- Move the queue_new flag clear to when the queue is selected
- Only select the non-first queue in cfq_get_best_queue(), if there's
a substantial difference between the best and first.
- Get rid of ->busy_rr
- Only select a close cooperator, if the current queue is known to take
a while to "think".
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
- Implement logic for detecting cooperating processes, so we
choose the best available queue whenever possible.
- Improve residual slice time accounting.
- Remove dead code: we no longer see async requests coming in on
sync queues. That part was removed a long time ago. That means
that we can also remove the difference between cfq_cfqq_sync()
and cfq_cfqq_class_sync(), they are now indentical. And we can
kill the on_dispatch array, just make it a counter.
- Allow a process to go into the current list, if it hasn't been
serviced in this scheduler tick yet.
Possible future improvements including caching the cfqq lookup
in cfq_close_cooperator(), so we don't have to look it up twice.
cfq_get_best_queue() should just use that last decision instead
of doing it again.
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
When testing the syslet async io approach, I discovered that CFQ
sometimes didn't perform as well as expected. cfq_should_preempt()
needs to better check for cooperating tasks, so fix that by allowing
preemption of an equal priority queue if the recently queued request
is as good a candidate for IO as the one we are currently waiting for.
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
There's a really rare and obscure bug in CFQ, that causes a crash in
cfq_dispatch_insert() due to rq == NULL. One example of the resulting
oops is seen here:
http://lkml.org/lkml/2007/4/15/41
Neil correctly diagnosed the situation for how this can happen: if two
concurrent requests with the exact same sector number (due to direct IO
or aliasing between MD and the raw device access), the alias handling
will add the request to the sortlist, but next_rq remains NULL.
Read the more complete analysis at:
http://lkml.org/lkml/2007/4/25/57
This looks like it requires md to trigger, even though it should
potentially be possible to due with O_DIRECT (at least if you edit the
kernel and doctor some of the unplug calls).
The fix is to move the ->next_rq update to when we add a request to the
rbtree. Then we remove the possibility for a request to exist in the
rbtree code, but not have ->next_rq correctly updated.
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
We have a 10-15% performance regression for sequential writes on TCQ/NCQ
enabled drives in 2.6.21-rcX after the CFQ update went in. It has been
reported by Valerie Clement <valerie.clement@bull.net> and the Intel
testing folks. The regression is because of CFQ's now more aggressive
queue control, limiting the depth available to the device.
This patches fixes that regression by allowing a greater depth when only
one queue is busy. It has been tested to not impact sync-vs-async
workloads too much - we still do a lot better than 2.6.20.
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
We currently check the FIFO once per slice. Optimize that a bit and
only do it as the first thing for a new slice, so we don't end up
doing a single request and then seek to the FIFO requests.
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
It must always be the active queue, otherwise it's a bug. So just
use the active_queue, don't pass it in explicitly.
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
If a slice uses less than it is entitled to (or perhaps more), include
that in the decision on how much time to give it the next time it
gets serviced.
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Right now we use slice_start, which gives async queues an unfair
advantage. Chance that to service_last, and base the resorter
on that.
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Two issues:
- The final return 1 should be a return 0, otherwise comparing cfqq is
a noop.
- bio_sync() only checks the sync flag, while rq_is_sync() checks both
for READ and sync. The latter is what we want. Expand the bio check
to include reads, and relax the restriction to allow merging of async
io into sync requests.
In the future we want to clean up the SYNC logic, right now it means
both sync request (such as READ and O_DIRECT WRITE) and unplug-on-issue.
Leave that for later.
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
The logic in cfq_allow_merge() wasn't clear enough - basically allow
merging for the same queues only. Do a fast check for 'rq and bio both
sync/async' before doing the cfqq hash lookup.
This is verified to work with the fixed elv_try_merge() from commit
bb4067e341.
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Currently we allow any merge, even if the io originates from different
processes. This can cause really bad starvation and unfairness, if those
ios happen to be synchronous (reads or direct writes).
So add a allow_merge hook to the io scheduler ops, so an io scheduler can
help decide whether a bio/process combination may be merged with an
existing request.
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
We need to do this, otherwise the io schedulers don't get access to the
sync flag. Then they cannot tell the difference between a regular write
and an O_DIRECT write, which can cause a performance loss.
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Replace all uses of kmem_cache_t with struct kmem_cache.
