Currently the forker thread can lose wake-ups which may lead to unnecessary
delays in processing bdi works. E.g., consider the following scenario.

1. 'bdi_forker_thread()' walks the 'bdi_list', finds out there is nothing to
   do, and is about to finish the loop.
2. A bdi thread decides to exit because it was inactive for long time.
3. 'bdi_queue_work()' adds a work to the bdi which just exited, so it wakes up
   the forker thread.
4. but 'bdi_forker_thread()' executes 'set_current_state(TASK_INTERRUPTIBLE)'
   and goes sleep. We lose a wake-up.

Losing the wake-up is not fatal, but this means that the bdi work processing
will be delayed by up to 5 sec. This race is theoretical, I never hit it, but
it is worth fixing.

The fix is to execute 'set_current_state(TASK_INTERRUPTIBLE)' _before_ walking
'bdi_list', not after.
Signed-off-by: NArtem Bityutskiy <Artem.Bityutskiy@nokia.com>
Reviewed-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJens Axboe <jaxboe@fusionio.com>

This patch fixes a very unlikely race condition on the bdi forker thread error
path: when bdi thread creation fails, 'bdi->wb.task' may contain the error code
for a short period of time. If at the same time someone submits a work to this
bdi, we can end up with an oops 'bdi_queue_work()' while executing
'wake_up_process(wb->task)'.

This patch fixes the issue by introducing a temporary variable 'task' and
storing the possible error code there, so that 'wb->task' would never take
erroneous values.

Note, this race is very unlikely and I never hit it, so it is theoretical, but
nevertheless worth fixing.

This patch also merges 2 comments which were previously separate.
Signed-off-by: NArtem Bityutskiy <Artem.Bityutskiy@nokia.com>
Reviewed-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJens Axboe <jaxboe@fusionio.com>

The write-back code mixes words "thread" and "task" for the same things. This
is not a big deal, but still an inconsistency.

hch: a convention I tend to use and I've seen in various places
is to always use _task for the storage of the task_struct pointer,
and thread everywhere else.  This especially helps with having
foo_thread for the actual thread and foo_task for a global
variable keeping the task_struct pointer

This patch renames:
* 'bdi_add_default_flusher_task()' -> 'bdi_add_default_flusher_thread()'
* 'bdi_forker_task()'              -> 'bdi_forker_thread()'

because bdi threads are 'bdi_writeback_thread()', so these names are more
consistent.

This patch also amends commentaries and makes them refer the forker and bdi
threads as "thread", not "task".

Also, while on it, make 'bdi_add_default_flusher_thread()' declaration use
'static void' instead of 'void static' and make checkpatch.pl happy.
Signed-off-by: NArtem Bityutskiy <Artem.Bityutskiy@nokia.com>
Reviewed-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJens Axboe <jaxboe@fusionio.com>

Trace queue/sched/exec parts of the writeback loop. This provides
insight into when and why flusher threads are scheduled to run. e.g
a sync invocation leaves traces like:

     sync-[...]: writeback_queue: bdi 8:0: sb_dev 8:1 nr_pages=7712 sync_mode=0 kupdate=0 range_cyclic=0 background=0
flush-8:0-[...]: writeback_exec: bdi 8:0: sb_dev 8:1 nr_pages=7712 sync_mode=0 kupdate=0 range_cyclic=0 background=0

This also lays the foundation for adding more writeback tracing to
provide deeper insight into the whole writeback path.

The original tracing code is from Jens Axboe, though this version is
a rewrite as a result of the code being traced changing
significantly.
Signed-off-by: NDave Chinner <dchinner@redhat.com>
Signed-off-by: NJens Axboe <jaxboe@fusionio.com>

Move all code for the writeback thread into fs/fs-writeback.c instead of
splitting it over two functions in two files.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJens Axboe <jaxboe@fusionio.com>

The wb_list member of struct backing_device_info always has exactly one
element.  Just use the direct bdi->wb pointer instead and simplify some
code.

