
|
|
|
|
|
2015-05-22 21:13:37 +00:00
|
|
|
/**
|
|
|
|
* wb_get_create - get wb for a given memcg, create if necessary
|
|
|
|
* @bdi: target bdi
|
|
|
|
* @memcg_css: cgroup_subsys_state of the target memcg (must have positive ref)
|
|
|
|
* @gfp: allocation mask to use
|
|
|
|
*
|
|
|
|
* Try to get the wb for @memcg_css on @bdi. If it doesn't exist, try to
|
|
|
|
* create one. The returned wb has its refcount incremented.
|
|
|
|
*
|
|
|
|
* This function uses css_get() on @memcg_css and thus expects its refcnt
|
|
|
|
* to be positive on invocation. IOW, rcu_read_lock() protection on
|
|
|
|
* @memcg_css isn't enough. try_get it before calling this function.
|
|
|
|
*
|
|
|
|
* A wb is keyed by its associated memcg. As blkcg implicitly enables
|
|
|
|
* memcg on the default hierarchy, memcg association is guaranteed to be
|
|
|
|
* more specific (equal or descendant to the associated blkcg) and thus can
|
|
|
|
* identify both the memcg and blkcg associations.
|
|
|
|
*
|
|
|
|
* Because the blkcg associated with a memcg may change as blkcg is enabled
|
|
|
|
* and disabled closer to root in the hierarchy, each wb keeps track of
|
|
|
|
* both the memcg and blkcg associated with it and verifies the blkcg on
|
|
|
|
* each lookup. On mismatch, the existing wb is discarded and a new one is
|
|
|
|
* created.
|
|
|
|
*/
|
|
|
|
struct bdi_writeback *wb_get_create(struct backing_dev_info *bdi,
|
|
|
|
struct cgroup_subsys_state *memcg_css,
|
|
|
|
gfp_t gfp)
|
2010-07-25 11:29:22 +00:00
|
|
|
{
|
2015-05-22 21:13:37 +00:00
|
|
|
struct bdi_writeback *wb;
|
|
|
|
|
2015-11-07 00:28:21 +00:00
|
|
|
might_sleep_if(gfpflags_allow_blocking(gfp));
|
2015-05-22 21:13:37 +00:00
|
|
|
|
|
|
|
if (!memcg_css->parent)
|
|
|
|
return &bdi->wb;
|
|
|
|
|
|
|
|
do {
|
|
|
|
rcu_read_lock();
|
|
|
|
wb = radix_tree_lookup(&bdi->cgwb_tree, memcg_css->id);
|
|
|
|
if (wb) {
|
|
|
|
struct cgroup_subsys_state *blkcg_css;
|
|
|
|
|
|
|
|
/* see whether the blkcg association has changed */
|
|
|
|
blkcg_css = cgroup_get_e_css(memcg_css->cgroup,
|
blkcg: rename subsystem name from blkio to io
blkio interface has become messy over time and is currently the
largest. In addition to the inconsistent naming scheme, it has
multiple stat files which report more or less the same thing, a number
of debug stat files which expose internal details which shouldn't have
been part of the public interface in the first place, recursive and
non-recursive stats and leaf and non-leaf knobs.
Both recursive vs. non-recursive and leaf vs. non-leaf distinctions
don't make any sense on the unified hierarchy as only leaf cgroups can
contain processes. cgroups is going through a major interface
revision with the unified hierarchy involving significant fundamental
usage changes and given that a significant portion of the interface
doesn't make sense anymore, it's a good time to reorganize the
interface.
As the first step, this patch renames the external visible subsystem
name from "blkio" to "io". This is more concise, matches the other
two major subsystem names, "cpu" and "memory", and better suited as
blkcg will be involved in anything writeback related too whether an
actual block device is involved or not.
As the subsystem legacy_name is set to "blkio", the only userland
visible change outside the unified hierarchy is that blkcg is reported
as "io" instead of "blkio" in the subsystem initialized message during
boot. On the unified hierarchy, blkcg now appears as "io".
