While updating locking, b2efa05265 "block, cfq: unlink
cfq_io_context's immediately" moved elevator_exit_icq_fn() invocation
from exit_io_context() to the final ioc put. While this doesn't cause
catastrophic failure, it effectively removes task exit notification to
elevator and cause noticeable IO performance degradation with CFQ.
On task exit, CFQ used to immediately expire the slice if it was being
used by the exiting task as no more IO would be issued by the task;
however, after b2efa05265, the notification is lost and disk could sit
idle needlessly, leading to noticeable IO performance degradation for
certain workloads.
This patch renames ioc_exit_icq() to ioc_destroy_icq(), separates
elevator_exit_icq_fn() invocation into ioc_exit_icq() and invokes it
from exit_io_context(). ICQ_EXITED flag is added to avoid invoking
the callback more than once for the same icq.
Walking icq_list from ioc side and invoking elevator callback requires
reverse double locking. This may be better implemented using RCU;
unfortunately, using RCU isn't trivial. e.g. RCU protection would
need to cover request_queue and queue_lock switch on cleanup makes
grabbing queue_lock from RCU unsafe. Reverse double locking should
do, at least for now.
Signed-off-by: Tejun Heo <tj@kernel.org>
Reported-and-bisected-by: Shaohua Li <shli@kernel.org>
LKML-Reference: <CANejiEVzs=pUhQSTvUppkDcc2TNZyfohBRLygW5zFmXyk5A-xQ@mail.gmail.com>
Tested-by: Shaohua Li <shaohua.li@intel.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
icq->changed was used for ICQ_*_CHANGED bits. Rename it to flags and
access it under ioc->lock instead of using atomic bitops.
ioc_get_changed() is added so that the changed part can be fetched and
cleared as before.
icq->flags will be used to carry other flags.
Signed-off-by: Tejun Heo <tj@kernel.org>
Tested-by: Shaohua Li <shaohua.li@intel.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
put_io_context() performed a complex trylock dancing to avoid
deferring ioc release to workqueue. It was also broken on UP because
trylock was always assumed to succeed which resulted in unbalanced
preemption count.
While there are ways to fix the UP breakage, even the most
pathological microbench (forced ioc allocation and tight fork/exit
loop) fails to show any appreciable performance benefit of the
optimization. Strip it out. If there turns out to be workloads which
are affected by this change, simpler optimization from the discussion
thread can be applied later.
Signed-off-by: Tejun Heo <tj@kernel.org>
LKML-Reference: <1328514611.21268.66.camel@sli10-conroe>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Now block layer knows everything necessary to create and associate
icq's with requests. Move ioc_create_icq() to blk-ioc.c and update
get_request() such that, if elevator_type->icq_size is set, requests
are automatically associated with their matching icq's before
elv_set_request(). io_context reference is also managed by block core
on request alloc/free.
* Only ioprio/cgroup changed handling remains from cfq_get_cic().
Collapsed into cfq_set_request().
* This removes queue kicking on icq allocation failure (for now). As
icq allocation failure is rare and the only effect of queue kicking
achieved was possibily accelerating queue processing, this change
shouldn't be noticeable.
There is a larger underlying problem. Unlike request allocation,
icq allocation is not guaranteed to succeed eventually after
retries. The number of icq is unbound and thus mempool can't be the
solution either. This effectively adds allocation dependency on
memory free path and thus possibility of deadlock.
This usually wouldn't happen because icq allocation is not a hot
path and, even when the condition triggers, it's highly unlikely
that none of the writeback workers already has icq.
However, this is still possible especially if elevator is being
switched under high memory pressure, so we better get it fixed.
Probably the only solution is just bypassing elevator and appending
to dispatch queue on any elevator allocation failure.
* Comment added to explain how icq's are managed and synchronized.
This completes cleanup of io_context interface.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
With kmem_cache managed by blk-ioc, io_cq exit/release can be moved to
blk-ioc too. The odd ->io_cq->exit/release() callbacks are replaced
with elevator_ops->elevator_exit_icq_fn() with unlinking from both ioc
and q, and freeing automatically handled by blk-ioc. The elevator
operation only need to perform exit operation specific to the elevator
- in cfq's case, exiting the cfqq's.
Also, clearing of io_cq's on q detach is moved to block core and
automatically performed on elevator switch and q release.