Also remove bdi_task_init which is now trivial to prepare for the next
patch.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJens Axboe <jaxboe@fusionio.com>

First remove items from work_list as soon as we start working on them.  This
means we don't have to track any pending or visited state and can get
rid of all the RCU magic freeing the work items - we can simply free
them once the operation has finished.  Second use a real completion for
tracking synchronous requests - if the caller sets the completion pointer
we complete it, otherwise use it as a boolean indicator that we can free
the work item directly.  Third unify struct wb_writeback_args and struct
bdi_work into a single data structure, wb_writeback_work.  Previous we
set all parameters into a struct wb_writeback_args, copied it into
struct bdi_work, copied it again on the stack to use it there.  Instead
of just allocate one structure dynamically or on the stack and use it
all the way through the stack.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJens Axboe <jaxboe@fusionio.com>

This was just an odd wrapper around writeback_inodes_wb.  Removing this
also allows to get rid of the bdi member of struct writeback_control
which was rather out of place there.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJens Axboe <jaxboe@fusionio.com>

Commit 69b62d01 fixed up most of the places where we would enter
busy schedule() spins when disabling the periodic background
writeback. This fixes up the sb timer so that it doesn't get
hammered on with the delay disabled, and ensures that it gets
rearmed if needed when /proc/sys/vm/dirty_writeback_centisecs
gets modified.

bdi_forker_task() also needs to check for !dirty_writeback_centisecs
and use schedule() appropriately, fix that up too.
Signed-off-by: NJens Axboe <jens.axboe@oracle.com>

noop_backing_dev_info is used only as a flag to mark filesystems that
don't have any backing store, like tmpfs, procfs, spufs, etc.
Signed-off-by: NJoern Engel <joern@logfs.org>

Changed the BUG_ON() to a WARN_ON(). Note that adding dirty inodes
to the noop_backing_dev_info is not legal and will not result in
them being flushed, but we already catch this condition in
__mark_inode_dirty() when checking for a registered bdi.
Signed-off-by: NJens Axboe <jens.axboe@oracle.com>

Pretty trivial helper, just sets up the bdi and registers it. An atomic
sequence count is used to ensure that the registered sysfs names are
unique.
Signed-off-by: NJens Axboe <jens.axboe@oracle.com>

I hit this when we had a bug in IDR for a few days. Basically sysfs would
fail to create new inodes since it uses an IDR and therefore class_create would
fail.

While we are unlikely to see this fail we may as well handle it instead of
oopsing.
Signed-off-by: NAnton Blanchard <anton@samba.org>
Signed-off-by: NJens Axboe <jens.axboe@oracle.com>

To touch task->flags directly is racy. thaw_process() still has race
(changing non_current->flags, but this is another issue) though, I think
it's much better off.

So, use thaw_process() instead.
Signed-off-by: NOGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: NJens Axboe <jens.axboe@oracle.com>

Unfreezes the bdi flusher task when the said task needs to exit.

Steps to reproduce this.
1) Mount a file system from MMC/SD card.
2) Unmount the file system. This creates a flusher task.
3) Attempt suspend to RAM. System is unresponsive.

This is because the bdi flusher thread is already in the refrigerator and will
remain so until it is thawed. The MMC driver suspend routine call stack will
ultimately issue a 'kthread_stop' on the bdi flusher thread and will block
until the flusher thread is exited. Since the bdi flusher thread is in the
refrigerator it never cleans up until thawed.
Signed-off-by: NRomit Dasgupta <romit@ti.com>
Signed-off-by: NJens Axboe <jens.axboe@oracle.com>

Commit 592b09a4 was different from
the tested path, in that it moved the bdi super_block prune from
unregister to destroy context. This doesn't fully fix the sync hang
bug on unexpected device removal, as need to prune the bdi cache
pointer before killing flusher thread.
Tested-by: NArtur Skawina <art.08.09@gmail.com>
Signed-off-by: NJens Axboe <jens.axboe@oracle.com>

When the bdi is being removed, we have to ensure that no super_blocks
currently have that cached in sb->s_bdi. Normally this is ensured by
the sb having a longer life span than the bdi, but if the device is
suddenly yanked, we have to kill this reference. sb->s_bdi is pointed
to freed memory at that point.