The patch was generated using the following script:
#!/bin/sh
#
# Replace one string by another in all the kernel sources.
#
set -e
for file in `find * -name "*.c" -o -name "*.h"|xargs grep -l $1`; do
quilt add $file
sed -e "1,\$s/$1/$2/g" $file >/tmp/$$
mv /tmp/$$ $file
quilt refresh
done
The script was run like this
sh replace kmem_cache_t "struct kmem_cache"
Signed-off-by: Christoph Lameter <clameter@sgi.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Conflicts:
drivers/infiniband/core/iwcm.c
drivers/net/chelsio/cxgb2.c
drivers/net/wireless/bcm43xx/bcm43xx_main.c
drivers/net/wireless/prism54/islpci_eth.c
drivers/usb/core/hub.h
drivers/usb/input/hid-core.c
net/core/netpoll.c
Fix up merge failures with Linus's head and fix new compilation failures.
Signed-Off-By: David Howells <dhowells@redhat.com>
- ->init_queue() does not need the elevator passed in
- ->put_request() is a hot path and need not have the queue passed in
- cfq_update_io_seektime() does not need cfqd passed in
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Pass the work_struct pointer to the work function rather than context data.
The work function can use container_of() to work out the data.
For the cases where the container of the work_struct may go away the moment the
pending bit is cleared, it is made possible to defer the release of the
structure by deferring the clearing of the pending bit.
To make this work, an extra flag is introduced into the management side of the
work_struct. This governs auto-release of the structure upon execution.
Ordinarily, the work queue executor would release the work_struct for further
scheduling or deallocation by clearing the pending bit prior to jumping to the
work function. This means that, unless the driver makes some guarantee itself
that the work_struct won't go away, the work function may not access anything
else in the work_struct or its container lest they be deallocated.. This is a
problem if the auxiliary data is taken away (as done by the last patch).
However, if the pending bit is *not* cleared before jumping to the work
function, then the work function *may* access the work_struct and its container
with no problems. But then the work function must itself release the
work_struct by calling work_release().
In most cases, automatic release is fine, so this is the default. Special
initiators exist for the non-auto-release case (ending in _NAR).
Signed-Off-By: David Howells <dhowells@redhat.com>
In very rare circumstances would we be pruning a merged request and at
the same time delete the implicated cfqq from the rr_list, and not readd
it when the merged request got added. This could cause io stalls until
that process issued io again.
Fix it up by putting the rr_list add handling into cfq_add_rq_rb(),
identical to how pruning is handled in cfq_del_rq_rb(). This fixes a
hang reproducible with fsx-linux.
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
When the ioprio code recently got juggled a bit, a bug was introduced.
changed_ioprio() is no longer called with interrupts disabled, so using
plain spin_lock() on the queue_lock is a bug.
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
If cfq_set_request() is called for a new process AND a non-fs io
request (so that __GFP_WAIT may not be set), cfq_cic_link() may
use spin_lock_irq() and spin_unlock_irq() with interrupts already
disabled.
Fix is to always use irq safe locking in cfq_cic_link()
Acked-By: Arjan van de Ven <arjan@linux.intel.com>
Acked-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
All on stack DECLARE_COMPLETIONs should be replaced by:
DECLARE_COMPLETION_ONSTACK
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Acked-by: Ingo Molnar <mingo@elte.hu>
Acked-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Give meta data reads preference over regular reads, as the process
often needs to get that out of the way to do the io it was actually
interested in.
Signed-off-by: Jens Axboe <axboe@suse.de>
CFQ implements this on its own now, but it's really block layer
knowledge. Tells a device queue to start dispatching requests to
the driver, taking care to unplug if needed. Also fixes the issue
where as/cfq will invoke a stopped queue, which we really don't
want.
Signed-off-by: Jens Axboe <axboe@suse.de>
No point in having a place holder list just for empty queues, so remove
it. It's not used for anything other than to keep ->cfq_list busy.
Signed-off-by: Jens Axboe <axboe@suse.de>
Currently it scales with number of processes in that priority group,
which is potentially not very nice as it's called quite often.
Basically we always need to do tail inserts, except for the case of a
new process. So just mark/detect a queue as such.
Signed-off-by: Jens Axboe <axboe@suse.de>
Kill a few inlines that bring in too much code to more than one location
Shrinks kernel text by about 300 bytes on 32-bit x86.
Signed-off-by: Jens Axboe <axboe@suse.de>
It's ok if the read path is a lot more costly, as long as inc/dec is
really cheap. The inc/dec will happen for each created/freed io context,
while the reading only happens when a disk queue exits.
