2008-01-29 14:53:40 +01:00
|
|
|
/*
|
|
|
|
* Functions related to io context handling
|
|
|
|
*/
|
|
|
|
#include <linux/kernel.h>
|
|
|
|
#include <linux/module.h>
|
|
|
|
#include <linux/init.h>
|
|
|
|
#include <linux/bio.h>
|
|
|
|
#include <linux/blkdev.h>
|
|
|
|
#include <linux/bootmem.h> /* for max_pfn/max_low_pfn */
|
include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h
percpu.h is included by sched.h and module.h and thus ends up being
included when building most .c files. percpu.h includes slab.h which
in turn includes gfp.h making everything defined by the two files
universally available and complicating inclusion dependencies.
percpu.h -> slab.h dependency is about to be removed. Prepare for
this change by updating users of gfp and slab facilities include those
headers directly instead of assuming availability. As this conversion
needs to touch large number of source files, the following script is
used as the basis of conversion.
http://userweb.kernel.org/~tj/misc/slabh-sweep.py
The script does the followings.
* Scan files for gfp and slab usages and update includes such that
only the necessary includes are there. ie. if only gfp is used,
gfp.h, if slab is used, slab.h.
* When the script inserts a new include, it looks at the include
blocks and try to put the new include such that its order conforms
to its surrounding. It's put in the include block which contains
core kernel includes, in the same order that the rest are ordered -
alphabetical, Christmas tree, rev-Xmas-tree or at the end if there
doesn't seem to be any matching order.
* If the script can't find a place to put a new include (mostly
because the file doesn't have fitting include block), it prints out
an error message indicating which .h file needs to be added to the
file.
The conversion was done in the following steps.
1. The initial automatic conversion of all .c files updated slightly
over 4000 files, deleting around 700 includes and adding ~480 gfp.h
and ~3000 slab.h inclusions. The script emitted errors for ~400
files.
2. Each error was manually checked. Some didn't need the inclusion,
some needed manual addition while adding it to implementation .h or
embedding .c file was more appropriate for others. This step added
inclusions to around 150 files.
3. The script was run again and the output was compared to the edits
from #2 to make sure no file was left behind.
4. Several build tests were done and a couple of problems were fixed.
e.g. lib/decompress_*.c used malloc/free() wrappers around slab
APIs requiring slab.h to be added manually.
5. The script was run on all .h files but without automatically
editing them as sprinkling gfp.h and slab.h inclusions around .h
files could easily lead to inclusion dependency hell. Most gfp.h
inclusion directives were ignored as stuff from gfp.h was usually
wildly available and often used in preprocessor macros. Each
slab.h inclusion directive was examined and added manually as
necessary.
6. percpu.h was updated not to include slab.h.
7. Build test were done on the following configurations and failures
were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my
distributed build env didn't work with gcov compiles) and a few
more options had to be turned off depending on archs to make things
build (like ipr on powerpc/64 which failed due to missing writeq).
* x86 and x86_64 UP and SMP allmodconfig and a custom test config.
* powerpc and powerpc64 SMP allmodconfig
* sparc and sparc64 SMP allmodconfig
* ia64 SMP allmodconfig
* s390 SMP allmodconfig
* alpha SMP allmodconfig
* um on x86_64 SMP allmodconfig
8. percpu.h modifications were reverted so that it could be applied as
a separate patch and serve as bisection point.
Given the fact that I had only a couple of failures from tests on step
6, I'm fairly confident about the coverage of this conversion patch.
If there is a breakage, it's likely to be something in one of the arch
headers which should be easily discoverable easily on most builds of
the specific arch.
Signed-off-by: Tejun Heo <tj@kernel.org>
Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
2010-03-24 09:04:11 +01:00
|
|
|
#include <linux/slab.h>
|
2008-01-29 14:53:40 +01:00
|
|
|
|
|
|
|
#include "blk.h"
|
|
|
|
|
|
|
|
/*
|
|
|
|
* For io context allocations
|
|
|
|
*/
|
|
|
|
static struct kmem_cache *iocontext_cachep;
|
|
|
|
|
block: make ioc get/put interface more conventional and fix race on alloction
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>
2011-12-14 00:33:38 +01:00
|
|
|
/**
|
|
|
|
* get_io_context - increment reference count to io_context
|
|
|
|
* @ioc: io_context to get
|
|
|
|
*
|
|
|
|
* Increment reference count to @ioc.
