2010-01-06 01:47:10 -07:00
|
|
|
/*
|
|
|
|
* padata.c - generic interface to process data streams in parallel
|
|
|
|
*
|
|
|
|
* Copyright (C) 2008, 2009 secunet Security Networks AG
|
|
|
|
* Copyright (C) 2008, 2009 Steffen Klassert <steffen.klassert@secunet.com>
|
|
|
|
*
|
|
|
|
* This program is free software; you can redistribute it and/or modify it
|
|
|
|
* under the terms and conditions of the GNU General Public License,
|
|
|
|
* version 2, as published by the Free Software Foundation.
|
|
|
|
*
|
|
|
|
* This program is distributed in the hope it will be useful, but WITHOUT
|
|
|
|
* ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
|
|
|
|
* FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
|
|
|
|
* more details.
|
|
|
|
*
|
|
|
|
* You should have received a copy of the GNU General Public License along with
|
|
|
|
* this program; if not, write to the Free Software Foundation, Inc.,
|
|
|
|
* 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
|
|
|
|
*/
|
|
|
|
|
|
|
|
#include <linux/module.h>
|
|
|
|
#include <linux/cpumask.h>
|
|
|
|
#include <linux/err.h>
|
|
|
|
#include <linux/cpu.h>
|
|
|
|
#include <linux/padata.h>
|
|
|
|
#include <linux/mutex.h>
|
|
|
|
#include <linux/sched.h>
|
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 02:04:11 -06:00
|
|
|
#include <linux/slab.h>
|
2010-01-06 01:47:10 -07:00
|
|
|
#include <linux/rcupdate.h>
|
|
|
|
|
|
|
|
#define MAX_SEQ_NR INT_MAX - NR_CPUS
|
2010-04-29 06:37:32 -06:00
|
|
|
#define MAX_OBJ_NUM 1000
|
2010-01-06 01:47:10 -07:00
|
|
|
|
|
|
|
static int padata_index_to_cpu(struct parallel_data *pd, int cpu_index)
|
|
|
|
{
|
|
|
|
int cpu, target_cpu;
|
|
|
|
|
|
|
|
target_cpu = cpumask_first(pd->cpumask);
|
|
|
|
for (cpu = 0; cpu < cpu_index; cpu++)
|
|
|
|
target_cpu = cpumask_next(target_cpu, pd->cpumask);
|
|
|
|
|
|
|
|
return target_cpu;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int padata_cpu_hash(struct padata_priv *padata)
|
|
|
|
{
|
|
|
|
int cpu_index;
|
|
|
|
struct parallel_data *pd;
|
|
|
|
|
|
|
|
pd = padata->pd;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Hash the sequence numbers to the cpus by taking
|
|
|
|
* seq_nr mod. number of cpus in use.
|
|
|
|
*/
|
|
|
|
cpu_index = padata->seq_nr % cpumask_weight(pd->cpumask);
|
|
|
|
|
|
|
|
return padata_index_to_cpu(pd, cpu_index);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void padata_parallel_worker(struct work_struct *work)
|
|
|
|
{
|
|
|
|
struct padata_queue *queue;
|
|
|
|
struct parallel_data *pd;
|
|
|
|
struct padata_instance *pinst;
|
|
|
|
LIST_HEAD(local_list);
|
|
|
|
|
|
|
|
local_bh_disable();
|
|
|
|
queue = container_of(work, struct padata_queue, pwork);
|
|
|
|
pd = queue->pd;
|
|
|
|
pinst = pd->pinst;
|
|
|
|
|
|
|
|
spin_lock(&queue->parallel.lock);
|
|
|
|
list_replace_init(&queue->parallel.list, &local_list);
|
|
|
|
spin_unlock(&queue->parallel.lock);
|
|
|
|
|
|
|
|
while (!list_empty(&local_list)) {
|
|
|
|
struct padata_priv *padata;
|
|
|
|
|
|
|
|
padata = list_entry(local_list.next,
|
|
|
|
struct padata_priv, list);
|
|
|
|
|
|
|
|
list_del_init(&padata->list);
|
|
|
|
|
|
|
|
padata->parallel(padata);
|
|
|
|
}
|
|
|
|
|
|
|
|
local_bh_enable();
|
|
|
|
}
|
|
|
|
|
2010-05-18 21:44:27 -06:00
|
|
|
/**
|
2010-01-06 01:47:10 -07:00
|
|
|
* padata_do_parallel - padata parallelization function
|
|
|
|
*
|
|
|
|
* @pinst: padata instance
|
|
|
|
* @padata: object to be parallelized
|
|
|
|
* @cb_cpu: cpu the serialization callback function will run on,
|
|
|
|
* must be in the cpumask of padata.
|
|
|
|
*
|
|
|
|
* The parallelization callback function will run with BHs off.