Because the q io_cq points to might be freed before RCU callback for
the io_cq runs, blk-ioc code should remember to which cache the io_cq
needs to be freed when the io_cq is released. New field
io_cq->__rcu_icq_cache is added for this purpose. As both the new
field and rcu_head are used only after io_cq is released and the
q/ioc_node fields aren't, they are put into unions.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Currently io_context and cfq logics are mixed without clear boundary.
Most of io_context is independent from cfq but cfq_io_context handling
logic is dispersed between generic ioc code and cfq.
cfq_io_context represents association between an io_context and a
request_queue, which is a concept useful outside of cfq, but it also
contains fields which are useful only to cfq.
This patch takes out generic part and put it into io_cq (io
context-queue) and the rest into cfq_io_cq (cic moniker remains the
same) which contains io_cq. The following changes are made together.
* cfq_ttime and cfq_io_cq now live in cfq-iosched.c.
* All related fields, functions and constants are renamed accordingly.
* ioc->ioc_data is now "struct io_cq *" instead of "void *" and
renamed to icq_hint.
This prepares for io_context API cleanup. Documentation is currently
sparse. It will be added later.
Changes in this patch are mechanical and don't cause functional
change.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Now that lazy paths are removed, cfqd_dead_key() is meaningless and
cic->q can be used whereever cic->key is used. Kill cic->key.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
cic is association between io_context and request_queue. A cic is
linked from both ioc and q and should be destroyed when either one
goes away. As ioc and q both have their own locks, locking becomes a
bit complex - both orders work for removal from one but not from the
other.
Currently, cfq tries to circumvent this locking order issue with RCU.
ioc->lock nests inside queue_lock but the radix tree and cic's are
also protected by RCU allowing either side to walk their lists without
grabbing lock.
This rather unconventional use of RCU quickly devolves into extremely
fragile convolution. e.g. The following is from cfqd going away too
soon after ioc and q exits raced.
general protection fault: 0000 [#1] PREEMPT SMP
CPU 2
Modules linked in:
[ 88.503444]
Pid: 599, comm: hexdump Not tainted 3.1.0-rc10-work+ #158 Bochs Bochs
RIP: 0010:[<ffffffff81397628>] [<ffffffff81397628>] cfq_exit_single_io_context+0x58/0xf0
...
Call Trace:
[<ffffffff81395a4a>] call_for_each_cic+0x5a/0x90
[<ffffffff81395ab5>] cfq_exit_io_context+0x15/0x20
[<ffffffff81389130>] exit_io_context+0x100/0x140
[<ffffffff81098a29>] do_exit+0x579/0x850
[<ffffffff81098d5b>] do_group_exit+0x5b/0xd0
[<ffffffff81098de7>] sys_exit_group+0x17/0x20
[<ffffffff81b02f2b>] system_call_fastpath+0x16/0x1b
The only real hot path here is cic lookup during request
initialization and avoiding extra locking requires very confined use
of RCU. This patch makes cic removal from both ioc and request_queue
perform double-locking and unlink immediately.
* From q side, the change is almost trivial as ioc->lock nests inside
queue_lock. It just needs to grab each ioc->lock as it walks
cic_list and unlink it.
* From ioc side, it's a bit more difficult because of inversed lock
order. ioc needs its lock to walk its cic_list but can't grab the
matching queue_lock and needs to perform unlock-relock dancing.
Unlinking is now wholly done from put_io_context() and fast path is
optimized by using the queue_lock the caller already holds, which is
by far the most common case. If the ioc accessed multiple devices,
it tries with trylock. In unlikely cases of fast path failure, it
falls back to full double-locking dance from workqueue.
Double-locking isn't the prettiest thing in the world but it's *far*
simpler and more understandable than RCU trick without adding any
meaningful overhead.
This still leaves a lot of now unnecessary RCU logics. Future patches
will trim them.
-v2: Vivek pointed out that cic->q was being dereferenced after
cic->release() was called. Updated to use local variable @this_q
instead.
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Vivek Goyal <vgoyal@redhat.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
ioprio/cgroup change was handled by marking the changed state in ioc
and, on the following access to the ioc, performing RCU-protected
iteration through all cic's grabbing the matching queue_lock.