|
|
|
|
*/
|
|
|
|
void get_io_context(struct io_context *ioc)
|
|
|
|
{
|
|
|
|
BUG_ON(atomic_long_read(&ioc->refcount) <= 0);
|
|
|
|
atomic_long_inc(&ioc->refcount);
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(get_io_context);
|
|
|
|
|
2011-12-14 00:33:42 +01:00
|
|
|
static void icq_free_icq_rcu(struct rcu_head *head)
|
|
|
|
{
|
|
|
|
struct io_cq *icq = container_of(head, struct io_cq, __rcu_head);
|
|
|
|
|
|
|
|
kmem_cache_free(icq->__rcu_icq_cache, icq);
|
|
|
|
}
|
|
|
|
|
block: exit_io_context() should call elevator_exit_icq_fn()
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>
2012-02-15 09:45:53 +01:00
|
|
|
/* Exit an icq. Called with both ioc and q locked. */
|
2011-12-14 00:33:42 +01:00
|
|
|
static void ioc_exit_icq(struct io_cq *icq)
|
block: exit_io_context() should call elevator_exit_icq_fn()
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>
2012-02-15 09:45:53 +01:00
|
|
|
{
|
|
|
|
struct elevator_type *et = icq->q->elevator->type;
|
|
|
|
|
|
|
|
if (icq->flags & ICQ_EXITED)
|
|
|
|
return;
|
|
|
|
|
|
|
|
if (et->ops.elevator_exit_icq_fn)
|
|
|
|
et->ops.elevator_exit_icq_fn(icq);
|
|
|
|
|
|
|
|
icq->flags |= ICQ_EXITED;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Release an icq. Called with both ioc and q locked. */
|
|
|
|
static void ioc_destroy_icq(struct io_cq *icq)
|
2011-12-14 00:33:42 +01:00
|
|
|
{
|
|
|
|
struct io_context *ioc = icq->ioc;
|
|
|
|
struct request_queue *q = icq->q;
|
|
|
|
struct elevator_type *et = q->elevator->type;
|
|
|
|
|
|
|
|
lockdep_assert_held(&ioc->lock);
|
|
|
|
lockdep_assert_held(q->queue_lock);
|
|
|
|
|
|
|
|
radix_tree_delete(&ioc->icq_tree, icq->q->id);
|
|
|
|
hlist_del_init(&icq->ioc_node);
|
|
|
|
list_del_init(&icq->q_node);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Both setting lookup hint to and clearing it from @icq are done
|
|
|
|
* under queue_lock. If it's not pointing to @icq now, it never
|
|
|
|
* will. Hint assignment itself can race safely.
|
|
|
|
*/
|
|
|
|
if (rcu_dereference_raw(ioc->icq_hint) == icq)
|
|
|
|
rcu_assign_pointer(ioc->icq_hint, NULL);
|
|
|
|
|
block: exit_io_context() should call elevator_exit_icq_fn()
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>
2012-02-15 09:45:53 +01:00
|
|
|
ioc_exit_icq(icq);
|
2011-12-14 00:33:42 +01:00
|
|
|
|
|
|
|
/*
|
|
|
|
* @icq->q might have gone away by the time RCU callback runs
|
|
|
|
* making it impossible to determine icq_cache. Record it in @icq.
|
|
|
|
*/
|
|
|
|
icq->__rcu_icq_cache = et->icq_cache;
|
|
|
|
call_rcu(&icq->__rcu_head, icq_free_icq_rcu);
|
|
|
|
}
|
|
|
|
|
2011-12-14 00:33:39 +01:00
|
|
|
/*
|
|
|
|
* Slow path for ioc release in put_io_context(). Performs double-lock
|
2011-12-14 00:33:41 +01:00
|
|
|
* dancing to unlink all icq's and then frees ioc.
|
2011-12-14 00:33:39 +01:00
|
|
|
*/
|
|
|
|
static void ioc_release_fn(struct work_struct *work)
|
2008-01-29 14:53:40 +01:00
|
|
|
{
|
2011-12-14 00:33:39 +01:00
|
|
|
struct io_context *ioc = container_of(work, struct io_context,
|
|
|
|
release_work);
|
2012-02-11 12:37:25 +01:00
|
|
|
unsigned long flags;
|
2011-12-14 00:33:39 +01:00
|
|
|
|
2012-02-11 12:37:25 +01:00
|
|
|
/*
|
|
|
|
* Exiting icq may call into put_io_context() through elevator
|
|
|
|
* which will trigger lockdep warning. The ioc's are guaranteed to
|
|
|
|
* be different, use a different locking subclass here. Use
|
|
|
|
* irqsave variant as there's no spin_lock_irq_nested().