|
|
|
|
* Note: Every object which is parallelized by padata_do_parallel
|
|
|
|
* must be seen by padata_do_serial.
|
|
|
|
*/
|
|
|
|
int padata_do_parallel(struct padata_instance *pinst,
|
|
|
|
struct padata_priv *padata, int cb_cpu)
|
|
|
|
{
|
|
|
|
int target_cpu, err;
|
|
|
|
struct padata_queue *queue;
|
|
|
|
struct parallel_data *pd;
|
|
|
|
|
|
|
|
rcu_read_lock_bh();
|
|
|
|
|
|
|
|
pd = rcu_dereference(pinst->pd);
|
|
|
|
|
|
|
|
err = 0;
|
|
|
|
if (!(pinst->flags & PADATA_INIT))
|
|
|
|
goto out;
|
|
|
|
|
|
|
|
err = -EBUSY;
|
|
|
|
if ((pinst->flags & PADATA_RESET))
|
|
|
|
goto out;
|
|
|
|
|
|
|
|
if (atomic_read(&pd->refcnt) >= MAX_OBJ_NUM)
|
|
|
|
goto out;
|
|
|
|
|
|
|
|
err = -EINVAL;
|
|
|
|
if (!cpumask_test_cpu(cb_cpu, pd->cpumask))
|
|
|
|
goto out;
|
|
|
|
|
|
|
|
err = -EINPROGRESS;
|
|
|
|
atomic_inc(&pd->refcnt);
|
|
|
|
padata->pd = pd;
|
|
|
|
padata->cb_cpu = cb_cpu;
|
|
|
|
|
|
|
|
if (unlikely(atomic_read(&pd->seq_nr) == pd->max_seq_nr))
|
|
|
|
atomic_set(&pd->seq_nr, -1);
|
|
|
|
|
|
|
|
padata->seq_nr = atomic_inc_return(&pd->seq_nr);
|
|
|
|
|
|
|
|
target_cpu = padata_cpu_hash(padata);
|
|
|
|
queue = per_cpu_ptr(pd->queue, target_cpu);
|
|
|
|
|
|
|
|
spin_lock(&queue->parallel.lock);
|
|
|
|
list_add_tail(&padata->list, &queue->parallel.list);
|
|
|
|
spin_unlock(&queue->parallel.lock);
|
|
|
|
|
|
|
|
queue_work_on(target_cpu, pinst->wq, &queue->pwork);
|
|
|
|
|
|
|
|
out:
|
|
|
|
rcu_read_unlock_bh();
|
|
|
|
|
|
|
|
return err;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(padata_do_parallel);
|
|
|
|
|
2010-05-18 21:44:27 -06:00
|
|
|
/*
|
|
|
|
* padata_get_next - Get the next object that needs serialization.
|
|
|
|
*
|
|
|
|
* Return values are:
|
|
|
|
*
|
|
|
|
* A pointer to the control struct of the next object that needs
|
|
|
|
* serialization, if present in one of the percpu reorder queues.
|
|
|
|
*
|
|
|
|
* NULL, if all percpu reorder queues are empty.
|
|
|
|
*
|
|
|
|
* -EINPROGRESS, if the next object that needs serialization will
|
|
|
|
* be parallel processed by another cpu and is not yet present in
|
|
|
|
* the cpu's reorder queue.
|
|
|
|
*
|
|
|
|
* -ENODATA, if this cpu has to do the parallel processing for
|
|
|
|
* the next object.
|
|
|
|
*/
|
2010-01-06 01:47:10 -07:00
|
|
|
static struct padata_priv *padata_get_next(struct parallel_data *pd)
|
|
|
|
{
|
|
|
|
int cpu, num_cpus, empty, calc_seq_nr;
|
|
|
|
int seq_nr, next_nr, overrun, next_overrun;
|
|
|
|
struct padata_queue *queue, *next_queue;
|
|
|
|
struct padata_priv *padata;
|
|
|
|
struct padata_list *reorder;
|
|
|
|
|
|
|
|
empty = 0;
|
|
|
|
next_nr = -1;
|
|
|
|
next_overrun = 0;
|
|
|
|
next_queue = NULL;
|
|
|
|
|
|
|
|
num_cpus = cpumask_weight(pd->cpumask);
|
|
|
|
|
|
|
|
for_each_cpu(cpu, pd->cpumask) {
|
|
|
|
queue = per_cpu_ptr(pd->queue, cpu);
|
|
|
|
reorder = &queue->reorder;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Calculate the seq_nr of the object that should be
|
2010-05-18 21:44:27 -06:00
|
|
|
* next in this reorder queue.