This fixes a problem with sync(1) hanging when a USB stick is pulled
without cleanly umounting it first.
Reported-by: NPavel Machek <pavel@ucw.cz>
Signed-off-by: NJens Axboe <jens.axboe@oracle.com>

The space is not script friendly, kill it.
Signed-off-by: NWu Fengguang <fengguang.wu@intel.com>
Signed-off-by: NJens Axboe <jens.axboe@oracle.com>

We cannot safely ensure that the inodes are all gone at this point
in time, and we must not destroy this bdi with inodes having off it.
So just splice our entries to the default bdi since that one will
always persist.
Signed-off-by: NJens Axboe <jens.axboe@oracle.com>

Now that bdi_writeback_all() no longer handles integrity writeback,
it doesn't have to block anymore. This means that we can switch
bdi_list reader side protection to RCU.
Signed-off-by: NJens Axboe <jens.axboe@oracle.com>

Also a debugging aid. We want to catch dirty inodes being added to
backing devices that don't do writeback.
Signed-off-by: NJens Axboe <jens.axboe@oracle.com>

This enables us to track who does what and print info. Its main use
is catching dirty inodes on the default_backing_dev_info, so we can
fix that up.
Signed-off-by: NJens Axboe <jens.axboe@oracle.com>

Add some debug entries to be able to inspect the internal state of
the writeback details.
Signed-off-by: NJens Axboe <jens.axboe@oracle.com>

This gets rid of pdflush for bdi writeout and kupdated style cleaning.
pdflush writeout suffers from lack of locality and also requires more
threads to handle the same workload, since it has to work in a
non-blocking fashion against each queue. This also introduces lumpy
behaviour and potential request starvation, since pdflush can be starved
for queue access if others are accessing it. A sample ffsb workload that
does random writes to files is about 8% faster here on a simple SATA drive
during the benchmark phase. File layout also seems a LOT more smooth in
vmstat:

 r  b   swpd   free   buff  cache   si   so    bi    bo   in    cs us sy id wa
 0  1      0 608848   2652 375372    0    0     0 71024  604    24  1 10 48 42
 0  1      0 549644   2712 433736    0    0     0 60692  505    27  1  8 48 44
 1  0      0 476928   2784 505192    0    0     4 29540  553    24  0  9 53 37
 0  1      0 457972   2808 524008    0    0     0 54876  331    16  0  4 38 58
 0  1      0 366128   2928 614284    0    0     4 92168  710    58  0 13 53 34
 0  1      0 295092   3000 684140    0    0     0 62924  572    23  0  9 53 37
 0  1      0 236592   3064 741704    0    0     4 58256  523    17  0  8 48 44
 0  1      0 165608   3132 811464    0    0     0 57460  560    21  0  8 54 38
 0  1      0 102952   3200 873164    0    0     4 74748  540    29  1 10 48 41
 0  1      0  48604   3252 926472    0    0     0 53248  469    29  0  7 47 45

where vanilla tends to fluctuate a lot in the creation phase:

 r  b   swpd   free   buff  cache   si   so    bi    bo   in    cs us sy id wa
 1  1      0 678716   5792 303380    0    0     0 74064  565    50  1 11 52 36
 1  0      0 662488   5864 319396    0    0     4   352  302   329  0  2 47 51
 0  1      0 599312   5924 381468    0    0     0 78164  516    55  0  9 51 40
 0  1      0 519952   6008 459516    0    0     4 78156  622    56  1 11 52 37
 1  1      0 436640   6092 541632    0    0     0 82244  622    54  0 11 48 41
 0  1      0 436640   6092 541660    0    0     0     8  152    39  0  0 51 49
 0  1      0 332224   6200 644252    0    0     4 102800  728    46  1 13 49 36
 1  0      0 274492   6260 701056    0    0     4 12328  459    49  0  7 50 43
 0  1      0 211220   6324 763356    0    0     0 106940  515    37  1 10 51 39
 1  0      0 160412   6376 813468    0    0     0  8224  415    43  0  6 49 45
 1  1      0  85980   6452 886556    0    0     4 113516  575    39  1 11 54 34
 0  2      0  85968   6452 886620    0    0     0  1640  158   211  0  0 46 54

A 10 disk test with btrfs performs 26% faster with per-bdi flushing. A
SSD based writeback test on XFS performs over 20% better as well, with
the throughput being very stable around 1GB/sec, where pdflush only
manages 750MB/sec and fluctuates wildly while doing so. Random buffered
writes to many files behave a lot better as well, as does random mmap'ed
writes.