|
|
|
|
*/
|
|
|
|
spin_lock_irqsave_nested(&ioc->lock, flags, 1);
|
2011-12-14 00:33:39 +01:00
|
|
|
|
2011-12-14 00:33:41 +01:00
|
|
|
while (!hlist_empty(&ioc->icq_list)) {
|
|
|
|
struct io_cq *icq = hlist_entry(ioc->icq_list.first,
|
|
|
|
struct io_cq, ioc_node);
|
2012-02-15 09:45:52 +01:00
|
|
|
struct request_queue *q = icq->q;
|
|
|
|
|
|
|
|
if (spin_trylock(q->queue_lock)) {
|
block: exit_io_context() should call elevator_exit_icq_fn()
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>
2012-02-15 09:45:53 +01:00
|
|
|
ioc_destroy_icq(icq);
|
2012-02-15 09:45:52 +01:00
|
|
|
spin_unlock(q->queue_lock);
|
|
|
|
} else {
|
|
|
|
spin_unlock_irqrestore(&ioc->lock, flags);
|
|
|
|
cpu_relax();
|
|
|
|
spin_lock_irqsave_nested(&ioc->lock, flags, 1);
|
2011-12-14 00:33:39 +01:00
|
|
|
}
|
|
|
|
}
|
2008-02-19 10:02:29 +01:00
|
|
|
|
2012-02-15 09:45:52 +01:00
|
|
|
spin_unlock_irqrestore(&ioc->lock, flags);
|
2011-12-14 00:33:39 +01:00
|
|
|
|
|
|
|
kmem_cache_free(iocontext_cachep, ioc);
|
2008-01-29 14:53:40 +01:00
|
|
|
}
|
|
|
|
|
2011-12-14 00:33:37 +01:00
|
|
|
/**
|
|
|
|
* put_io_context - put a reference of io_context
|
|
|
|
* @ioc: io_context to put
|
|
|
|
*
|
|
|
|
* Decrement reference count of @ioc and release it if the count reaches
|
2012-02-07 07:51:30 +01:00
|
|
|
* zero.
|
2008-01-29 14:53:40 +01:00
|
|
|
*/
|
2012-02-07 07:51:30 +01:00
|
|
|
void put_io_context(struct io_context *ioc)
|
2008-01-29 14:53:40 +01:00
|
|
|
{
|
2011-12-14 00:33:39 +01:00
|
|
|
unsigned long flags;
|
2012-03-14 15:34:48 +01:00
|
|
|
bool free_ioc = false;
|
2011-12-14 00:33:39 +01:00
|
|
|
|
2008-01-29 14:53:40 +01:00
|
|
|
if (ioc == NULL)
|
2011-12-14 00:33:37 +01:00
|
|
|
return;
|
2008-01-29 14:53:40 +01:00
|
|
|
|
2011-12-14 00:33:37 +01:00
|
|
|
BUG_ON(atomic_long_read(&ioc->refcount) <= 0);
|
2008-01-29 14:53:40 +01:00
|
|
|
|
2011-12-14 00:33:39 +01:00
|
|
|
/*
|
2012-02-07 07:51:30 +01:00
|
|
|
* Releasing ioc requires reverse order double locking and we may
|
|
|
|
* already be holding a queue_lock. Do it asynchronously from wq.
|
2011-12-14 00:33:39 +01:00
|
|
|
*/
|
2012-02-07 07:51:30 +01:00
|
|
|
if (atomic_long_dec_and_test(&ioc->refcount)) {
|
|
|
|
spin_lock_irqsave(&ioc->lock, flags);
|
|
|
|
if (!hlist_empty(&ioc->icq_list))
|
|
|
|
schedule_work(&ioc->release_work);
|
2012-03-14 15:34:48 +01:00
|
|
|
else
|
|
|
|
free_ioc = true;
|
2012-02-07 07:51:30 +01:00
|
|
|
spin_unlock_irqrestore(&ioc->lock, flags);
|
2011-12-14 00:33:39 +01:00
|
|
|
}
|
2012-03-14 15:34:48 +01:00
|
|
|
|
|
|
|
if (free_ioc)
|
|
|
|
kmem_cache_free(iocontext_cachep, ioc);
|
2008-01-29 14:53:40 +01:00
|
|
|
}
|
2011-12-14 00:33:39 +01:00
|
|
|
EXPORT_SYMBOL(put_io_context);
|
2008-01-29 14:53:40 +01:00
|
|
|
|
2012-03-05 22:15:26 +01:00
|
|
|
/**
|
|
|
|
* put_io_context_active - put active reference on ioc
|
|
|
|
* @ioc: ioc of interest
|
|
|
|
*
|
|
|
|
* Undo get_io_context_active(). If active reference reaches zero after
|
|
|
|
* put, @ioc can never issue further IOs and ioscheds are notified.