Signed-off-by: Jens Axboe <axboe@suse.de>
cfq_exit_lock is protecting two things now:
- The per-ioc rbtree of cfq_io_contexts
- The per-cfqd linked list of cfq_io_contexts
The per-cfqd linked list can be protected by the queue lock, as it is (by
definition) per cfqd as the queue lock is.
The per-ioc rbtree is mainly used and updated by the process itself only.
The only outside use is the io priority changing. If we move the
priority changing to not browsing the rbtree, we can remove any locking
from the rbtree updates and lookup completely. Let the sys_ioprio syscall
just mark processes as having the iopriority changed and lazily update
the private cfq io contexts the next time io is queued, and we can
remove this locking as well.
Signed-off-by: Jens Axboe <axboe@suse.de>
A collection of little fixes and cleanups:
- We don't use the 'queued' sysfs exported attribute, since the
may_queue() logic was rewritten. So kill it.
- Remove dead defines.
- cfq_set_active_queue() can be rewritten cleaner with else if conditions.
- Several places had cfq_exit_cfqq() like logic, abstract that out and
use that.
- Annotate the cfqq kmem_cache_alloc() so the allocator knows that this
is a repeat allocation if it fails with __GFP_WAIT set. Allows the
allocator to start freeing some memory, if needed. CFQ already loops for
this condition, so might as well pass the hint down.
- Remove cfqd->rq_starved logic. It's not needed anymore after we dropped
the crq allocation in cfq_set_request().
- Remove uneeded parameter passing.
Signed-off-by: Jens Axboe <axboe@suse.de>
Get rid of the cfq_rq request type. With the added elevator_private2, we
have enough room in struct request to get rid of any crq allocation/free
for each request.
Signed-off-by: Jens Axboe <axboe@suse.de>
Right now, every IO scheduler implements its own backmerging (except for
noop, which does no merging). That results in duplicated code for
essentially the same operation, which is never a good thing. This patch
moves the backmerging out of the io schedulers and into the elevator
core. We save 1.6kb of text and as a bonus get backmerging for noop as
well. Win-win!
Signed-off-by: Jens Axboe <axboe@suse.de>
Obviously, cfq_cic_link() shouldn't free a just allocated cfq_io_context?
The dead key is from __cic, so drop that.
Signed-off-by: Oleg Nesterov <oleg@tv-sign.ru>
Signed-off-by: Jens Axboe <axboe@suse.de>
The CIC_SEEKY() test really wants to use the minimum of either:
- 2 msecs (not jiffies)
- or, the pending slice time
So code it like that.
Signed-off-by: Jens Axboe <axboe@suse.de>
They all duplicate macros to check for empty root and/or node, and
clearing a node. So put those in rbtree.h.
Signed-off-by: Jens Axboe <axboe@suse.de>
- Remember to set ->last_sector so that the cfq_choose_req() logic
works correctly.
- Remove redundant call to cfq_choose_req()
Signed-off-by: Jens Axboe <axboe@suse.de>
This is a collection of patches that greatly improve CFQ performance
in some circumstances.
- Change the idling logic to only kick in after a request is done and we
are deciding what to do. Before the idling included the request service
time, so it was hard to adjust. Now it's true think/idle time.
- Take advantage of TCQ/NCQ/queueing for seeky sync workloads, but keep
it in control for sync and sequential (or close to) workloads.
- Expire queues immediately and move on to other busy queues, if we are
not going to idle after the current one finishes.
- Don't rearm idle timer if there are no busy queues. Just leave the
system idle.
Signed-off-by: Jens Axboe <axboe@suse.de>
Patch originally from Vasily Tarasov <vtaras@sw.ru>
If you set io-priority of process 1 using sys_ioprio_set system call by
another process 2 (like ionice do), then cfq_init_prio_data() function
sets priority of process 2 (current) on queue of process 1 and clears
the flag, that designates change of ioprio. So the process 1 will work
like with priority of process 2.
I propose not to call cfq_init_prio_data() on io-priority change, but
only mark queue as queue with changed prority. Every time when new
request comes cfq-scheduler checks for this flag and atomaticaly changes
priority of queue to new value.
Signed-off-by: Jens Axboe <axboe@suse.de>
A process flag to indicate whether we are doing sync io is incredibly
ugly. It also causes performance problems when one does a lot of async
io and then proceeds to sync it. Part of the io will go out as async,
and the other part as sync. This causes a disconnect between the
previously submitted io and the synced io. For io schedulers such as CFQ,
this will cause us lost merges and suboptimal behaviour in scheduling.