|
2010-01-06 01:47:10 -07:00
|
|
|
*/
|
|
|
|
overrun = 0;
|
|
|
|
calc_seq_nr = (atomic_read(&queue->num_obj) * num_cpus)
|
|
|
|
+ queue->cpu_index;
|
|
|
|
|
|
|
|
if (unlikely(calc_seq_nr > pd->max_seq_nr)) {
|
|
|
|
calc_seq_nr = calc_seq_nr - pd->max_seq_nr - 1;
|
|
|
|
overrun = 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (!list_empty(&reorder->list)) {
|
|
|
|
padata = list_entry(reorder->list.next,
|
|
|
|
struct padata_priv, list);
|
|
|
|
|
|
|
|
seq_nr = padata->seq_nr;
|
|
|
|
BUG_ON(calc_seq_nr != seq_nr);
|
|
|
|
} else {
|
|
|
|
seq_nr = calc_seq_nr;
|
|
|
|
empty++;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (next_nr < 0 || seq_nr < next_nr
|
|
|
|
|| (next_overrun && !overrun)) {
|
|
|
|
next_nr = seq_nr;
|
|
|
|
next_overrun = overrun;
|
|
|
|
next_queue = queue;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
padata = NULL;
|
|
|
|
|
|
|
|
if (empty == num_cpus)
|
|
|
|
goto out;
|
|
|
|
|
|
|
|
reorder = &next_queue->reorder;
|
|
|
|
|
|
|
|
if (!list_empty(&reorder->list)) {
|
|
|
|
padata = list_entry(reorder->list.next,
|
|
|
|
struct padata_priv, list);
|
|
|
|
|
|
|
|
if (unlikely(next_overrun)) {
|
|
|
|
for_each_cpu(cpu, pd->cpumask) {
|
|
|
|
queue = per_cpu_ptr(pd->queue, cpu);
|
|
|
|
atomic_set(&queue->num_obj, 0);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
spin_lock(&reorder->lock);
|
|
|
|
list_del_init(&padata->list);
|
|
|
|
atomic_dec(&pd->reorder_objects);
|
|
|
|
spin_unlock(&reorder->lock);
|
|
|
|
|
|
|
|
atomic_inc(&next_queue->num_obj);
|
|
|
|
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
2010-05-18 21:43:14 -06:00
|
|
|
queue = per_cpu_ptr(pd->queue, smp_processor_id());
|
|
|
|
if (queue->cpu_index == next_queue->cpu_index) {
|
2010-01-06 01:47:10 -07:00
|
|
|
padata = ERR_PTR(-ENODATA);
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
|
|
|
padata = ERR_PTR(-EINPROGRESS);
|
|
|
|
out:
|
|
|
|
return padata;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void padata_reorder(struct parallel_data *pd)
|
|
|
|
{
|
|
|
|
struct padata_priv *padata;
|
|
|
|
struct padata_queue *queue;
|
|
|
|
struct padata_instance *pinst = pd->pinst;
|
|
|
|
|
2010-05-18 21:44:27 -06:00
|
|
|
/*
|
|
|
|
* We need to ensure that only one cpu can work on dequeueing of
|
|
|
|
* the reorder queue the time. Calculating in which percpu reorder
|
|
|
|
* queue the next object will arrive takes some time. A spinlock
|
|
|
|
* would be highly contended. Also it is not clear in which order
|
|
|
|
* the objects arrive to the reorder queues. So a cpu could wait to
|
|
|
|
* get the lock just to notice that there is nothing to do at the
|
|
|
|
* moment. Therefore we use a trylock and let the holder of the lock
|
|
|
|
* care for all the objects enqueued during the holdtime of the lock.
|
|
|
|
*/
|
2010-01-06 01:47:10 -07:00
|
|
|
if (!spin_trylock_bh(&pd->lock))
|
2010-05-18 21:43:14 -06:00
|
|
|
return;
|
2010-01-06 01:47:10 -07:00
|
|
|
|
|
|
|
while (1) {
|
|
|
|
padata = padata_get_next(pd);
|
|
|
|
|
2010-05-18 21:44:27 -06:00
|
|
|
/*
|
|
|
|
* All reorder queues are empty, or the next object that needs
|
|
|
|
* serialization is parallel processed by another cpu and is
|
|
|
|
* still on it's way to the cpu's reorder queue, nothing to
|
|
|
|
* do for now.
|
|
|
|
*/
|
2010-01-06 01:47:10 -07:00
|
|
|
if (!padata || PTR_ERR(padata) == -EINPROGRESS)
|
|
|
|
break;
|
|
|
|
|
2010-05-18 21:44:27 -06:00
|
|
|
/*
|
|
|
|
* This cpu has to do the parallel processing of the next
|
|
|
|
* object. It's waiting in the cpu's parallelization queue,
|
|
|
|
* so exit imediately.