|
|
|
|
*/
|
|
|
|
void put_io_context_active(struct io_context *ioc)
|
2008-01-29 14:53:40 +01:00
|
|
|
{
|
block: exit_io_context() should call elevator_exit_icq_fn()
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>
2012-02-15 09:45:53 +01:00
|
|
|
unsigned long flags;
|
2012-03-05 22:15:26 +01:00
|
|
|
struct io_cq *icq;
|
2008-01-29 14:53:40 +01:00
|
|
|
|
2012-03-05 22:15:26 +01:00
|
|
|
if (!atomic_dec_and_test(&ioc->active_ref)) {
|
block: exit_io_context() should call elevator_exit_icq_fn()
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>
2012-02-15 09:45:53 +01:00
|
|
|
put_io_context(ioc);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Need ioc lock to walk icq_list and q lock to exit icq. Perform
|
|
|
|
* reverse double locking. Read comment in ioc_release_fn() for
|
|
|
|
* explanation on the nested locking annotation.
|
|
|
|
*/
|
|
|
|
retry:
|
|
|
|
spin_lock_irqsave_nested(&ioc->lock, flags, 1);
|
hlist: drop the node parameter from iterators
I'm not sure why, but the hlist for each entry iterators were conceived
list_for_each_entry(pos, head, member)
The hlist ones were greedy and wanted an extra parameter:
hlist_for_each_entry(tpos, pos, head, member)
Why did they need an extra pos parameter? I'm not quite sure. Not only
they don't really need it, it also prevents the iterator from looking
exactly like the list iterator, which is unfortunate.
Besides the semantic patch, there was some manual work required:
- Fix up the actual hlist iterators in linux/list.h
- Fix up the declaration of other iterators based on the hlist ones.
- A very small amount of places were using the 'node' parameter, this
was modified to use 'obj->member' instead.
- Coccinelle didn't handle the hlist_for_each_entry_safe iterator
properly, so those had to be fixed up manually.
The semantic patch which is mostly the work of Peter Senna Tschudin is here:
@@
iterator name hlist_for_each_entry, hlist_for_each_entry_continue, hlist_for_each_entry_from, hlist_for_each_entry_rcu, hlist_for_each_entry_rcu_bh, hlist_for_each_entry_continue_rcu_bh, for_each_busy_worker, ax25_uid_for_each, ax25_for_each, inet_bind_bucket_for_each, sctp_for_each_hentry, sk_for_each, sk_for_each_rcu, sk_for_each_from, sk_for_each_safe, sk_for_each_bound, hlist_for_each_entry_safe, hlist_for_each_entry_continue_rcu, nr_neigh_for_each, nr_neigh_for_each_safe, nr_node_for_each, nr_node_for_each_safe, for_each_gfn_indirect_valid_sp, for_each_gfn_sp, for_each_host;
type T;
expression a,c,d,e;
identifier b;
statement S;
@@
-T b;
<+... when != b
(
hlist_for_each_entry(a,
- b,
c, d) S
|
hlist_for_each_entry_continue(a,
- b,
c) S
|
hlist_for_each_entry_from(a,
- b,
c) S
|
hlist_for_each_entry_rcu(a,
- b,
c, d) S
|
hlist_for_each_entry_rcu_bh(a,
- b,
c, d) S
|
hlist_for_each_entry_continue_rcu_bh(a,
- b,
c) S
|
for_each_busy_worker(a, c,
- b,
d) S
|
ax25_uid_for_each(a,
- b,
c) S
|
ax25_for_each(a,
- b,
c) S
|
inet_bind_bucket_for_each(a,
- b,
c) S
|
sctp_for_each_hentry(a,
- b,
c) S
|
sk_for_each(a,
- b,
c) S
|
sk_for_each_rcu(a,
- b,
c) S
|
sk_for_each_from
-(a, b)
+(a)
S
+ sk_for_each_from(a) S
|
sk_for_each_safe(a,
- b,
c, d) S
|
sk_for_each_bound(a,
- b,
c) S
|
hlist_for_each_entry_safe(a,
- b,
c, d, e) S
|
hlist_for_each_entry_continue_rcu(a,
- b,
c) S
|
nr_neigh_for_each(a,
- b,
c) S
|
nr_neigh_for_each_safe(a,
- b,
c, d) S
|
nr_node_for_each(a,
- b,
c) S
|
nr_node_for_each_safe(a,
- b,
c, d) S
|
- for_each_gfn_sp(a, c, d, b) S