Remove PF_SYNCWRITE completely from the fsync/msync paths, and let
the O_DIRECT path just directly indicate that the writes are sync
by using WRITE_SYNC instead.
Signed-off-by: Jens Axboe <axboe@suse.de>
We cannot update them if the user changes nr_requests, so don't
set it in the first place. The gains are pretty questionable as
well. The batching loss has been shown to decrease throughput.
Signed-off-by: Jens Axboe <axboe@suse.de>
The color is now in the low bits of the parent pointer, and initializing
it to 0 happens as part of the whole memset above, so just remove the
unnecessary RB_CLEAR_COLOR.
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
* git://git.infradead.org/~dwmw2/rbtree-2.6:
[RBTREE] Switch rb_colour() et al to en_US spelling of 'color' for consistency
Update UML kernel/physmem.c to use rb_parent() accessor macro
[RBTREE] Update hrtimers to use rb_parent() accessor macro.
[RBTREE] Add explicit alignment to sizeof(long) for struct rb_node.
[RBTREE] Merge colour and parent fields of struct rb_node.
[RBTREE] Remove dead code in rb_erase()
[RBTREE] Update JFFS2 to use rb_parent() accessor macro.
[RBTREE] Update eventpoll.c to use rb_parent() accessor macro.
[RBTREE] Update key.c to use rb_parent() accessor macro.
[RBTREE] Update ext3 to use rb_parent() accessor macro.
[RBTREE] Change rbtree off-tree marking in I/O schedulers.
[RBTREE] Add accessor macros for colour and parent fields of rb_node
We don't clear the seek stat values in cfq_alloc_io_context(), and if
->seek_mean is unlucky enough to be set to -36 by chance, the first
invocation of cfq_update_io_seektime() will oops with a divide by zero
in do_div().
Just memset the entire cic instead of filling invididual values
independently.
Signed-off-by: Jens Axboe <axboe@suse.de>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
There's a race between shutting down one io scheduler and firing up the
next, in which a new io could enter and cause the io scheduler to be
invoked with bad or NULL data.
To fix this, we need to maintain the queue lock for a bit longer.
Unfortunately we cannot do that, since the elevator init requires to be
run without the lock held. This isn't easily fixable, without also
changing the mempool API. So split the initialization into two parts,
and alloc-init operation and an attach operation. Then we can
preallocate the io scheduler and related structures, and run the attach
inside the lock after we detach the old one.
This patch has survived 30 minutes of 1 second io scheduler switching
with a very busy io load.
Signed-off-by: Jens Axboe <axboe@suse.de>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Now that we select busy_rr for possible service, insert entries at the
back of that list instead of at the front.
Signed-off-by: Jens Axboe <axboe@suse.de>
There's a small window from when the timer is entered and we grab
the queue lock, where cfq_set_active_queue() could be rearming the
timer for us. Seen in the wild on a 12-way ppc box. Fix this by
just using mod_timer(), which will do the right thing for us.
Signed-off-by: Jens Axboe <axboe@suse.de>
If the hardware is doing real queueing, decide that it's worthless to
idle the hardware. It does reasonable simultaneous io in that case
anyways, and the idling hurts some work loads.
Signed-off-by: Jens Axboe <axboe@suse.de>
If we are anticipating a sync request from this process and we are
waiting for that and see an async request come in, expire that slice
and move on.
Signed-off-by: Jens Axboe <axboe@suse.de>
For just one busy queue (like async write out), we often overlooked
that we could queue more io and decided we were idle instead. This causes
us quite a bit of performance loss.
Signed-off-by: Jens Axboe <axboe@suse.de>
- Drop cic from the list when seen as dead.
- Fixup the locking, just use a simple spinlock.
Signed-off-by: Jens Axboe <axboe@suse.de>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
They were abusing the rb_color field to mark nodes which weren't currently
on the tree. Fix that to use the same method as eventpoll did -- setting
the parent pointer to point back to itself. And use the appropriate
accessor macros for setting and reading the parent.
Signed-off-by: David Woodhouse <dwmw2@infradead.org>
In current code, we are re-reading cic->key after dead cic->key check.
So, in theory, it may really re-read *after* cfq_exit_queue() seted NULL.
To avoid race, we copy it to stack, then use it. With this change, I
guess gcc will assign cic->key to a register or stack, and it wouldn't
be re-readed.
Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Jens Axboe <axboe@suse.de>
When queue dies, we set cic->key=NULL as dead mark. So, when we
traverse a rbtree, we must check whether it's still valid key. if it
was invalidated, drop it, then restart the traversal from top.
Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Jens Axboe <axboe@suse.de>
On rmmod path, cfq/as waits to make sure all io-contexts was
freed. However, it's using complete(), not wait_for_completion().
I think barrier() is not enough in here. To avoid the following case,
this patch replaces barrier() with smb_wmb().
cpu0 visibility cpu1
[ioc_gnone=NULL,ioc_count=1]
ioc_gnone = &all_gone NULL,ioc_count=1
atomic_read(&ioc_count) NULL,ioc_count=1
wait_for_completion() NULL,ioc_count=0 atomic_sub_and_test()
NULL,ioc_count=0 if ( && ioc_gone)
[ioc_gone==NULL,
so doesn't call complete()]
&all_gone,ioc_count=0
Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Jens Axboe <axboe@suse.de>
Detect whether a given process is seeky and if so disable (mostly) the
idle window if it is. We still allow just a little idle time, just enough
to allow that process to submit a new request. That is needed to maintain
fairness across priority groups.
In some cases, we could setup several async queues. This is not optimal
from a performance POV, since we want all async io in one queue to perform
good sorting on it. It also impacted sync queues, as async io got too much
slice time.
Signed-off-by: Jens Axboe <axboe@suse.de>
this is a small optimization to cfq_choose_req() in the CFQ I/O scheduler
(this function is a semi-often invoked candidate in an oprofile log):
by using a bit mask variable, we can use a simple switch() to check
the various cases instead of having to query two variables for each check.
Benefit: 251 vs. 285 bytes footprint of cfq_choose_req().
Also, common case 0 (no request wrapping) is now checked first in code.
Signed-off-by: Andreas Mohr <andi@lisas.de>
Signed-off-by: Jens Axboe <axboe@suse.de>
On setups with many disks, we spend a considerable amount of time
looking up the process-disk mapping on each queue of io. Testing with
a NULL based block driver, this costs 40-50% reduction in throughput
for 1000 disks.
Signed-off-by: Jens Axboe <axboe@suse.de>
Modify well over a dozen mempool users to call mempool_create_slab_pool()
rather than calling mempool_create() with extra arguments, saving about 30
lines of code and increasing readability.
Signed-off-by: Matthew Dobson <colpatch@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
We don't need to pin ->key down; ->cfqq->cfqd will do that for us.
Incidentally, that stops the leak we had - that reference was never
dropped.
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
If somebody does a hash lookup for cfq_queue while ioprio of an async queue
is elevated, they shouldn't end up stuck with lowered ioprio when we go back.
Fix is to use ->org_ioprio{,class} in hash lookups.
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
During testing of SLES10, we encountered a hang in the CFQ io scheduler.
Turns out the deferred slice expiry logic is buggy, so remove that for
now. We could be left with an idle queue that would never wake up. So
kill that logic, always expire immediately. Also fix a potential timer
race condition.
Patch looks bigger than it is, because it moves a function.
Signed-off-by: Jens Axboe <axboe@suse.de>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
the patch below marks various read-only variables in block/* as const,
so that gcc can optimize the use of them; eg gcc will replace the use by
the value directly now and will even remove the memory usage of these.
Signed-off-by: Arjan van de Ven <arjan@infradead.org>
Signed-off-by: Jens Axboe <axboe@suse.de>
Some leftover comments referring to drivers/block that are now block/.
They don't add any information we don't already have, so kill them.
Signed-off-by: Coywolf Qi Hunt <qiyong@fc-cn.com>
Signed-off-by: Jens Axboe <axboe@suse.de>
When cfq slice expires, remainder of slice is calculated and stored in
cfqq->slice_left. Current code calculates the opposite of remainder -
how many jiffies the cfqq has used past slice end. This patch fixes
the bug.
Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jens Axboe <axboe@suse.de>
cfq forced dispatching might not return all requests on the queue.
This bug can hang elevator switchinig and corrupt request ordering
during flush sequence.
Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jens Axboe <axboe@suse.de>
drivers/block/ is right now a mix of core and driver parts. Lets move
the core parts to a new top level directory. Al will move the fs/
related block parts to block/ next.
Signed-off-by: Jens Axboe <axboe@suse.de>
Jens Axboe