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Li Zefan <lizefan@huawei.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: cgroups@vger.kernel.org
Signed-off-by: Jens Axboe <axboe@fb.com>
2015-08-18 21:55:29 +00:00
|
|
|
&io_cgrp_subsys);
|
2015-05-22 21:13:37 +00:00
|
|
|
if (unlikely(wb->blkcg_css != blkcg_css ||
|
|
|
|
!wb_tryget(wb)))
|
|
|
|
wb = NULL;
|
|
|
|
css_put(blkcg_css);
|
|
|
|
}
|
|
|
|
rcu_read_unlock();
|
|
|
|
} while (!wb && !cgwb_create(bdi, memcg_css, gfp));
|
|
|
|
|
|
|
|
return wb;
|
|
|
|
}
|
2010-07-25 11:29:22 +00:00
|
|
|
|
2015-07-02 14:44:34 +00:00
|
|
|
static int cgwb_bdi_init(struct backing_dev_info *bdi)
|
2015-05-22 21:13:37 +00:00
|
|
|
{
|
2015-07-02 14:44:34 +00:00
|
|
|
int ret;
|
|
|
|
|
2015-05-22 21:13:37 +00:00
|
|
|
INIT_RADIX_TREE(&bdi->cgwb_tree, GFP_ATOMIC);
|
|
|
|
bdi->cgwb_congested_tree = RB_ROOT;
|
|
|
|
atomic_set(&bdi->usage_cnt, 1);
|
2015-07-02 14:44:34 +00:00
|
|
|
|
|
|
|
ret = wb_init(&bdi->wb, bdi, 1, GFP_KERNEL);
|
|
|
|
if (!ret) {
|
2016-01-14 23:20:56 +00:00
|
|
|
bdi->wb.memcg_css = &root_mem_cgroup->css;
|
2015-07-02 14:44:34 +00:00
|
|
|
bdi->wb.blkcg_css = blkcg_root_css;
|
|
|
|
}
|
|
|
|
return ret;
|
2010-07-25 11:29:22 +00:00
|
|
|
}
|
|
|
|
|
2015-05-22 21:13:37 +00:00
|
|
|
static void cgwb_bdi_destroy(struct backing_dev_info *bdi)
|
|
|
|
{
|
|
|
|
struct radix_tree_iter iter;
|
2015-10-13 22:14:19 +00:00
|
|
|
struct rb_node *rbn;
|
2015-05-22 21:13:37 +00:00
|
|
|
void **slot;
|
|
|
|
|
|
|
|
WARN_ON(test_bit(WB_registered, &bdi->wb.state));
|
|
|
|
|
|
|
|
spin_lock_irq(&cgwb_lock);
|
2015-07-02 00:53:37 +00:00
|
|
|
|
2015-05-22 21:13:37 +00:00
|
|
|
radix_tree_for_each_slot(slot, &bdi->cgwb_tree, &iter, 0)
|
|
|
|
cgwb_kill(*slot);
|
2015-07-02 00:53:37 +00:00
|
|
|
|
2015-10-13 22:14:19 +00:00
|
|
|
while ((rbn = rb_first(&bdi->cgwb_congested_tree))) {
|
|
|
|
struct bdi_writeback_congested *congested =
|
|
|
|
rb_entry(rbn, struct bdi_writeback_congested, rb_node);
|
|
|
|
|
|
|
|
rb_erase(rbn, &bdi->cgwb_congested_tree);
|
2015-07-02 00:53:37 +00:00
|
|
|
congested->bdi = NULL; /* mark @congested unlinked */
|
|
|
|
}
|
|
|
|
|
2015-05-22 21:13:37 +00:00
|
|
|
spin_unlock_irq(&cgwb_lock);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* All cgwb's and their congested states must be shutdown and
|
|
|
|
* released before returning. Drain the usage counter to wait for
|
|
|
|
* all cgwb's and cgwb_congested's ever created on @bdi.
|
|
|
|
*/
|
|
|
|
atomic_dec(&bdi->usage_cnt);
|
|
|
|
wait_event(cgwb_release_wait, !atomic_read(&bdi->usage_cnt));
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* wb_memcg_offline - kill all wb's associated with a memcg being offlined
|
|
|
|
* @memcg: memcg being offlined
|
|
|
|
*
|
|
|
|
* Also prevents creation of any new wb's associated with @memcg.
|
2010-08-29 17:22:30 +00:00
|
|
|
*/
|
2015-05-22 21:13:37 +00:00
|
|
|
void wb_memcg_offline(struct mem_cgroup *memcg)
|
|
|
|
{
|
|
|
|
LIST_HEAD(to_destroy);
|
|
|
|
struct list_head *memcg_cgwb_list = mem_cgroup_cgwb_list(memcg);
|
|
|
|
struct bdi_writeback *wb, *next;
|
|
|
|
|
|
|
|
spin_lock_irq(&cgwb_lock);
|
|
|
|
list_for_each_entry_safe(wb, next, memcg_cgwb_list, memcg_node)
|
|
|
|
cgwb_kill(wb);
|
|
|
|
memcg_cgwb_list->next = NULL; /* prevent new wb's */
|
|
|
|
spin_unlock_irq(&cgwb_lock);
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* wb_blkcg_offline - kill all wb's associated with a blkcg being offlined
|
|
|
|
* @blkcg: blkcg being offlined
|
|
|
|
*
|
|
|
|
* Also prevents creation of any new wb's associated with @blkcg.