+ for_each_gfn_sp(a, c, d) S
|
- for_each_gfn_indirect_valid_sp(a, c, d, b) S
+ for_each_gfn_indirect_valid_sp(a, c, d) S
|
for_each_host(a,
- b,
c) S
|
for_each_host_safe(a,
- b,
c, d) S
|
for_each_mesh_entry(a,
- b,
c, d) S
)
...+>
[akpm@linux-foundation.org: drop bogus change from net/ipv4/raw.c]
[akpm@linux-foundation.org: drop bogus hunk from net/ipv6/raw.c]
[akpm@linux-foundation.org: checkpatch fixes]
[akpm@linux-foundation.org: fix warnings]
[akpm@linux-foudnation.org: redo intrusive kvm changes]
Tested-by: Peter Senna Tschudin <peter.senna@gmail.com>
Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Signed-off-by: Sasha Levin <sasha.levin@oracle.com>
Cc: Wu Fengguang <fengguang.wu@intel.com>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Gleb Natapov <gleb@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 02:06:00 +01:00
|
|
|
hlist_for_each_entry(icq, &ioc->icq_list, ioc_node) {
|
block: exit_io_context() should call elevator_exit_icq_fn()
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>
2012-02-15 09:45:53 +01:00
|
|
|
if (icq->flags & ICQ_EXITED)
|
|
|
|
continue;
|
|
|
|
if (spin_trylock(icq->q->queue_lock)) {
|
|
|
|
ioc_exit_icq(icq);
|
|
|
|
spin_unlock(icq->q->queue_lock);
|
|
|
|
} else {
|
|
|
|
spin_unlock_irqrestore(&ioc->lock, flags);
|
|
|
|
cpu_relax();
|
|
|
|
goto retry;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
spin_unlock_irqrestore(&ioc->lock, flags);
|
|
|
|
|
2012-02-07 07:51:30 +01:00
|
|
|
put_io_context(ioc);
|
2008-01-29 14:53:40 +01:00
|
|
|
}
|
|
|
|
|
2012-03-05 22:15:26 +01:00
|
|
|
/* Called by the exiting task */
|
|
|
|
void exit_io_context(struct task_struct *task)
|
|
|
|
{
|
|
|
|
struct io_context *ioc;
|
|
|
|
|
|
|
|
task_lock(task);
|
|
|
|
ioc = task->io_context;
|
|
|
|
task->io_context = NULL;
|
|
|
|
task_unlock(task);
|
|
|
|
|
|
|
|
atomic_dec(&ioc->nr_tasks);
|
|
|
|
put_io_context_active(ioc);
|
|
|
|
}
|
|
|
|
|
2011-12-14 00:33:42 +01:00
|
|
|
/**
|
|
|
|
* ioc_clear_queue - break any ioc association with the specified queue
|
|
|
|
* @q: request_queue being cleared
|
|
|
|
*
|
|
|
|
* Walk @q->icq_list and exit all io_cq's. Must be called with @q locked.
|
|
|
|
*/
|
|
|
|
void ioc_clear_queue(struct request_queue *q)
|
|
|
|
{
|
|
|
|
lockdep_assert_held(q->queue_lock);
|
|
|
|
|
|
|
|
while (!list_empty(&q->icq_list)) {
|
|
|
|
struct io_cq *icq = list_entry(q->icq_list.next,
|
|
|
|
struct io_cq, q_node);
|
|
|
|
struct io_context *ioc = icq->ioc;
|
|
|
|
|
|
|
|
spin_lock(&ioc->lock);
|
block: exit_io_context() should call elevator_exit_icq_fn()
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>
2012-02-15 09:45:53 +01:00
|
|
|
ioc_destroy_icq(icq);
|
2011-12-14 00:33:42 +01:00
|
|
|
spin_unlock(&ioc->lock);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2012-03-05 22:15:24 +01:00
|
|
|
int create_task_io_context(struct task_struct *task, gfp_t gfp_flags, int node)
|
2008-01-29 14:53:40 +01:00
|
|
|
{
|
2011-06-06 05:11:34 +02:00
|
|
|
struct io_context *ioc;
|
2012-05-31 13:39:05 +02:00
|
|
|
int ret;
|
2008-01-29 14:53:40 +01:00
|
|
|
|
2011-12-14 00:33:37 +01:00
|
|
|
ioc = kmem_cache_alloc_node(iocontext_cachep, gfp_flags | __GFP_ZERO,
|
|
|
|
node);
|
|
|
|
if (unlikely(!ioc))
|
2012-03-05 22:15:24 +01:00
|
|
|
return -ENOMEM;
|
2011-12-14 00:33:37 +01:00
|
|
|
|
|
|
|
/* initialize */
|
|
|
|
atomic_long_set(&ioc->refcount, 1);
|
block: uninitialized ioc->nr_tasks triggers WARN_ON