A separate thread is added to sync the super blocks. In the long term,
adding sync_supers_bdi() functionality could get rid of this thread again.
Signed-off-by: NJens Axboe <jens.axboe@oracle.com>

This is a first step at introducing per-bdi flusher threads. We should
have no change in behaviour, although sb_has_dirty_inodes() is now
ridiculously expensive, as there's no easy way to answer that question.
Not a huge problem, since it'll be deleted in subsequent patches.
Signed-off-by: NJens Axboe <jens.axboe@oracle.com>

Commit 1faa16d2 accidentally broke
the bdi congestion wait queue logic, causing us to wait on congestion
for WRITE (== 1) when we really wanted BLK_RW_ASYNC (== 0) instead.
Signed-off-by: NJens Axboe <jens.axboe@oracle.com>

This makes sure that we never wait on async IO for sync requests, instead
of doing the split on writes vs reads.
Signed-off-by: NJens Axboe <jens.axboe@oracle.com>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

It really makes no sense to have it in readahead.c, so move it where
it belongs.
Signed-off-by: NJens Axboe <jens.axboe@oracle.com>

The background dirty and dirty limits are better defined with type
specifiers of unsigned long since negative writeback thresholds are not
possible.

These values, as returned by get_dirty_limits(), are normally compared
with ZVC values to determine whether writeback shall commence or be
throttled.  Such page counts cannot be negative, so declaring the page
limits as signed is unnecessary.
Acked-by: NPeter Zijlstra <peterz@infradead.org>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Christoph Lameter <cl@linux-foundation.org>
Signed-off-by: NDavid Rientjes <rientjes@google.com>
Cc: Andrea Righi <righi.andrea@gmail.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Impact: fix lockdep false positives

Classify percpu_counter instances similar to regular lock objects --
that is, per instantiation site.

The networking code has increased its use of percpu_counters, which
leads to false positives if they are treated as a single class.
Signed-off-by: NPeter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

On second thoughts, this is just going to disturb people while telling us
things which we already knew.

Cc: Peter Korsgaard <jacmet@sunsite.dk>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Kay Sievers <kay.sievers@vrfy.org>
Cc: David Woodhouse <dwmw2@infradead.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Devices which share the same queue, like floppies and mtd devices, get
registered multiple times in the bdi interface, but bdi accounts only the
last registered device of the devices sharing one queue.

On remove, all earlier registered devices leak, stay around in sysfs, and
cause "duplicate filename" errors if the devices are re-created.

This prevents the creation of multiple bdi interfaces per queue, and the
bdi device will carry the dev_t name of the block device which is the
first one registered, of the pool of devices using the same queue.

[akpm@linux-foundation.org: add a WARN_ON so we know which drivers are misbehaving]
Tested-by: NPeter Korsgaard <jacmet@sunsite.dk>
Acked-by: NPeter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: NKay Sievers <kay.sievers@vrfy.org>
Cc: David Woodhouse <dwmw2@infradead.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

There is a race from when a device is created with device_create() and
then the drvdata is set with a call to dev_set_drvdata() in which a
sysfs file could be open, yet the drvdata will be NULL, causing all
sorts of bad things to happen.

This patch fixes the problem by using the new function,
device_create_vargs().

Many thanks to Arthur Jones <ajones@riverbed.com> for reporting the bug,
and testing patches out.

Cc: Kay Sievers <kay.sievers@vrfy.org>
Cc: Arthur Jones <ajones@riverbed.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Miklos Szeredi <mszeredi@suse.cz>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

Move BDI statistics to debugfs:

   /sys/kernel/debug/bdi/<bdi>/stats

Use postcore_initcall() to initialize the sysfs class and debugfs,
because debugfs is initialized in core_initcall().

Update descriptions in ABI documentation.
Signed-off-by: NMiklos Szeredi <mszeredi@suse.cz>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Add "max_ratio" to /sys/class/bdi.  This indicates the maximum percentage of
the global dirty threshold allocated to this bdi.