|
|
|
|
*/
|
2010-01-06 01:47:10 -07:00
|
|
|
if (PTR_ERR(padata) == -ENODATA) {
|
2010-05-18 21:43:14 -06:00
|
|
|
del_timer(&pd->timer);
|
2010-01-06 01:47:10 -07:00
|
|
|
spin_unlock_bh(&pd->lock);
|
2010-05-18 21:43:14 -06:00
|
|
|
return;
|
2010-01-06 01:47:10 -07:00
|
|
|
}
|
|
|
|
|
|
|
|
queue = per_cpu_ptr(pd->queue, padata->cb_cpu);
|
|
|
|
|
|
|
|
spin_lock(&queue->serial.lock);
|
|
|
|
list_add_tail(&padata->list, &queue->serial.list);
|
|
|
|
spin_unlock(&queue->serial.lock);
|
|
|
|
|
|
|
|
queue_work_on(padata->cb_cpu, pinst->wq, &queue->swork);
|
|
|
|
}
|
|
|
|
|
|
|
|
spin_unlock_bh(&pd->lock);
|
|
|
|
|
2010-05-18 21:44:27 -06:00
|
|
|
/*
|
|
|
|
* The next object that needs serialization might have arrived to
|
|
|
|
* the reorder queues in the meantime, we will be called again
|
|
|
|
* from the timer function if noone else cares for it.
|
|
|
|
*/
|
2010-05-18 21:43:14 -06:00
|
|
|
if (atomic_read(&pd->reorder_objects)
|
|
|
|
&& !(pinst->flags & PADATA_RESET))
|
|
|
|
mod_timer(&pd->timer, jiffies + HZ);
|
|
|
|
else
|
|
|
|
del_timer(&pd->timer);
|
2010-01-06 01:47:10 -07:00
|
|
|
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
2010-05-18 21:43:14 -06:00
|
|
|
static void padata_reorder_timer(unsigned long arg)
|
|
|
|
{
|
|
|
|
struct parallel_data *pd = (struct parallel_data *)arg;
|
|
|
|
|
|
|
|
padata_reorder(pd);
|
|
|
|
}
|
|
|
|
|
2010-01-06 01:47:10 -07:00
|
|
|
static void padata_serial_worker(struct work_struct *work)
|
|
|
|
{
|
|
|
|
struct padata_queue *queue;
|
|
|
|
struct parallel_data *pd;
|
|
|
|
LIST_HEAD(local_list);
|
|
|
|
|
|
|
|
local_bh_disable();
|
|
|
|
queue = container_of(work, struct padata_queue, swork);
|
|
|
|
pd = queue->pd;
|
|
|
|
|
|
|
|
spin_lock(&queue->serial.lock);
|
|
|
|
list_replace_init(&queue->serial.list, &local_list);
|
|
|
|
spin_unlock(&queue->serial.lock);
|
|
|
|
|
|
|
|
while (!list_empty(&local_list)) {
|
|
|
|
struct padata_priv *padata;
|
|
|
|
|
|
|
|
padata = list_entry(local_list.next,
|
|
|
|
struct padata_priv, list);
|
|
|
|
|
|
|
|
list_del_init(&padata->list);
|
|
|
|
|
|
|
|
padata->serial(padata);
|
|
|
|
atomic_dec(&pd->refcnt);
|
|
|
|
}
|
|
|
|
local_bh_enable();
|
|
|
|
}
|
|
|
|
|
2010-05-18 21:44:27 -06:00
|
|
|
/**
|
2010-01-06 01:47:10 -07:00
|
|
|
* padata_do_serial - padata serialization function
|
|
|
|
*
|
|
|
|
* @padata: object to be serialized.
|
|
|
|
*
|
|
|
|
* padata_do_serial must be called for every parallelized object.
|
|
|
|
* The serialization callback function will run with BHs off.