This patch moves the changed state to each cic. When ioprio or cgroup
changes, the respective bit is set on all cic's of the ioc and when
each of those cic (not ioc) is accessed, change is applied for that
specific ioc-queue pair.
This also fixes the following two race conditions between setting and
clearing of changed states.
* Missing barrier between assign/load of ioprio and ioprio_changed
allowed applying old ioprio.
* Change requests could happen between application of change and
clearing of changed variables.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Make the following changes to prepare for ioc/cic management cleanup.
* Add cic->q so that ioc can determine the associated queue without
querying cfq. This will eventually replace ->key.
* Factor out cfq_release_cic() from cic_free_func(). This function
assumes that the caller handled locking.
* Rename __cfq_exit_single_io_context() to cfq_exit_cic() and make it
take only @cic.
* Restructure cfq_cic_link() for future updates.
This patch doesn't introduce any functional changes.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Ignoring copy_io() during fork, io_context can be allocated from two
places - current_io_context() and set_task_ioprio(). The former is
always called from local task while the latter can be called from
different task. The synchornization between them are peculiar and
dubious.
* current_io_context() doesn't grab task_lock() and assumes that if it
saw %NULL ->io_context, it would stay that way until allocation and
assignment is complete. It has smp_wmb() between alloc/init and
assignment.
* set_task_ioprio() grabs task_lock() for assignment and does
smp_read_barrier_depends() between "ioc = task->io_context" and "if
(ioc)". Unfortunately, this doesn't achieve anything - the latter
is not a dependent load of the former. ie, if ioc itself were being
dereferenced "ioc->xxx", it would mean something (not sure what tho)
but as the code currently stands, the dependent read barrier is
noop.
As only one of the the two test-assignment sequences is task_lock()
protected, the task_lock() can't do much about race between the two.
Nothing prevents current_io_context() and set_task_ioprio() allocating
its own ioc for the same task and overwriting the other's.
Also, set_task_ioprio() can race with exiting task and create a new
ioc after exit_io_context() is finished.
ioc get/put doesn't have any reason to be complex. The only hot path
is accessing the existing ioc of %current, which is simple to achieve
given that ->io_context is never destroyed as long as the task is
alive. All other paths can happily go through task_lock() like all
other task sub structures without impacting anything.
This patch updates ioc get/put so that it becomes more conventional.
* alloc_io_context() is replaced with get_task_io_context(). This is
the only interface which can acquire access to ioc of another task.
On return, the caller has an explicit reference to the object which
should be put using put_io_context() afterwards.
* The functionality of current_io_context() remains the same but when
creating a new ioc, it shares the code path with
get_task_io_context() and always goes through task_lock().
* get_io_context() now means incrementing ref on an ioc which the
caller already has access to (be that an explicit refcnt or implicit
%current one).
* PF_EXITING inhibits creation of new io_context and once
exit_io_context() is finished, it's guaranteed that both ioc
acquisition functions return %NULL.
* All users are updated. Most are trivial but
smp_read_barrier_depends() removal from cfq_get_io_context() needs a
bit of explanation. I suppose the original intention was to ensure
ioc->ioprio is visible when set_task_ioprio() allocates new
io_context and installs it; however, this wouldn't have worked
because set_task_ioprio() doesn't have wmb between init and install.
There are other problems with this which will be fixed in another
patch.
* While at it, use NUMA_NO_NODE instead of -1 for wildcard node
specification.
-v2: Vivek spotted contamination from debug patch. Removed.
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Vivek Goyal <vgoyal@redhat.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
* int return from put_io_context() wasn't used by anybody. Make it
return void like other put functions and docbook-fy the function
comment.
* Reorder dummy declarations for !CONFIG_BLOCK case a bit.
* Make alloc_ioc_context() use __GFP_ZERO allocation, take init out of
if block and drop 0'ing.
* Docbook-fy current_io_context() comment.
This patch doesn't introduce any functional change.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Move the variables to do think time check to a sepatate struct. This is
to prepare adding think time check for service tree and group. No
functional change.
Signed-off-by: Shaohua Li <shaohua.li@intel.com>
Acked-by: Vivek Goyal <vgoyal@redhat.com>
Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
This adds annotations for RCU operations in core kernel components
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Al Viro <viro@ZenIV.linux.org.uk>
Cc: Jens Axboe <jens.axboe@oracle.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Josh Triplett <josh@joshtriplett.org>
Remove ->dead_key field from cfq_io_context to shrink its size to 128 bytes.