[mszeredi@suse.cz]

 - fix parsing in max_ratio_store().
 - export bdi_set_max_ratio() to modules
 - limit bdi_dirty with bdi->max_ratio
 - document new sysfs attribute
Signed-off-by: NPeter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: NMiklos Szeredi <mszeredi@suse.cz>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Under normal circumstances each device is given a part of the total write-back
cache that relates to its current avg writeout speed in relation to the other
devices.

min_ratio - allows one to assign a minimum portion of the write-back cache to
a particular device.  This is useful in situations where you might want to
provide a minimum QoS.  (One request for this feature came from flash based
storage people who wanted to avoid writing out at all costs - they of course
needed some pdflush hacks as well)

max_ratio - allows one to assign a maximum portion of the dirty limit to a
particular device.  This is useful in situations where you want to avoid one
device taking all or most of the write-back cache.  Eg.  an NFS mount that is
prone to get stuck, or a FUSE mount which you don't trust to play fair.

Add "min_ratio" to /sys/class/bdi.  This indicates the minimum percentage of
the global dirty threshold allocated to this bdi.

[mszeredi@suse.cz]

 - fix parsing in min_ratio_store()
 - document new sysfs attribute
Signed-off-by: NPeter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: NMiklos Szeredi <mszeredi@suse.cz>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Provide a place in sysfs (/sys/class/bdi) for the backing_dev_info object.
This allows us to see and set the various BDI specific variables.

In particular this properly exposes the read-ahead window for all relevant
users and /sys/block/<block>/queue/read_ahead_kb should be deprecated.

With patient help from Kay Sievers and Greg KH

[mszeredi@suse.cz]

 - split off NFS and FUSE changes into separate patches
 - document new sysfs attributes under Documentation/ABI
 - do bdi_class_init as a core_initcall, otherwise the "default" BDI
   won't be initialized
 - remove bdi_init_fmt macro, it's not used very much

[akpm@linux-foundation.org: fix ia64 warning]
Signed-off-by: NPeter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Kay Sievers <kay.sievers@vrfy.org>
Acked-by: NGreg KH <greg@kroah.com>
Cc: Trond Myklebust <trond.myklebust@fys.uio.no>
Signed-off-by: NMiklos Szeredi <mszeredi@suse.cz>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

this call should use the array index j, not i.  But with this approach, just
one int i is enough, int j is not needed.
Signed-off-by: NDenis Cheng <crquan@gmail.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Scale writeback cache per backing device, proportional to its writeout speed.

By decoupling the BDI dirty thresholds a number of problems we currently have
will go away, namely:

 - mutual interference starvation (for any number of BDIs);
 - deadlocks with stacked BDIs (loop, FUSE and local NFS mounts).

It might be that all dirty pages are for a single BDI while other BDIs are
idling. By giving each BDI a 'fair' share of the dirty limit, each one can have
dirty pages outstanding and make progress.

A global threshold also creates a deadlock for stacked BDIs; when A writes to
B, and A generates enough dirty pages to get throttled, B will never start
writeback until the dirty pages go away. Again, by giving each BDI its own
'independent' dirty limit, this problem is avoided.

So the problem is to determine how to distribute the total dirty limit across
the BDIs fairly and efficiently. A DBI that has a large dirty limit but does
not have any dirty pages outstanding is a waste.

What is done is to keep a floating proportion between the DBIs based on
writeback completions. This way faster/more active devices get a larger share
than slower/idle devices.

[akpm@linux-foundation.org: fix warnings]
[hugh@veritas.com: Fix occasional hang when a task couldn't get out of balance_dirty_pages]
Signed-off-by: NPeter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: NHugh Dickins <hugh@veritas.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Provide scalable per backing_dev_info statistics counters.
Signed-off-by: NPeter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

These patches aim to improve balance_dirty_pages() and directly address three
issues:
  1) inter device starvation
  2) stacked device deadlocks
  3) inter process starvation

1 and 2 are a direct result from removing the global dirty limit and using
per device dirty limits. By giving each device its own dirty limit is will
no longer starve another device, and the cyclic dependancy on the dirty limit
is broken.

In order to efficiently distribute the dirty limit across the independant
devices a floating proportion is used, this will allocate a share of the total
limit proportional to the device's recent activity.

3 is done by also scaling the dirty limit proportional to the current task's
recent dirty rate.

This patch:

nfs: remove congestion_end().  It's redundant, clear_bdi_congested() already
wakes the waiters.
Signed-off-by: NPeter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Trond Myklebust <trond.myklebust@fys.uio.no>
Cc: "J. Bruce Fields" <bfields@fieldses.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>