|
|
|
|
*/
|
|
|
|
void padata_do_serial(struct padata_priv *padata)
|
|
|
|
{
|
|
|
|
int cpu;
|
|
|
|
struct padata_queue *queue;
|
|
|
|
struct parallel_data *pd;
|
|
|
|
|
|
|
|
pd = padata->pd;
|
|
|
|
|
|
|
|
cpu = get_cpu();
|
|
|
|
queue = per_cpu_ptr(pd->queue, cpu);
|
|
|
|
|
|
|
|
spin_lock(&queue->reorder.lock);
|
|
|
|
atomic_inc(&pd->reorder_objects);
|
|
|
|
list_add_tail(&padata->list, &queue->reorder.list);
|
|
|
|
spin_unlock(&queue->reorder.lock);
|
|
|
|
|
|
|
|
put_cpu();
|
|
|
|
|
|
|
|
padata_reorder(pd);
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(padata_do_serial);
|
|
|
|
|
2010-05-18 21:44:27 -06:00
|
|
|
/* Allocate and initialize the internal cpumask dependend resources. */
|
2010-01-06 01:47:10 -07:00
|
|
|
static struct parallel_data *padata_alloc_pd(struct padata_instance *pinst,
|
|
|
|
const struct cpumask *cpumask)
|
|
|
|
{
|
|
|
|
int cpu, cpu_index, num_cpus;
|
|
|
|
struct padata_queue *queue;
|
|
|
|
struct parallel_data *pd;
|
|
|
|
|
|
|
|
cpu_index = 0;
|
|
|
|
|
|
|
|
pd = kzalloc(sizeof(struct parallel_data), GFP_KERNEL);
|
|
|
|
if (!pd)
|
|
|
|
goto err;
|
|
|
|
|
|
|
|
pd->queue = alloc_percpu(struct padata_queue);
|
|
|
|
if (!pd->queue)
|
|
|
|
goto err_free_pd;
|
|
|
|
|
|
|
|
if (!alloc_cpumask_var(&pd->cpumask, GFP_KERNEL))
|
|
|
|
goto err_free_queue;
|
|
|
|
|
2010-04-29 06:41:36 -06:00
|
|
|
cpumask_and(pd->cpumask, cpumask, cpu_active_mask);
|
|
|
|
|
|
|
|
for_each_cpu(cpu, pd->cpumask) {
|
2010-01-06 01:47:10 -07:00
|
|
|
queue = per_cpu_ptr(pd->queue, cpu);
|
|
|
|
|
|
|
|
queue->pd = pd;
|
|
|
|
|
2010-04-29 06:41:36 -06:00
|
|
|
queue->cpu_index = cpu_index;
|
|
|
|
cpu_index++;
|
2010-01-06 01:47:10 -07:00
|
|
|
|
|
|
|
INIT_LIST_HEAD(&queue->reorder.list);
|
|
|
|
INIT_LIST_HEAD(&queue->parallel.list);
|
|
|
|
INIT_LIST_HEAD(&queue->serial.list);
|
|
|
|
spin_lock_init(&queue->reorder.lock);
|
|
|
|
spin_lock_init(&queue->parallel.lock);
|
|
|
|
spin_lock_init(&queue->serial.lock);
|
|
|
|
|
|
|
|
INIT_WORK(&queue->pwork, padata_parallel_worker);
|
|
|
|
INIT_WORK(&queue->swork, padata_serial_worker);
|
|
|
|
atomic_set(&queue->num_obj, 0);
|
|
|
|
}
|
|
|
|
|
|
|
|
num_cpus = cpumask_weight(pd->cpumask);
|
|
|
|
pd->max_seq_nr = (MAX_SEQ_NR / num_cpus) * num_cpus - 1;
|
|
|
|
|
2010-05-18 21:43:14 -06:00
|
|
|
setup_timer(&pd->timer, padata_reorder_timer, (unsigned long)pd);
|
2010-01-06 01:47:10 -07:00
|
|
|
atomic_set(&pd->seq_nr, -1);
|
|
|
|
atomic_set(&pd->reorder_objects, 0);
|
|
|
|
atomic_set(&pd->refcnt, 0);
|
|
|
|
pd->pinst = pinst;
|
|
|
|
spin_lock_init(&pd->lock);
|
|
|
|
|
|
|
|
return pd;
|
|
|
|
|
|
|
|
err_free_queue:
|
|
|
|
free_percpu(pd->queue);
|
|
|
|
err_free_pd:
|
|
|
|
kfree(pd);
|
|
|
|
err:
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void padata_free_pd(struct parallel_data *pd)
|
|
|
|
{
|
|
|
|
free_cpumask_var(pd->cpumask);
|
|
|
|
free_percpu(pd->queue);
|
|
|
|
kfree(pd);
|
|
|
|
}
|
|
|
|
|
2010-05-18 21:44:27 -06:00
|
|
|
/* Flush all objects out of the padata queues. */
|
2010-05-18 21:43:46 -06:00
|
|
|
static void padata_flush_queues(struct parallel_data *pd)
|
|
|
|
{
|
|
|
|
int cpu;
|
|
|
|
struct padata_queue *queue;
|
|
|
|
|
|
|
|