Hi,
I'm using the old-fashioned 'dump' backup tool, and I noticed that it spews the
below warning as of 3.5-rc1 and later (3.4 is fine):
[ 10.886893] ------------[ cut here ]------------
[ 10.886904] WARNING: at include/linux/iocontext.h:140 copy_process+0x1488/0x1560()
[ 10.886905] Hardware name: Bochs
[ 10.886906] Modules linked in:
[ 10.886908] Pid: 2430, comm: dump Not tainted 3.5.0-rc7+ #27
[ 10.886908] Call Trace:
[ 10.886911] [<ffffffff8107ce8a>] warn_slowpath_common+0x7a/0xb0
[ 10.886912] [<ffffffff8107ced5>] warn_slowpath_null+0x15/0x20
[ 10.886913] [<ffffffff8107c088>] copy_process+0x1488/0x1560
[ 10.886914] [<ffffffff8107c244>] do_fork+0xb4/0x340
[ 10.886918] [<ffffffff8108effa>] ? recalc_sigpending+0x1a/0x50
[ 10.886919] [<ffffffff8108f6b2>] ? __set_task_blocked+0x32/0x80
[ 10.886920] [<ffffffff81091afa>] ? __set_current_blocked+0x3a/0x60
[ 10.886923] [<ffffffff81051db3>] sys_clone+0x23/0x30
[ 10.886925] [<ffffffff8179bd73>] stub_clone+0x13/0x20
[ 10.886927] [<ffffffff8179baa2>] ? system_call_fastpath+0x16/0x1b
[ 10.886928] ---[ end trace 32a14af7ee6a590b ]---
Reproducing is easy, I can hit it on a KVM system with a very basic
config (x86_64 make defconfig + enable the drivers needed). To hit it,
just install dump (on debian/ubuntu, not sure what the package might be
called on Fedora), and:
dump -o -f /tmp/foo /
You'll see the warning in dmesg once it forks off the I/O process and
starts dumping filesystem contents.
I bisected it down to the following commit:
commit f6e8d01bee036460e03bd4f6a79d014f98ba712e
Author: Tejun Heo <tj@kernel.org>
Date: Mon Mar 5 13:15:26 2012 -0800
block: add io_context->active_ref
Currently ioc->nr_tasks is used to decide two things - whether an ioc
is done issuing IOs and whether it's shared by multiple tasks. This
patch separate out the first into ioc->active_ref, which is acquired
and released using {get|put}_io_context_active() respectively.
This will be used to associate bio's with a given task. This patch
doesn't introduce any visible behavior change.
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Vivek Goyal <vgoyal@redhat.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
It seems like the init of ioc->nr_tasks was removed in that patch,
so it starts out at 0 instead of 1.
Tejun, is the right thing here to add back the init, or should something else
be done?
The below patch removes the warning, but I haven't done any more extensive
testing on it.
Signed-off-by: Olof Johansson <olof@lixom.net>
Acked-by: Tejun Heo <tj@kernel.org>
Cc: stable@kernel.org
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2012-08-01 12:17:27 +02:00
|
|
|
atomic_set(&ioc->nr_tasks, 1);
|
2012-03-05 22:15:26 +01:00
|
|
|
atomic_set(&ioc->active_ref, 1);
|
2011-12-14 00:33:37 +01:00
|
|
|
spin_lock_init(&ioc->lock);
|
2011-12-14 00:33:41 +01:00
|
|
|
INIT_RADIX_TREE(&ioc->icq_tree, GFP_ATOMIC | __GFP_HIGH);
|
|
|
|
INIT_HLIST_HEAD(&ioc->icq_list);
|
2011-12-14 00:33:39 +01:00
|
|
|
INIT_WORK(&ioc->release_work, ioc_release_fn);
|
2008-01-29 14:53:40 +01:00
|
|
|
|
2011-12-25 14:29:14 +01:00
|
|
|
/*
|
|
|
|
* Try to install. ioc shouldn't be installed if someone else
|
|
|
|
* already did or @task, which isn't %current, is exiting. Note
|
|
|
|
* that we need to allow ioc creation on exiting %current as exit
|
|
|
|
* path may issue IOs from e.g. exit_files(). The exit path is
|
|
|
|
* responsible for not issuing IO after exit_io_context().