|
|
|
|
*/
|
|
|
|
void wb_blkcg_offline(struct blkcg *blkcg)
|
|
|
|
{
|
|
|
|
LIST_HEAD(to_destroy);
|
|
|
|
struct bdi_writeback *wb, *next;
|
|
|
|
|
|
|
|
spin_lock_irq(&cgwb_lock);
|
|
|
|
list_for_each_entry_safe(wb, next, &blkcg->cgwb_list, blkcg_node)
|
|
|
|
cgwb_kill(wb);
|
|
|
|
blkcg->cgwb_list.next = NULL; /* prevent new wb's */
|
|
|
|
spin_unlock_irq(&cgwb_lock);
|
|
|
|
}
|
|
|
|
|
|
|
|
#else /* CONFIG_CGROUP_WRITEBACK */
|
|
|
|
|
2015-07-02 14:44:34 +00:00
|
|
|
static int cgwb_bdi_init(struct backing_dev_info *bdi)
|
|
|
|
{
|
|
|
|
int err;
|
|
|
|
|
|
|
|
bdi->wb_congested = kzalloc(sizeof(*bdi->wb_congested), GFP_KERNEL);
|
|
|
|
if (!bdi->wb_congested)
|
|
|
|
return -ENOMEM;
|
|
|
|
|
|
|
|
err = wb_init(&bdi->wb, bdi, 1, GFP_KERNEL);
|
|
|
|
if (err) {
|
|
|
|
kfree(bdi->wb_congested);
|
|
|
|
return err;
|
|
|
|
}
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2015-05-22 21:13:37 +00:00
|
|
|
static void cgwb_bdi_destroy(struct backing_dev_info *bdi) { }
|
|
|
|
|
|
|
|
#endif /* CONFIG_CGROUP_WRITEBACK */
|
2010-08-29 17:22:30 +00:00
|
|
|
|
2007-10-17 06:25:47 +00:00
|
|
|
int bdi_init(struct backing_dev_info *bdi)
|
|
|
|
{
|
2015-10-02 18:47:05 +00:00
|
|
|
int ret;
|
|
|
|
|
2008-04-30 07:54:32 +00:00
|
|
|
bdi->dev = NULL;
|
|
|
|
|
2008-04-30 07:54:35 +00:00
|
|
|
bdi->min_ratio = 0;
|
2008-04-30 07:54:36 +00:00
|
|
|
bdi->max_ratio = 100;
|
2012-05-24 16:59:11 +00:00
|
|
|
bdi->max_prop_frac = FPROP_FRAC_BASE;
|
2009-09-02 07:19:46 +00:00
|
|
|
INIT_LIST_HEAD(&bdi->bdi_list);
|
2015-10-02 18:47:05 +00:00
|
|
|
INIT_LIST_HEAD(&bdi->wb_list);
|
2015-05-22 21:13:58 +00:00
|
|
|
init_waitqueue_head(&bdi->wb_waitq);
|
2009-09-09 07:08:54 +00:00
|
|
|
|
2015-10-02 18:47:05 +00:00
|
|
|
ret = cgwb_bdi_init(bdi);
|
|
|
|
|
|
|
|
list_add_tail_rcu(&bdi->wb.bdi_node, &bdi->wb_list);
|
|
|
|
|
|
|
|
return ret;
|
2007-10-17 06:25:47 +00:00
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(bdi_init);
|
2010-08-29 17:22:30 +00:00
|
|
|
|
2015-05-22 21:13:31 +00:00
|
|
|
int bdi_register(struct backing_dev_info *bdi, struct device *parent,
|
|
|
|
const char *fmt, ...)
|
|
|
|
{
|
|
|
|
va_list args;
|
|
|
|
struct device *dev;
|
2010-08-29 17:22:30 +00:00
|
|
|
|
2015-05-22 21:13:31 +00:00
|
|
|
if (bdi->dev) /* The driver needs to use separate queues per device */
|
|
|
|
return 0;
|
2010-08-29 17:22:30 +00:00
|
|
|
|
2015-05-22 21:13:31 +00:00
|
|
|
va_start(args, fmt);
|
|
|
|
dev = device_create_vargs(bdi_class, parent, MKDEV(0, 0), bdi, fmt, args);
|
|
|
|
va_end(args);
|
|
|
|
if (IS_ERR(dev))
|
|
|
|
return PTR_ERR(dev);
|
2007-10-17 06:25:50 +00:00
|
|
|
|
2015-05-22 21:13:31 +00:00
|
|
|
bdi->dev = dev;
|
2007-10-17 06:25:47 +00:00
|
|
|
|
2015-05-22 21:13:31 +00:00
|
|
|
bdi_debug_register(bdi, dev_name(dev));
|
|
|
|
set_bit(WB_registered, &bdi->wb.state);
|
|
|
|
|
|
|
|
spin_lock_bh(&bdi_lock);
|
|
|
|
list_add_tail_rcu(&bdi->bdi_list, &bdi_list);
|
|
|
|
spin_unlock_bh(&bdi_lock);
|
|
|
|
|
|
|
|
trace_writeback_bdi_register(bdi);
|
|
|
|
return 0;
|
2007-10-17 06:25:47 +00:00
|
|
|
}
|
2015-05-22 21:13:31 +00:00
|
|
|
EXPORT_SYMBOL(bdi_register);
|
2007-10-17 06:25:47 +00:00
|
|
|
|
2015-05-22 21:13:31 +00:00
|
|
|
int bdi_register_dev(struct backing_dev_info *bdi, dev_t dev)
|
2007-10-17 06:25:47 +00:00
|
|
|
{
|
2015-05-22 21:13:31 +00:00
|
|
|
return bdi_register(bdi, NULL, "%u:%u", MAJOR(dev), MINOR(dev));
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(bdi_register_dev);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* 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);
|
2007-10-17 06:25:47 +00:00
|
|
|
|
2015-05-22 21:13:31 +00:00
|
|
|
synchronize_rcu_expedited();
|
|
|
|
}
|
2008-04-30 07:54:32 +00:00
|
|
|
|
block: don't release bdi while request_queue has live references
bdi's are initialized in two steps, bdi_init() and bdi_register(), but
destroyed in a single step by bdi_destroy() which, for a bdi embedded
in a request_queue, is called during blk_cleanup_queue() which makes
the queue invisible and starts the draining of remaining usages.