for_each_cpu(cpu, pd->cpumask) {
|
|
|
|
queue = per_cpu_ptr(pd->queue, cpu);
|
|
|
|
flush_work(&queue->pwork);
|
|
|
|
}
|
|
|
|
|
|
|
|
del_timer_sync(&pd->timer);
|
|
|
|
|
|
|
|
if (atomic_read(&pd->reorder_objects))
|
|
|
|
padata_reorder(pd);
|
|
|
|
|
|
|
|
for_each_cpu(cpu, pd->cpumask) {
|
|
|
|
queue = per_cpu_ptr(pd->queue, cpu);
|
|
|
|
flush_work(&queue->swork);
|
|
|
|
}
|
|
|
|
|
|
|
|
BUG_ON(atomic_read(&pd->refcnt) != 0);
|
|
|
|
}
|
|
|
|
|
2010-05-18 21:44:27 -06:00
|
|
|
/* Replace the internal control stucture with a new one. */
|
2010-01-06 01:47:10 -07:00
|
|
|
static void padata_replace(struct padata_instance *pinst,
|
|
|
|
struct parallel_data *pd_new)
|
|
|
|
{
|
|
|
|
struct parallel_data *pd_old = pinst->pd;
|
|
|
|
|
|
|
|
pinst->flags |= PADATA_RESET;
|
|
|
|
|
|
|
|
rcu_assign_pointer(pinst->pd, pd_new);
|
|
|
|
|
|
|
|
synchronize_rcu();
|
|
|
|
|
2010-05-18 21:43:46 -06:00
|
|
|
padata_flush_queues(pd_old);
|
2010-01-06 01:47:10 -07:00
|
|
|
padata_free_pd(pd_old);
|
|
|
|
|
|
|
|
pinst->flags &= ~PADATA_RESET;
|
|
|
|
}
|
|
|
|
|
2010-05-18 21:44:27 -06:00
|
|
|
/**
|
2010-01-06 01:47:10 -07:00
|
|
|
* padata_set_cpumask - set the cpumask that padata should use
|
|
|
|
*
|
|
|
|
* @pinst: padata instance
|
|
|
|
* @cpumask: the cpumask to use
|
|
|
|
*/
|
|
|
|
int padata_set_cpumask(struct padata_instance *pinst,
|
|
|
|
cpumask_var_t cpumask)
|
|
|
|
{
|
|
|
|
struct parallel_data *pd;
|
|
|
|
int err = 0;
|
|
|
|
|
|
|
|
mutex_lock(&pinst->lock);
|
|
|
|
|
2010-04-29 06:42:30 -06:00
|
|
|
get_online_cpus();
|
|
|
|
|
2010-01-06 01:47:10 -07:00
|
|
|
pd = padata_alloc_pd(pinst, cpumask);
|
|
|
|
if (!pd) {
|
|
|
|
err = -ENOMEM;
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
|
|
|
cpumask_copy(pinst->cpumask, cpumask);
|
|
|
|
|
|
|
|
padata_replace(pinst, pd);
|
|
|
|
|
|
|
|
out:
|
2010-04-29 06:42:30 -06:00
|
|
|
put_online_cpus();
|
|
|
|
|
2010-01-06 01:47:10 -07:00
|
|
|
mutex_unlock(&pinst->lock);
|
|
|
|
|
|
|
|
return err;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(padata_set_cpumask);
|
|
|
|
|
|
|
|
static int __padata_add_cpu(struct padata_instance *pinst, int cpu)
|
|
|
|
{
|
|
|
|
struct parallel_data *pd;
|
|
|
|
|
|
|
|
if (cpumask_test_cpu(cpu, cpu_active_mask)) {
|
|
|
|
pd = padata_alloc_pd(pinst, pinst->cpumask);
|
|
|
|
if (!pd)
|
|
|
|
return -ENOMEM;
|
|
|
|
|
|
|
|
padata_replace(pinst, pd);
|
|
|
|
}
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2010-05-18 21:44:27 -06:00
|
|
|
/**
|
2010-01-06 01:47:10 -07:00
|
|
|
* padata_add_cpu - add a cpu to the padata cpumask
|
|
|
|
*
|
|
|
|
* @pinst: padata instance
|
|
|
|
* @cpu: cpu to add
|
|
|
|
*/
|
|
|
|
int padata_add_cpu(struct padata_instance *pinst, int cpu)
|
|
|
|
{
|
|
|
|
int err;
|
|
|
|
|
|
|
|
mutex_lock(&pinst->lock);
|
|
|
|
|
2010-04-29 06:42:30 -06:00
|
|
|
get_online_cpus();
|
2010-01-06 01:47:10 -07:00
|
|
|
cpumask_set_cpu(cpu, pinst->cpumask);
|
|
|
|
err = __padata_add_cpu(pinst, cpu);
|
2010-04-29 06:42:30 -06:00
|
|
|
put_online_cpus();
|
2010-01-06 01:47:10 -07:00
|
|
|
|
|
|
|
mutex_unlock(&pinst->lock);
|
|
|
|
|
|
|
|
return err;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(padata_add_cpu);
|
|
|
|
|
|
|
|