(64 bytes for 32-bit hosts)
Use lower bit in ->key as dead-mark, instead of moving key to separate field.
After this for dead cfq_io_context we got cic->key != cfqd automatically.
Thus, io_context's last-hit cache should work without changing.
Now to check ->key for non-dead state compare it with cfqd,
instead of checking ->key for non-null value as it was before.
Plus remove obsolete race protection in cfq_cic_lookup.
This race gone after v2.6.24-1728-g4ac845a
Signed-off-by: Konstantin Khlebnikov <khlebnikov@openvz.org>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Modify the Block I/O cgroup subsystem to be able to be built as a module.
As the CFQ disk scheduler optionally depends on blk-cgroup, config options
in block/Kconfig, block/Kconfig.iosched, and block/blk-cgroup.h are
enhanced to support the new module dependency.
Signed-off-by: Ben Blum <bblum@andrew.cmu.edu>
Cc: Li Zefan <lizf@cn.fujitsu.com>
Cc: Paul Menage <menage@google.com>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Lai Jiangshan <laijs@cn.fujitsu.com>
Cc: Vivek Goyal <vgoyal@redhat.com>
Cc: Jens Axboe <jens.axboe@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
On 64 bit builds when CONFIG_BLK_CGROUP=n (the default) this removes 8
bytes of padding from structure io_context and drops its size from 72 to
64 bytes, so needing one fewer cachelines and allowing more objects per
slab in it's kmem_cache.
Signed-off-by: Richard Kennedy <richard@rsk.demon.co.uk>
----
patch against 2.6.33
compiled & test on x86_64 AMDX2
regards
Richard
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
With CLONE_IO, parent's io_context->nr_tasks is incremented, but never
decremented whenever copy_process() fails afterwards, which prevents
exit_io_context() from calling IO schedulers exit functions.
Give a task_struct to exit_io_context(), and call exit_io_context() instead of
put_io_context() in copy_process() cleanup path.
Signed-off-by: Louis Rilling <louis.rilling@kerlabs.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
o This is basic implementation of blkio controller cgroup interface. This is
the common interface visible to user space and should be used by different
IO control policies as we implement those.
Signed-off-by: Vivek Goyal <vgoyal@redhat.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
async cfq_queue's are already shared between processes within the same
priority, and forthcoming patches will change the mapping of cic to sync
cfq_queue from 1:1 to 1:N. So, calculate the seekiness of a process
based on the cfq_queue instead of the cfq_io_context.
Signed-off-by: Jeff Moyer <jmoyer@redhat.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Currently io_context has an atomic_t(32-bit) as refcount. In the case of
cfq, for each device against whcih a task does I/O, a reference to the
io_context would be taken. And when there are multiple process sharing
io_contexts(CLONE_IO) would also have a reference to the same io_context.
Theoretically the possible maximum number of processes sharing the same
io_context + the number of disks/cfq_data referring to the same io_context
can overflow the 32-bit counter on a very high-end machine.
Even though it is an improbable case, let us make it atomic_long_t.
Signed-off-by: Nikanth Karthikesan <knikanth@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
SLAB_DESTROY_BY_RCU is not a direct substitute for normal call_rcu()
freeing, since it'll page freeing but NOT object freeing. So change
cfq to do the freeing on its own.
Signed-off-by: Fabio Checconi <fabio@gandalf.sssup.it>
Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
It's cumbersome to browse a radix tree from start to finish, especially
since we modify keys when a process exits. So add a hlist for the single
purpose of browsing over all known cfq_io_contexts, used for exit,
io prio change, etc.
This fixes http://bugzilla.kernel.org/show_bug.cgi?id=9948
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
The io context sharing introduced a per-ioc spinlock, that would protect
the cfq io context lookup. That is a regression from the original, since
we never needed any locking there because the ioc/cic were process private.
The cic lookup is changed from an rbtree construct to a radix tree, which
we can then use RCU to make the reader side lockless. That is the performance
critical path, modifying the radix tree is only done on process creation
(when that process first does IO, actually) and on process exit (if that
process has done IO).
As it so happens, radix trees are also much faster for this type of
lookup where the key is a pointer. It's a very sparse tree.
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