A request_queue's user can access the congestion state of the embedded
bdi as long as it holds a reference to the queue. As such, it may
access the congested state of a queue which finished
blk_cleanup_queue() but hasn't reached blk_release_queue() yet.
Because the congested state was embedded in backing_dev_info which in
turn is embedded in request_queue, accessing the congested state after
bdi_destroy() was called was fine. The bdi was destroyed but the
memory region for the congested state remained accessible till the
queue got released.
a13f35e87140 ("writeback: don't embed root bdi_writeback_congested in
bdi_writeback") changed the situation. Now, the root congested state
which is expected to be pinned while request_queue remains accessible
is separately reference counted and the base ref is put during
bdi_destroy(). This means that the root congested state may go away
prematurely while the queue is between bdi_dstroy() and
blk_cleanup_queue(), which was detected by Andrey's KASAN tests.
The root cause of this problem is that bdi doesn't distinguish the two
steps of destruction, unregistration and release, and now the root
congested state actually requires a separate release step. To fix the
issue, this patch separates out bdi_unregister() and bdi_exit() from
bdi_destroy(). bdi_unregister() is called from blk_cleanup_queue()
and bdi_exit() from blk_release_queue(). bdi_destroy() is now just a
simple wrapper calling the two steps back-to-back.
While at it, the prototype of bdi_destroy() is moved right below
bdi_setup_and_register() so that the counterpart operations are
located together.
Signed-off-by: Tejun Heo <tj@kernel.org>
Fixes: a13f35e87140 ("writeback: don't embed root bdi_writeback_congested in bdi_writeback")
Cc: stable@vger.kernel.org # v4.2+
Reported-and-tested-by: Andrey Konovalov <andreyknvl@google.com>
Link: http://lkml.kernel.org/g/CAAeHK+zUJ74Zn17=rOyxacHU18SgCfC6bsYW=6kCY5GXJBwGfQ@mail.gmail.com
Reviewed-by: Jan Kara <jack@suse.com>
Reviewed-by: Jeff Moyer <jmoyer@redhat.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2015-09-08 16:20:22 +00:00
|
|
|
void bdi_unregister(struct backing_dev_info *bdi)
|
2007-10-17 06:25:47 +00:00
|
|
|
{
|
writeback: move backing_dev_info->wb_lock and ->worklist into bdi_writeback
Currently, a bdi (backing_dev_info) embeds single wb (bdi_writeback)
and the role of the separation is unclear. For cgroup support for
writeback IOs, a bdi will be updated to host multiple wb's where each
wb serves writeback IOs of a different cgroup on the bdi. To achieve
that, a wb should carry all states necessary for servicing writeback
IOs for a cgroup independently.
This patch moves bdi->wb_lock and ->worklist into wb.
* The lock protects bdi->worklist and bdi->wb.dwork scheduling. While
moving, rename it to wb->work_lock as wb->wb_lock is confusing.
Also, move wb->dwork downwards so that it's colocated with the new
->work_lock and ->work_list fields.
* bdi_writeback_workfn() -> wb_workfn()
bdi_wakeup_thread_delayed(bdi) -> wb_wakeup_delayed(wb)
bdi_wakeup_thread(bdi) -> wb_wakeup(wb)
bdi_queue_work(bdi, ...) -> wb_queue_work(wb, ...)
__bdi_start_writeback(bdi, ...) -> __wb_start_writeback(wb, ...)
get_next_work_item(bdi) -> get_next_work_item(wb)
* bdi_wb_shutdown() is renamed to wb_shutdown() and now takes @wb.
The function contained parts which belong to the containing bdi
rather than the wb itself - testing cap_writeback_dirty and
bdi_remove_from_list() invocation. Those are moved to
bdi_unregister().
* bdi_wb_{init|exit}() are renamed to wb_{init|exit}().
Initializations of the moved bdi->wb_lock and ->work_list are
relocated from bdi_init() to wb_init().
* As there's still only one bdi_writeback per backing_dev_info, all
uses of bdi->state are mechanically replaced with bdi->wb.state
introducing no behavior changes.
Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Wu Fengguang <fengguang.wu@intel.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2015-05-22 21:13:30 +00:00
|
|
|
/* make sure nobody finds us on the bdi_list anymore */
|
|
|
|
bdi_remove_from_list(bdi);
|
|
|
|
wb_shutdown(&bdi->wb);
|
2015-05-22 21:13:37 +00:00
|
|
|
cgwb_bdi_destroy(bdi);
|
2011-11-11 12:29:04 +00:00
|
|
|
|
2015-01-20 21:05:00 +00:00
|
|
|
if (bdi->dev) {
|
|
|
|
bdi_debug_unregister(bdi);
|
|
|
|
device_unregister(bdi->dev);
|
|
|
|
bdi->dev = NULL;
|
|
|
|
}
|
block: don't release bdi while request_queue has live references
bdi's are initialized in two steps, bdi_init() and bdi_register(), but
destroyed in a single step by bdi_destroy() which, for a bdi embedded
in a request_queue, is called during blk_cleanup_queue() which makes
the queue invisible and starts the draining of remaining usages.
A request_queue's user can access the congestion state of the embedded
bdi as long as it holds a reference to the queue. As such, it may
access the congested state of a queue which finished
blk_cleanup_queue() but hasn't reached blk_release_queue() yet.
Because the congested state was embedded in backing_dev_info which in
turn is embedded in request_queue, accessing the congested state after
bdi_destroy() was called was fine. The bdi was destroyed but the
memory region for the congested state remained accessible till the
queue got released.
a13f35e87140 ("writeback: don't embed root bdi_writeback_congested in
bdi_writeback") changed the situation. Now, the root congested state
which is expected to be pinned while request_queue remains accessible
is separately reference counted and the base ref is put during
bdi_destroy(). This means that the root congested state may go away
prematurely while the queue is between bdi_dstroy() and
blk_cleanup_queue(), which was detected by Andrey's KASAN tests.
The root cause of this problem is that bdi doesn't distinguish the two
steps of destruction, unregistration and release, and now the root
congested state actually requires a separate release step. To fix the
issue, this patch separates out bdi_unregister() and bdi_exit() from
bdi_destroy(). bdi_unregister() is called from blk_cleanup_queue()
and bdi_exit() from blk_release_queue(). bdi_destroy() is now just a
simple wrapper calling the two steps back-to-back.
While at it, the prototype of bdi_destroy() is moved right below
bdi_setup_and_register() so that the counterpart operations are
located together.
Signed-off-by: Tejun Heo <tj@kernel.org>
Fixes: a13f35e87140 ("writeback: don't embed root bdi_writeback_congested in bdi_writeback")
Cc: stable@vger.kernel.org # v4.2+
Reported-and-tested-by: Andrey Konovalov <andreyknvl@google.com>
Link: http://lkml.kernel.org/g/CAAeHK+zUJ74Zn17=rOyxacHU18SgCfC6bsYW=6kCY5GXJBwGfQ@mail.gmail.com
Reviewed-by: Jan Kara <jack@suse.com>
Reviewed-by: Jeff Moyer <jmoyer@redhat.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2015-09-08 16:20:22 +00:00
|
|
|
}
|
2015-01-20 21:05:00 +00:00
|
|
|
|
block: don't release bdi while request_queue has live references
bdi's are initialized in two steps, bdi_init() and bdi_register(), but
destroyed in a single step by bdi_destroy() which, for a bdi embedded
in a request_queue, is called during blk_cleanup_queue() which makes
the queue invisible and starts the draining of remaining usages.
A request_queue's user can access the congestion state of the embedded
bdi as long as it holds a reference to the queue. As such, it may
access the congested state of a queue which finished
blk_cleanup_queue() but hasn't reached blk_release_queue() yet.
Because the congested state was embedded in backing_dev_info which in
turn is embedded in request_queue, accessing the congested state after
bdi_destroy() was called was fine. The bdi was destroyed but the
memory region for the congested state remained accessible till the
queue got released.
a13f35e87140 ("writeback: don't embed root bdi_writeback_congested in
bdi_writeback") changed the situation. Now, the root congested state
which is expected to be pinned while request_queue remains accessible
is separately reference counted and the base ref is put during
bdi_destroy(). This means that the root congested state may go away
prematurely while the queue is between bdi_dstroy() and
blk_cleanup_queue(), which was detected by Andrey's KASAN tests.
The root cause of this problem is that bdi doesn't distinguish the two
steps of destruction, unregistration and release, and now the root
congested state actually requires a separate release step. To fix the
issue, this patch separates out bdi_unregister() and bdi_exit() from
bdi_destroy(). bdi_unregister() is called from blk_cleanup_queue()
and bdi_exit() from blk_release_queue(). bdi_destroy() is now just a
simple wrapper calling the two steps back-to-back.
While at it, the prototype of bdi_destroy() is moved right below
bdi_setup_and_register() so that the counterpart operations are
located together.
Signed-off-by: Tejun Heo <tj@kernel.org>
Fixes: a13f35e87140 ("writeback: don't embed root bdi_writeback_congested in bdi_writeback")
Cc: stable@vger.kernel.org # v4.2+
Reported-and-tested-by: Andrey Konovalov <andreyknvl@google.com>
Link: http://lkml.kernel.org/g/CAAeHK+zUJ74Zn17=rOyxacHU18SgCfC6bsYW=6kCY5GXJBwGfQ@mail.gmail.com
Reviewed-by: Jan Kara <jack@suse.com>
Reviewed-by: Jeff Moyer <jmoyer@redhat.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2015-09-08 16:20:22 +00:00
|
|
|
void bdi_exit(struct backing_dev_info *bdi)
|
|
|
|
{
|
|
|
|
WARN_ON_ONCE(bdi->dev);
|
writeback: move backing_dev_info->wb_lock and ->worklist into bdi_writeback
Currently, a bdi (backing_dev_info) embeds single wb (bdi_writeback)
and the role of the separation is unclear. For cgroup support for
writeback IOs, a bdi will be updated to host multiple wb's where each
wb serves writeback IOs of a different cgroup on the bdi. To achieve
that, a wb should carry all states necessary for servicing writeback
IOs for a cgroup independently.