static int __padata_remove_cpu(struct padata_instance *pinst, int cpu)
|
|
|
|
{
|
|
|
|
struct parallel_data *pd;
|
|
|
|
|
|
|
|
if (cpumask_test_cpu(cpu, cpu_online_mask)) {
|
|
|
|
pd = padata_alloc_pd(pinst, pinst->cpumask);
|
|
|
|
if (!pd)
|
|
|
|
return -ENOMEM;
|
|
|
|
|
|
|
|
padata_replace(pinst, pd);
|
|
|
|
}
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2010-05-18 21:44:27 -06:00
|
|
|
/**
|
2010-01-06 01:47:10 -07:00
|
|
|
* padata_remove_cpu - remove a cpu from the padata cpumask
|
|
|
|
*
|
|
|
|
* @pinst: padata instance
|
|
|
|
* @cpu: cpu to remove
|
|
|
|
*/
|
|
|
|
int padata_remove_cpu(struct padata_instance *pinst, int cpu)
|
|
|
|
{
|
|
|
|
int err;
|
|
|
|
|
|
|
|
mutex_lock(&pinst->lock);
|
|
|
|
|
2010-04-29 06:42:30 -06:00
|
|
|
get_online_cpus();
|
2010-01-06 01:47:10 -07:00
|
|
|
cpumask_clear_cpu(cpu, pinst->cpumask);
|
|
|
|
err = __padata_remove_cpu(pinst, cpu);
|
2010-04-29 06:42:30 -06:00
|
|
|
put_online_cpus();
|
2010-01-06 01:47:10 -07:00
|
|
|
|
|
|
|
mutex_unlock(&pinst->lock);
|
|
|
|
|
|
|
|
return err;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(padata_remove_cpu);
|
|
|
|
|
2010-05-18 21:44:27 -06:00
|
|
|
/**
|
2010-01-06 01:47:10 -07:00
|
|
|
* padata_start - start the parallel processing
|
|
|
|
*
|
|
|
|
* @pinst: padata instance to start
|
|
|
|
*/
|
|
|
|
void padata_start(struct padata_instance *pinst)
|
|
|
|
{
|
|
|
|
mutex_lock(&pinst->lock);
|
|
|
|
pinst->flags |= PADATA_INIT;
|
|
|
|
mutex_unlock(&pinst->lock);
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(padata_start);
|
|
|
|
|
2010-05-18 21:44:27 -06:00
|
|
|
/**
|
2010-01-06 01:47:10 -07:00
|
|
|
* padata_stop - stop the parallel processing
|
|
|
|
*
|
|
|
|
* @pinst: padata instance to stop
|
|
|
|
*/
|
|
|
|
void padata_stop(struct padata_instance *pinst)
|
|
|
|
{
|
|
|
|
mutex_lock(&pinst->lock);
|
|
|
|
pinst->flags &= ~PADATA_INIT;
|
|
|
|
mutex_unlock(&pinst->lock);
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(padata_stop);
|
|
|
|
|
2010-04-29 06:40:10 -06:00
|
|
|
#ifdef CONFIG_HOTPLUG_CPU
|
2010-03-29 02:15:31 -06:00
|
|
|
static int padata_cpu_callback(struct notifier_block *nfb,
|
|
|
|
unsigned long action, void *hcpu)
|
2010-01-06 01:47:10 -07:00
|
|
|
{
|
|
|
|
int err;
|
|
|
|
struct padata_instance *pinst;
|
|
|
|
int cpu = (unsigned long)hcpu;
|
|
|
|
|
|
|
|
pinst = container_of(nfb, struct padata_instance, cpu_notifier);
|
|
|
|
|
|
|
|
switch (action) {
|
|
|
|
case CPU_ONLINE:
|
|
|
|
case CPU_ONLINE_FROZEN:
|
|
|
|
if (!cpumask_test_cpu(cpu, pinst->cpumask))
|
|
|
|
break;
|
|
|
|
mutex_lock(&pinst->lock);
|
|
|
|
err = __padata_add_cpu(pinst, cpu);
|
|
|
|
mutex_unlock(&pinst->lock);
|
|
|
|
if (err)
|
|
|
|
return NOTIFY_BAD;
|
|
|
|
break;
|
|
|
|
|
|
|
|
case CPU_DOWN_PREPARE:
|
|
|
|
case CPU_DOWN_PREPARE_FROZEN:
|
|
|
|
if (!cpumask_test_cpu(cpu, pinst->cpumask))
|
|
|
|
break;
|
|
|
|
mutex_lock(&pinst->lock);
|
|
|
|
err = __padata_remove_cpu(pinst, cpu);
|
|
|
|
mutex_unlock(&pinst->lock);
|
|
|
|
if (err)
|
|
|
|
return NOTIFY_BAD;
|
|
|
|
break;
|
|
|
|
|
|
|
|
case CPU_UP_CANCELED:
|
|
|
|
case CPU_UP_CANCELED_FROZEN:
|
|
|
|
if (!cpumask_test_cpu(cpu, pinst->cpumask))
|
|
|
|
break;
|
|
|
|