|
|
|
|
*/
|
block: make ioc get/put interface more conventional and fix race on alloction
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>
2011-12-14 00:33:38 +01:00
|
|
|
task_lock(task);
|
2011-12-25 14:29:14 +01:00
|
|
|
if (!task->io_context &&
|
|
|
|
(task == current || !(task->flags & PF_EXITING)))
|
block: make ioc get/put interface more conventional and fix race on alloction
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>
2011-12-14 00:33:38 +01:00
|
|
|
task->io_context = ioc;
|
2011-12-14 00:33:40 +01:00
|
|
|
else
|
block: make ioc get/put interface more conventional and fix race on alloction
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>
2011-12-14 00:33:38 +01:00
|
|
|
kmem_cache_free(iocontext_cachep, ioc);
|
2012-05-31 13:39:05 +02:00
|
|
|
|
|
|
|
ret = task->io_context ? 0 : -EBUSY;
|
|
|
|
|
block: make ioc get/put interface more conventional and fix race on alloction
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>
2011-12-14 00:33:38 +01:00
|
|
|
task_unlock(task);
|
2012-03-05 22:15:24 +01:00
|
|
|
|
2012-05-31 13:39:05 +02:00
|
|
|
return ret;
|
2008-01-29 14:53:40 +01:00
|
|
|
}
|
|
|
|
|
block: make ioc get/put interface more conventional and fix race on alloction
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>
2011-12-14 00:33:38 +01:00
|
|
|
/**
|
|
|
|
* get_task_io_context - get io_context of a task
|
|
|
|
* @task: task of interest
|
|
|
|
* @gfp_flags: allocation flags, used if allocation is necessary
|
|
|
|
* @node: allocation node, used if allocation is necessary
|
|
|
|
*
|
|
|
|
* Return io_context of @task. If it doesn't exist, it is created with
|
|
|
|
* @gfp_flags and @node. The returned io_context has its reference count
|
|
|
|
* incremented.
|
2008-01-29 14:53:40 +01:00
|
|
|
*
|
block: make ioc get/put interface more conventional and fix race on alloction
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>
2011-12-14 00:33:38 +01:00
|
|
|
* This function always goes through task_lock() and it's better to use
|
2011-12-14 00:33:40 +01:00
|
|
|
* %current->io_context + get_io_context() for %current.
|
2008-01-29 14:53:40 +01:00
|
|
|
*/
|
block: make ioc get/put interface more conventional and fix race on alloction
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>
2011-12-14 00:33:38 +01:00
|
|
|
struct io_context *get_task_io_context(struct task_struct *task,
|
|
|
|
gfp_t gfp_flags, int node)
|
2008-01-29 14:53:40 +01:00
|
|
|
{
|
block: make ioc get/put interface more conventional and fix race on alloction
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>
2011-12-14 00:33:38 +01:00
|
|
|
struct io_context *ioc;
|
2008-01-29 14:53:40 +01:00
|
|
|
|
block: make ioc get/put interface more conventional and fix race on alloction
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>
2011-12-14 00:33:38 +01:00
|
|
|
might_sleep_if(gfp_flags & __GFP_WAIT);
|
|
|
|
|
2011-12-14 00:33:40 +01:00
|
|
|
do {
|
|
|
|
task_lock(task);
|
|
|
|
ioc = task->io_context;
|
|
|
|
if (likely(ioc)) {
|
|
|
|
get_io_context(ioc);
|
|
|
|
task_unlock(task);
|
|
|
|
return ioc;
|
|
|
|
}
|
block: make ioc get/put interface more conventional and fix race on alloction
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>
2011-12-14 00:33:38 +01:00
|
|
|
task_unlock(task);
|
2012-03-05 22:15:24 +01:00
|
|
|
} while (!create_task_io_context(task, gfp_flags, node));
|
block: make ioc get/put interface more conventional and fix race on alloction
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>
2011-12-14 00:33:38 +01:00
|
|
|
|
2011-12-14 00:33:40 +01:00
|
|
|
return NULL;
|
2008-01-29 14:53:40 +01:00
|
|
|
}
|
block: make ioc get/put interface more conventional and fix race on alloction
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>
2011-12-14 00:33:38 +01:00
|
|
|
EXPORT_SYMBOL(get_task_io_context);
|
2008-01-29 14:53:40 +01:00
|
|
|
|
2011-12-14 00:33:42 +01:00
|
|
|
/**
|
|
|
|
* ioc_lookup_icq - lookup io_cq from ioc
|
|
|
|
* @ioc: the associated io_context
|
|
|
|
* @q: the associated request_queue
|
|
|
|
*
|
|
|
|
* Look up io_cq associated with @ioc - @q pair from @ioc. Must be called
|
|
|
|
* with @q->queue_lock held.
|
|
|
|
*/
|
|
|
|
struct io_cq *ioc_lookup_icq(struct io_context *ioc, struct request_queue *q)
|
|
|
|
{
|
|
|
|
struct io_cq *icq;
|
|
|
|
|
|
|
|
lockdep_assert_held(q->queue_lock);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* icq's are indexed from @ioc using radix tree and hint pointer,
|
|
|
|
* both of which are protected with RCU. All removals are done
|
|
|
|
* holding both q and ioc locks, and we're holding q lock - if we
|
|
|
|
* find a icq which points to us, it's guaranteed to be valid.