This patch moves bdi->wb_lock and ->worklist into wb.
* The lock protects bdi->worklist and bdi->wb.dwork scheduling. While
moving, rename it to wb->work_lock as wb->wb_lock is confusing.
Also, move wb->dwork downwards so that it's colocated with the new
->work_lock and ->work_list fields.
* bdi_writeback_workfn() -> wb_workfn()
bdi_wakeup_thread_delayed(bdi) -> wb_wakeup_delayed(wb)
bdi_wakeup_thread(bdi) -> wb_wakeup(wb)
bdi_queue_work(bdi, ...) -> wb_queue_work(wb, ...)
__bdi_start_writeback(bdi, ...) -> __wb_start_writeback(wb, ...)
get_next_work_item(bdi) -> get_next_work_item(wb)
* bdi_wb_shutdown() is renamed to wb_shutdown() and now takes @wb.
The function contained parts which belong to the containing bdi
rather than the wb itself - testing cap_writeback_dirty and
bdi_remove_from_list() invocation. Those are moved to
bdi_unregister().
* bdi_wb_{init|exit}() are renamed to wb_{init|exit}().
Initializations of the moved bdi->wb_lock and ->work_list are
relocated from bdi_init() to wb_init().
* As there's still only one bdi_writeback per backing_dev_info, all
uses of bdi->state are mechanically replaced with bdi->wb.state
introducing no behavior changes.
Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Wu Fengguang <fengguang.wu@intel.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2015-05-22 21:13:30 +00:00
|
|
|
wb_exit(&bdi->wb);
|
2007-10-17 06:25:47 +00:00
|
|
|
}
|
block: don't release bdi while request_queue has live references
bdi's are initialized in two steps, bdi_init() and bdi_register(), but
destroyed in a single step by bdi_destroy() which, for a bdi embedded
in a request_queue, is called during blk_cleanup_queue() which makes
the queue invisible and starts the draining of remaining usages.
A request_queue's user can access the congestion state of the embedded
bdi as long as it holds a reference to the queue. As such, it may
access the congested state of a queue which finished
blk_cleanup_queue() but hasn't reached blk_release_queue() yet.
Because the congested state was embedded in backing_dev_info which in
turn is embedded in request_queue, accessing the congested state after
bdi_destroy() was called was fine. The bdi was destroyed but the
memory region for the congested state remained accessible till the
queue got released.
a13f35e87140 ("writeback: don't embed root bdi_writeback_congested in
bdi_writeback") changed the situation. Now, the root congested state
which is expected to be pinned while request_queue remains accessible
is separately reference counted and the base ref is put during
bdi_destroy(). This means that the root congested state may go away
prematurely while the queue is between bdi_dstroy() and
blk_cleanup_queue(), which was detected by Andrey's KASAN tests.
The root cause of this problem is that bdi doesn't distinguish the two
steps of destruction, unregistration and release, and now the root
congested state actually requires a separate release step. To fix the
issue, this patch separates out bdi_unregister() and bdi_exit() from
bdi_destroy(). bdi_unregister() is called from blk_cleanup_queue()
and bdi_exit() from blk_release_queue(). bdi_destroy() is now just a
simple wrapper calling the two steps back-to-back.
While at it, the prototype of bdi_destroy() is moved right below
bdi_setup_and_register() so that the counterpart operations are
located together.