mutex_lock(&pinst->lock);
|
|
|
|
__padata_remove_cpu(pinst, cpu);
|
|
|
|
mutex_unlock(&pinst->lock);
|
|
|
|
|
|
|
|
case CPU_DOWN_FAILED:
|
|
|
|
case CPU_DOWN_FAILED_FROZEN:
|
|
|
|
if (!cpumask_test_cpu(cpu, pinst->cpumask))
|
|
|
|
break;
|
|
|
|
mutex_lock(&pinst->lock);
|
|
|
|
__padata_add_cpu(pinst, cpu);
|
|
|
|
mutex_unlock(&pinst->lock);
|
|
|
|
}
|
|
|
|
|
|
|
|
return NOTIFY_OK;
|
|
|
|
}
|
2010-04-29 06:40:10 -06:00
|
|
|
#endif
|
2010-01-06 01:47:10 -07:00
|
|
|
|
2010-05-18 21:44:27 -06:00
|
|
|
/**
|
2010-01-06 01:47:10 -07:00
|
|
|
* padata_alloc - allocate and initialize a padata instance
|
|
|
|
*
|
|
|
|
* @cpumask: cpumask that padata uses for parallelization
|
|
|
|
* @wq: workqueue to use for the allocated padata instance
|
|
|
|
*/
|
|
|
|
struct padata_instance *padata_alloc(const struct cpumask *cpumask,
|
|
|
|
struct workqueue_struct *wq)
|
|
|
|
{
|
|
|
|
struct padata_instance *pinst;
|
|
|
|
struct parallel_data *pd;
|
|
|
|
|
|
|
|
pinst = kzalloc(sizeof(struct padata_instance), GFP_KERNEL);
|
|
|
|
if (!pinst)
|
|
|
|
goto err;
|
|
|
|
|
2010-04-29 06:42:30 -06:00
|
|
|
get_online_cpus();
|
|
|
|
|
2010-01-06 01:47:10 -07:00
|
|
|
pd = padata_alloc_pd(pinst, cpumask);
|
|
|
|
if (!pd)
|
|
|
|
goto err_free_inst;
|
|
|
|
|
2010-03-03 22:30:22 -07:00
|
|
|
if (!alloc_cpumask_var(&pinst->cpumask, GFP_KERNEL))
|
|
|
|
goto err_free_pd;
|
|
|
|
|
2010-01-06 01:47:10 -07:00
|
|
|
rcu_assign_pointer(pinst->pd, pd);
|
|
|
|
|
|
|
|
pinst->wq = wq;
|
|
|
|
|
|
|
|
cpumask_copy(pinst->cpumask, cpumask);
|
|
|
|
|
|
|
|
pinst->flags = 0;
|
|
|
|
|
2010-04-29 06:40:10 -06:00
|
|
|
#ifdef CONFIG_HOTPLUG_CPU
|
2010-01-06 01:47:10 -07:00
|
|
|
pinst->cpu_notifier.notifier_call = padata_cpu_callback;
|
|
|
|
pinst->cpu_notifier.priority = 0;
|
2010-04-29 06:40:10 -06:00
|
|
|
register_hotcpu_notifier(&pinst->cpu_notifier);
|
|
|
|
#endif
|
2010-01-06 01:47:10 -07:00
|
|
|
|
2010-04-29 06:42:30 -06:00
|
|
|
put_online_cpus();
|
|
|
|
|
2010-01-06 01:47:10 -07:00
|
|
|
mutex_init(&pinst->lock);
|
|
|
|
|
|
|
|
return pinst;
|
|
|
|
|
|
|
|
err_free_pd:
|
|
|
|
padata_free_pd(pd);
|
|
|
|
err_free_inst:
|
|
|
|
kfree(pinst);
|
2010-04-29 06:42:30 -06:00
|
|
|
put_online_cpus();
|
2010-01-06 01:47:10 -07:00
|
|
|
err:
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(padata_alloc);
|
|
|
|
|
2010-05-18 21:44:27 -06:00
|
|
|
/**
|
2010-01-06 01:47:10 -07:00
|
|
|
* padata_free - free a padata instance
|
|
|
|
*
|
2010-05-18 21:44:27 -06:00
|
|
|
* @padata_inst: padata instance to free
|
2010-01-06 01:47:10 -07:00
|
|
|
*/
|
|
|
|
void padata_free(struct padata_instance *pinst)
|
|
|
|
{
|
|
|
|
padata_stop(pinst);
|
|
|
|
|
|
|
|
synchronize_rcu();
|
|
|
|
|
2010-04-29 06:40:10 -06:00
|
|
|
#ifdef CONFIG_HOTPLUG_CPU
|
2010-01-06 01:47:10 -07:00
|
|
|
unregister_hotcpu_notifier(&pinst->cpu_notifier);
|
2010-04-29 06:40:10 -06:00
|
|
|
#endif
|
2010-05-18 21:43:46 -06:00
|
|
|
padata_flush_queues(pinst->pd);
|
2010-01-06 01:47:10 -07:00
|
|
|
padata_free_pd(pinst->pd);
|
2010-03-03 22:30:22 -07:00
|
|
|
free_cpumask_var(pinst->cpumask);
|
2010-01-06 01:47:10 -07:00
|
|
|
kfree(pinst);
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(padata_free);
|