|
|
|
|
*/
|
|
|
|
rcu_read_lock();
|
|
|
|
icq = rcu_dereference(ioc->icq_hint);
|
|
|
|
if (icq && icq->q == q)
|
|
|
|
goto out;
|
|
|
|
|
|
|
|
icq = radix_tree_lookup(&ioc->icq_tree, q->id);
|
|
|
|
if (icq && icq->q == q)
|
|
|
|
rcu_assign_pointer(ioc->icq_hint, icq); /* allowed to race */
|
|
|
|
else
|
|
|
|
icq = NULL;
|
|
|
|
out:
|
|
|
|
rcu_read_unlock();
|
|
|
|
return icq;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(ioc_lookup_icq);
|
|
|
|
|
2011-12-14 00:33:42 +01:00
|
|
|
/**
|
|
|
|
* ioc_create_icq - create and link io_cq
|
2012-03-05 22:15:24 +01:00
|
|
|
* @ioc: io_context of interest
|
2011-12-14 00:33:42 +01:00
|
|
|
* @q: request_queue of interest
|
|
|
|
* @gfp_mask: allocation mask
|
|
|
|
*
|
2012-03-05 22:15:24 +01:00
|
|
|
* Make sure io_cq linking @ioc and @q exists. If icq doesn't exist, they
|
|
|
|
* will be created using @gfp_mask.
|
2011-12-14 00:33:42 +01:00
|
|
|
*
|
|
|
|
* The caller is responsible for ensuring @ioc won't go away and @q is
|
|
|
|
* alive and will stay alive until this function returns.
|
|
|
|
*/
|
2012-03-05 22:15:24 +01:00
|
|
|
struct io_cq *ioc_create_icq(struct io_context *ioc, struct request_queue *q,
|
|
|
|
gfp_t gfp_mask)
|
2011-12-14 00:33:42 +01:00
|
|
|
{
|
|
|
|
struct elevator_type *et = q->elevator->type;
|
|
|
|
struct io_cq *icq;
|
|
|
|
|
|
|
|
/* allocate stuff */
|
|
|
|
icq = kmem_cache_alloc_node(et->icq_cache, gfp_mask | __GFP_ZERO,
|
|
|
|
q->node);
|
|
|
|
if (!icq)
|
|
|
|
return NULL;
|
|
|
|
|
|
|
|
if (radix_tree_preload(gfp_mask) < 0) {
|
|
|
|
kmem_cache_free(et->icq_cache, icq);
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
|
|
|
icq->ioc = ioc;
|
|
|
|
icq->q = q;
|
|
|
|
INIT_LIST_HEAD(&icq->q_node);
|
|
|
|
INIT_HLIST_NODE(&icq->ioc_node);
|
|
|
|
|
|
|
|
/* lock both q and ioc and try to link @icq */
|
|
|
|
spin_lock_irq(q->queue_lock);
|
|
|
|
spin_lock(&ioc->lock);
|
|
|
|
|
|
|
|
if (likely(!radix_tree_insert(&ioc->icq_tree, q->id, icq))) {
|
|
|
|
hlist_add_head(&icq->ioc_node, &ioc->icq_list);
|
|
|
|
list_add(&icq->q_node, &q->icq_list);
|
|
|
|
if (et->ops.elevator_init_icq_fn)
|
|
|
|
et->ops.elevator_init_icq_fn(icq);
|
|
|
|
} else {
|
|
|
|
kmem_cache_free(et->icq_cache, icq);
|
|
|
|
icq = ioc_lookup_icq(ioc, q);
|
|
|
|
if (!icq)
|
|
|
|
printk(KERN_ERR "cfq: icq link failed!\n");
|
|
|
|
}
|
|
|
|
|
|
|
|
spin_unlock(&ioc->lock);
|
|
|
|
spin_unlock_irq(q->queue_lock);
|
|
|
|
radix_tree_preload_end();
|
|
|
|
return icq;
|
|
|
|
}
|
|
|
|
|
2008-02-18 13:45:53 +01:00
|
|
|
static int __init blk_ioc_init(void)
|
2008-01-29 14:53:40 +01:00
|
|
|
{
|
|
|
|
iocontext_cachep = kmem_cache_create("blkdev_ioc",
|
|
|
|
sizeof(struct io_context), 0, SLAB_PANIC, NULL);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
subsys_initcall(blk_ioc_init);
|