Signed-off-by: Tejun Heo <tj@kernel.org>
Fixes: a13f35e87140 ("writeback: don't embed root bdi_writeback_congested in bdi_writeback")
Cc: stable@vger.kernel.org # v4.2+
Reported-and-tested-by: Andrey Konovalov <andreyknvl@google.com>
Link: http://lkml.kernel.org/g/CAAeHK+zUJ74Zn17=rOyxacHU18SgCfC6bsYW=6kCY5GXJBwGfQ@mail.gmail.com
Reviewed-by: Jan Kara <jack@suse.com>
Reviewed-by: Jeff Moyer <jmoyer@redhat.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2015-09-08 16:20:22 +00:00
|
|
|
|
|
|
|
void bdi_destroy(struct backing_dev_info *bdi)
|
|
|
|
{
|
|
|
|
bdi_unregister(bdi);
|
|
|
|
bdi_exit(bdi);
|
|
|
|
}
|
2007-10-17 06:25:47 +00:00
|
|
|
EXPORT_SYMBOL(bdi_destroy);
|
|
|
|
|
2010-04-22 09:37:01 +00:00
|
|
|
/*
|
|
|
|
* For use from filesystems to quickly init and register a bdi associated
|
|
|
|
* with dirty writeback
|
|
|
|
*/
|
2015-01-14 09:42:32 +00:00
|
|
|
int bdi_setup_and_register(struct backing_dev_info *bdi, char *name)
|
2010-04-22 09:37:01 +00:00
|
|
|
{
|
|
|
|
int err;
|
|
|
|
|
|
|
|
bdi->name = name;
|
2015-01-14 09:42:32 +00:00
|
|
|
bdi->capabilities = 0;
|
2010-04-22 09:37:01 +00:00
|
|
|
err = bdi_init(bdi);
|
|
|
|
if (err)
|
|
|
|
return err;
|
|
|
|
|
2013-07-03 22:04:56 +00:00
|
|
|
err = bdi_register(bdi, NULL, "%.28s-%ld", name,
|
|
|
|
atomic_long_inc_return(&bdi_seq));
|
2010-04-22 09:37:01 +00:00
|
|
|
if (err) {
|
|
|
|
bdi_destroy(bdi);
|
|
|
|
return err;
|
|
|
|
}
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(bdi_setup_and_register);
|
|
|
|
|
2006-10-20 06:28:16 +00:00
|
|
|
static wait_queue_head_t congestion_wqh[2] = {
|
|
|
|
__WAIT_QUEUE_HEAD_INITIALIZER(congestion_wqh[0]),
|
|
|
|
__WAIT_QUEUE_HEAD_INITIALIZER(congestion_wqh[1])
|
|
|
|
};
|
2015-05-22 21:13:41 +00:00
|
|
|
static atomic_t nr_wb_congested[2];
|
2006-10-20 06:28:16 +00:00
|
|
|
|
2015-05-22 21:13:41 +00:00
|
|
|
void clear_wb_congested(struct bdi_writeback_congested *congested, int sync)
|
2006-10-20 06:28:16 +00:00
|
|
|
{
|
2009-04-06 12:48:01 +00:00
|
|
|
wait_queue_head_t *wqh = &congestion_wqh[sync];
|
2015-05-22 21:13:41 +00:00
|
|
|
enum wb_state bit;
|
2006-10-20 06:28:16 +00:00
|
|
|
|
2015-05-22 21:13:26 +00:00
|
|
|
bit = sync ? WB_sync_congested : WB_async_congested;
|
2015-05-22 21:13:41 +00:00
|
|
|
if (test_and_clear_bit(bit, &congested->state))
|
|
|
|
atomic_dec(&nr_wb_congested[sync]);
|
2014-03-17 17:06:10 +00:00
|
|
|
smp_mb__after_atomic();
|
2006-10-20 06:28:16 +00:00
|
|
|
if (waitqueue_active(wqh))
|
|
|
|
wake_up(wqh);
|
|
|
|
}
|
2015-05-22 21:13:41 +00:00
|
|
|
EXPORT_SYMBOL(clear_wb_congested);
|
2006-10-20 06:28:16 +00:00
|
|
|
|
2015-05-22 21:13:41 +00:00
|
|
|
void set_wb_congested(struct bdi_writeback_congested *congested, int sync)
|
2006-10-20 06:28:16 +00:00
|
|
|
{
|
2015-05-22 21:13:26 +00:00
|
|
|
enum wb_state bit;
|
2006-10-20 06:28:16 +00:00
|
|
|
|
2015-05-22 21:13:26 +00:00
|
|
|
bit = sync ? WB_sync_congested : WB_async_congested;
|
2015-05-22 21:13:41 +00:00
|
|
|
if (!test_and_set_bit(bit, &congested->state))
|
|
|
|
atomic_inc(&nr_wb_congested[sync]);
|
2006-10-20 06:28:16 +00:00
|
|
|
}
|
2015-05-22 21:13:41 +00:00
|
|
|
EXPORT_SYMBOL(set_wb_congested);
|
2006-10-20 06:28:16 +00:00
|
|
|
|
|
|
|
/**
|
|
|
|
* congestion_wait - wait for a backing_dev to become uncongested
|
2009-07-09 12:52:32 +00:00
|
|
|
* @sync: SYNC or ASYNC IO
|
2006-10-20 06:28:16 +00:00
|
|
|
* @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.
|
|
|
|
*/
|
2009-07-09 12:52:32 +00:00
|
|
|
long congestion_wait(int sync, long timeout)
|
2006-10-20 06:28:16 +00:00
|
|
|
{
|
|
|
|
long ret;
|
2010-10-26 21:21:41 +00:00
|
|
|
unsigned long start = jiffies;
|
2006-10-20 06:28:16 +00:00
|
|
|
DEFINE_WAIT(wait);
|
2009-07-09 12:52:32 +00:00
|
|
|
wait_queue_head_t *wqh = &congestion_wqh[sync];
|
2006-10-20 06:28:16 +00:00
|
|
|
|
|
|
|
prepare_to_wait(wqh, &wait, TASK_UNINTERRUPTIBLE);
|
|
|
|
ret = io_schedule_timeout(timeout);
|
|
|
|
finish_wait(wqh, &wait);
|
2010-10-26 21:21:41 +00:00
|
|
|
|
|
|
|
trace_writeback_congestion_wait(jiffies_to_usecs(timeout),
|
|
|
|
jiffies_to_usecs(jiffies - start));
|
|
|
|
|
2006-10-20 06:28:16 +00:00
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(congestion_wait);
|
2007-10-17 06:25:50 +00:00
|
|
|
|
2010-10-26 21:21:45 +00:00
|
|
|
/**
|
|
|
|
* wait_iff_congested - Conditionally wait for a backing_dev to become uncongested or a zone to complete writes
|
|
|
|
* @zone: A zone to check if it is heavily congested
|
|
|
|
* @sync: SYNC or ASYNC IO
|
|
|
|
* @timeout: timeout in jiffies
|
|
|
|
*
|
|
|
|
* In the event of a congested backing_dev (any backing_dev) and the given
|
|
|
|
* @zone has experienced recent congestion, this waits for up to @timeout
|
|
|
|
* jiffies for either a BDI to exit congestion of the given @sync queue
|
|
|
|
* or a write to complete.
|
|
|
|
*
|
2015-12-11 21:40:32 +00:00
|
|
|
* In the absence of zone congestion, a short sleep or a cond_resched is
|
|
|
|
* performed to yield the processor and to allow other subsystems to make
|
|
|
|
* a forward progress.
|
2010-10-26 21:21:45 +00:00
|
|
|
*
|
|
|
|
* The return value is 0 if the sleep is for the full timeout. Otherwise,
|
|
|
|
* it is the number of jiffies that were still remaining when the function
|
|
|
|
* returned. return_value == timeout implies the function did not sleep.
|
|
|
|
*/
|
|
|
|
long wait_iff_congested(struct zone *zone, int sync, long timeout)
|
|
|
|
{
|
|
|
|
long ret;
|
|
|
|
unsigned long start = jiffies;
|
|
|
|
DEFINE_WAIT(wait);
|
|
|
|
wait_queue_head_t *wqh = &congestion_wqh[sync];
|
|
|
|
|
|
|
|
/*
|
|
|
|
* If there is no congestion, or heavy congestion is not being
|
|
|
|
* encountered in the current zone, yield if necessary instead
|
|
|
|
* of sleeping on the congestion queue
|
|
|
|
*/
|
2015-05-22 21:13:41 +00:00
|
|
|
if (atomic_read(&nr_wb_congested[sync]) == 0 ||
|
2014-10-09 22:28:17 +00:00
|
|
|
!test_bit(ZONE_CONGESTED, &zone->flags)) {
|
2015-12-11 21:40:32 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Memory allocation/reclaim might be called from a WQ
|
|
|
|
* context and the current implementation of the WQ
|
|
|
|
* concurrency control doesn't recognize that a particular
|
|
|
|
* WQ is congested if the worker thread is looping without
|
|
|
|
* ever sleeping. Therefore we have to do a short sleep
|
|
|
|
* here rather than calling cond_resched().
|
|
|
|
*/
|
|
|
|
if (current->flags & PF_WQ_WORKER)
|
2016-02-05 23:36:30 +00:00
|
|
|
schedule_timeout_uninterruptible(1);
|
2015-12-11 21:40:32 +00:00
|
|
|
else
|
|
|
|
cond_resched();
|
2010-10-26 21:21:45 +00:00
|
|
|
|
|
|
|
/* In case we scheduled, work out time remaining */
|
|
|
|
ret = timeout - (jiffies - start);
|
|
|
|
if (ret < 0)
|
|
|
|
ret = 0;
|
|
|
|
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Sleep until uncongested or a write happens */
|
|
|
|
prepare_to_wait(wqh, &wait, TASK_UNINTERRUPTIBLE);
|
|
|
|
ret = io_schedule_timeout(timeout);
|
|
|
|
finish_wait(wqh, &wait);
|
|
|
|
|
|
|
|
out:
|
|
|
|
trace_writeback_wait_iff_congested(jiffies_to_usecs(timeout),
|
|
|
|
jiffies_to_usecs(jiffies - start));
|
|
|
|
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(wait_iff_congested);
|
2012-07-31 23:41:52 +00:00
|
|
|
|
|
|
|
int pdflush_proc_obsolete(struct ctl_table *table, int write,
|
|
|
|
void __user *buffer, size_t *lenp, loff_t *ppos)
|
|
|
|
{
|
|
|
|
char kbuf[] = "0\n";
|
|
|
|
|
2013-09-11 21:22:44 +00:00
|
|
|
if (*ppos || *lenp < sizeof(kbuf)) {
|
2012-07-31 23:41:52 +00:00
|
|
|
*lenp = 0;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (copy_to_user(buffer, kbuf, sizeof(kbuf)))
|
|
|
|
return -EFAULT;
|
|
|
|
printk_once(KERN_WARNING "%s exported in /proc is scheduled for removal\n",
|
|
|
|
table->procname);
|
|
|
|
|
|
|
|
*lenp = 2;
|
|
|
|
*ppos += *lenp;
|
|
|
|
return 2;
|
|
|
|
}
|